some aspects of maithili syntax: a transformational-generative approach

482
Central Institute of Indian Laguages S S O OM ME E A AS S P PE EC C T TS S O OF F M M A AI IT TH HI I L LI I S S Y YN N T TA AX X: : A TRANSFORMATIONAL-GENERATIVE APPROACH By U U D D A A Y YA A N N A A R R A A Y Y A A N N A A S S I IG G H H A A MICROFILMED By C C E EN NT TR RA AL L I I N NS S T TI I T TU UT TE E O OF F I I N ND DI I A AN N L L A AN NG GU UA AG GE ES S Mysore – 570006

Upload: others

Post on 11-Sep-2021

10 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

SSSOOOMMMEEE AAASSSPPPEEECCCTTTSSS OOOFFF MMMAAAIIITTTHHHIIILLLIII SSSYYYNNNTTTAAAXXX:::

A TRANSFORMATIONAL-GENERATIVE APPROACH

By

UUUDDDAAAYYYAAA NNNAAARRRAAAYYYAAANNNAAA SSSIIIGGGHHHAAA

M I C R O F I L M E D

By

CCCEEENNNTTTRRRAAALLL IIINNNSSSTTTIIITTTUUUTTTEEE OOOFFF IIINNNDDDIIIAAANNN LLLAAANNNGGGUUUAAAGGGEEESSS

Mysore – 570006

Page 2: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

SOME ASPECTS OF MYTHILI SYNTAX; A TRANSFORMATIONAL - GENERATIVE APPROACH

THESIS SUBMITTED TO THE UNIVERSITY OF DELHI FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

By UDAYA NARAYANA SINGH

DEPARMENT OF LINGUISTICS UNIVERSITY OF DELHI

DELHI

AUGUST, 1979

Page 3: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

CERTIFICATES I certify that the work ‘Some aspects of Maithili Syntex: A

transformational- Generative approach’ was done by Udaya Narayana Singh

uder my supervision and guidance. This work was not submitted, in part or full,

for any other degree or diploma of this or any other University. This thesis

conforms to the standards of University of Delhi.

(Prof.R.N.SRIVASTAVA) (Dr. K.V. SUBBARAO) Head Supervisor Department of Linguistics Department of Linguistics, University of Delhi University of Delhi Delhi-110007 Delhi-110007

Page 4: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

The research work embodied in this thesis was carried out at the Department of

Linguistics, University of Delhi, Delhi-110007, under the guidance of Dr. K.V.

Subbarao, Reader in Linguistics, University of Delhi. The work is original and has not

been submitted in part or full any other degree or diploma of this or any other

University.

(Udaya Narayan Singh)

Page 5: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

To

M y G r a n d m o t h e r

Smt. Jaya Singh

Who taught me the most

Page 6: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

ACKNOWLEDGEMENTS

This dissertation owes a lot to the late Professor P.B. Pandit—a debt that can never be paid back. I am very grateful to my supervisor, Dr. K.V.Subbarao for his valuable suggestions, insightful comments and pointed criticisms. It has really been great fun working with him. I must thank Professors R.N. Srivastava, C.Ramarao, C.J. Daswani, Munishwar Jha, B.P.Mallick, P.C. Nagundar, and S. Pandey—all of whom have taught me linguistics at different levels. Their doors were always open to me, and I have learnt a lot from them in course of discussions that I had with them over these years. I am indebted to Professors Yamuna Kachru, A Chandrasekar, D.N. Basu, Dr. Suhas Chatterjee, Dr. Pron Singh, and Dr. Peter, N. Hook for their encouragement during the preparation of different versions of this dissertative. Dr. David B. Johnson had kindly sent me materials on Relational Grammar when I was in urgent need of these. Although Professors Bernard Courie, Jerry Sadock, Georgia Green and George Cardona taught me for a very short period, nevertheless, the courses that they gave at the Linguistics Institute 1978 at the University of Illinois were very useful. My friends have been very helpful to no in many ways during my overdue research career, and they include Pradip Day, Dipti Misra, B.C. Chowdhury, Asish Kumar Dey, Atanu Raha, Dipankar Basu,

Page 7: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

6 S. Panigrahi, R.Gargesh, Pijush Battacharjee, Abul Hasan, Sudesh Sawhney, Kiran Parmar, and Anjan Sen. My friends Omprakesh Arora, Harbir Arora and Swapan Kumar Bandyopadhyay deserve special mention in this respect. I am grateful to my parents for letting no grow in my own way Dr. lla Rani Singh, Dr. Pron Shankar Singh, Dr. Subhadra Jha, Dr. Virendra Mallick, Ramlochan Thakur, Mahendra Jha, Srikant Nandal, Ajay Singh, Prabodh Narayan Singh, Vinay Shankar Prasad, Mrs. Urmila Jha, Mrs. Kamalakshi Jha and Mrs. Jaya Singh voiced their institutions on Maithilli sentences and suggested many improvements. I am thankful to them. Mr. Prem Chand has shown great patience in typing out this long work with appreciable accuracy, and I am very much obliged to him. Finally, no work can be completed successfully without a constant source of love, inspiration and good wishes. I shall be failing in my duty if I do not thank the lady who gave me all these in abundance for the last seven years. Thank you,Suchita.

Page 8: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

CONTENTS Dedication 4 Acknowledgment 5-6 Contents 7 Abbreviations and symbols 8-14 I. Introduction 15-21 II. The Maithili Language 22-37 III. Grammatical Relations and the Theory of Grammar 38-173 IV. Relativisation, complementation and Interrogation 174-233 V. Verb Agrement 234-264 VI. Compound Verbs in Maithili 265 –378 VII. Pragmatics, Linearisation Rules and Maithili Grammar 379-438 VIII. Summary and Conclusion 439-453 References (A) In English 454-474 (B) In Bengali 475 (C) In Maithili 475 Vita 476-478

Page 9: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

ABBREVIATION AND SYMBOLS

ABBREVIATION A,Adj, ADJ : Adjective Abs : Absolutive Adv, ADV : Adverb Adv man : Manner adverb Adv pl : Place adverb Adv reas : Reason Adverb Adv t : Time adverb Advp, ADVP Adverbial Phrase Aff : Affix Agr : Agreement Agr (1p-> 2p), etc : Agreement marker showing 1st person in The subject and 2nd person in the noun - subject position. AH : Accessibility Hierarchy APG : Are Pair grammar Aux, AUX : Auxiliary b-sentence : “Basis” sentences b-subject : Subject of a b-sentence CH : Chamour Class : Classifier (a dot, here) Comit- : Comitative COMP : Complement of CONJ,CONJUNCT : Conjunctive marker Conj v : Conjunct verb CPF : Conjunctive Participle Formation (also Called conjunction Reduction by some) CPL : Complementiner CP…V : Verbal Conjuncts CORNL,CORRNL,Corel : Correlative pronoun OV : Compond Verb DAT : Dativd

Page 10: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

9 DES, Des : Desinence DERIV : Derived relation Det, dot : Determinor Di-tr, dt : Di-transitive DO (I) Direct Object of ; (ii) Dunny Kao in CPF IH : Irgative Hierarchy EMPH Emph : Emphatic particle EQUI,Equal : Equi- or Identical- NP deletion rule IRG,Irg : Irgative Fem : Feminine (gender) Fut : Future (tense) G : Gender Gen : Genitive Ger : gerund GR : Grammatical Relations H : Honorificity Habit : Habitual Hon,hon : Honorfic (pronoun) IC : Immediate Constituents Imp, imp : Imperative Inf,inf : Infinitive Intran,Intr : Intransitive IO : Indirect object of Loc : Locative Masc : masculine (gender) Med : Medium (level of pronouns) N : (I) Number, (II) Noun, nominal Neg, neg : Negative marker NH, Non-hon : Non-honerific (pronoun) Nom, nom : Nominative NP : Noun Phrase NP d : ‘neved’ or ‘Displaced’NP NP t : ‘Affected’ or “Trace’ NP Obl, obl, ohl : Oblique OO : Oblique Object of

Page 11: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

10 Opt : Optative OSV : Object- Subject-Verb OVS : Object- verb – subject (order) P : (I) person, (ii) Pre-or Post- Positional Phrase PART, Part, prt : Particle (PTCL) Pass : Passive Pft : Perfective (aspect) Pl: : plural (number) Poss, poss : Possessive PP : Pre-or Post-Positional Phrase p.ptcpl : Post Participle Pred P : predicate Phrase Prop : Proposition Pres, pr : Present (tense) PRO : Pronoun Prog : Progressive (aspect) PSH : Promotion- to- subject Hierarchy Pst : Past ( tense) Pte, pte : Participle PTCL Participle (-PART) Ptcpl, ptcpl : Participle RC : Relational Change RCs : Relative Clauses RCP : Relative Clauses Formation RCR : Relation Changing Rule RD : Relational Description Redup : Reduplication REL, rel : Relative (pronoun or marker) RG : Relational Grammer RH : Relational Hierarchy RN : Relational Network S 1 : Matrix Sentence S 2, S’ : Embodied Sentence SC : Structural Change SD : Structural Description E : Embodied sentence

Page 12: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

11 Self : Reflexive pronoun Sg : Singular (number) ST : Intransitive subject S m : Matrix Sentence SCV : Subject-Object-Verb (order) TP : Transitive “patient” when possivised Spec : Specifier SPL : Subject Properties List ST : Transitive Subject SU, subj, SUBJ : Subject of SVO : Subject-verb-object (order) TC : Transformational-Generative (grammar) Tr, Tran : Transitive (verb) T-rule : Transformational rule UG : Universal Grammer UNS : Udaya Narayana Singh V : Verb or Verbal Complex V2 : Aspectual or Vector Verb Van : Verb Agreement Hierarchy Vb : Verb or verbal complecx VDCT : Vector (Verb) V n : Main Verb Voc : Vocative VOS : Verb-Object-subject (order) VP : Verb Phrase V pol : Polar Verb VSO : Verb-subject-object (order) V vee : Vector Verb 1p, 1st p : First personal (pronoun) 2p, 2nd p : Second personal (pronoun) 3p, 3rd p : Third personal (pronoun I : Subject – relation II : Direct Object- relation III : Indirect Object- relation IV : Oblique Object of

Page 13: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

12

Symbols # : Word boundary -,+ : Concationating symbols between norphs ? ,? : Rewriiten as = : Is identical to; has the property of # : Is not identical to; does not have the property of [ ] : Phrase of Sentence boundaries ( ) : parentheses denoting a optionality of occurrence { } : Branches denoting a choice between more than one where the occurance of any one is a must > : Higher is the hierarchy

??Lesser than or equal to in accessibility

: Two nodes higher than X ( a variable) : One node higher than x( a variable) ?? : Null or deleted element

: Completely ungramatical : More or less unacceptable : Odd : Very Odd …. : Etc. A,B,W, X,Y,E : Variables (used mainly in writing syntactic pulse) -r : Negative value for a feature _r : Positive value for a feature / : In the context of -------- : Actual position of occurance of a change taking place subscript : Marked to show identity or differences.

Page 14: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

13

Phonetic Symbols Vowels

i,ii, i : Short, long and half-short front unrounded high vowels u, un, u : Short, long and half-short back rounded high vowels e : Front unrounded high-aid vowel N : Front unrounded low-aid vowel O : Back rounded high-aid vowel A : Central unrounded high-mid vowel Aa : Central unrounded low vowel

COMMENTS (i) Steps Unvoice Unvoiced Voiced Voiced Nasal Unaspirated Aspirated Unaspirated Aspirated Bilabial P ph b bh a Alveedentals t th d dh a Retroflax r th d th (N) Velar k kh g gh D (ii) Affricates e ch j jh - (iii) Lateral r lateral (voiced) (iv) Frill and Planned : R I Alveelar trill (voiced) X I Flapped or retroflex (voiced; the combination Dh is also found

Page 15: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

14 CHAPTER I

INTRODUCTION 1.1.1. At this juncture of the development of linguistic science, it becomes obligatory for a linguist describing a particular language. (1) to test and provide evidence for and against proposals of theoretical, universal and of typological interest from vittals the language under consideration. b. to provide a thorough description and analysis of as may aspects of a language as possible, and c. to come up with tentative but interesting generalisations about the language being described which may be valid for and applicable to other languages related to its genealogically, typologically or a really. It may also be recalled that Chensky (1965) in his anuubava had expressed a hope that 2. “perhaps the day will come when the kind of date to now can obtain is abundance will be insufficient to resolve deeper questions commencing the structure of language” (Chensky 1965,21) 1.1.2. That the day has come become evident if one goes through the volumes of work is the field of syntax and sometime of natural languages that here been coming to light in recent years. It is a healthy trend that now hardly anybody believes is entrusting ‘ met a throgires’ of language singly as the basis of a handful of well-known languages. The linguistics theoreticians here now realised that ‘ at this juncture of the development of linguistic science, it appears that one of the most productive directions of

Page 16: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

16 research lies in the collection of valuable facts from a diverse cross-sections of languages and the discovery of generalizations based on them’ (Id,1976,xi) 1.1.2. While it is true that the more language-structures are unveiled the more shall we come to know about the nature of the human mind, at the same time, it must also be made possible to capture the varied cross-linguistics generalisations in a simpler way. That is, our theory should be so well equipped as to express such generalisations is an intuitive, unambiguous and direct manner. There seem to be two ways in which change, modification and improvement could be brought is to be particular theory. The way is to show the pitfalls in the organisation that has been followed in a particular theory of grammar on yearly theoretical grounds. Another way of evaluating a theory would here to be based on its operational ca[abilities. In the latter case the probations could be passed and questioned asked on the basis of the analysis of the linguistics elements of particular languages, and of the structure, relations and behaviour of these elements in larger environments. 1.1.4. However, unless a proposal for modification in based on both of the argument-types, it cannot be considered seriously. Otherwise, as fillmere (1977; 87) has rightly pointed out, there can be no end to the ‘criticisms that take the form,’I can do anything that you can do’- the arguments of more-notationals variances” For instance, the initial attempts made (of. Meghwley, 1968” 1970a) to free syntex from the well known states regarding the underlyingly assigned linear precedence relations ( popularly known

Page 17: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

17

as the phrases structure Rules) was argument of the type mentioned by fillnere (1957:67) of late, such tendencies to build castles in the air have received the right kind of reprimed’. And instead the syntastitians have begun to pace theoretical questions depending uponthe insights that they gained from expirisal observations’. 1.2.1. It is only with these aims any perspectives in mind that this work on Maithili syntax has been prepared. The ‘transformational-generative approach’ has been taken as more a research strategy them accepting a particular brand of Generative grammar as a fool proof viable model of synactestic disaristics. In the following paragraphs of this chapter, a brief introduction to the specific topics distribution discussed in different chapters is given. In the Chapter that immediately follows, an outline of Maithili grammar will be presented. It will include a brief description of the structure of the Maithili language. This has been done to make the renders of this week familiar with the sound patterns and the morphological systems of Maithili. The conversion that have been followed throughout the work in transcribing the Maithili date have been explained here. This chapter also includes the background informations regarding the language under consideration. In the third chapter. The theoretical postulates that underlie this week will be presented. This chapter will include a review of the factors that gave therein that the study of grammatical relations has disounded directly from the practice of and dissatisfaction with the Transformational generative grammatical traditions. An attempt will be made there is incorporative the action of grammatical relations

Page 18: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

18

into a modified framework of Transformational Generative Grammar thereby accepting both grammatical ‘categories and grammatical ‘relations’ as basic ingredients of language description. It will also be argued there that the relational motions are “non- discrete” or “fusay” emotion and that Maithili provides a good number of evidences in favour of this hypothesis. 1.2.2. The subsequent chapters deal with different aspects of Maithili syntax in detail. For instance, the fourth chapter would comprise of the oft-discussed syntactic processes such as Relative clause formation (RCP), complementation and interrogation. This chapter would thus include the transformations that do not alter grammatical relations but may refer to them. It would have a detailed discussion on the marker(s) of relativisation, complementation, and interrogation, and on their distribution. The RCP-strategies (of. Keenam & Courie, 1972: Johnson, 1977b) of different kinds as attested in Maithili, the motion of Q-clause Quadrangle (of. Dasgupta, In preparation) and the interrelationship of these grammatical processes are one of the topics to be covered in this chapter. 1.2.3. Chapter V touches on differs in types of agreement systems attended in human languages and the language universals concerning verb agreement that have come to light in the recent years, especially the no-called “verb-agreement Law” Agreement Hierarchy” (Singh, 1976c). The verb agreement pattern of the earliest stages of the Eastern NIA will also be discussed at length in this chapter.

Page 19: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

19

1.2.3. The next two chapters are somewhat related in that they both take into account native speakers’ intuitions to find out and describe the various business of the Maithili language, Chapter VI is limited to the description of the compound verbs in Maithili. In particular, it tries to answer the following questions: what are the formal criteria that define certain “verb-verb” sequences as compound Verbs as against the others ? What are the merits and defects in the traditional tests such as Kar- incertain, Negation, and By-the- time-test and whether there could be supplemented with the Verb Interrogation test (of. Singh & Bandyopadhyay, 1978)? What are the different kinds of Verbal sequences found in Maithili? What is the nature of compound Verb formation in the earlier stages of Maithili? Which are the probable candidates for the vector-ship in Maithili? Are the tests to establish perfective aspect (tests such as the Phrasal verb test, Apprehension test, Contradiction test, compound verb Reduction test, and the Incompletion test) applicable to find out how many of these “probable” vectors are really (aspectual, explicator, operator or) vector verbs in Maithili? How could one classify the polar or main verb in Maithili on the basis of their combinatory possibilities with these vectors? 1.2.5. Chapter VII would shed light on the various optionalitics in the application and selection of particular rules and the pragmatic considerations involved. Including a few non-linearisation rules such as raising, equi, adverb-creating rule, unspecified agent deletion and comparative reduction, the chapter discusses in great

Page 20: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

20

detail (with examples from novels, short stories and plays written in Modern Maithili) the various types of linearisation or reordering rules, and pragmatic motivation behind the selection of each one of these. The linearisation rules include negative adverb proposing, verb proposing, compound verb reordering, extraposition-to –the-right, extraposition-to-the-left, heavy-KP shift, sentence-shift, shifting, adverb movement, right dislocation, topicalisation and some more. This chapter is followed by a short chapter on conclusion and references.

Page 21: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

21

Notes To Chapter X 1. Sayer (1974) gives a number of arguments against Modaulay’s (1970a) hypothesis concerning English as a VSC language. See Johnson (1974a :X1-X2) for a different kind of country-argument to Modaulay’s proposal. One could show that the arguments that Modaulay (1970a) gives in support of a VSC Order for English are equally good (or bad) arguments that could be advanced in favour of an NCV order for the same. Recently, however, Modaulay (1978) himself admitted that his earlier claims regarding the phrase structure rules of English were faulty. 2. In the recent years, a good number of studies, mainly language descriptions, have come up that have shaken the universalistic claims that were made on the basis of more speculations. The works such as Dixon (1972, 1977), keenam (1972), Keenam & Courie (1976) has made a case for more and more studies of such types instead of indulging in the theoretical wrangling of the type witnessed in the not-so-recent post. 3. While going through this work, it would become evident that in certain cases fluctuations of variations (such as hana raa ~ “me, ham a kaa hamkaa “him” (hon) “ ) have been allowed to remain in the text because they occur within the standard colloquial, and at times even in the speech of the same person.

Page 22: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

CHAPTER II

The Maithili Language 2.0 Mithilan 1, also known as Videha, has Bhojpuri, Magahi, Bengali, and Nepali as its neighbouring speech committees. It has been said in there scriptures that Maithilaa was named after Mithi, a ruler of the Videha family. In the ancient times it is said to have included three kingdoms; Videha, Vaisaali and Anga 2 . maithilaaa, at one time, was also called Tir 2nxta. However, now this term refers only to the districts of Darbhanga, Madhubani andnusaffarpar. 2.1. Maithili 3 as a language has been mentioned even in Asin-I-Akbarii. However, according to Orierson (1906b), the first Europian scholar to make a note of its existence was N.T. celebrate (1801). In different stages of development, this language was called as ‘desila ba anna, 4 Avahatiya, Tirhutiaa, Touritiana and Tirhuti 5. Maithili is spoken in the districts if Parbhanga, Madhubani, Nusaffarpur, Bhagalpur, Saharsa, puruia, and in some parts of Champaron and Santhal Parganas of Bihar 6, in some pockets of Malda and West Dinajpur of west Bengal, and in the southern districts of Nepal. The dialest survey of Maithili was never taken up seriously ?. Therefore, the dialectal divisions mentioned in the works of the scholars who studied Maithili are totally impressionistic. For instance, Subhadra Jha (1953;5) has a list of seven dialects such as the standard, Southern, Eastern, Chikaachikii, Western, Jel hil, and the Central Colloquial. Although he gives a geographical account of each of these dialects, it is evident

Page 23: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

23 that some of these are social dialects rather than being geographical. This is especially true for Jel a hii and the standard dialect. However, neither social for geographical divisions are explored fully by him. Bholanath Jha (1970;29) has a sin-way division based mainly on the district division; Darbhanga, Dooghar, Manghye, Purnia, Nussaffarpur and Bhagalpur. Although he tried to justify this classification by giving a few examples of these dialects, the specimens are far from convicting. Others, such as Dinabandhu Jha (1950), Govinda Jha (1960), and Ramanath Jha (1955) do not even care to mention any dialect division. Again, some have also claimed that any study of Maithili dialects should not only include south Nepal but also areas such as Norang, Saptari, Mahattari, Palpa, Bhairahva, Tunhang, Kathmndu, Bhatgaon, Lalitpur and others (Pekhrel, 1972;3). 2.3.1. Historically, Maithili can be divided into three periods; (1) a. Early Maithili (the language of the Caryaapadas, Sarvaamanda, praakrtopalngala and that of Vidyaapati’s works such a ‘Kiirtilutan and ‘Kiirtipataakaa’) (2) b. Middle Maithili (the language of Jyotiriiswara’s ‘Varaaratuaakara’, written in the 14th century, down to ‘Kranajana’, 18th century). c. Modern Maithili (the language and literature written after 10th century) 2.3.2. Genealogically, Maithili has been described (see Chatterji; 1926) as a Central Maaghadhan language. Maithili is very closely related to Megahi, Oriya, Assamese and Bengali. It is, as though, a link between the Eastern Maangadhili tongues and the western dialects such as Bhejpuri, Avadhi, Bagheli and Chattisgarhi.

Page 24: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

24 2.4. Maithili is written in three scripts; Tir a hutas, Kaithii and Davanagarii. Tir a hutas was once by far the most popular script, although it was mainly used by the Brahmanas, Kaayasthan and other higher castes. Kaithii on the other hand, was used by the business communities and was also popular among the socialiterates and among the women who wrote their household accounts in this script. Davanaagarii is the script in which almost all the books and journals appear nowadays. In the speech area adjacent to West Bengal the Bengali character is also used. 2.5.1 Modern Maithili has 13 vowels and 31 consonants. The vowel system of Maithili is popular in a sense because it seems to be the only Eastern Nia language that has developed three vowel-lengths. In the word-final position and in between the consonantal clusters to mark the syllable boundaries the very short vowels could be seen to be appearing in the sentences quoted in the work as examples. Phonemically they are short vowels. Of the 31 consonants, there are 19 steps, 4 affrientes, one lateral, 2 trills, 2 sibililants one glottal frictive, and 2 semi- vowels. 2.5.2. The total inventory of the sound system is presented below; (2) Maithili Vowels; I, ii, I u,un, n O, a,a o, B o aa

Page 25: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

25

(3) Maithili Consonants: P, ph, b bh, t, th, d, dh, T, th, D, Dh k, kh, g, gh N n y C, ch, j, jh S r r W l y 2.6.1. The most common pluraliser in Maithili is : Sah. For the same purpose, Lokan 1 is also used but it is selected only when the referent is an honorific pronoun or noun. At times, plurals are formed also by proposing quantifier words such as bakut. Subhadra Jha (1958) reports that there are two more processes of making plurals and they are: by reduplication and by using the nouns standing in apposition to the pronouns of the pronouns of the first and second person as pluralisers. Consider the following examples; (4)kan aras brashmanak to yik shal I+abj Braknin=gen PART this+ Imph way “This is the way of ours, of the Brahmins.” (4) han gaon gaon ghun a lakn n I village- village room+po, t+agr “ I roamed aromi in villages.” In (4), han a ran in apposition to brashnanak pluralises the latter, while in (5), two occurrences of Haon make it understood to be in plural. 2.6.2.1.Maithili has almost lost the OIA gender system. Only in a few places the traces of it could be found. One such case is the non-definitive adjectives agreeing with nouns belonging to

Page 26: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

26 Masculine and feminine genders. Thus, balkas kas “theeldest unclu”, but bankii daas ‘the eldest sister”. The verbal forms derived from the participial bases also show gender distinction when the subject is in the third person. This is true both for the 1-forms that mark the past tense and for the t-forms which indicate the future 9. Consider the following examples; (6) raan ga-l-ash a/ jas-t-ash a Ran ge+pst+agr/ ge+fut+ agr “ran went/ will go’. (7) Siitaa ge-l-lih a/ jac-t-lih a sita ge-pst+agr/ ge+fut+agr “sita went/will go.” 2.6.22 This distinction is maintained even in the auxiliarise. For instance; (8) manukka a swatal a shal a man sleep+p, ptapl+agr be+pst+agr “The man was asleep,” (9) nangli swatal I shal I woman sleep+p, ptcpl+ agr be+pst+agr “ The woman was asleep”. 2.6.3.1 The case relations in Modern Maithili is expressed either by independent postpositions or by the order o elements in the sentence. The postpositions are as follows: (10) NOK ACC INSTR DAT APL GEN LOC -o -kz 2 -oO2 -kp 2 -oO2 -ak -ab -kaa -(ak) -kaa -ker 2 dwaaraa -oO n

Page 27: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

27

2.6.3.2. A few examples will be illustrative as to how the word-order plays an important role in determining the case-relations; (11) O bangh a naar I kao ghar 2 gel a he (NH) tiger kill+ptopl do house ge+pst+agr “No, having killed the tiger, want home”. (12) raat I etms san night here come+imp “come here at night.” The first of these sentences show how one has to depend on the order of elements for the correct interpretation. The order here is subject first and object next and in name of the cases the NPs concerned received any marker. The sentence (12) has an oblique object in the sentence-initial position. Note that if the adverb kaat I is shifted to any other position it has to receive a suffix –KL n. For instance, (12) etaa raat I kt 2 nan ‘Come here at night”. 2.6.4.1. It may be mentioned at this point that the pronouns being a relatively closed system show some differences in the declausional patterns. The prenominal pattern 10 of Maithili is as follows : (13) SUBJECT OBJECT POSSESSIVE

1p sg hen hamraa hamar 1p pl hansab hanraasabkin kanraasabhak 2p Non 1ag apaR apnokN 2 apnak 2p Non1 pl apuNIoken I apaKiokanikp apnNLokanik

Page 28: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

28

2p Non 2 sg shan n shasnkbn shaa nk 2p Non 2 pl shaa n sab shaa nnabhabs shaa 2 sabhak 2p NH/N eg to 2 teraa tohar 2p NH/n pl to n sab teraasabkapn toraasabhak 3p Non ag o hum 2 kaa humak/hum akar 3p Non pl osab ok araasabkB2 hum a kaasabhak 3p NH eg O Ok araa okar 3p NH pl esab ok 2raasabkN2 ok araasabhak 2.6.4.2. A few points about the way these pronouns work in Maithili may be made at this point. First, ???? is used mainly in the cases where the social distance between the addresser and the addresses is relatively great. For instance, ???? would be used while talking to a very honorable and/or to an elderly person, with unknown but seemingly upper class gentleman and oven with one’s father-in-law and mother-in-law. The person who uses this pronoun does not normally get it back. The honorific pronoun to show mutual respect is shaa n. Secondly, lakan I as a pluraliser is employed to show more respect to persons being talked to or about. This postpositive can be used with shaa 2 and I as well, But it cannot be attached to the various other levels of pronouns. Thus, *to n-lakan I “You (medium/ Non-Honorific) and *e-lakan I “They (Non-Honorific) would be ill-formed. Similarly, since the first person is neutral is respect to honorificity, *hon-lakan I “we” will be equally ungrammatical. It may, however, be sentenced that one can occasionally use the pluraliser sab with ???? (as in ???? sab). This would then be, in terms of the degree of honorificity, a little lower than the form ansT-lokani, but still be higher than shan 2-sab. Thirdly, although Maithili has to n for second person

Page 29: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

29 pronouns at the medium as well s at non-honorific levels, the verb agreement shows that they are honophorous. Fourthly, in certain varieties of the standard colloquial Maithili, one may optionally from possessive plurals from a corresponding possessive singular base instead of taking the object singular as the base form. Thus, though rare, one may also come across forms such as honersabhak “one”, techrsabhak “your (pl)” and ekersabhak “Their” in place of honrasabhak, teraasabhak and ek 2readabhak, respectively. Finally, the relative, correlative and the interrogative pronouns of Modern Maithili ( i.e, se, and ke, respectively) show the same changes in the shape of the surface of these pronouns. This is shown in (14) that follows; (14) Relative Correlative Genitive DIRDOT je se ke ONL. (Non) jan I- tan I- kin 2- ONL. (NH) ja- ta- ka- OBL. (pron. Adj.) jash I tash I ka-/ken a GEN (Non) janik a tinik a kinik a GEN (NK) jokar tokar kokar 2.6.4.3. There are a host of other pronouns and pronominal adjectives and adverbs in Maithili which, among others, include the following ;kee “somebody, kodan a “someone unknown”, san a “ other s , sar /sar 2 “other (aggregative)”, sab/sabh “all”, kich” “some”, sab kich’ “everything”. Ie kee “whoever”, is kich” “whatsoever”, etc.

Page 30: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

30

2.7.1 The verbal complex in Maithili has an order of elements such as follows: (15) Vs + (participle + Auxiliary ) Tense + Agreement 12 It may be mentioned here that although the past and the future forms show suffixes, the simple present is formed by using the appropriate participial suffixes followed by an auxiliary aah- “to be”. Whenever an auxiliary verb is used, the main verb receives either of the following suffixes; (16) Simple Present : -(a)it a Simple Past : {-el a} : {-el a} : {-al a}

Past Absolutive : {-(a)nB} : {-(a) LB}

2.7.2. Maithili has the following nodes : indicative, imperative, conditional, optative, and presumptive. In indicative, a number of auxiliaries could be seen to be used all of which mean “to be”, and these include aah, ha-, bha-, an rah-, The conjugational pattern of the indicative in different aspects and tenses is as follows;

Page 31: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

31

(17) PRESENT PAST FUTURE INFINITE; -(A)IT a+ach -al a/-al a/-al -at 2/-ab a PERFECT; -al a -al a -al a -al a -al a -al a -al a+ach- -al a +ach+al a -al a+rah+{at a} {ab a} -(a)nB _(a)nb -(a)nN -(a)lB -(a)lb -(a)lB

PROGRESSIVE -(a)it a + {rah-} -(a)it a+{rah}+ala -(a)it a+rah+{at a} {ach} {ach} {ab a} HABITUAL -(a) it a+rah-sit a +ach-(a)it a+rah+al a -(a)it a+rah+{at a} {ab a} 2.7.5. Note that although in the non-futuritive progressive, one can use either of the auxiliaries, rah or ask, one normally uses rah to make a distinction between imperfect and progressive in the present tense, because the use if ach is a must in the former. Those suffixes are followed by the person-gender-honorific markers (for a discussion of these, see Chapter V). Here, the present perfect seems to be the only case where these agreement suffixes can optionally be added directly to the participle and in such a case the auxiliary ach has a constant shape, i.e., ach i. In the conditional, one may or may not use the participle plus verb construction. The marker for the conditional, -it- proceeds the agreement suffixes. In case of presumptive, he- “to be” occurs as an auxiliary after the appropriate participial markers. Optative is expressed by a set of suffixes

Page 32: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

32

attached after the main or the auxiliary verb and these suffixes also express the categorise of person, and honorificity. The main verb is followed by the present participial suffix in case an auxiliary is used. The optative suffixes include the following; (18) I,1st person & 2nd person (Non ) : -ii ii,2nd person (Medium) : -ah a iii. 2nd person (Non-honerific) : -H 2 iv. 3rd person (Honorific) : -atk I v. 3rd person (Non-honorific) : -aB 2.8.1 In case of adjectives, just as in nouns, the gender-distinction is maintained only in the reference to animate objects. But even this system is gradually breaking down and in some dialects, the distinction is totally lost. In the colloquial standard, the feminine forms are used with feminine nouns and with the names of the disinative forms of adjectives. Consider the following constructions (19) karikkaa balled aa karikkii gve nah I niik a black bull and black cow net good “The black bull and the black cow are not good”. (20) han a raa cher a kii tharii nah I, hex a kan thaaRa caahii 1p ag+obj small plate net big plate require “I require a big plate, not a small one.” 2.8.2.1. Modern Maithili adjectives do not change their shape according to number and case of nouns, as they did in OIA. The comparative degree is expressed by the ablative postposition –ac n .

Page 33: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

33

For examples, (21) hun a kaa oC n humak patnil niik chath I 3p pro+obl OOM 3p pre+cen wife good be+pres+Agr “His wife is better than him.” 2.6.2.2.1 The superlative is formed generally by Sabac n, e.g.,

(22) o oh I laat a nb sab sc niik a chath i. 3p pre this group lee SUPL good be+pres=agr “He is the best in this group.” 2.8.3.1. The pronominal forms used as adjectives include among others the following :kate(k 2 )/ kat 2 kan ‘how much or how many”, jate(k 2 )/ jat 2 has “as much as”, tate (k 2 ) tat2 has ‘so much”, ate (ke) / at 2 has “this much”, ete (k2) / et2 has “that much”, kat2 nii “ what a little quantity of”, et2nii “only this much”, et2 nii / tat2 nii. Mainly that much / so much” kehan “what sort of” schen “this sort of “ chen “that sort of”. Jehan “such a - - - - . etc. 2.6.3.2 Like the pronominal adjective, even adverbs of all types could be formed from pronouns. A few examples may be given here: etap?et “ here” , stoa/etO ‘there’, iatoa/iato ‘where” kataO/katO “where (interrogative)”, onto “elsewhere’, ekhan “now”, iakhan “then”, tahina “that day”, kahina “when (interrogative )” ones “thus, iccan “in that way (relative)”, Kenan/kama “how”, etc

Page 34: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

34

2.9. The ‘normal’ word order in Maithili is subject-Object-verb (SOV, in short). As it has been pointed out by Nasier (1974), it is only under special kind of emphases that the SOV languages of the Indian Linguistic area (languages such as Hindi-Urdu, Bengali, Telugu, Malayalam, Santhali and a host of others ) change their word order. A few examples could be illustrative.

(23) kar chuke13 dueraa upnay (Hindi) do Pft another way “you have (already) exhausted the other way.” (24) bho gel2 shaa2k kaaj2 ( Maithili)14 be+ytopl go+pst+Agr You+paas work “your work has indeed been done”.

Page 35: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

35

Notes to Chapter II 1. The name of the speech community has been spelled as Mithilan throughout the

work, whereas the name of the language has been spelled as Maithili as against a normal transliteration of this word is known as Maithilii.

2. See Subhadra Jha (1958;I). 3. Maithili was also spelled as Mithilee or Mythili by the eastern scholars. 4. Vidyaapati is his “Kiirrtilataa” gives the name to Maithili. 5. Mishra (1976) reports that “ the Syhad-Vissu- Puranna (C.5th century A.D.) defines the boundries of Mithila as “surrounded by the Kondikii river on the Best, by the Gondaki, on the west, by the Ganga on the South, and by the Himalaya, on the North” (Mishra, 1976 : I, fa, 3). Radhakrishna Choudhary (1976 : X) says : The Mithilamahatswa describes the boundary of Mithila in the following manner : “Between the Himalayas and the Ganges, interrupted by fifteen rivers, lies the most holy land known as Tirubhukti: beginning from the Kasi and running up to the Gundaki, its length is declared to be twenty-four yojanas…; beginning from the Gangas and extending up to the Himalayan forests, its breadth is sixteen yojanas…There is situated the city of Mithila.” “M. Aquique (1974 I) talks about a similar area of mithilaa between 600 B .C. and 1097 A.B. 6. The table of “Major language of Bihar “ prepared by Paul Brass (1974 : 65) on the basis of consume of India reports (1911-1961 ) could be summarised and presented here. In the following table , the number of Hindi speakers are not the same as shown in the census of India reports of 1911-1951, became in these reports the number for the bihari languages was included in the hindi speaking population . Therefore, to keep the total same, this number has been subtracted from the figures for Hindi. In the 1951 census, the figures for Urdu were counted in Hindi, and the figures for Bihari languages (with included Maithili, Magahi, Bhejpuri and others ) were not given at all. What is interesting in these figures is that the relative percentage of Bihari-speaking population is gradually decreasing from 0.524 in 1911 ro 0.396 in 1921 and 0.296 in 1951, and this number suddenly jumps up to 33.396 in 1961. This would definitely puzzle all statisticians of population growth in India. In fact, this shows the unreliability of the census data regarding Maithili.

Page 36: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

36

The comparative table showing the figures and relative percentages of the languages spoken in Bihar is presented below (Here, N.B,M,Hg,Bh, OB,U, Tr, Oth stand for Hindi, Bihari, Maithili, Magahhi, Bhejpuri, other Bihari tongues listed Urdu, Tribal languages , and other languages, respectively): Lgs. 1911 1921 1931 1951 1961 Numb:23607245 24160402 27193393 31316884 20580643 K ? : 83.994 86.246 86.706 81.096 44.306 Numb :00164536 00097554 (Not incl.) 00111767 16442067 2 ? :0.586 0.396 ( “ ) 0.296 35.396 numb : ___ 00001641 ( “ ) 00007674 04902673 K ? : 0.00,6 0.016 ( “ ) 0.236 10.736 Numb :00000165 ____ ( “) 00003723 02816492 Ms ? : 0.006 0.006 ( “ ) 0.096 16.006 Numb :00000007 ____ ( “) 00001699 07642722 Bh ? :0.006 0.006 ( “ ) 0.096 16.006 Numb:00164364 00095915 ( “) 00018666 00790258 OB ? : 0.526 0.346 ( “) 0.097 1.726 Numb : 00324686 00256364 (In K) 02646437 04143243 V ? :1;196 0.896 ( “) 6.836 8.996 Numb 02422314 02351064 02742493 02957306 03460341 Tr ? :8.596 8.306 8.796 7.626 7.4976 Numb ;01798135 01281804 01415006 01753706 01823294 Os ? :6.326 4.306 4.526 4.526 3.926 Total Numb : 26616916 29129208 31349092 30766184 46455610 It may be noted here that following a different method of calculation, Brass (1974 :64 ) came to the conclusion that as in 1961, the estimated figures for Maithili, Magahi and Bhejpuri would be 16,565,477,11,075,845, and 11,629,425 respectively.

Page 37: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

37

7. Even mishra (1976 : 2, fn I ) admits that “an objective linguistic survey is …still necessary.’ 8. The symbol that have been used here and throughout this work do not follow any established system of phonetic code (such as the IPA or Block & Trigger (1942)- system). This has been done mainly because of typographical reasons, in this regard, this is very much similar to Peter Neck’s (1974) symbols for Hindi sounds and Punya sleka Ray’s (1966) diacritics to transcribe Bengali sounds. In the examples used in the text, a raised a indicates that the vowel before it is nasalised. In case the vowel concerned is long, the raised nasal appears after the second number of a home-vocalic cluster, The raised vowels I, a and u are very short vowels, phonetically very much different from their counterparts I,a and u. The capital U and C mark the front and back half –open vowel, respectively. The capital U is the palatal sibilant, and T, Th, D, Dh, (N) are retroflex stops. Similarly, the trills also hare three varieties; r, R Bh. In certain cases, to represent the old texts faithfully all the three sibilants were to be used. There B, r and s were used as dental, palatal and cerebral sibilants. 9. To be mere precise, the gender distinction is now confined to intransitive sentences which are in a tense other that the present. And there too, these sentences must have a subject nominal that is relatively high in the honorific hierarchy. 10. In (13), p denoted persons, sg and pl are singular and plural, Ken is honorific, N is medium and N refers to non-honorific. 11. See Chapter IV for details regarding markers used in relative clauses. 12. See Chapter VI, especially (6) a-c for different kinds of structure found in the verbal category in Maithili. 13. Euk is considered by many as a vector or explientar number of a compound verb in Hindi (of . Neck, 1974). See Chapter VI for reasons for eliminating this verb from the list of vectors. 14. For a detailed discussion of various kinds of reordering of elements, and their conversational implication, see chapter VII.

Page 38: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

CHAPTER III

GRAMMATICALS RELATIONS AND THE THEORY OR GRAMMER

3.0 Introduction; 3.0.1/ In the chapter, I shall first attempt to summarise the arguments advanced by the advocates of the study of “grammatical relations ” against the different brands of transformational-generative grammar in the recent years. I will then outline the initial attempts to formalise a grammatical theory based on ‘grammatical relations” and show its weaknesses. I will, then, go on to describe more serious attempts at formalising what is now known as “Relational Grammar”. And shall briefly touch upon the attempts to define or characterise the relational notions such as ‘Subject-of’. ‘direct-object-of, ‘indirect-object-of, etc. Finally, I will suggest and outline a methodological framework for incorporating the basic insights of relational grammar within a theory of Generative Grammar, wherein both grammatical categorise and grammatical relations could be taken as “printives” of grammatical analysis. The root of this dissertation would show how such a theory could be translated into an analysis of some language-particular facts of Maithili grammar. In course of working out a methodology for such an analysis, I will also explain as to why room must be made for pragmatic considerations of sentential structure s of any language used in a particular speech community, and will further argue that “Grammar” should not be considered as something that is “discrete”.

Page 39: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

39

3.1. A TG Definition of Grammatical Relations: 3.1.1. There are quite a few notions and beliefs that transformationalists inherited from the “Descriptive “ or :structural “ linguists. Some of these, such as turning a blind eye to the notion of meaning were severely criticized in the late sixties by the semanticists belonging to different groups. One such assumption is that the “Phrase structure” competent is an essential part of TG grammar, and that by applying the methods of “Substitution” (Harris, 1946;161-3) and “Expansion” (Wells,1947:25) one can reach notions such as NP,V,VP,ADJ, and ABV, etc. Under such an assumption, these categories are to be accepted as “Printives” for a syntactic analysis. Notice that “substitution”and “expansion” are known in the TG tradition as the “Paradigmatic” and syntactic” relationships(respectively) that one finds among the categorial elements in a sentence. The theoretical position of the TG grammarians, namely, that the meaningful elements “must”be “ordered” in the “deep” structure and the grammatical relations which are definable “automatically” by the deep structural configuration of the word classes were too reductive for a syntactitian to question the basic of such assumptions. The notions of “grammatical relations’ was thus given a back seat by the TG theories in that they were described as useful things only in the field of semantic interpretation of sentence structures (kata and Postal,1964). Choneky (1971:2) would say, for instance, that a syntactic transformation “is independent of the grammatical relations or meanings expressed in these grammatical relations’. It was, however, felt by some that the relational notions such as

Page 40: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

40

‘ Subject’ and ‘object’ should not only be defined in a language-indendent way, but that ‘this characterisation must be accomplished within the general theory of linguistics description” (Kate and Postal, 1964:!^$). But the definition that they come up with snacked - - - -their obsession with generalisation on the basis of handful of languages showing a particular syntactic structure. 3.1.2. I shall come back to this point later while summarising the criticisms of this position of the TG grammarians by the proponents of the Relational Grammar, Before that, I would describe in brief the way TG theorists tried to resolve the problem of defining grammatical relations. The context-free Phrase structure rules in Chemsky (1965) included the following:

(1) a. S ? NP VP b. VP ? { Copula Predicate (NP) (PP) (PP) Manner}} S/ Predicate c. NP ? (Dot) N (S) A typical syntactic structure generated by those PS –rules could be shown by a tree diagram such as follows (2) 3.1.3.1. The

relational notions were defined in the following ways. (2) a. SUBJECT OF [ NP, S] b. DIRBOT OBJBOT OF [ NP, VP] c. INDIRBOT OBJBOT OF [PP, VP]

Page 41: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

41

3.1.3.2. The definitions in (3) need, perhaps, little more elaboration. (3) a states that the noun phrase which is immediately dominated by a S-mode is the subject of the sentence. Under this definition, NP, in (2) is the subject of S1. NP2 is the direct object according to (3) b because the NP directly dominated by VP-mode is to be called direct object. (3) o would then shoe the PP dominated directly by VP in (2) to be the indirect object of this sentence structure, An example of (2) would make the definitions in (3) more explicate: (3) The fact that he can load his team to victory has given the right kind of push to Gavaskar 3.1.3.5. The sentence in (4) has the same structure as in (2). A tree diagram of (4) would look like (5): (5) 3.1.3.4. Notice that definitions given in (3) automatically and correctly predict that NP1 in (5) i.e. the fact that he can lead his team to victory is the subject of the sentence in (4). In the same way, they also predict that the right kind of team and to Gavaskar are direct and indirect objects, respectively.

Page 42: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

42

one can, therefore, a draw a conclusion from the above demonstration that Grammatical categories are the “prinitives” of any generative theory of syntax and that Grammatical relations are derivable and hence secondary facts in a syntactic theory. However, such a generalisation would be erroneous on many counts. A theoretical position such as this would also miss a number of important generalisations about the syntactic structure of human languages. In the following paragraphs the arguments showing lacma in the Chanakyn treatment of grammatical relations are presented. 3.2.1. Arguments against the Deep word Order: 3.2.3.1 Chunky (1965) conceived of the definitions of grammatical relations as given in (#) as “universal definitions .” Notice that they were based on ‘dominance’ and not on ‘precedence’ relations. These definitions also depend on the universality of the notions such as S,VP, NP, etc. But of one stops generalising about ‘Language’ only by considering a few non-verb initial languages, and takes in it consideration VBC (Johnson, 1974 a ; 11.9) or VOS (Pullen, 1977:234-9) languages, the definitions given in (3) run into problem. If the Vp is a universal category, and ‘dominance’ and precedence’ are basic relations the PS-rules for these languages have to be the following (6) VBO-languages; a. s? VP+KP b. VP?V+NP,etc. (7) VOS-languages 3: a.S ? VP+NP b.VP?V+NP

Page 43: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

43

3.2.1.2. If we did to this the Object-initial languages found in the Brazilian Penninsula by Partyshire and Pullan (1978)- the existence of which was described earlier as improbability by Pullan (1977) himself, then the PB-rules for these would lock like as follows; (8) OVS- language: a. S ? NP+NP b. VP?NP+V (9) CSV-language : a. S?NP+VP b. VP?NP+v 3,2.1.3. These four types of languages would thus have the following tree structures generated by their PS-rules as given in (10)-(13):

Now, given the definitions in (3), the np 1 should ‘automatically’ be the subject of a sentence in the languages showing structures (10) through (13). But though it is true for (11) and (12), in case of (10) and (13) it is the NP 2 which has the subjecthood.

Page 44: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

44

Chenaky’s definitions in (3) thus fail to predict the subject relationship in VSC and OSV languages. 3.2.1.4. If we now turn to the definitions of direct and indirects objects, these will also be found to be faulty. First of all, there are languages where the indirect object does not receive a pre-or a post-position. For these languages, thus, (3) e is a false definition. One can also very well imagine that there may be languages where both direct and indirect objects receive pre- or post-positions. In that case, (3)b is a false statement and (3) c is equally vague. It also shows that Ppmay not be a universal category. Also recall Chenaky’s (1965) expansion of VP given in (1)b. The VP, according to this definition, does not dominate a “Unique” PP which could be described as the indirect object. There is thus no way of knowing as to which PP would ‘automatically” be assigned the relation ‘indirect object of ‘ by Chunaky (1965). It shows that either (1)b is incorrect or that (3) c is false. Even if we accept (1)b to be wrong, to save (#) c we have to assume that in all languages the NPs to be defined as indirect objects must have at a deeper level a pre-or a post-position which in some languages are deleted obligatorily. And, one can also hold that languages such as Japanese, where both the objects show particles, the marker attached to the direct object should not be taken note of in writing the PE-rules for these languages, but that it is introduced by a transformation at a latter stage. However, the futility of such a position is too clear to be pointed out.

Page 45: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

45

3.2.1.3. There seems to be only one serious alternative to the definition of grammatical relations in terms of grammatical categories and their configuration in the underlying structure. This system takes care of almost all the criticism against a definition such as in (3). It will basically be a follow -up of the proposals made by Fillners (1968). McGawley (1970a) and Lakoff (!1970a), although, it may be noted at this point that the proponents of such a theory have in the recent years abandoned their own position (of. Fillners, 1977: MaGawlay, 1978). In this system, the grammatical relations will be defined only by the “precedence’ relations. The claim is that the grammatical categories may have different orders in the surface, but at a deeper level they are ordered in a unique manner in all the languages of the world. Such an underlying order could be VSO (MaGawley,1970a). although all other word order possibilities could be equally good claimants to such an honour. The VP-node which is the source of most of the trouble faced by definitions in (3) could be dispensed with 4 under such an analysis. All languages would have the following underlying representation : (14)

To cover these cases where there is a particle either proceeding or following the noun, one can expand the NP-nodes in (14) in either of the following way:

Page 46: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

46

3.2.1.6. Under this analysis, then, the relational notions may be defined in the following manner 3; (16) a. NP/V__X = Subject of c. NP/V,NP__ X = Direct Object of d. NP/V,NP, NP__ X = Indirect Object of According to (16), NP 1, HP 2, and NP 3 in (14) would be subject. Direct object and indirect object, respectively. 3.2.1.7. However attractive this proposal may seen at the first instance, upon a second consideration, it does not lock that Incrative a preposition. Notice first of all that there is no reason why the relational notions should be defined in the way they have been in (16), because one may take a position that all languages have an SOV or SVO order in the underlying representation. In that case (16) would have to be taken as an arbitrary definition 6.Secondly, if one accepts (14) as the deep word order for all the languages of the world, one has to posit various types of reordering rules some of which may apply prior to any other rule. That is, for the languages which do not show VSO as an alternative order in the “surface’, such reordering become a must. This definition, therefore, does not have any explanatory values and is inadequate although it may point out the right relationship that an NP boars to a sentence in which it is used. Thirdly, (16) is a circular definition in that if one asked, for instance, which NP is the

Page 47: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

47

“subject” of a particular sentence? The answer would be –the NP that immediately follows the verb. Obviously, the next question is : Now does one know whether a partic ular NP is to be ordered in the deep structure in such a way that it comes immediately after the verb? The answer would be as follows; Simply find out the subject and place it after the verb. Evidently, the answer is simple, but it is circular too. Fourthly, even if one holds that one should utilize both ‘dominance’ and ‘precedence’ relationship, and that categories such as Vpand PP are not universal, i.e., if one conceives of tree structures such as (17) through (20) instead of (10) through (13), one would face the same problems as mentioned above:

If one also adds SCV and SVO languages to this list, one would get the following structures:

Such a position as above would run into more difficulties than any other proposal, because here one cannot even use precedence relations to point out grammatical relations correctly. Notice that there is no differences between the structural configurations

Page 48: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

48

(16) and (18). Similarly, (19) and (21), and (20) and (22) have the same structure. But in (17), (21) and (22), the NP that occurs first is subject, while in the other three, it is the second NP that is subject. Therefore, one cannot give a universal definition of the relational notions within a framework (or an extension there of) outlined by MaGawlay-Fillners-Lakoff. 3.2.2. The Arguments Against TC Definitions of Surface Relations: 3.2.2.1 One of the main defects in the definition of relational notions given by the transformationalise is that they left it vague as to how superficial grammatial relations should be defined. This may be due to a belief that they do not play important role in the surface structure. Chuncky (1965 ;220-1) admitted, however, that in case o topicalisation, one would find the S-node dominating two NPs at a time but that in such cases one could use the notion of ’precedence’ to define “subject” relationship. Johnson (1974a :11-14) reports that one way to got out of this is to imagine that topicalised NP has been Chuncky adjoined in the following way:

3.2.2.2. One may guess that to define or point out the two objects in a sentence that has undergone dative movement, one

Page 49: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

49

may have to have a structure such as follows:

(23) and(24) show that the ‘neved NPs’ must be marked in some way or the other 9 or that the configurations have to be changed to accommodate these into the cases covered by (3). But ad houners in such solutions is quite obvious even to a layman in syntactic theory. Moreover, since ‘precedence’ relations are language-specific and because Chencky (1965) opts for such a notion to define surface grammatical relations, one must conclude that TG grammarians do not think that relational terms should be defined universally at the superficial level too. Now, if one can show that surface grammatical relations defined universally enable one to capture certain generalisations which would otherwise be missing when one does not do so, it would surely make te TG position more vulnerable. This is a point that energies clearly from the discussion of the argument that comes next. 3.2.3. The Arguments Based on language universals ; 3.2.3.1. Keenam and Courie (1972) reports that languages belonging to different languages-families behave in a remarkably similar fashion in the manner they allow various “surface” grammatical relations to take a relative clause. If we assume

Page 50: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

50

that relations such as subject, direct object, indirect object, etc. are defined in a uniform way at the superficial level, only then we would be able to explain as to why relativisation phenomenon behaves in a certain manner. They claim that these surface relations form a hierarchy as in (23) and that if a language employs a relative clause formation strategy to a relation occupying a particular position in the hierarchical other relations that are higher than this relation must also show relative clause formation in that language. (25) Subject? Direct object < Indirect object < oblique Object < Poss-NP < object of comparative partials 10 Thus, if a language showed relativisation at the level of, say indirect object, it must also show the same to be occuring at the levels of subjects and direct object, and so on and so forth. 3.2.3.2. Courie (1974) shows that human languages have a tendency universal in the way they form the causative constructions, Here again, the hierarchy in (25), better known as ‘accessibility Hierarchy’ (henceforth, AN) is followed. The general tendency is that when a causer is introduced as an agent of a sentence, the original subject loses its subjecthood and gets denoted to the next higher position in the sentence which is lying vacant 11. Thus, if the sentence concerned is intransitive, i.e., if it has originally only a subject, with the introduction of a causes or an agent, the original subject would become the direct object. If the sentence is transitive thereby involving a subject and a (direct) object, the causer would push this subject NP as low as the indirect object position. Similarly, the subject will be

Page 51: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

51

denoted to an oblique object position, if the sentence concerned is di-transitive, and so on and so forth. This is a further proof that grammatical relations (even at a superficial level) could be used in the study of language universals. It would be impossible to translate these and many other universals based on relational notions in terms of grammatical categories. 3.2.3.3. Johnson (1974b) shows that rules that give rise to derived subjects also follow the same hierarchy of grammatical relations. His “Advancement-to-subject- claiming constraint” shows that if, in a particular language, one is able to derive subject from indirect object position, it can not be the case that in the same language, direct objects cannot be made subjects. Notice that it does not say that all languages must have rules to create subjects from all position. He also shows (of Johnson, 1977b) that rules that promote Indirect objects, Benefactives, Instrumentals, and commitatives also follow the relational hierarchy. 3.2.3.4. Perlmutter and Postal (1974) had proposed a number of universals that utilised the notions such as subject and object. They included reflexivisation and careference Deletion Laws and a number of other phenomena. Similarly, permutter and postal (1978) also talks about a number of generalisations which they call the oblique law, stratal uniqueness law, ebounour law, motivated change law, final l law and the nuclear dummy law. We would not go into the details of these and other generalisation about ‘Language’, but what all these show is that

Page 52: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

52

grammatical relations are to be taken more seriously than they were in the TG grammar. 3.2.4. The Arguments Regarding the Characterisation of Rules : 3.2.4.1. The Transformational-generative grammar has the following assumption that “… in all cases, the rule applies blindly, earing nothing for these differences (“differences of relational notions ; UNS). Thus, in an important sense, the rules are structure- dependent and only structure – dependent. Technically, they are rules that apply to abstract labeled brackettings of sentences…, not to systems of grammatical or semantic 12 relations” (Chuncky, 1971 : 31). Such a characterisation of syntactic rules can be shown to be false if it could be proved that a syntactic rule written in terms of grammatical relations captures new insights about ‘Language’. By locking into the nature of one rule-passive formation, one can show why 26 charaterisation of this rule is not acceptable. To capture the notion of passivisation, Ot grammarians employed a number of criteria to suit the indissynaries in the passive rule of different languages. These criteria included ‘word order’, ‘case’, and ‘verb marking’. Notice that none if these could be used consistently and inversely. For instance, one can think of at least the following types of word order changes, if not more, when Passivisation applies;

Page 53: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

53

(26) a. No change : e.g., Cebuane (Bell,1976), Chuj (Johnson, in preparation) c. NP 1 -V- NP 2--? NP 2-V-NP 1 :e.g., English (Chuncky,1957). Albanian (Johnson, in preparation) c.Np 1 NP 2 –V ? NP2-NP 1 _V; e.g., Turkish (Perlmutter and Postal, 1977), Japanese (Kune,1975) d. V- NP 2 – NP 1 ? V-Np 1 – NP 2 : e.g., Malagamy (Keenam, 1972), and e. V-NP 1 –NP 2 ? V- NP 2- Np ! e.g., Nitinst (klekied, 1975) In (26)b to O, NP 1 in the left column is the original subject and NP 2 is the direct object,. In all such cases in the other column, NP 2 is the (derived) subject and NP 1 and oblique object of some kind. Also note that (26) is somewhat simplified in that the passive marker in the verb and other phenomena have not been shown here. From (26), one can easily understand the futility of using ‘word order’ as a criterion to characterise a universal passive rule. 3.2.4.2. Similarly, the criterion of ‘case’ also fails to cover all cases, because one faces at least the following situations with regard to case markings; (27) a. No overt case marking : e.g., English, Indonesian (Perlmutter and Postal, 1977). c. Nominative –Accusative system, where :

NP non- NP Acc ? NP obl-NP non :e.g., Latin (where the oblique is ablative ), and Russian (where the oblique is instrumental),

Page 54: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

54

d. pragative –Absolutive system, where:

NP Rrg – NP Abc ? NP obl –NP abc : e.g., West Greenlandic Bakine (Rischel, 1975; where the oblique is instrumental), and

e. Pragative- Absolutive system, where even in passive. The case marking remains unchanged : e.g., Pasque (Lafitte, 1962)

The types noted in (27) would multiply if one takes into account the word order changes that accompany the change in case marking. For instance, in Bukime, the case-marking for absolute does not change, but it is fronted. Similar things happen in Latin and Russian. The position of V is another thing that has been ignored here for the sake of simplification. In spite of this, it is quite clear from the typology of case-marking in passive- formation that ‘case’ cannot be a criterion in defining passivisation as a universal phenomenon.

3.2.4.3. If we turn towards ‘Verb morphology’ now, the situation would seen more hopeless. This is because verb morphology may involve prefination, infixation, suffixation or an introduction of a non-finite form plus on ANX, and it is not clear as to how one can think of a language-independent ‘passive morphology’. Then, there are languages like mandarin Chinese (Commins, 1976) where the shape of the verb remains the same when an active sentence undergoes passivisation. Here the object NP is simply fronted and a particle bet “by” is placed before the deer. All these point out that TG grammar fails miserably in characterising passive rule in a language- independent manner. The reason may be either that they used wrong criteria so far or that

Page 55: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

55

passive IB a language-particular rule. Perlmutter and postal (1977) argue that the former is true in this case 13. They hold that the right criterion would be the notion of “grammatical relations”. By using such a criterion, they come up with the following generalisation about passivisation and the way it works in different languages. (30), they claim would be a consequence of (28) and (29); (28) An active ‘direct object’ becomes a ‘derived (or superficial) subject’ in the passive clause, (29) The ‘(Original) subject’ of this active clause is neither the subject nor the ‘(Superficial) direct object’ of the corresponding passive sentence. (30) In the absence of a ‘Direct-object creating rule’ in a particular language, which can fill in the gap left over by the promotion of direct object to the subjectedhood, the passive clause in that language would be a ‘(Superficially) intransitive clause’. (28) and (29) could be roughly characterised in the following way (of Johnson, 1977b 134): (31) Passive (in relational terms) (DO?S) a. DO (a, V 2 )? S (A, V I) b. S (B, V 1 )? X (b, Vi ) 3.3. Relatioanl Grammar on outline: 3.3.1.1 Relational Grammar (henceforth, RG) originated from a generative tradition of grammar-writing because of a general dissatisfaction of the way syntactic facts were handled by the

Page 56: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

56

transformationalists. But within a span of less than five years, one witnessed a number of changes in the formulation of RG. As if to lock very different from the traditional TG grammar, the proponents of RG came up with what is now known as Are pair Grammar (henceforth, APG). Both RG and APG introduced a whole lot of now terms and conventions, and presented the (already confused) analyst with a set of postulates, devices and laws. I shall discuss then very briefly in a rather diluted and non-technical way in the following paragraphs.

3.3.1.2. The theory of Relational Grammar, to my mind, was developed by one-time generativists as well as by the syntactitians interested in the comparison of various syntactic processes across different language-families, and this is why RG has had two distinct stream of development. While the former linguists went ton to develop RG into APG showing as keen an interest in rigerous formalises as the TG grammarians, the latter seen to have lost faith in exploiting the notion of “grammatical relations” to such an extent. These two approaches may be called generative and universalistic, respectively. I shall first outline the basic assumptions and hypotheses of RG which developed in the early seventies with the studies such as Perlmutter and Postal (1974). Johnson (1974ab), Keeman (1972, 1973 ), Keenam and Courie (1972), Trithart (1975), Courie (1974), Bell (1974) and others.

3.3.1.2.1. When keenam and Courie (1972) worked on a number of languages across different language-families, they were trying to do several things :(I) They wanted that the attention of linguists should go to a wide variety of languages rather than

Page 57: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

57

being restricted to the nature of a few well-known Western languages; (2) they were interested in comparing languages by employing certain criteria which were independent of a particular model of grammar and (3) they wanted to show that (1) and (2) may help us in gaining now insights about universal grammar. Thus, at that point, they did not envisage that their research night load a number of linguists to construct the theory of RG. They pointed out that with reference to grammatical relations, languages behave in a strikingly similar way. Their Accessibility Hierarchy (AH) as given in (25) received support from the works done by Jakson (1974b) and Trithart (1975) who showed that certain other rules also maintain such a hierarchy. In the Linguistics Institute in the summer of 1974, Purlmutter and postal come up with some concrete proposals for an alternative to TG grammar. In this newly constructed model ( which they called Relational Grammar, ) they called subject (SU), Direct subject (DG),and Indirect object (IC) as terms out of which only the first two were prioncerice, whereas someterms included all types of oblique objects such as instrumental, ablative, locative, benefactive, possessive, and others. All types of oblique objects (OO) and IO would, then, be called non-primerice. Grammar was thought to involve two different representatives (in the pattern of deep and surface structure in the TG tradition) of grammatical relations, and a number of syntactic rules were conceived of as the bridges between the ‘underlying’ and ‘surface grammatical relations’. These relation-changing rules were also believed to be cyclic 14. The underlying representation of grammatical relations do not involve the notice such as ‘dominance’ and

Page 58: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

58

“Precedence”, and hence there is nothing parallel to PE-rules of TG In the RG theory. This also means that grammatical relations such as SV, DC, DO, IO., and OO were taken as basis, undefined and undefinable 15 “prinitives” in RG, and that there was no place for the notion of ‘grammatical categories’ in this theory 16. The surface orders were through to be results of the application of post-cyclic movement rules of different types. All derivations in a particular language and grammar of all languages were thought to be conditioned by a set of universal principles 17 (called “laws” by the authors). It was believed that this set of laws would also define the “possible grammar of natural languages”. They were. Thus. Interested in finding out as to what rules would be found in human languages and given a set of possible rules, which ones would be selected by particular languages and what would constraint such a selection?

3.3.2. The “laws” included the following, among other principles;

(32)(60) Ovalicity Law: operations that alter torched status of NPs are cyclic. (i.e., Relation-changing rules are cyclic.)

(33)(61) Relational Annihilation Law: NPs whose grammatical relations have been taken ever buy another (as it happens in passivisation, see 36 and 37 cause to bear any grammatical relation to their verb.(In that case they are ‘denoted’ to non-tern or “chensur” status. Notice that in the passive clause, where the DO becomes the SU, the original SU does not become DO: rather it becomes an OO).

(34)(62) The Agreement law : Only terms can trigger verb agreement (i.e., Oos camet control the same).

Page 59: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

59

(35)(63) Reflexivisation Law: only terms can trigger reflexivisation (i.e., Oos cannot trigger reflexivisation).

(36)(64) Geraferential Relation Law; Only terms can trigger Gereferential deletion.

(37)(65) The Berangtug Law ; A rule that allters the status of an NP with respect to torched must increase the rank of that NP (DG rank is equivalent to TG in the relational hierarchy. This law states that relation-changing rues would either promote an NP into a higher clause or increase its relational status. It also implies that such a rule can never denote an SU to DO or IO, or a DO to IO from a higher clause to a lower clause).

(38)(66) Advance Laginers Law: An NP undergoing advancement will advance to the lowest point in the hierarchy permitted by universal and language-particular conditions (Advancement rules change the ‘rask ‘ of an NP, and are different from incision rules 18 where a part of an NP usurps the ‘rank’ of the NP which embeds the farmer, e.g., subject-to subject raising. Passivisation, subjectivalisation of the dative, etc are examples of advancement rules. this law states that a rule which advances an IO to, a position higher than IO would automatically advance IO to dc and DU both, unless the language concerned has restrictions on, say, DO-formation).

3.3.3.1. Johnson (1974a) had an idea of ‘relationally-based’ grammar similar to that of Perlmutter and postal (1974). Johnson (1974a) assumed that each sentence has an ‘Underlying Relational Network’, on which ‘TranRelational Rules’ would apply, followed by a set of ‘Linearisation Rules’ to generate the surface sentence

Page 60: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

60

structures. A relational network (HN) is either a lexical category (such as N and V) or a set of relational networks in which there is a ‘Governor’- a grammatically independent unit on which depend all other members. If a relational network is part of another relational network, then the former would be described as a ‘Relational sub-network’. Trans-relational rules would “nap” RHs into RHs by changing the status of an NP in a particular clause. All such rules would be in a format of ‘Relational Description’ (RD, parallel to SD in TG;UNS0 and ‘Relational Change’ (RC-SC in TG tradition;UNS). There rules may ‘promote’ (through accession or advancement) the ranks of NPs and such promotions may necessitate ‘demotions’ of the NP (holding a status which was assigned to another NP). A principle similar to that in (35) would denote this affected NP to Oolevel. Such a demotion is called ‘X-rating’, i.e., the denoted NPs which had a SU, DO or IO status, would now have x or OH (‘Chemsur) status. The linearisation rules in each language would follow two general principles, and they are as follows:

(39) All linearisation would occur with respect to the domain of a governor.

(40) X-rated elements could be linearised in any order with respect to each other.

3.3.3.2. The latter principle suggests that if there are three X-rated elements, A,B, and C, then they may have any of the following arrangements ;ABC,ACB,BAC,DCA,CAB, and CBA. However, all such orders could be found only if A,B,and C Belonged to the same relational sub-network governed by the same verb, i.e., if they are of the same rank. Thus, although one can

Page 61: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

61

here (41)b and c, (41)d is an impossibility;

(41) (69) a. Hary believed that John photographed Bekis.

c. Boxic was believed by hary ki to have been rhetacrashed by John kj.

d. Boxix was believed to have been whetegraphed by John kj by hary ki.

e. *Boxie was believed to have been photographed by Hary by John.

In (414, by Hary, an X-rated element has been allowed to be ordered before by John which I an X-rated element of a different clause-level.

3.3.3.3. Johnson (1974a0 had also thought of a way of translating the final relational network into a constituent structure showing different grammatical categories on the tree. The non-relation-changing rules would apply on this structure. His basic organisation of a relationally-based grammar was as follows;

(42)

Page 62: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

62

3.3.3.4. Notion that the trans-relational rules are cyclic and among the linearisation and non-relation changing rules, some rules may have “priority” (*”ordering” in TG:UNS) over the other. And it is possible that all non-relation-changing rules are actually linearisation rules, and that there is no distinction between the two. It may also be noted that the ‘tail’ of the Accessibility Hierarchy in (25) was blunted by the RG theorists 19 who proposed a “Relational Hierarchy” (NH) in the following way:

(43) SU> DG> IO> OO

The NH was considered to be a universal statement which all laws of RG would refer to. The implication of a universal such as in (45) is that all of these relations are distinguished by all languages. ALSO NOTE THAT IN (42) above, both relational and non-relational component may show in addition to universal constraints on then many more conditions which are language particular. We shall look into one such condition known as “Subjecthood Hierarchy” (Keenam, 1976: Kachru ot al, 1976) with special reference to Maithili.

3.3 Relational-changing rules subjecthood Hierarchy and Maithili

3.4.1. In this section, we will discuss the nature if certain syntactic rules that change grammatical relational with special reference to subjecthood Hierarchy in maithili. In particular, the trans-relational rules that would be touched upon here include Raising, Passivisation, Dative Subject formation, Equi and conjunction Reduction (or CPT).

Page 63: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

63

3.4.2.1. Since Loss (1960) noticed that certain constructions could not be branded as passives, and Resenbaun (1967) came up with a rule of ‘Premium Replacement’, the rule of raising has been a controversial issue. In the next phase, Lakoff (1967) and Ross (1967) were seen trying to concatenate different types of raising operations by pointing out defects in Recenbaun’s (1967) analysis. Although most of their arguments were shown to be false by Orinder (19700 and postal (1974), they had successfully shown that Recenbaun’s ordering of rules generated many unacceptable sentences and that with certain verbs (such as barin in English) his pronoun Replacement rule would have to be obligatory because these typically showed only SV-to SU raising. By then, many (of.Stockwell, schachtor and Partee,1960) had abandoned the goal of finding a uniform charaterisation for SU-to SV and SV-to-Obj raising. But McGawlay (1970a) gave a unitary formulation of these two operations by assuming that the word order for English was VSO. However, the futility of such an underlying word order for English was realised very soon and hence his formulation fell through. Chencky (1972b) rejected the rule of SU-to-Obj raising because according to him the so-called raised and non-raised sentences were structurally unrelated. But he did accept the necessity of having a rule of SU-to Su raising. His work was followed by attempts to show that even the latter type of raising did not exist as Equi could be used to capture such cases. But the Equi-hypothesis could not be used to explain certain cases of raising (where the predicates such as seen in English were involved), and therefore was rejected.

3.4.2.2. In the recent years, syntacticians were clearly

Page 64: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

64

divided on whether a rule such as raising existed. Postal (1974) came up with about fifteen now arguments in favour of a rule of raising in English, all of which were questioned by Brosman (1976a). the traditional tests that were quite often applied to show the existence of raising involves passivisation, reflexivisation and reciprocity. One could give some examples from Maithili to illustrate this point. Consider the following sentences: (44) a. han bujhait n shalaku n je e niik leg shath I I think-pet-Agr that he good man be-pres-Agr ‘I thought that he was a good man “

b. * o hauraa niik leg shath I bujhaait a chalaak a/chl athiink a he no-DC good man be-pres-Agr think-pass ptspl be-pst-Agr “He was thought by no (that) was a good man”

c. hen kunkaa niik leg bujhait a chalianm I

I him-DC good man think-pst-Agr I thought him to be a good man”

d. O hauraa niik leg bujhaait n shal a thiink a

He no-DO good man think-pass ptepl be-pst-Agr “No was thought by no to be a good man”.

(45) a. harandra I bhujhait a shath I je I (harandra) budhisar Narendra think-pres-Agr that he (*Narendra) intelli shath I Goat be-pres-Agr “Narendra thinks that he (Narendra) is intelligent”.

b. *narendra I bujhait a shath I jo anp I budhiaar shathi c. *narendra I kumkas I (narendra-ks n0 budhiaar bujhait a

narendra him-DC Narendra- DC intelligent think- chath 6 pres-Agr “Narendra thinks/considers him (Narendra )to be intelligent”.

Page 65: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

65

d.narendra ayan-kn n budhiaar bujhait a chath I

In (44)a, I secure as the SU of a lower sentence, while in (44)a the some has been made the DC of the higher sentence and hence assumed a shape humkaa. The passive rule changed the relative status of SU and DC of (44)c as in shown in (44)d (along with a change in the verb morphology which I am emitting here ) without facing any problem and this shown that hen and kunkaa were SU and DO in the same clause in (44)c and d. But since they were net in the same clause (and since o was not Dc of the higher sentence) in (44)a and b, application of passive rule on (44) a generated wrong results as in (44)b. This shows that compared to (44)a, in (44) c the status of the subordinate SU has been raised and it has been made a superordinate DC. This is thus a case of intersentential promotion/ That it was a case of raising could be shown also with the help of reflexivisation in (45). Reflexivisation rule applies on a sentence structure only when there are two identical NPs in the same clause. In (45)a the identity condition is not but not the clause-mateness condition, and therefore an attempt to reflexivise the second occurrence of narendra or o in (45)b by replacing it with ernas ‘self’ (with appropriate ase-marking) has made it an unacceptable string. But since reflexivisation has to apply obligatorily when the

Page 66: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

65

d. Narendra apnaa-kN2 budhiaar bujhait a shath I In (44)a, o course as the SU of a lower sentence, while in (44)c the same has been made the DO of the higher sentence and hence assumed a shape hunkaa. The passive rule changed the relative statue of SU and DO of (44)c as in shown in (44)d (along with a change in the verb morphology which I am emitting here) without facing any problem, and this shown that hen and humkaa were SU and DO in the same clause in (44) c and d. But since they were not in the same clause (and since o was not DO of the higher sentence) in (44) a and b, application of passive rule on (44)a generated wrong results as in (44)b. This shows that compared to (44)a, in (44)c the status of the subordinate SU has been raised and it has been made a superordinate DO. This is thus a case of intersentential promotion. That it was a case of raising could be shown also with the help of reflexivisation in (45). Reflexivisation rule applies on a sentence structure only when there are two identical NPs in the same clause. In (45)a the identity condition is not but not the clause-mateness condition, and therefore an attempt to reflexivise the second occurrence of narendra or o in (45)b by replacing it with senas ‘self” (with appropriate case-marking) has made it an unacceptable string. But since reflexivisation has to apply obligatorily when the conditions are not and since (45)c meets both identity and clause-mateness conditions, non-application of reflexivisation has made it unacceptable (in the intended sense). One has to apply this rule to get the corresponding acceptable sentence as in (45)d. Both these tests show that in (44)c and (45)d, some syntactic rule

Page 67: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

66

must have altered the status of NPs in these sentences. Reciprocity test does not work satisfactorily in Maithili and hence it would not be discussed here. The typical examples that are given in English include the following sentences where (46)a shows that the use if each other in a that –complement clause has been faulty, and each other has to be raised as in (46)b to make it acceptable: (46) a. * They believed (that0 each other were honest.

b. They believed each other to be honest. 3.4.2.3. Before we show that the arguments given by postal (1974) in support of a raising rule in English do not work in most of the cases for Maithili, it would be interesting to know whether there are arguments within Maithili grammar to support raising in Maithili. It turns out that there is an argument of this type, which has to do with the scope of the verb agreement rule in Maithili. In Maithili, subjects control verb agreement rule as much as any non--subjects do. Therefore, if a sentence is transitive or di-transitive (or di-transitive(or even if it is intransitive) where more than one human noun occurs, the verb would agree with both of them (for details, see chapter V). In that case, the agreement suffixes would be selected depending upon the relative person-honorific status if the two or more human nouns concerned. Now, in case of non-raised sentences only the subject noun phrase would control verb agreement. But if a rule of raising introduced another (human) noun as an object of this higher sentence, agreement suffixes in the verb must show this fact. Consider the following sentence where the agreement suffixes are italicized.

Page 68: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

67

47 a. han socait a chalakx n jo a sundar shath I

I think-pst-Agr tat he handsome be-pres-Agr

‘I thought that he was handsome”.

b. *hen humkaa sundar socait a chalahn n

“I thought him to be handsome”.

c. hen humkaa sundar socait a chalianh I

“I thought him to be handsome”.

(48) a. han gel a chalaku a (intransitive : one human nominal)

I go-pst be-pst-agr

“I had gone”

b.hen humkar gaon nP gel a chalisah I (intransitive; two human nominal)

his (hen) village in

“I had gone to his village”

c. han hunkaa kitaab done chailoah I (Di-Transitive; two human nominal)

book give-pst

“I had given his the book”.

d.hen humkaa naaraliash I (transitive; two human nominal)

I hin 9hen) kill-pst-Agr

“I killed him’.

If one compares (48)a with (48)b,c, and d, one can easily find out that in the latter the agreement marker is iank I while in the former it is abu n, In (48)b, c, and d, humkar or humkas boars the relations OO(here genitive), IO, and DC, respectively. These sentences thus show that if the subject is a first person pre-form, and if one of the (human) object is in third person honorific, the marker has to be insh i. It follows from the fact that if one wanted to replace iaoh I with abn n in (480b, c, and d, the result would be unacceptable strings. Now, if we turned our attention to sentences in (470, it would very soon be realised that (47)b is

Page 69: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

68 Ungrammatical only because the agreement marker did not show that there was a human non-subject present in the higher sentence. In (47 c the agreement marker conforms to this pattern, and hence this sentence is grammatical. Thus, for all practical purposes humkaa in (47)c was behaving like an object nominal. But notice that in (47) a it is the subject of the lower sentence and hence the higher sentence verb selects an agreement marker ahn a. Apart from the traditional arguments in support of SU-to-Obj raising, in Maithili, verb agreement scene to clearly prove that this language does have a rule of raising. 3.4.2.4. Another evidence to show that in the raised sentences a particular nominal behaves as derived object comes from case marking. It is difficult in languages such as English to provide such morphological evidences except for the pronouns, but for the Indo-Aryan languages where case marking for objects is more or loss regular this becomes a very strong argument to show that a particular nominal is an object. While the pronouns would change their shape (when raised to superordinate object status) in accordance with the paradigm given in chapter II, other nominals would invariably take the object case suffix ; kn n. Thus normally more infinitivisation of the lower sentence (without attaching this case suffix) would make a raised sentence unacceptable. To make such sentences acceptable, one has either to select k N n or retain the finite shape of the lower clause. Consider the following sentences:

Page 70: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

69

(49) a. han dekh a lahu n [jo raan basjaar jas rahal a shath I] “I saw that Ram was going to the market”.

b. *hen raan deth a lahu n baajaar jacit a c. hen [raam baajar ina rakal a shath I] dekh a lahu n d. han [raam] dekh a lahu n [basjaar jas rahal a shath I] e. [raam]dokh a lahu a [baajaar jaa rahel a shath I] f. han raamki n {dakh a lahu n/ dekh a lianh I} baajaar inait a

(49) a is an instance of an embedded ie- complement clause. In (49)b, the lower sentence has been infinitivised and the original subject of this embodied sentence has been raised to a superordinate object position, but the result is ungrammatical because this raised nominal has not received the case marking for object. In (49) f, the case marking is present but in one of the options wrong agreement pattern was used this making that particular option unacceptable. In (49) c, the whole embodied sentence has been shifted leftward (or, one could alternatively say that 49a was an instance of rightward shift or extraposition: but this discussion would be redundant given the framework of relational grammar.) (49)d and e show that the subject-NP of the lower clause has been shifted to the left. (49)d is a little odd probably because two nominatives are too close here. The same sentence is perfectly acceptable when han is deleted this leaving behind the verb agreement marking to understand what the deleted subject was. However, consider the following sentences: (49)g han raankb n (dokh a lahu n/dothh a lianh I} baajaar jaa rahal a shath I (49)g is somewhat puzzling because here apparently the subject

Page 71: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

70

of the lower sentence seems to be raised but at the same time the lower sentence retains its finite shape. However, the higher sentence verb could show or choose to ignore the presence of raank n which seems to no to be the most surprising of all. Finally, there is no pause after the matrix verb to suggest that the finite clause that follows is a different sentence. I think that the explanation lies in the fact that (49) g comes from a source where ram occurred twice-once as a higher sentence object, and another time as a lower sentence subject .The first one is retained as this sentence shows, and the latter is dropped under an identical NP deletion (or an equi) rule. Still, allowance of the marker-eku n remains a puzzle. 3.4.2.5. Many of the arguments given by postal (1974) are language-specific and hence they do not apply to Maithili and other NIA languages. The arguments which seen to be ineffective as for as maithili is include the following nah-initial NP argument, contraction argument quantifier fleet argument , etc. the sentences that show nah-initial kps in English could not be agreeable sentences in maithili. Alma –final are here parallel in nasty a initial NPS is maithili. But since nester2 as karal to nester in both subject and (derived) object pastimes, as evidence in support of raising could be final here. Consider the following samples where in (90) a and these occur in the subjects position, and in (50) a and a these occur is the subject position, and is (50) a and 4 time elements occur in the derived object positions.

Page 72: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

71 (50) a. {nasty a/koval} raam-c ii beat pareaar I sakait a shath I only ram-cmph this word spread can be-pres-Agr “Ram alone can spread such removers”. b. han nasty a /koval} hun a g-e jannait a chianh I I only him-omph know-pres-Agr “I know him alone’.

d.O dokh a lamh I jo [nasty a /koval} han-hii a ii kaaj kaslahu n He see-pst-Agr that only I-coph this work do-pst-Agr “He saw that I alone did this work.”

e. o {nasty a/koval} hanra-hi li kaaj karo dokh a lamh I he only no-cmph this work de-inf see-pst-Agr “He found no alone to have done this work”.

maithili does not have contraction, and the quantifiers that it has can occur freely in both subject and object positions thus providing no destinations between raised and nonraised sentences of maithili. These grantifiers included sak ‘all’, eradak ‘each’’imm both (similarly, time, three, yeere,'four’, and as an and so forth. etc. 3.4.2.6. the weakness of me of the rather strong and language-independent arguments of postal (1974) could be shown here. A strong argument of postal (1974) could be shown here. A strong argument that postal had given involved complex-nh-shift which , according to his does not operate on an kp if it was subject at point of the application . this rule which postpones an application kp does not separate(I) if the kp emerged is not complex as is when by the ungrammatically if(31)b, and by subsequent grammatically of (31)4, and (11) if the hp occurred was a subject grammatically of (31) and (11) if the hp occurred was a subject (of a lower sentence). The second position which crucial for

Page 73: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

72

proving the existing does not held good in maithili in a number of insentences, notice that while the sentences is (31) follow the first condition ,these in (32), especially (52), do not the obey postals constraint, consider the following examples. 51 a. o raank B n nirded promanit kaelanh I he ran0DC innocent prove make-pst-agr “He proved Ram innocent” b. o nirdeS pramannit kaslanh I ranaki n

f. o rannkN n ie attack mati chalack a nirded prananit kaelah I Who Sita-gan husband be-pst-Agr “He proved Ram who was Sita’s husband innocent”. g. o nirted pranaanit keslamh I ramsku n is siitack mati chalack a

(52) a o pranaanit kaslamh I je samati a beaner-sonas is yuidh a he prove de-pst-agr that all monky-army who fight kamav rahath I nirded chath I de-ptcpl pst-Agr innocent be-pres-Agr “He proved that the whole monkey-contingent that fought was innocent”. B o nirdozohath I) pranaanit kaelamh I- sanast a banner-3 nnas is yudh a kamalrahath C. o prananit laslamh I jo mnirded chath I sanast a bannar- sanas is yuddh a kamal rahatath I Brosman (1976;486-7) had attempted to show that Pistal’s constraint on complex NP-shift fails to prodict in English grammatically of certain sentences where apparently a subject Np was postponed. Similarly, it could not explain ungrammaticality of some examples where a complex where a complex non-subject NP was shifted in each case. But with the Maithili sentences given above show that

Page 74: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

73

there is virtually no difference between a raised and a non-raised sentence as regards to complex NP postponing. Just as a non-subject complex NP could be postponed as in (51)d, one could, in the same way, shift a subject complex NP as in (54)b and c. Against Brosman’s criticism, one may still held that in all her examples, either the sentences chosen involved complex NPs which were subjects of higher sentences or they had a quantifier in the sentence-initial position. Thus, one could try to save Postal’s argument by taking a position that his constraint regarding complex-NP-shift applied only when the NP concerned was a derived object coming from a lower sentential subject position. I have no idea as to how far one could go trying to save his argument in this case but even if one forgets Brosman’s criticisms and looks into counter examples from Maithili instead, one can easily find out the weakness in his argument based on complex-NP-shift.

3.4.2.7. There is one peculiarity about raising rule in Maithili which may be mentioned here. If the lower sentence concerned is equational in nature involving a subject nominal (to be raised) and an adjectival compliment, it is always the case that the adjectival is shifted along with the raised nominal. For instance, if one takes up sentences such as (44)c and (47)o, one cannot apply the rule of raising on these structures without moving the adjectival compliments along with the derived object. If these are not moved, the result would be odd strings such as follows:

Page 75: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

74

(44) e’. tt han humkaa bujhait n shalianh I nick leg (45) e’. tt han kumkaa socait a shalianh I sundar If one moved neither the derived object nor the adjectival, the results would be worse as the following sentences would shown (44) e n. tc han bujhait n chalianj I kumkaa niik leg (46) e n. tc han secait a chaliank I kumkaa sunder what is interesting to note is that if the length of compliments increased, the structures as above do not sound so odd. Consider the following examples: (55) a. han humkaa bujhait a chalinnh I shaa n k sealhurk bharis lineal (your (hen) wife’s-sister’s -brother-husband -in law “ I took him to be your wife’s brother-in –law’s sister’s husband’ b.han kumkaa socait a chalianj a aalamarar-byalhiik sha-kitaab Alamagar-palace- poss younger lord I took him to be the Chetaasshab of the Alamnagar Palace”. However, compared to these, the raised sentences with, say, simple intransitive structure do not require that the predicate by moved along with the raised nominal. This is evident from the fact that both (54)b and o are equally acceptable in Maithili: (54) a. han doth a lahu n jo o baajaar inn ruhel a shath I I see-pst-Agr that he market ge-preg-pst-Agr “I saw that he was going to the market’ b. has humkaa haajaar jaait a doth a lianh I ‘I saw him going to the market”. c. han kumkaa dokh a lisah I baajaar jaait a “I saw him going to the market”.

Page 76: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

75

3.4.2.8. So far we have looked into instances of SU-to-Obj raising. Now, consider the following sentences: (55) a. raadhaa hamaraa ba x a priya lagnit n shath I Radha no-obj very dear soon-pres-Agr ‘Radha seems very dear to me’.

b. sahaar Nb taanas hamraa sonas-san buhaait a aoh I darkness in brass no-obj gold-like appear-pres-Agr “In darkness, brass appears to me to be gold”. c. aanhar logak lel a laathii upayogii bho sakait a aahi blind person-for stick useful be-inf may be-pres-Agr “The stick may happen to be useful for blind man”. d. raan kak a rah n sc n muskil-c sc n khus hoit a shath I ram someone by difficulty-emph with happy be-pres-Agr

Ram is difficult to please by someone”.

e. kanii kanii beat par baabujii sahajah I krudha hoit a chat I little little matter on father easily angry be –pres-Agr “Father is easy to be angry over little things”.

Notice that all of these sentences have semantic parallels in the following sentences; (56) a. hamraa lagait a ach I jo raadhaa bai a priya chath I ‘(It) seems to no that Radha is very dear “.

b. amhaar n P human bujaait achi jo tasman sonas-can thikaik a “(It0 appears to no in the darkness that brass is gold-like”. c. ii bhe sakait a aah I jo sanhar logak lel a laathii upayogii thiika ̀ “It is possible that for blind man the stick is useful”.

d. kak a rah a lel a raakin khu S karab muskil-e thiik a “For someone to please Ram is difficult”. e. kanii kanii beat oar baabujiik N n kraddh a karab sahaj thiik a.. “(It) is easy to make father angry over little things”.

Page 77: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

76

If we compare sentences in (55) and (56), we would find out (I) that there is a change in the assignment of grammatical relations in the nominals that are underlined, (ii) that this change in relation-assignment in effect “raises” the nominals concerned, (iii) that the verbs of the lower sentences undergo deletion along with the raising of the nominal, and (iv) that when the raised nominal is [human], the higher sentence verbs show changes in decencies. These suggest that these are also instances of raising of some kind, but that the raising here differs substantially from the cases discussed in earlier paragraphs. The main points of difference are as follows; (I) the raising talked about earlier showed a SU-to-obj raising, whereas in (550 the raised nominals attain subjecthood status, and (iii) in the cases described earlier lower sentence verbs assumed infinitival shapes, whereas here they are deleted. Notice that (55) and (56) included two different types of raising into SU status,. In the first type, represented by a, b, and c sentences it is a case of SU-to SV raising, and in case of d and e sentences, it is an Obj-to-SU raising. I shall not go into the details of how the kar-verbs of (56) and o change to hi-verbs in (55)d and c, and how thiik a finds a replacement in the appropriate agreement bearing shath I here. What concerns no more at this point is the possibility of collapsing different kinds of raising under one rule which is what grammarians of English have been trying since Recenbaum (1967). My impression is that they were asking wrong questions to themselves, vis., how to conentenate SU-to-Obj and SU-to-SV raising? The right question would, to my mind be whether to collapse them. We have already soon that Maithili had, in

Page 78: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

77

addition to these two types, also an Obj-to SU raising. We have also seen that in terms of what happens when a nominal is raised, these three types of raisings differed substantially. All these lead one to an inevitable conclusion that these three cause of raising could not be structurally deserted unitarily, but that they are related only in terms of status-improvement of grammatical relations. Thus, if we assume that the relative status of relational notions across a clause boundary is as in (57), different kinds of raisings look similar in that the direction of status alternation showed only an upward movement as (57)’ would show: (57) The relative status of relational notions: STATUS-LEVEL I ; -------------- SUBJ S n STATUS-LEVEL II ; OBJ ____________ STATUS-LEVEL III; SUBJ S o STATUS-LEVEL IV OBJ

(57)’ Status-Alternation in different kinds of Raising I ; _________________________ II : OBJ III : _________________________ SUBJ S o IV : OBJ It may be mentioned here that so far as status- alternation is concerned, it is not necessary that a language should have only three or two types of raising. Although Maithili has III>I, III> II, and IV >I types, there is nothing inherently wrong in an alternations-type IV>II that it cannot be found occurring

Page 79: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

78

in any language. Various universals proposed within RG suggest that there could be a hierarchy of status-improvement of the type as in (57)’ in that of a language showed a particular status-improvement rule k. it would show all rules that are x +i…+n. i.e., it cannot be the case that a language has rules x and x+5, and not n+1 to z +4, and so on and so for. A possible hierarchy could be suggested here in the following terms; A Possible (57) Hierarchy of status-improvement rules : Operation Instances a. II? I Passive Quantifier fleat (t) b. IV? III Same as above c. III ? I Tough movement d. III? II Subject –(to-Object) raising, e. IV? I Object –(to-subject) raising f. IV?II ? Notice that (57)” is not without problems. This is especially true if we accept c-type rules higher in the hierarchy compared to d-types. Because tough movement and similar rules do not seen to be as popular with natural language grammars as rules such as raising. But we still not go into the merits and demerits of such a rule-hierarchy here. Instead we would discuss in detail a few more relation-changing rules.

Page 80: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

79

3.4.3.1. In this and the following paragraphs of this section, I shall look into a few more trans-relational rules of Maithili such as Passive, Dative subject Formation and conjunction Reduction to test a claim by some that the relational notion of “subject” is not a discrete relational category (of. Kashru et al, 1976: notice that towards the end of this section as well as in 3,10, 1, we shall reach the same conclusion) as against others (of.M.H. Klainon, 1979) that it is a discrete category no far as Hindi is concerned. Kachru et al (1976:94) claim that in Hindi-Urdu (also applicable to Punjabi and Kashmiri) different kinds of subjects form a hierarchy such as follows; (58) Hierarchy of Subjects in Hindi-Urdu : SI SP < S DAT < S BL < SP (Where S= Subjects, I =Intransitive, T= Transitive, D= Dativer, S Obl = Transitive Agent when passivised, and SU * Transitive Patient when passivised) The hierarchy is based on the relative performance (or potentiality ?) of these different kinds of subjects in controlling Equi, conjunction Reduction and Reflexivisation as well as in being accessible to Equi, Raising and conjunction Reduction rules. If (58) is well-founded, it would also claim that syntactic rules such as Passivisation and Dative subject Formation do not, in all languages, promote or denote grammatical relations in such a way that the derived relational terms receive exactly the same status as the underlying or non-derived

Page 81: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

80

grammatical relations, i.e., relation-changing rules also behave in a non-discrete or fussy mannack. 3.4.3.2. I shall first present examples of a few syntactic rules such as passive, Dative subject formation, Equi and conjunction Reduction in Maithili, and would then go to discuss the subjecthood Hierarchy as in (56). Consider the following sentences; (59) a. shaa n sabdin din-e haa n kab you-SV always tell-talk-rayh call-fut-Agr “You (shall) always talk tall”.

b. anarbaabun kotaari khaa rahal a chath I anar-Mr. Sugarcane eat prog be-pres-Agr “Amarbabu is eating sugarcane”. c. banjaaraasab ksbilank sardaark N n naar a lak a vandorev-pl settlement-gem leader-obj hit/kill-pst-Agr “The wanderers hit/killed the leader of the settlement d. kyee beat pradhaannantriikn n paku n coael a Someone matter prime minister of the matter’. e. o hanraa niik jakaa n khuaclak a he no good way eat-cause-pst-Agr “He fed no nicely”

These sentences could be passivised in the following way : (60) a. shaa n {sha n /evaaraa} sabiin biin-e haa n kal a hapt a ??’ Tall will always be talked by you”.

b. anarbeaku,m {OO n/cynaraa} kotaarii khaael a inn rahal a ash I “sugarcane is being eaten by Amarbabu”. c. banjaaraasabh {ak haatha n} kabilack sardaar maarl a gel a banjaaraasab dwaraa “The leader of the settlement was hit/killed by the vanderors”.

Page 82: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

81

d.kak a rak u {*sr n/dwaaraa} beat pradhan mantrii {-kp n/-dhar I} paku n eal a “The matter reached the Prime Minister (through someone)’. e. *k a raa {/oo n /dwaaraa} han niik jakaa n khunaael a gelahu “I was nicely fed by him”.

A quick glance through these passive sentences would indicate that the passive rule in Maithili introduces oo n, dwaaraa or haatha n (in the lst case, along with a genitive suffix :ak) as an oblique marker immediately after the Chenesur, although there seems to be certain restrictions on the selection of any one of these markers in a particular case. That is why, in certain cases selection of ao n or dwaaraa makea a string a little odd. In addition to this, passive typically show a verb inn- (as in 60b, c, and o) or a he-(as in 60a) while the original verb receives a marker –al a –al a thereby becoming a participle. If and when necessary, the agreement marking in the verb changes as is clear from a comparison of (59)b and (60)b : shath I in (59)b became ash I in (60)b; Similarly, futuritive suffix –ab became –Rt a in the a-sentence. In (60)d and o, the pre-forms in changes assumed an oblique shape following the pronominal paradigm presented earlier in Chapter II. Finally, in all the sentences in (60), the derived subject received the nominative case marking, which is –l. Notice that since semantically the IO in (60)d- pradhanamantrii is a gen nominal in (60)d, -dhar I “up to” is preferred to –ky n “to (obj marker)”. Notice also that a shift of relationally affected nominals

Page 83: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

82

(affected by the passive rule. I.e., derived SU, and Sv-in change) is not necessary in Maithili, although one could interchange the terms positionally. It may also be noted that the nominals presented by a passive rule in Maithili affected not only actor or agent as in (60)a and b, but also patient or victim as in (60)c, recipient as in (60)c, and goal as in (60)d. One could go on providing examples of either types of nominals being promoted by passivisation but that would not provide any new information about passive formation in Maithili. In the terminology used in (56), among the examples that we have given from Maithili, (60)b would be S ONL, and (60)c sp. In relational terms, one would write the passive rule in the following way: (61) Passive Rule (DO? SU) i. DO (x,vi) +{KB n? }? SU (x,v I) +0 ii. SU (y,vi) +0 ? X (y,vi) + sc n/dwaaraa / nk haathe n} (Where the v I +dissonance ? v + ptcpl,- ins+desinance 3.4.3.3. Examples of Dative subjects appear also in the section 3.5.1. of this chapter from Bengali. A few typical instances of Dative subject formation could be given here from Maithili. The sentences in (62) have alternate structures as in (63), where in the latter once, the rule of Dative subject formation has been applied. Consider the following examples.

Page 84: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

83

(62) a. han humkaa niik bujhait a chaliash I I him good think-pst-Agr “I thought him to be good.”

b. hen sankar nB Vishnu Bhaguaan kb n non paal a liaah I I crisis in Vishnu god –obj mind fall-cause-pst-agr “I remembered God Vishnu in crisis” c. han siitanphal pasand a karait shii I pumpkin like de-pres-Agr “I like pumpkin”. d. han seyar-markit np paai lagnelahu n I share-market in money invest-pst-Agr “I invested money in the share-market”. e. hen ii baat bejane lagaolaku n I this matter bad like-pst-Agr “I felt bad about this matter”.

Dative subject formation could be applied on these sentences to donate the underscored nominals into SU in change where instead of the normal-0 suffix associated with a nominative subject, these forms receive various dative markers. This rule also promote the Bos in these sentences into subjecthood status as the following examples would show : (63) a. hanran o niik bujhaait a chalack a “He so good to be”.

b. hanran sankaf np Vishnu bhagwaan non paar a lack a “In crisis, God Vishnu struck my mind”. c. hanraa sitaaphal pasand a (hodt a) ash I “To me, pumpkin is dear “. d. hanraa seyar-maarkit np paai laagal a (ash I) “Money was invested in the share-market to (why)me”. e. hanraa ii beat bejaae laagal a “To me, (I) felt bad about this matter”.

Page 85: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

84

Apart from change in the grammatical relations of the nominals. In Dative subject construction, the verbs change their agreement suffixes and take up direct bases as against causative bases (as shown in b,d,and e sentences). Moreover, DO verbs (i.e., kar in 620) are either deleted or replaced by Rrvwerbs (e.g., ho in 630) in the corresponding Derived subject constructions. The reason behind the deletion of connotative-indicator DO-verb replacement is in “non-volitional” nature (of. Klaiman, in preparation) of the Dative Subject sentences. Because of the same reason, (63)c is quite akay whereas (62)c is completely ungrammatical. The Dative Subject Formation in Maithili could be tentatively written in the following manner: (64) DATIVE SUBJECT FORMATION(SU?X) i. DO (x,v I) + {kb n/kaa/ 0}? SU (x,v I) + 0 ii. SU (y,v I ) + 0? x (y,v I ) +{ kb n/ raa/ kaa/…} (where [vi /[+cause]/[+ DC] ? [v I [-cause]/[+ BN ] 3.4.3.4. The Equi rule deletion a subordinate “subject” nominal if it is identical with any of the relational terms in the immediately superordinate sentence. This rule would explain the defection of the subject nominals in lower sentences in (65) as well as in (66);

Page 86: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

85

(65) a. laaje neihar jane cachait n shath I laje parent’s go-inf want-pres-Agr “laje wants to go to (her) parent’s home”.

b. laaje khus hono caahait a chath I Laje happy be-inf want-pres-Agr ”Laje wants to be happy”

(66) a. humak saayan sum I kac kamraa babe gulaam aalii non par a laah a his siing hoar-conj ptcpl no Senior Gulam ali mind fall-pst-Agr “Having heard his singing, I was realised of Gulam Ali (Senior)

b. ctas haath nb haath dhar baisa; n rah aklan sc n kaaj here hand in hand place-conj ptcpl sit+p.ptcpl prog from work nah I calat a not move-pres-Agr (showing 2nd P SU) “By sitting on here placing hand in hand, (your) work would not be done”.

The sentences in (65) and (66) result among other things from application of two rules; Equi and conjunction Reduction. Notice that (65) a and b come from a source where laahe occurs twice in both the sentences as the following would show; (65) a. laaje eaahait a shath I je laaje naither jaath s laje wanpres-Agr that laje parent’s place go-ops-pres-Agr “Laje wants that Laje goes to (her) parent’s place”.

b. laje caahait a chath I je laaje khu s hoth I Laje want-pres-Agr that Laje happy be-opt-pres-Agr “Laje wants that Laje be happy”.

Equi rule deletes the underscored occurrence of identical lower sentence subject, and in the absence of a subject, the predicate is infinitivised. The complement marker i.e. is deleted when the lower sentence loses its sentential status (and is nominalised).

Page 87: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

86

The nominalised predicate is then shifted to the ideal position for DO which proverbial in Maithili. The same rule also operates on (66), but here the rule of conjunction Reduction is also at work. The sentences in (66) have the following parallels : (66) a. han humak gaayan sum a lahu a as han buro gukaam aalii kn n I his singing hear-pst-agr and I senior Gulam Ali Obj Non paar alahu a Mind fall-cause-pst-Agr “I heard him singing and I remembered Gulam Ali (Senior)”.

b. shaa n haath nb haath dharab aa shaa n stac baisal a You hand in hand place-fut-Agr and you here sit-p. ptcpl Rahab eh I nC n (shaa n kaaj nah I calaa (-sak-)ab Be-fut-Agr this from (you)work not move-cause-(can)-fut-Agr

“If you place hand in hand, and (if) you remain seated here, (then) with this you cannot do work”. The underscored has in (66)’ a is deleted by an Equi rule under the identity of has in the next sentence. If instead, the second occurrence of han was deleted, one could not got (66)a, because Dative Subject Formation could not apply on (66) a to bring in the required changes. In that case the result would have been as follows; (66) a. han humak gaayaa sun I kas babe gulaam aalii kb n non paar a lahu n However, even in this case, the rule of conjunction Reduction has been at work which participalised sun-a lahu n by introducing CP marker –I +(has). I have put –koo within parentheses because at times it is not used as in the case of (66) b. (66)b shows that dhar-ab has been changed only to dhar here instead of the usual dho+hae. The concatenation of the

Page 88: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

87

conditional clause with the higher sentence in (66) b has given rise to rah-laa sc n from rah-laa sc n from rah+desinance-ah I-ac n by a conjowing rule that is fairy regular in Maithili. Under the identity of parenthetical shaa n in the higher clause, the underscored occurrences of shaa n in the higher clause, the underscored occurrences of shaa n were deleted by the Equi rule. Notice that finally, coming to (66)b, the higher sentence shaa n was also deleted. But this deletion has nothing to do with Equi or similar rule. There is a fairly general rule in Maithili (as it is in classical Sanskrit and in many of the Nanda languages) to delete the subject pre-form if it is recoverable from the agreement suffixes attested in the verb. But before the application of this optional subject pronoun deletion rule, (66)’b has undergone a Dative Subject Formation rule too which made kaai the derived SU and altered the markers in the verb by making Oal- non-casual and non-volitional (or +BB) as (66)b would show. The above discussion shows that a number of syntactic rules including Equi, conjunction reduction, conditional S-conjoining, Dative subject formation, and optional subject-pronoun deletion relate the sentences (66) with these in (66)’. 3.4.3.5. Having looked into raising, passive, dative subject formation, Equi and conjunction Reduction rules in detail, we can now turn our attention to the subjecthood hierarchies proposed for Hindi-Urdu, Kashmiri and Punjabi, as given in (58). There scene to bbe three ways to prove that the motion of “subject of “ (SU) is non-discrete. One way would be to show that different kinds of subjects behave differently syntactically as Kashru et al

Page 89: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

88

(1976) have done. Another way would be to show that for many of these syntactic operations, no clear picture emerges in respect of one or more types of subjects ads not all S OBL or S DAY, etc. behave in the same way. A third and final way would be to show that subjecthood hierarchies differed in different languages. In this section, I shall demonstrate that the notion of subject is non-discrete in Maithili on all counts. If we apply the same tests that were used by kachru et al (1876), the result in the fol lowing: (67) The behaviour of subjects under different syntactic environments in Maithili:

ENVIRONMENTS ___________________________________________________________ ????? Reflexive Equi-contr Equi access CPP-contr CPP-acess Raising ____________________________________________________________ SU,intr : yes yes yes yes yes yes SU,tran : yes yes yes yes yes yes SU,di-tr : yes yes yes yes yes yes SU,dat : yes yes yes yes Yes (?) T SU,OBL: yes/an yes yes yes T T SU,deriv : yes yes yes yes Yes yes Maithili differs from Hindi-urdu and other languages that Kachru et al (1976) discussed in the environments that have been underlined in (67). This table of correspondence clearly shows that so far as six environments above are concerned, intransitive transitive and di-transitive subjects do not differ from each other. This , however, does not mean that they would not differ an these or on any other count in other languages. If (67) is correct, it also shows that the so-called derived subjects or NP

Page 90: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

89

are probably to be put higher in the subjecthood hierarchy than the Dative Subjects in Maithili. This should not be very astonishing because both S DAT are results if demotion rules if some kind, whereas SP is derived by a promotion rule. Thus, a language-particular hierarchy that places UNDERLYING SU (including intransitive, transitive and di-transitive), DERIVED SU, and X-RATED SU in this order is what one would normally expect. In the following paragraphs, I shall discuss in detail about the Derived and denoted subjects in Maithili. 3.4.3.6. Dative Subjects in Maithili (like that in Hindi and many other NIA languages) control reflexivisation, Equi and conjunction Reduction, and are accessible to Equi rule as the following examples would indicate: (66) a. hanraa I span I sankaf nb visnu bhagwaan non par alaah a (of 63b) “In my own crisis, God Vishnu struck my ,mind”.

b. hanraa tahiyaa niik lag a baak baat aaiya non ash I no then fine feel-in gun matter today-still mind bpres-AGR “Till tiday, I remember feeling good long back”. c. hen sanat nB Vishnu bhagwaan men paro dao I crisis in Vishnu god mind fall-inf Yamraaj as city a gapy a dunm-k T beteen chalioas I About Yam-king and both-obj inform-ptcpl be-pst-Agr “I had told both Yama, the king and chitragupta about my having remembered God vishnu at the time of crisis”. d. bar akaa jhanjhat np pak I kac hanran vishnu bhagwaan big trouble in fall-conj ptcpl non nar a laah a “Having fallen in a big trouble, God Vishnu struck my mind”.

Page 91: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

90

The sentences given above show that Dative subject controls Reflexivisation (as in 66 a), Equi (68b), and conjunction reduction (60d), and in itself accessible to an Equi rule (60c). The underscored sentences in(66)b through d are higher clauses. In a number of sentences with Dative Subject, Raising would generate ungrammatical; strings, but there are, nevertheless, some cases where an application of Raising is allowed. At this stage, I do not know why this difference should exist. Each sentences are given in (68)e and f. In the same way. Dative subjects seen to be accessible to conjunction Reduction in a number of cases as (68)g and h would show. Thus, the following examples would clearly show that Maithili Dative Subjects are nearer to underlying subjects than their counterpart inHindi;

(68) e. raam hanraa ceriik iljaan laguael a paolaah I Ram no theft-gun blame give-pass-ptcpl find-pst-Agr “Ram found no with a theft-charge against (no)”.

e. baabuujii hanraa hrist heit a dekhal athanh a father-hen no happy be-inf see-pst-Agr “Father found no (being) happy”. f. sankaT nb vishnu bhagwaan man par I kao han tar a samahu n fall-conj ptcpl I cross –abc go-pst-Agr “By remembering God Vishnu in crisis, I attained salvation”. g. siitaaphal pasand abho kao hauraa lel a nashat bho gel o no for importance be-abc go-pst-agr “By liking pumpkin, it was almost bounteous for me”.

The raised Dative Subjects have been underscored in (68)e and f, and in the next two examples, the higher sentences have been

Page 92: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

91

italicised. Notice that to a native speaker of Maithili, (68) h is better than (68)h is better than (68)g probably because the head nominal of the higher sentence in the former is non-subject whereas the same is subject in the latter. There are a number of examples where Dative Subjects cannot be raised or be accessible to COF, but I shall not give such examples here because they would serve no purpose. For our purposes, it is enough to know that under these environments, all Dative subjects do not behave in the same way. 3.4.3.7. Instrumental and other oblique objects in Maithili can control reflexivisation , Equi and conjunction Reduction in the same way as is done in hindi, punjabi and Kashmiri. Consider the following examples. (69) a. ahaan so n sabdin eban I dhel-e-taa piitab hact you by always own drum-emph-dst boat-inf be-pass-fut-agr. “Your own drum will always be beaten by you”.

b. ek araa sc n has niik jokaa n span khaanan khuaasl a gelahu a him by I good way own food food-cause-pass-go-pst-Agr “I was fed by him (my/his) own food”. c. amarbaabuu oC n kstaarii khaack a jasbaak baat nuukaael a Amarbabu by someone et-ptcpl pase-inf-gum matter Nak I galach I Hide-pst not go-pst-agr “The story of eating sugarcane could be kept undisclosed by Amarbabu”> d. shum umar I nB paki n v I has amarbaabuu sc n ketaarii khaaci a this –emph age in reach –conj ptcpl iaa rahal a chaahi “Even after reaching this age, sugarcane was being eaten by Amarbabu”.

Page 93: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

92

That oblique subjects do control reflexivisation is evident from apas and its autocedenmt in (69) a. But the next sentence is ambiguous in that span here could be interpreted as referring either to ek araa sc n or to hen. The explanation for this ambiguity that comes to my mind is that it has to do with the fact that (69) b is transitive with two human ominals whereas (69) a is transitive with only one human noun. The italicised part in both (69) e and d indicate the higher clause. Notice that in (69)c, amarbaabhu sc n of the higher clause has deleted an identical noun in the lower sentences by an Equi rule. In case of (69)d, amarbaabuu in the conjoined sentence has been deleted under conjunction reduction. In Maithili, unlike in Hindi, oblique objects are accessible to Equi rule as (69)c and fwould show. In these sentences, amarbaabuu nc n and shaa n sc n have been deleted under the identity of higher sentence amarbaabuu and shaa n respectively. As (69) g and h would show, oblique subjects in Maithili can be raised if the higher sentence verb is dekh- “see”, but in most cases, these nominals cannot be raised. Similarly, (69)I and j would point out that although oblique subjects are not accessible to CPF, the resultant sentences could be made loss odd if the controller nominal was a non-subject. Consider the following examples : (69) e. amarbaabuu budhases np kotaarii khaael ajaaob a nosand a old-age in like karait a shalack a de-ptcpl be-pst-Agr “Amarbabu liked to be able to eat sugarcane at the old-age”.

Page 94: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

93

f. chaan cohun dist1 opn Bhol piiTala jacob1 you four sides own drum beat-ptopl pass-inf onehait1 chalahun want-pptopol be-pst-Agr “you wanted (your own drum to be beated in all four sides”. g. han kunkan dvaaran din haan kal2 jamit2 dekhaliaah1 I him by tall-talk cell-pte pass-inf see-pst-agr “I found him taking tall”. h. han kunkaa nOn dhaRaandhaRa madhur khassel a jamit a I him by successively sweetmeat eat-upon pass-inf Dekhaliaah2 See-pst-Agr “I saw him eating sweetmeat repeatedly:. i. sabdin e-Tan haa a kal a bhO kao {? Ahaan kB kon laabh/ bhol a ? / “ahaa kon laabh/keelabu n } “What did you gain by always talking tall?” j. kotaatii khasel n jaa kao { Thumkaa aOn kan mahaan kaaj bholaah I I / kon mahaan kaaj kaalash/ he +instr /SU What great work do/ do-pst-Agr “By eating sugarcane, what great work was done by himself”.

3.4.3.6. Maithili differs from Hindi, Kashmiri and Punjabi most in the Derived subject or SP behaviour. The derived possivised patient or victim of an action in Maithili does not normally receive any case marking. Hence, in the examples given below, the Derived (Patient) Subject kabiliasj sarhaar “the leader of settlement” appeared without kb2 or any such case suffixes. These

Page 95: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

94

Subjects in Hindi only controlled Reflexivisation and Equi, whereas the sentences in (70) suggest that they also controlled OPP and allowed raising, and were accessible to OPP and Equi rules. Consider the following examples: (70) a (banjaaraasab dwaaran ) kabilaak sardaar:apon, Diih a par wanderers by

settlement+gen leader own house-site on maaral a gel a kill – pte go-pte –Agr “The leader of the settlement was killed at his own place (by the wanderers)”.

b. eteo jaa kaI kabilaak sardaar maaral a gela there go-Conj ptopl “Having gone there, the leader of the settlement was killed. c. maaral a jaa kaO kabilaak sardaar tar 1 gel a kill-pte pass-inf conj ptopl cross-pte go – pst –Agr “Having been killed, the leader of the settlement attained salvation”. d. kabilaak sardaar pachbaRiyaak pokhar 1 lag jaa kao backyard-gen tank near go – conj, ptopl maaral a gel a kill – pass. Pte go-pst –Agr “Having gone near the tank at the backyard, the leader of the settlement was killed”. e. kabilaak sardaar maaral a jase caahait a chal a tea kii (kill-pte go(pass)-jaf want-pte be – pst-Agr so that baan kii sabhak jaan becao rest all-gem life save –opt – Agr “The leader of the settlement wanted to be killed so that the life of the rest is spared”. f. han kabilaak sardaar-cBn maarala jaait dekhup chaliianh 1 go-(pass) pte see- pte be-pst-Agr mudaa DorRn kich u nah 1 kar a liaik a but fear from something not do –pst-Agr

“I saw the leader of the settlement being killed, but I did not do anything our of fear”.

Page 96: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

95

In these sentences, one could clearly see that Derived Subjects or NP allows Raising (as in 70f), controls Reflexivisation (70a), Equi (70b), Conjunction Reduction (70d), and is accessible to equi 970d) as well as to Conjunction reduction (70c). What all those and previously discussed examples point out is a necessity for a different subjecthood Hierarchy for Maithili, which is as follows (; the SDER here is equivalent to SP in 58); (58)’ Hierarchy of Subjects in Maithili SI SP Sdt ? S BER ? S DAT ? S OBJ 3.4.4. To conclude this discussion on some relation changing rules in Maithili and their behaviour in different syntactic environments, it may be mentioned that there are syntactic evidence in Maithili Grammar to show that the notion of Subject (both underlying and superficial) could not be characterized in a unique and unitary was as was attempted by the TG theoreticians. Even in the proceeding section where an outline of Relational Grammar was presented (of section 3.3.), it could be observed that even the proponents of RG took it for granted that the relational notions were discrete primitives. That in reality was not so pointed out with a large number of instances from Maithili is correct, then one would expect the same thing to be true for certain other languages too. But to show that the relational notions are fuzzy in other languages does not fall within the scope of this work and hence it has not been attempted here. However, (despite the claim by Klainan, 1979 that this was not

Page 97: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

96

true for Hindi, at least ) if the claim of Kachru et al (1976) regarding the subjecthood Hierarchy (given in 58) is true for Hindi, Kashmiri and Punjabi, it would althemore be evident that different languages were capable of rearranging different kinds of subjects of 58 and 58’ would show this difference between these languages and Maithili). Ofcourse, this does not mean that natural languages would not show any significant similarity in their subjecthood hierarchies. On the other hand, at this stage of research, (71) seems to be a general tendency. (71) A Probable Constraint regarding patterning of Subjecthood Hierarchies: All languages would tend to show underlying Sus to be higher in the Subjecthood Hierarchy and X-rated to be Sus comparatively lower in such a hierarchy. The outline of RG as presented earlier in 3.3. did not only suffer from this defect of not being able to incorporate “fussiness” within the definition of relational notions, it had some more shortcomings too which are to be shown in the section that follows. Later, in the section 3.6, Keenam’s (1976) attempt to characterize the notion of “Subject of” (while trying to incorporate “non-discreteness” associated with this relational notion) would be presented along with the criticism of this position by the Formal Relational Grammarians such as Johnson (1977a) 3.5. A Critique of Relational Grammar 3.5.1. There are several important ways in which one can show that Relational Grammar as it has been outlined in the section 3.3.

Page 98: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

97

is deviant in some respects. Notice that RG and its principles are based on the Relational Hierarchy (NN). Therefore, the best way to begin would be to show that NN and the assumptions RG makes about it are lacunar. Gary and Keenan (1977) show that in Kinyarwanda, a Bantu language, the relational term IO does not exist, and that the NPs that are semantically “Reciepient” and “Benefactive” behave syntactically in the same way as Dos do. For instance, Dos have the following properties in Kinyarwanda, which are shared by these two semantic classes: This language has SU-V-DO-OO order, where SU and Dos do not show case-marking while Oos take propositions: only SU triggers verb agreement; Dos are reflexivizable while Oos are not; both SU and DO show a shadow pronoun in the verb under pronominalisation, whereas certain types of Oos which are prenominalizable take independent form of pronouns; the optional passive rule promotes Dos to subjecthood and denotes Sus to OO levels; and, SU and DO are accessible to relativisation, while Oos are not. How, the recipient and the benefactive in Kinyarwanda behave exactly in the same way as the DO, and in most of the instances they occur before the DO, but after the verb. They are as different from Oos and Dos are. If this analysis is correct, then, it shows that all the languages may not show all the positions on NN, and that a language may show doubling at a particular level. This leads to a conclusion that NN is not a universal syntactic fact. Another instance of syntactic doubling 20 comes from the causativization rule (of. Courie, 1974:11-19). With the introduction of a causer, the causes SU gets denoted to the next available

Page 99: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

98 Vacant position is such constructions. But some languages do allow the original SU either to remain SV, or to denote this cause to another level in the NN which is ready occupied by another NP. For instance, Park, (1973) and kune (1973) give examples from Korean and Japanese, respectively, of multiple subject constructions. On the DO level, doublings are found in Latin, Ancient Greek, Hebrew, Arabic and Sanskrit (of. Centre, 1974). IO-level doublings were found to be commonest, and it occurs in Punjabi, French, Tagalog, Georgian, and in some dialects of Turkish. All these show again that the relational annihilation law as defined in (33) is not a universal principle. Also, such denetions to termhood violate the reranking law as formulated in (37). Notice that it is only logical to expect doubling on the term levels (i.e., on SU, DO, and IO), as OO in NN has already been shown to assimilate a number of categories such as locatives, ablatives, instrumentals, etc. Finally, another sources of doubling is the impersonal passive construction attested in Spanish, Latin, German, Polis, Dutch, Finnish and other languages but rather subjects are denoted to other termhood status. If this is the only way of forming passivisation even in one language, then (31) would no more remain a “Universal” passive rule. It also shows that in the superficial levels, SU need not be expected in all the constructions of human languages. Lastly the X-Rating Convention (Johnson, 1974a) which puses the “usurped” NPs to 00 level does not seem to work here and two Dos

Page 100: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

99

or Ios are allowed to stay in surface when such a passive rule operates. An example of denotion without promotion near home would be Bengali constructions such as in (72)c as against (72)b such as follows: (72) a. asmi etaeke bhaale zone keri I + SU it +NO good mind_lee_de_pres_agr “I think it to be good’ b. ammaar eTan bhsale zone h0E I+00 (Pres) it +SV “It is thought to be good by no” c. ammaar eTaake bhsale zone h0e I+00 it+DO “It is though to be good be me” In (72)a, ammi bears the SV-relation and eTan-ke to the DO, but in (72)b the latter is promoted to subjecthood when the former is denoted to 00-level. But notice that in (72)c, ammi of (72)a is denoted to 00-level, without promoting eTan-ke to SU-level. 3.3.2. Johnson (1974)a had himself shown that Ergative Absolutive languages such as Dyirbal followed a hierarchy which was very different from AN (ef. 23) and NN (ef. 43), and it has a hierarchy such as follows: (73) The Ergative Hierarcy (NN) SU of intransitive /DO ? SU of transitive ? XO ? 00 --- Without going into details of NN, one can only say that given the fact that it exists, it would imply that NN in (43) is not universal. Hence, the laws that make use of NN are also of typological nature.

Page 101: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

100

3.3.3. There is also evidence that some of these laws are not part of the “legal” code of particular languages. One such “law is the agreement law21 as in (34). Lawler (1977:225) reports that “Perlmatter and Postal’s original claim, which seems to be supported by evidence from most languages with verb agreement, was that such agreement could only be triggered by terms. ….. and further, that these must be cyclic terms of the clause in which agreement takes place ….” This position follows from an assumption on the part of the RG theorists that the rules in the relational component are cyclic, whereas the non-relation changing rules are probably ordered but not cyclic, and that the former rules apply on blade before the latter. In Lawler (1975, 1977) there was ample proof that at least in one language, Achonese, verbs agree with the pre-cyclic terms, i.e., the Sus, Dos and Ido of the Underlying Relational Network. It is very important to note at this point that the Achonese data suggests that in some languages the organisation of grammar may be such that some of the non-relation changing rules can or must be ordered before any of the relation-changing rules apply. This, in turn, implies that the organisation of RG in (42) is at best typological statement, and is no more a universal. Recently, in Singh (1978b) it was shown that in Naithili, Oos could control and trigger verb agreement in the same way as any other relational term would do. Thus, it is claimed there that RG should be satisfied with a relatively weaker principle then (34), which says that (74) The NPs which control verb agreement, if any, include b-subjects (Keenan, 1976:316).

Page 102: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

101

3.3.4 The evidences presented in the above paragraphs show quite clearly that there are many loopholes in the formulation of RG as sketched earlier. At the same time, it is also true that a definition of relations on the basis of grammatical categories given in Chonsky (1965) runs into a number of problems pointed out earlier in section 3.2. In what follows I would present a solitary attempt at defining the relational notices within the framework of RG and the subsequent criticisms thereof. I shall thereafter attempt to find out as to what the Generative Semanticists and the lexicalists had to say about relational notices. 3.6. An Attempted Definition of Relational Notices: 3.6.1.1 Gary and Keeman (1977) have opined that Relational Grammar has a serious drawback in that the relational notions are not defined properly. Their standpoint has been “universalist” as against the “generative” or “fermalistic” viewpoint of their counterparts, represented mainly by the works of Perlmutter. Postal and Johnson. It is with this viewpoint that Keeman (1976) made an attempt to define the notion of “subject of” (SU). This is an important exercise as it not only characterises the most basic relation in the RN, AN or even in NN, but it seems to be the lone viable alternative to Chemsky’s (1965) definition of the same22. According to Keeman (1976), it is necessary to define SU and other relations because he thinks that the universals proposed in terms of these notions need to be verified in languages belonging to different families. Such cross-linguistic verifications are possible only if these notions can be defined

Page 103: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

102

once for all in a universal manner. Otherwise, he thinks, one would have to depend on the “intuition” of the analyst of a particular languages to be able to find out whether an NP bears a subject relationship to the verb in a particular sentence. The dangers involved in such a ‘dependence of “Basic” sentences (er, b-sentences) and then tries to identify a set of properties (called “subject properties list” or SPL) that would be universally valid for identifying the subject NPs. The NPs showing a “clear preponderance” of SPL are “b-subjects”. Notice that he does not rule out that different languages would have additional idiosyncratic properties for the bp-subjects too. Sus of the non-basic sentences would be defined in terms of their predominance of the priorities of b-subjects. According to him, if the meaning of a sentence depends on another, the latter is “semantically mere basic” than the former, and a basic-sentence is “more” basic than all other complete sentences of a language. The basic-sentences “will present the greatest range of tense, need, aspect, mode, and voice distinctions. They will have the greatest privileges of occurrence: they will accept the greatest range of verbal and sentential modifiers; they will be the easiest to embed and adjoin to other sentences, the easiest to nominalice and internally reorder, the easiest to relativise, question and topicalise out of, the easiest to prenominalise and delete into, etc” (Keeman, 1976:309). These b-sentence will be semantically “unambiguous”, and in most of the cases they will be declarative and affirmative, and semantically “least informative” sentences

Page 104: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

103

of a language. Information-wise, the non-basic sentences will always be “richer” than the basic, and the former will actually modify, elaborate, constrain or extend the meaning of the latter. They will also be “relatively” free from presupposition. Finally, it is not necessary that b-sentences of one language will be translatable into another always as b-sentences. Given such a characterisation of basic-sentences, one can now present a list of properties that subjects of such sentences (in other words, basic-subjects) would exhibit. 3.6.1.2 Keenam (1976:311-23) presents about thirty-odd properties of pragmatic,

semantic as well as syntactic nature. I would summarise then below: (74) AUTONOMY PRPERTIES:

a. The b-subject is likely to exist independently of the action or property of the predicate. b. A non-subject is more easily deletable than a b-subject. c. The reference of b-subject is “autonomous” in a sense that the addresses cannot depend on the non-subject NPs to be able to determine its identity. If two NPs have the same reference, the non-subject is more likely to get “marked:. Related to this are the following autonomy properties. d. In general, b-subject are possible controllers of reflexivisation and cereferential deletion in most of the cases. e. The control of “switch reference” is largely limited to b-subjects f. The NPs which control verb-agreement, if any, include b-subjects.

Page 105: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

104

g. The NPs which undergo reflexivisation or cereferential deletion across clause boundaries will always include b-subjects. It applies to the coordinate cerefenential deletion also. h. Generally, if the b-sentence is “true”, then the generate or abstract entity that has been referred to by the b-subject is also true. (Notice that this criterion is inapplicable in case a b-sentence has a “dummy” subject). i. The existence of the entity referred to by the b-subject is harder to demy under negation, interrogation, or conditionalisation than that expressed by the non-subjects. j. Netaphoric idioms which “suspend” the reference or existence implication or NPs affect b-subjects rarely. k. B-subjects are normally the “topic” of the sentence-an information shared by the addresser and the addressee. l. Highly referential elements such as personal pronouns, proper nouns and demonstrative can more freely occur as b-subjects. m. “advancement-to-subject” rules are more commonly found than advancement to object positions. n. “ b-subjects have wider “scope”than the non-subjects in most cases. o. B-subjects are “normally” the leftness occurring NP in b-sentences. p. The NPs which can be relativised, questioned, and left include b-subjects. q. The NPs whose possessors can be relativised, questioned and cleft include b-subjects. r. If a language has a relativisation strategy to leave a pronoun in place of the “affected” NP, such as NP is rarely a b-subject. s. A b-subject can be “raised” in all languages. t. Norphologically independent ( and emphatic) pronouns can always occupy b-subject position. u. NPs which “launch” floating quantifiers include b-subjects.

Page 106: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

104

(75) CARD MARKING PROPERTIES:

a. b-subjects of intransitive sentences are usually not case-marked. b. The NPs that change their case marking under causativisation include b-subjects. c. The NPs that change their case marking under action nominalization include b-subjects

(76) SEMANTIC ROLE:

a. The semantic role of the reference of a b-subject is predictable from the main verb, and semantic restrictions on non-subjects are usually more than that on the b-subjects. b. B-subjects normally show an “agent” relationship vis-à-vis the action. c. Subjects normally express the addresses phrase of imperative. d. B-subjects show the same position, case marking and verb agreement as does the causer NP in most of the cases.

(77) IMMEDIATE DONINANCE: The b-subject is immediately “dominated’ by the root S-mode. (Notice that only if the VP is considered to be a universal category does this criterion hold. But even in that case, it is at best a typological or probabilistic statement). 3.6.1.3. Before I present criticisms by the later relational theoreticians of Keeman’s attempt to define the notion of “subject of”, let me briefly mention as to what Keeman thought could be the utility of such a definition. Apart from the obvious utility such as a universal way of being able to characterise the “b-subjects” in any and all the languages of the world, it also helps in verifying already proposed generalisations that refer to the notion of subject as well as in making fresh cross-

Page 107: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

106

linguistic generalisations. Keeman also assumes that since not all the subject-properties could be attested in the b-subjects of all the languages of the world, there must exist a hierarchy among the subject-properties, i.e., some properties must be harder for the derived subjects to acquire and consequently some others must be easier for the subject-to-be-in chamage to losea. This idea led Keeman to propose the following hierarchy: (78) The Promotion-to-Subject Hierarchy (PSH0 I.

CODING PROPERTIES > i. Position > ii. Case Marking > iii. Verb Agreement

II

BEHAVIOUR AND CONTROL> PROPERTIES

Deletion, Movement, Case Changing, Central of Cross-Reference Properties, and others

III.

SEMANTIC PROPERTIES Agency, Autonomous Existence, Selectional Restrictions, and others

What (79) implies is that the derived subjects will tend to take the higher properties in the hierarchy, and if in a particular language they take up any of the properties at the lowest level (i.e., the semantic properties), it must be the case that all the higher properties have already been acquired by the derived

Page 108: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

107

subject NP concerned. Also note that within the coding properties, position, case marking and verb agreement form a hierarchy of a similar kind. In this case too, the higher the property, the easier it is for a detonable subject NP to lose. 3.6.2.1. The definition of the notion “subject of” essentially depends on the characterization of what Keeman calls “b-sentence”. Although Keeman talks about a number of “qualities” of a b-sentence, while defining the latter, he chose the following: (00) For any Language L,

a. a syntactic structure x is essentially more basic then a syntactic structure v if, and only if, the KRANING and v DEPENDS on that of x. That is, to UNDERSTAND the meaning of v it is necessary to understand the meaning of x. b. a sentence in L is a basic sentence (in L) if, and only if, no(other) COMPLETE sentence in L is MORE BASIC THAN it. (Keeman, 1976 :307; italics UNS)

3.6.2.2. The italicised words in the definition are not “unambiguous” or “intuition-free”as Keeman would like them to be. And for (80)b, Johnson (1976b: published as 1977a) rightly argues that it “has the unfortunate result that any paraphrase of a given sentences N is more than S”. This is because in that case two sentences would have the same incoming, and knowledge of the meaning of N1 would automatically imply that one also understands the meaning of S2. Keeman also talks about a ‘syntactic simplicity’, which in itself is a not-so-well-understood notion. No is, moreover, freed to take recourse to a weaker definition. No would thus say that “an NP in a b-sentence

Page 109: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

108

( in any L) is a subject of that sentence to the extent that it has the properties in the properties list” as in (75) through (76). When he fails to identify the ‘necessary’ and “sufficient” criteria from those properties, it is difficult to understand as to why then is he trying to agree that a general criterion of “elder preponderance” would make that defect up. Moreover, he does not specify with concrete examples from natural languages as to how such a criterion of “clear” preponderance” would actually work. Johnson (1977a) argues that this criterion can be understood in at least three different ways (01) a. the b-subject must have at least a majority of (say, 16) properties,

b. it must have more of the subject properties than any other NP in the same clause, c. it must have a specific number of properties (does not matter as to what that number is).

3.6.2.3. Keenan, at one place, categorically stated that he would understood the motion of “clear preponderance” in the sense of (SO) b. In that case, it is possible that B-subjects in two different languages are called b-subjects in comparison to other NPs, when they employ two different sets of criteria. This would, then,violate Keenam’s own assumption that “ if we use different criteria to identify subjects in different Le then “subject” is simply not a universal category and apparently universal generalisations stated in terms of that motion are not generalisations at all” (Keenam, 1976 : 305). Johnson rejects

Page 110: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

109

the relative interpretation (as in Sob) on these grounds. But there seems to be another reasons why this interpretation is to be rejected. Recall the hierarchy of subject properties as in (79) and imagine a situation where an NP y has meet or all of the properties at the ending, and behavior and central level, whereas the NPx, which is “otherwise” the real subject, has only some or all of the comentic properties. My virtue of sheer number of properties, NP y will outnumber NP X, whereas PHH as stated in (79) would suggest that the comentic criteria seem to be more “basic” in that a dometable subject would lose than very rarely, and for a promotee subject NP they are the hardest to acquire. It seems, therefore, that the assumptions behind (79) and 80)b are contradictory. If we accept the universal statement regarding PSM, we are fareed to conclude that (EO) is a wrong or undesirable interpretation of the term “clear preponderancies”. 3.6.2.4. Johnson (1977a;679) also points out that keenam’s definition of “subject” would not recognise the dummy subjects in English (e.g., it and there) and other language as subjects, just because they would not have the majority of the properties in SPL. Similarly, the “Idiom chunk” and “ sentential subjects” would also be excluded by such a definition. Moreover, the distinction between the universal list (UL) of subject properties (such as PH and other Laws or Principles) and the specific subject Properties list (SPL) is not clearly maintainable. One could always argue for some of the criteria in SPL to be a part of

Page 111: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

110

UL, or vice versa. For instance, the verb agreement property in (75)f and the reflexivisation and cereferential deletion properties in (75)g could have been put in UL 23 rather than in SPL. The same is the case with a number of other criteria as in (75)p, q,s,t and in (76)b. One may also add in this respect that while a number of properties in the SPL may help one in correctly IDENTIFYING the subjects in natural languages, they are NOT USEFUL for the purposes of DEFINING the motion of subject, because they give rise to CIRCULARITY. (For details, see Johnson: 1977a). if one wants to take off these properties that make the definition circular, for the SPL, one would be left with very few criteria. 3.6.2.4. Notice that Johnson (1977a: II) stretches his point to an extreme when he claims that comentic or pragmatic properties in the SPL are ‘irrelevant’. He says, and I quote: (01) ‘… in most current theories with intermediate grammatical stages, there would be, in general, no comentic or pragmatic properties associated with intermediate stage N.P. Thus, intermediate stages would not..have SUBJ. Since there is GOOD EVIDANCE FOR INTERMEDIATE STAGES, it appears that a predominantly non-syntactic approach would FAIL to define SUBJ in a significant class of cases” (Johnson, 1977a: II : EMPHASIS :UNS). (02) can be criticised in a number of ways and by doing so, one can show that Keenam (1976) is not at fault when he used non-syntactic criteria coupled with syntactic ones in defining SUBJ. As I can see it, ( 02) has the following defects: First, the very use of an expression such as “Significant number of cases”

Page 112: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

111

shows that even Johnson believes that the semantic/pragmatic criteria can at times be of help24. Secondly, if one could prove that in some languages, the correct identification of Subject is necessary for applying a rule properly. Johnson’s statement that “intermediate stages would not . . . . have SUBJ” could be shown to be false. Thirdly, by taking the claims of SONE of the MOST current theories as almost GIVEN25, Johnson is actually making use of the well-known TG position in this respect. 3.6.2.5. However, within the traditions of TG grammar, if we take Jackendoff’s (1972) views into consideration, the so-called non-syntactic approach (along with syntactic one) does not seem to be that unjustifies. In Jackendoff’s scheme of things, a model of grammar would lack like something such as follows.

Notice that although the main purpose of constructing such a model of grammar was to save the level of “Deep Structure” (from the generative semanticists’ criticisms of such a level), the merit in this particular organisation is in the recognition

Page 113: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

112

of the fact that different levels of semantic representation (comprising various semantic/ pragmatic facts) must be referred to by the “intermediate” stages of derivation. If one can imagine of a model where Semantic Representation is not necessarily an interpretative output only, that is, where they can also at times food the formulation of syntactic rules, the arrows from Transformational Component to Semantic Component would be bi-directional. If one considers the motion of Scale an unnecessary burden on Grammar (as is evident from its exclusion in APG), and think only in terms of ordered rules in the syntactic congement, and granted that Base Component does not exist and that Beep and Surface structures are to be replaced by Underlying and Superficial Relational Networks, thereby leaving the task of surface ordering on Idnearisation Rules, a translation of Jackendoff’s model of grammar into relational terms would lack like the followings

Page 114: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

113

3.6.2.7 All these arguments point out to a single fast, viz., that semantic and pragmatic criteria can and must be used to “identify” the b-subjects in human languages. Note that I did not use “define”, because the conclusion that was drawn in Johnson (1977a) is, I think, still remains. From the failure of Keenan’s (1976) attempt at “defining” the notion of “subject of” and also from Chemesky’s (1965) failure in doing the same, Johnson (1977a) concludes that Relational notions such as Subject and Object, etc. should be taken as “undefined” and “undefinable” primitives in any viable grammatical theory. 3.6.2.8. That the TG theoreticisms either did not think it important to precisely define relational notions (as in the case of Generative Semanticists) or defined them wrongly (as in Chemesky, 1965 as well as in Chemsky, 1970 and in Jackendeff, 1974) may not be enough to make such a judgement about deniability of these as was done by Johnson (1977a), but the failure of a follow Relational Grammarian in defining the same definitely throws some weight in favour of Johnson’s conclusions. In the following paragraphs, I shall discuss in brief as to how Generative Semanticists viewed at this problem. 3.7. Generative Semanticists and the Notion of Grammatical Relations: 3.7.1.1 In a tiny little paper (resembling a manifests). Lakeoff and Ross (1973)26 once declared that they “believe semantics may be generative, for the following reasons “ 1. There has never been any argument that semantics must be interpretive and syntax generative, and (2) As far as we know,

Page 115: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

114

there are only four types of arguments for the existence between semantic representation and surface structures of an intermediate level which has any theoretical significance” (Lakeff & Rose, 1975:1; 1976:160). 3.7.1.2. The arguments referred to in the second reason were though to have coincided in defining the level of Deep Structure. These conditions included the following: 85. A. The simplest base for all transformations to work on. B. The place where occurrences and selectional restrictions are defined. C. The place where basic grammatical relations are defined. D. The place where lexical items are inserted from the lexicon. 3.7.1.3. MacGawley (1968a) in “Lexical Insertion in a Transformational Grammar Without Deep Structure” and in many other places had shown that D above was false, as insertion takes place at different stages of derivation. A recent case that has come to my notice is Angami Naga complemetiser placement and ‘non-lexical’ pre-form selection. In Angami, the same forms occur as complementisers as well as pre-forms, and the selection of the lexical forms have to be delayed until the reordering rules are applied. This is because the lexical insertion here is dependent on the surface ordering of elements27. Similarly, B has been shown by McGawley (1968b) in his paper entitled ‘concerning the Base Component of Transformational Grammar’ to be guided by semantic fasters rather than by syntactic once.

Page 116: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

115

According to Lakeff and Rose (9175). A surely defines a level of representation, but this level in itself has no systematic interconnection with any other facts of language, because the T-Rules do not operate at this level.

3.7.1.3. However, what is most interesting is their observation regarding C, especially about the necessity and function of Grammatical Relations as they saw it. I present their viewpoint in their own words: (86) “. . . . subject of and object of are not directly relevant for semantic interpretation. The only reason they necessary is so that there is some way to keep apart dog bites man and man bites dog. For as has been realised for a long time, there are many kinds of interpretation of subjects and objects . . . . All that is necessary is that subject and object be kept distinct. But they are in surface structure (through different cases) or in the semantic representation, and we see no reason for singling out the level fixed by A above for special merit- the subject and object defined at this level seem to have no semantic or syntactic relevance, so why base a theory on them” (Lakeff and Rose, 1973:2 1976 : 161). 3.7.1.5. It does not take much time to understand from a policy statement of the Generative Semanticists such as above that their idea of relational notions was restricted to the following. (2) In their theory, Grammatical Relations would be required only to differentiate between the different case roles that the same NP and acquire26. This is important for them because they were still suffering from an illusion of REDUCTIVSM. This started with Lakeff (1965, later published as 1970) and continued with Bach (1968), Garden (1968, 1975), Ross (1969), McGawley (1970a, b and 1972) till almost all the syntactic categories fall one after

Page 117: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

another29, and it was decided that even word order typologies30 are accessible to such a reductivism. At that point it was believed by the generative semanticists that tenses, quantifiers and auxiliaries are actually main verbs, and even propositions and nouns are underlying predicates. The VP was also thought to be a derived category (following Fillmore, 1960). All these exercises were thought to be important steps towards bringing in “Symbolic Logic” to bear upon Grammatical Analysis (See Brame, 1976). Since they believed in the heart of their hearts at that point that “Unordered” base is THE replacement for an ordered Chanakya base component, the notion of grammatical relations taken by them as some sort of rele-marking criteria. This is clear from their man and dog examples. This, they thought could be the ONLY reason for having such notions as Subjects of and Object of. (2) Secondly, as a consequence to this, they hold that it was enough if the relational notions are “kept distinct” (which I interpret as somehow “identifiable”, and not as “definable”) in the surface structure. The falsification of their position regarding (1) and (2) above are enough indications that relational notions need to be taken a little more seriously than was done by these linguistics. The failure of those semanticists is equating language structure with logical structure which gave rise to the study of pragmatics in the recent years is also a pointer to the same direction. (3) Thirdly, the statement that defining relational notions at the underlying level has “no semantic or syntactic relevance” need not be argued against here, as it has already been shown earlier in this chapter

Page 118: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

117

as to how one is able to capture a number of syntactic generalisations if those relational notions are built into a theory of grammar in such a way that they make comparisons of language descriptions possible. That is, those notions may be taken as “trivially” universal insefaras they become a part of the theory of grammar. 3.7.2.1. There seems to be one danger in challenging the Generative Semanticist’s position concerning the importance of grammatical relations or the role they may play in the descriptions and interpretation of the syntactic and semantic facts of natural languages. One may raise a methodological objection to the selection (or singling our) of Lakeff & Ross (1973) as a material that truly represents the Generative Semantics position. In other words, one may argue that there is a “Resetta Stone Strategy” (Botha, 1976) adopted here to deliberately mislead readers regarding the Generative Semantics position in that (I) we did not select the works of Generative Semanticists which are highly relevant, (2) concentrated on material (s) that could more or less be called irrelevant, and (3) to suppress this defect, we camouflaged the non valid argument by assigning overdue importance to the work that we selected. One may recall that Botha (1976) had first accused Ray Dougherty (1975) of adopting this strategy when he had taken up irrelevant works of Postal (1966) and Lakeff & Ross (1970 to criticize Generative Semanticists’ works on methodological grounds. Botha (1976:3) suggested that to maintain the minimum ‘standards of adequacy’.

Page 119: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

118

a methodological analysis must meet the following cretinism: (87) “Materials for methodological analysis should be selected in a non-arbitrary

and unbiased manner” (Emphasis:UNS) 3.7.2.2 While one can appreciate Botha’s (1976) concern for what can be termed as ‘judicial impartiality’ on the part of a linguistic theoriticism, it would be some extend be difficult for a methodologist to select in a nonarbitrary and unbiased way one or more works of Generatife Semanticists as he himself agrees that “generative semantics does not represent a monolithic, static approach to the study of natural language” (Botha, 1976:3) and that “the theoretical viewpoints of none of these scholars have remained semantics” (Botha, 1976:3). If what Botha said about the Generative Semanticists was true ( and the fact that Lakeff and McGawley, while talking to Norman Farret (1974), had admitted31 this to be true makes it althomore convincing), it would really be difficult for anyone to be nonarbitrary or unbiases because there is always a danger that Lakeff or a Ross would interject an analyst to announce that they have changed their position or that the material included for analysis was irrelevant. This, to my mind, leaves the methodologist chasing a will-o-the-wisp. While some can deny the moral value of the sermon in (87), there must be a practical way in which a methodologist can act while selecting materials for analysis. It is not for me here to suggest such ways, but I feel I must defend myself for accepting a

Page 120: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

119

statement such as in (86) as a true representation of the Generative Semanticists’ thinking on grammatical relations and the role these notions play in constructing a theory of grammar. 3.7.2.3 There seem to be two ways to prove that my selection of (86) as a ‘policy statement’ on the part of the Generative Semantics was NOT a “Resetta Stone Strategy”. One of the ways is to give documented evidence as to what the follow Generative Semanticists thought of Lakeff & Ross (1975) when it was originally written (i.e. around March, 1967), and how they looked at this work later. Another way would be to show from the works (done by this group) that appeared later and those which are rated very high by Botha (1976) or other ‘unbiased’ methodologists that the Generative Semanticists’ position (regarding the notion of grammatical relations) remained the same (at least, upto recent times) inspite of a massive overhauling of their own theory by the subsequent developments that took place within this school of thought32. 3.7.2.4. I shall take up the first task first. In support of my claim, I shall give two large quotes from McGawley (1976b): (88) “This paper (Lakeff & Ross, 1975, UNS) originated as part of a letter from Lakeff and Ross to Arnold Gwicky, written March 1967, and was circulated in duplicated form at that time. It figured in much discussion at the conference on language universals and generative grammar that was held at the University

Page 121: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

120

of Texas at Austin in April 1967 . . . . , and it greatly influenced the “Festoscript” section of my contribution to that conference (McGawle, 1968b). Indeed, Lakeff and Ross’s paper is what turned no from revisionist interpretive semanticist into a generative semanticist. The program that it announces for the description of language . . . .. directly influenced most of the research done by generative semanticists in the late 1960s. “ (McGawley’s (1976:159) editorial comment on Lakeff & Ross (1973) which was reprinted in McGawley, 1976b). (88) “The paper that first raised these questions (- i. ‘Whether anything was to be gained by distinguishing between syntax and semantics”; ii. ‘whether transformations did not simply amount to rules for specifying how semantic structures are related to surface structures”; iii. Whether “base rules did not simply amount to rules specifying what semantic structures were possible”) seriously is Lakeff and Ross’s “Is Deep Structure Necessary ?”. . . . which can be said to mark the beginning of the variety of transformational grammar known as generative semantics” (McGawley, 1976b:10, in “Introduction”). These words of praise would, I believe, do to remove the doubts that I selected an obscure and irrelevant work as representing Generative Semanticists’ opinion regarding certain notions.

Page 122: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

121

3.7.2.3 Now we can turn towards the other task, namely, that of finding other ‘classical’ or ‘important’ (whatever that may mean) works where Generative Semanticists have said essentially the same thing as in (86). When one tries to do this task, one faces a different kind of problem altogether. It turns out that in most of the important works done by the Generative Semanticists, there is either no mention of the relational notions. Or at best a passing reference of the same. On the other hand, some of the relatively unimportant studies within this framework refer to these notions. But handle them in an unsatisfactory manner. I shall give one example of the latter type of work. In a “fairly widely read” underground study (which “has been given little attention in the above ground linguistic literature33) written in 1967. Anna Wierabicka (1976) takes such a position. She in one place argued that “it is scarcely necessary to argue that verbs like sit down, get up ,walk, etc are semantically not less “transitive” than seat , raise , carry, send, etc….. Verbs like sit down, simply contain their object in their own meaning, this “object”, but what is actually contained in the meaning of those verbs is the subject of another underlying sentence, i.e., the sentence that describes the fact caused by the will of the person whose name constitutes the subject of the first underlying sentences)13. (Wierabieka, 1976:141). She makes her argument more explicit in the footnote 13 where she proposes to do away with the relational notion such as “Object”. I quote what she has to say about it:

Page 123: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

122

(89) ‘ Semantically, the notion of “object is senseless ( or at least superfluous); the object of an action is simply the subject of some state of affairs that (the state of affairs) is caused by some other state of affairs . . . Semantically, there are subjects . . . ., and there are properties –predicates. Elements that are suitable as subjects of sentences are not suitable as predicates, and vice versa”. (Wierabicka, 1976: 151:2). 3.7.2.5. I don’t think I should try and show as to how erroneous are her conclusions. One way, however say that this should not be taken seriously as the original paper was written in 1967. But the reason I still quote Wierabicka is that she still maintains this position. Consider the “1974 COMMENTS” appended to the original paper published in McGawley (1976b:152-7), where in the very beginning she says: (90) “. . . . the main point of the paper- - - seems to me valid. I also remain in agreement with the claims concerning the syntax of semantic representation, in particular with the thesis of the notion of “object” is irrelevant to deep syntax, and that the only categories that are relevant are these of subject and predicate” (Wierabicka, 1976: 133) 3.7.2.7. Compare with this Lakeff’s (1971) “On Generative Semantics” or McGawley’s (1976a) collection of twenty-two papers on ‘Grammar and Meaning’ or many more works of Rose, Postal and George and Robin Lakeff. As far as my reading of

Page 124: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

123

their works goes, I found that they took these notions for granted, and never tried to define them, nor did they seriously try to evaluate them. But whenever they felt it necessary to use them to write to rule, or draw a tree diagram or to explain different kinds of linguistic facts, they did. For instance, Lakeff (1971:270-2), while reproducing Postal’s (1970) REMIND-argument, frequently uses the relational notions in a schemata34 or in the diagrams of deep and various intermediary structures25. Although Rose (1972,1974) arrives at a number of remarkable conclusions regarding the non-discrete nature of various things such as syntactic rule, grammatical categories and their membership, grammatical islands, grammatical environments, hierarchies of grammatical elements, and speaker-judgement regarding acceptability (of. Lakeff, 1973b: 271), he does not extend the notion of “fuzziness” further to cover some conceivable assumptions such as follows36. (92) a. It is not the case that grammatical elements simply bear or do not bear a particular grammatical relation to the verb of a sentence; rather, they are subjects or objects to a degree.

b. Grammatical relations form “hierarchies” which are constant to a degree (thus leaving a scope for variability from language to language open)

c. Language universals stable in terms of relational (or other) notions are universals to a degree.

Page 125: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

If Ross could talk about nominees, sentenciness or clausenateeness, there apparently seems to be no reason as to why he should not include (92), especially (92)a. The only explanation that comes to my mind readily is that he believed what he and Lakeff said in (86) even in the early seventies.

3.7.2.8. While reviewing Sebeck’s (1966) ‘Current Trends in Linguistics’, Vol.3 where Chemsky (1966b) was reprinted, McGawley (1968c; repr. In 1976a; 169) in a footnote had expressed a doubt whether it was meaningful to say that the ‘object relation’ was something unique that remains constant no matter what is selected as the predicate (and also, what kind of nominal is selected). No says: “ As far as I can tell Chemsky’s ‘grammatical relations’ are as much semantic as syntactic: his reasons for asserting a particular grammatical relation to hold in an example appears to be simply that there are other items containing taboos of the same LEXICAL ITEMS standing in the same SEMANTIC RELATIONS as in the examples just quoted3 “ (McGawley, 1976a: 169). No goes on to say in this factuals 3 that “ one major omission in Chemsky’s account of ‘grammatical relation” is justification of the assertion that all the things he treates as examples of . e.g. the ‘object relation’, being together. While it is easy to convince oneself that John and Please are related the same way in John is easy to please and No one can please John. It is not clear even what would be meant by saying that John, it is not clear even what would be meant by saying that John and please are related in those sentences in the same way as are build and seven birdhouses in John has built seven birdhouses

Page 126: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

125

(McGawley, 1976a: 169, fa 3). Apart from those stray comments and some more, we did not find McGawley trying to define those notions and incorporate them as basic tools for linguistic description as some of his generative semanticist compatrists (such as Postal and Pearlmutter) did in the latter days. 3.7.2.9. Lakeff, however, seems to have changed his position quite early, as long back as in Lakeff (1973b). More, while presenting a typology of functions of ‘correspondence rules37 – functions such as ‘trigger’, affected elements’, ‘domain specifier’, and ‘controller, he wrote the following words which showed that in 1973, he hold a different view about the relational notions: (93) “The ‘input’ and ‘output’ conditions in a rule will specify the relations holding amount the functional components of the rule. In classical correspondence grammars the relations were PRECEDES, DOMINATES, COMMANDS, etc. However, I would now accept the observations of Keeman, Postal and Perlmutter and take grammatical relations rather than precedence relations as basic . . . . I would therefore include such grammatical relations as SUBJECT-OF, DIRBOT-OBJNOT-OF, INDIRBOT-OBJNOT-OF, and COMPLEMENT-OF” (Lakeff, 1973b, 1973b:278). The above statement, I think, makes it very clear that the generative semanticists never took the relational notions seriously before they started formalising their ideas concerning non—discrete grammar. All these therefore show that in selecting

Page 127: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

126

(86) as a true generative semanticist position regarding those notions, I did not use a Resetta Stone Strategy. 3.8. Another Attempt to Save TG Defination of Relational Notions: 3.8.1. In the following paragraphs, I shall discuss one more attempt at defining relational notions by Chemsky (1970) and his associates. It could be pointed out that there is a general awareness by now even among the “lexicalist” camp of the importance of the role of grammatical relations in syntactic theory. Chemsky (1977:73) notes that applications of universal conditions on rules such as Specified Subject Constraint “in a language depends on the characterisation of the notion “subject” in this language . .. . For some case languages, one might want to chatacterise “subject” in terms of such notions as orgative, absolutive or nonoblique. Hale (176) proposes certain conditions on what can be taken as subject in the syntactically “unmarked” situation; in accordance with his approach, a language might characterize the notion “subject” differently, but at a cost in the grammar, in accordance with the logic of markedness. One would expect that current work in “relational grammar” will shed much light on these questions”. 3.8.2.1 Chemsky (1970) was an I-depth study of a problem that could not be solved if one maintained the characterisation of transformations and that of the lexical component of TG model as proposed in Aspects (Chemsky, 1965). The problem was concerning the ‘derived’ nominals (e.g., criticism, gift, etc.) which were discovered to be a very different entity – different

Page 128: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

127

from the ‘gerundive’ nominals (e.g., criticising, giving, etc.), in that while the latter could be generated transfernationally, the former could not be derived in the same manner. If one tried to generate the former transformationally, one would end up with very difficult, unnatural, idiosyneratic, cumbersome and a large number of language-particular rules. This was due to the ‘fact’ that the former behaved like nouns, whereas the latter were more sentence-like. In particular, the points of difference were the followings. (94) Conditions Derived Gerundive a: Occurrence with aspectual36 verbs such as have No Yes b. In partitive constructions such as with some of,

many Yes No c. Greater range of adverbs such as the often, in a

dubious way No yes d. Modifiers become adjective (as against

remaining adverbials) Yes No e. The use of the before the nominal Yes No f. May undergo subj-raising No Yes g. May undergo dative-shift No Yes h. May undergo particle-movement No Yes 3.8.2.2. Apart from these differences, Chemsky (1970) found out some more, such as non-translatability of certain construction with gerundive nominals into ones with derived nominals (e.g., its being easy ? “the easiness), and vice versa (e.g., his advice ? “his advising). Another argument was that relationship

Page 129: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

128

between gerundive nominals and their corresponding verbs was regular, whereas in many cases the derived nominals are idiosyneratically related to their verbal counterparts-semantically as well as formally. His examples for the latter type of relationship included laughter, marriage, construction, actions, activities, etc. No proposed that the gerundive nominals must have a nominal head in the deep structure configuration. But at the same time, he was interested in capturing their similarity in respect of common selectional restrictions by somehow trying to relate ‘noun’ (criticism) and ‘verb’ (critisising) in the lexicon. Such a relation could be brought about by either using a syntactic (distinctive) feature or a bundle of features that could collapse both noun and verb under one (syntactic) category. No used a complex category symbol for this purpose which he called X, and allowed one to two bars to be placed on this abstract symbol X (thereby getting two more categories of still more abstract order such as T and T). No introduced two more higher order symbols: Spee (which would include specifiers of different categories, in particular: aux , dot, degree phrases, etc. ) and Comp (which again includes complements of different kind). For example, his ‘base rules’ would include the following39 (95) a. S ? T- T (initial rule) b. T ? [spee, X] – T ? c. T ? I – Comp, . . . . where (95) abbreviates the following obvious rules:

Page 130: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

129

(95’) b. V ? Specv-V T ? Speca –N A ? Speca – A F ? Specp – P Similarly, (95)c would also have four different readings. 3.8.2.3. Having known something about the abstract cover symbol T that collapses all other category symbols, now we can turn towards the problem of finding out appropriate syntactic feature system that could possibly be used to capture different combinations, such as follows (96) Combinations : V N A P (i) : + + + + (ii) : + + + - (iii) : + + - - (iv) : - - + + (v) : - + + - (vi) : + - - + (viii) : + - + - 3.8.2.4. To resolve the problem of derived and gerundive nominals, finding out a feature that would combine N and V would have been enough, i.e. one could try to find only (96)iii. But that would be, in the first place, an ad hoc solution, and that would require that in all other cases, the transformations would have to be used to relate lexical items. We may note at this point that although we should “expect there to exist rules whose structural descriptions refer to a range of structures including more than one value of Xn, i..e. rules that affect V and N or those that might affect I and N, “we do not expect

Page 131: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

130

to find rules whose domains include nodes at different levels, for example F and X. (Jackendeff, 1974:11). It also would not be very surprising if we do not find all the possible domains referred to in (96) in any one particular language. 3.8.2.3. Being inspired by the traditional method of cross-classification, i.e. by a set of distinctive features, Jackendeff (1974: 12-4) introduces features such as [ ? Subj], [? Obj] and [? Comp] which combines and differentiates all the syntactic categories as the following matrix would show: (97)

+Subj -Subj

+ Comp -Comp +Comp -Comp

+Obj Verb Aux/medal Proposition Particle

-Obj Noun Quantfier article Adjective adverb

Notice that one could also use notations such as [ - Subj] to combine the traditional class of categories such as PP, PrtP, AP and AdvP. If a category allowed ‘Subject of ‘ to be identified, it could naturally be called [ +Subj] under this scheme. Similarly, [ +Obj ] would include only these categories that include an NP “direct object” after the head. The feature [? Comp] was required to differentiate between adjectives and adverbs mainly. In (97) the feature [-Comp] was used only for these mophological categories which resembled their [+Comp]40 counterparts closely, but lacked the full syntactic possibilities of the latter.

Page 132: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

Subj

T

?

131

3.8.2.6. Given such an oversimplified description of the T-convention as above, now we could find out as to how one could define the grammatical relation “Subject of” within this framework. It may be mentioned here that the proponents of this convention showed keen interest in redefining not only this particular relation, but also other structural relations like ‘Commands’ and “Predicate of’ , etc. The generalised subject relation in Jackendeff’s (1974:14-6) scheme of things looked like as follows: (98) “Subject-of” – [N, [+Subj]]41 The “Object” relation was defined as in (99): (99) “Object” – a. in V, it is: [V-NP-Y] b. in N, it is : [N-(Prt) – NP-Y] (99) a and b could be combined into a statement such as follows: (100) “Object” – in , it is : [ x-(Prt)- NP-Y]48

The relation “predicate’ could also defined in the following manner. (101) “ Predicate” – in V, it is : [V-(NP)- (Prt)- NP – Y ]43 3.8.2.7. Now, these definations need to be explained as to how in terms of tree configuration, one can identify grammatical relations automatically. Notice that Chemsky (1965) had the same aim, and the failure of his earlier attempts could not make him give up this theoritical position. Rather, he and Jackendeff

Page 133: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

132

thought that the real fault was with the base rules that generated wrong tree configurations to define those notions on. But we will soon see that the oversimplified picture of I-convention presented earlier needs to be made more complex and the simple-lacking base rules have to be made more rigorous to maintain the definations of grammatical relation as in (96)-(101). Jackendeff (1974) accomplishes this task by trying to modify the tree configurations of simple finite clauses and derived nominals (first to explain 98) as given in Chemsky (1970. (102) a. Finite Clause (Chemsky’s expansion):

Page 134: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

133

(103) a. Derived Nominals (Chemsky’s Expansion):

b. Derived Nominals (Jackendeff’s expansion ):

3.8.2.6. We would now enumerate the reasons for such modifications and mention as to what Jackendeff thinks he gains by such modifications. Notice first of all that according to (98) an element is “Subject” in terms of grammatical relation only if it has either of the following configurations: (104) a. (104) follows from the fact that both N and V are [+Subj]44

Page 135: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

134

feature-wise, and a double-bar on this feature is only an abbreviation for N and V with double-bar. New, if (104) is THE configuration for subject relation, Chemsky’s (1970) expansions fail to capture the fact that John is subject in both (102)a and (103)a. This is because the configurations there are as follows (if simplified), and they do not match with the ones in (104) (105) a. 3.8.2.9. One can easily see the merit in Jackendeff’s expansion now, His expansion

in (103)b matches exactly with one of the subject-relation configuration, i.e. with (104a His expansion in (102)b had originally a configuration such as follows:

(106) But anything that ‘dominates’ V with a single bar is to be denoted as (or equated with) a V with double bar, and hence (106) could alternatively be written as (106’) (106’)

Page 136: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

135

This now matches with the other possible subject-relation configuration (i.e., with 104b) exactly. These show that if “subject-of” is to be defined as in (98). And the lexicalists see no other option, then Jackendeff’s (1974) expansions in (102)b and (13)b are correct. 3.8.2.10. Jackendeff (1974) provides however some ‘ syntactic arguments’ too as to why ‘b’-configurations are preferable to those in ‘a’ in (102)-(103). In brief, they are as follows. First, he suggests that one should parse several of John’s proofs of the theorem in such a way that several and the rest of this sentence become NP and PP, respectively45. One could get rid of the Specv –mode from (103)a this way. Secondly, the affix ‘a is added to any NP in the subject nominal position; so why not introduce ‘a buy a late syntactic rule? The advantage with this is that we can get rid of the undesirable Pess-mode from (103)a. Thirdly, in (102)a, the mode V would be interpreted as a Predicate phrase. But instead of an expansion such as in (017)a, if we use the one in (107)b, we can get rid of the Predicate phrase interpretation of V, because the latter symbol (i.e., Predicate Phrase) is no longer needed in (102)a: (107) a. S ? NP ProdP b. S? NP Aux V Under this analysis, V could now be interpreted as the same as S in (103)b, because anything two nodes above V would be liable to be described as V, as noted earlier. Now one should also eliminate the Specv-mode from (102)a, because otherwise if one

Page 137: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

136

adopts (107)b, as against (107)a (which was used in 102a), one has to have the following tree diagram instead of one in (102)b: (108) Since there can be no justification for combining on N with Aux ( as for example, it would be equally absurd to combine a V with Dot or Art), one must exclude Specv – node from (106). This way, one can modify (102)a and reach (102)b. Now, finally, if one does not consider Specv as a mode, there is no reason why Specn should be considered a mode either. By excluding it, we are able to reach (103)b from (103)a now. 3.8.2.11. In the light of the changes suggested by Jackendeff (1974), the base rules

also needed to be rewritten in the following way46: (109) a. V ? N – Aux – V

b. N ? {Nart } – N c.Aux ? T – (N)

d. T ? (have-en) – (be-ing) – (Advp)- V- (NP) – (Prt)- -({NP AP} ) – ({PP Advp}) – (s)

e. N ? (AP) –N – (PP) – (s) f. I ? (Adc) A – (PP) – ({PP/S }) g. Adv ? (Adv) – Adv – (PP)

Page 138: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

h. P ? (Adv) – P – { (NP / (S) – (PP) }

i. Prt ? (Adv) – Prt I shall not go into the details of the definition of other relations (given in 100-101) for two reasons :First, if we could argue against (90), other definations will fall through; and, secondly, justification of (100)-(101) involves arguments of the same type as have been noted in the earlier paragraphs. 3.9. A Critics of Lexicalist Approach to the definition of ‘subject ef’ 3.9.1 Recall that X-convention originated from an interesting discovery by Chemsky (1970) concerning the systematic differences of derived and gerundive nominals. Therefore, one can start from there and ask questions such as these: Do these constructions differ in the same manner universally T is it the case that all languages have derived and gerundive nominals? And finally, is or is not there a language where gerunds show more noun-like properties than Chemsky (1970) or Jackendeff (1974) thought they would. I can not answer the second question because to do that one needs to undertake a survey of a representative number of languages of the world. But I shall answer the first and the third question with examples from Maithili and Bengali. The idea is this. If one could show that there exist languages where (94) is violated to a great extent (if not completely), or if one could show that one has to post base rules that would be different from these in (109) not only in terms of word order, but also in terms of what can dominate what, one could probably

Page 139: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

138

conclude that I-convention as developed by Chemsky and other follow lexicalists was only at best a blunt device with a lot of leakage that could be used to describe only a limited number of languages of the world, and also that a defilation of relational terms with the help of such a device was uninteresting, if not incorrect. 3.9.2.1. Maithili data which are crucial for invalidation of (94) as a possible account of systematic difference between derived and gerundive nominals would include the following verbal constructions. (110) VERB GERUNDIVE DERIVED GLOSS NOMINAL NOMINAL sunjh- sunjhab sunjh1 come to mind hapak- hap2kab hap2kii swalllow (eat indecently) cal- calab caal1 move haar- haarab haar1 defeat jan- jaayab gaman go gan- gaayab giit sing hakaar- hakaarab hakaar invite/call baaj- baajab vaktavya say 3.9.2.2. Chapter VIII of this dissertation show (as it has also been pointed out by some linguistics47 studying Hindi aspectual system) that the second member of a compound verb (which we will call the vector) marks perfective aspect in Maithili, and is therefore rightly called ‘aspectual’ verb by some48. These aspectual (or vector) verbs include verbs

Page 140: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

139

like isa “go”, aa “come”, with “rise”, doR “fall”, do “give”, lo “take”, rakh “keep”, and a few more all of which lose their original meaning when used to compound verbs and behave as semantically empty aspectual markers. An example would make it very much evident (110) a. rameS thaara par raakhala modhur kheel “think” Ramesh big-plate on keep-p, ptopl sweat eat-put-Agr “Ramesh ate sweats kept on the plate”. b. rameS thaaRa par raakhala madhur khee galasha 49 eat-abs go-put-Agr “Ramesh ate up sweat kept on the plate”. The verb khee “eat” is in non-perfective in (111)a, whereas the aspectual verb isi “go” added to khee “eat” makes it perfective in (111)b. Now, if we take derived and gerundive nominals in Maithili, we should expect that if (94) was more-er-less a language independent generalization, the derived nominals could not combine with aspectual verbs, whereas the gerundive nominals could take aspectual verbs. But the following examples would show that the data from Maithili shows exactly the opposite trend: (112)a. pirakiyaa nuu2aa aB o kee hap2kal2thinka he(Non) sweetmeat mouth in take-ing swallow-per-Agr “No, having taken the sweetmeat in his mout, swallowed it”. B pirakiyaa num2aa nB (0 kee hapak1 lelthinka swallow -abc take-put-Agr “No, having taken the sweatneat in his mouth gobbled it in”.

Page 141: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

140

(113) a. kunst pirakiyaa nmu2ha nP to kee kap2kab---- he(non)-poss Swallow -ing “His swallowing of the sweetmeat having taken it in his mouth . . .” b. “kunak pirakiyan num2ha nB to dee kapak1 lab- ---- (112)a is in non-perfective (or is unmarked with respect to perfective) and (112)b is in perfective aspect. One would naturally expect that both of these could be gerundivised. But (113)b shows that one cannot get a gerundive nominal from a “Verb plus aspectual verb” construction in Maithili. What makes the situation complicated is that the derived nominal CAN occur in some such contructions. Consider the following sentence which has an identical environment such as in (113)b: (114) kunak pirrkiyaa nun2ha nP to kee kap2kii lab - - - - At times, the verbs that combine with the derived nominals to give an aspectual reading different from the ones used with their gerundive counterparts30. Here is one such case which is also remarkable in that here both derived and gerundive nominals can occur with aspectual verbs: (115) a. juaa nP baRkaa kan-k haar maanab ----- (Derived ) dice is elder uncle-poss defear concode – ger “The defeat of the elder uncle in dice----“ b. juaa nP balkan kaa-k haar1 jacob---- (gerundive) lose-she ge-ger “Elder uncle’s losing in dice----“ It may be mentioned here that there are indeed some aspectual

Page 142: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

141

verbs such as ask “completeve” and ask “abilitative” with which only gerundives can occur. I do not want to give an impression that (94)a is not true even partially. Rather the purpose of giving these examples was to show that the situation is much more complex in Maithili, and it could be still more complecated than what proponents of (94)a would like us to believe. 3.9.2.3 Turning to (94)b, we find the following contexts with partitive elements kichu (kick)u “some and ask “many” where gerundives give grammatical results: (116) a. kicju hear chon heit2 ach1 jo jiitahu on some defeat such be-pte-aux-pres

that win-Nuph from bakh1 kae cross-ptopl EO “Some defeats are such that they are better than victories”. b. kicku haarab chon heita ach1 jo jiitahu on barh1 kaa

(117) a. kunak meek vektayya han2raa poslin nah1 he(non)-poss many remark/say me-to like not “I do not like many of his remarks.” b. kunak anak besiah51 har2raa posin nah1

But I am aware that there are contexts which permit only derived nominals to occur, Here is one example: (118) a. kunak kichu (kichu) glit amar bhO gelash1

he(Non) –poss some songs eternal be-ptopl go-pst-Arr “Some of his songs became eternalb. b. * kunak kicku (kicku) gaavab amar bhO golash1

Page 143: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

142

3.9.2.4. Similarly, there are examples where both gerundives and derived nominals could occur with adverbials in partial violationof (94)d as in (119). And, one also comes across instances where if one wants to use an adverb-plus-gerundive construction bringing it on par with an adjective-plus-derived-nominal construction, the result is disastrous. (120) would thus clearly violate (94)d: (119) a. mahesraak bandar jakaan caal1---- Maheswar-poss sankey manner movement “Maheswar’s monkey-like movement----“ b. mahesraak bandar jakaan calab ------- (120) a. chen samay mE hunak niik suu jj1 – ye-Taa par

such time at he (Hon)-poss good suggestion –Eaph-on-class bharos chal hope be-pst-Agr “At such a time (of distress), one could depend only on his good suggestion”. b. *chen samay par hunak niik jakaan suujhab---- suujhaavab socab good manner strike/suggest/think---- “At such a time (of distress), one could depend on his coming to mind/suggesting/thinking nicely”.

3.9.2.5. Both noun-classifier-test and positional occurrence of subjects show that gerundives and derived nominals are not

Page 144: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

143

distinguished on this count. Use of a determiner Tea (found also in Bengali) is parallel to a dot the is English. The following sentences would show that (94) was not universally attested. (121) a. mahesraak bandar jakaa caal1-tea houraa

maheswar –poss bombay manner movement-dot me-to posiin nah1 chala like nto be-pot-Agr “I did not like Maheswar’s movement like a monkey”. b.mahesraak bandar jakkan salab-Tea----

(122) a. gaman kii ettak2 sahaj chalaik2 ?

go-ing what so easy be-per-Agr “was departure so easy?” b. inavab kii etteka sahaj chalaika?

What is interesting here is that in violation of (94)c, derived nominal place in the subject position gives a worse sentence as in (122)a compared with (122)b which has gerundive isavab “going” in the subject nominal position. 3.9.2.6. One could go on giving more and more examples from Maithili (and probably from other New Indo-Aryan languages too) to show that (94) is not a universal grammatical trait (or is not so in the way it has been stated by Chemsky, 1970 or Jackendeff, 1974). But I would stop giving such examples with a last one where one could see that in violation of (94)f, subject-raising rule (here, subject-to-subject raising) would not treat derived and gerundive nominals differently. Consider

Page 145: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

144

the following sentences: (123) a. li miseit chala jo kaakaa haar2tasha it sure be-pst-Agr that uncle lose-fut-

Agr “It was sure that uncle would lose”. b.kaakaak {hear/haarab} miscit chala uncle-poss less/losing sure be-put-Agr “Uncle’s less/losing was sure”.

Notice that in contructions such as this, Raising was to be applied obligaterily. Otherwise, one would generate ungrammatical sequences such as follows: (123)e. *ii miseit chal a (jo) kaakaak {hear/haarab} 3.9.3.1 Now, I would briefly mention about at least one attempt in the analysis of India languages to show that gerundives were a nominal category. Probal Dasgupta (1978) in a stimulating paper has recently pointed out that in Bengali –Vas/-no forms added to verbal bases give both gerundives and participials, and that a lexicalist analysis would require that we treat gerundives as N (with a ‘gerundive phrase’ being its head NP Or T) rather than treating them as V dominated by an s-mode (which would, in Jackendeff’s notation be the same as V). An essentially transformational analysis (such as that of Jayanti Chattopadhyay, 1976), which he calls the ‘Clause hypothesis’ would require that the so-called gerundives were instances of –er. . . –an complementiser (parallel to –kaa ---- naa or infinite complementiser in Hindi as Subbarao, 1974 would name it ) placement. Both

Page 146: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

145

hypothesis could account for the following data52 with a difference that Dasgupta (1978) would relate the sentences in ‘a’ and ‘b’ (as follow) by lexical rule, and a Chattopadhyay (1976)-type analysis would pesit an Equi rule. (124) a. samaar eiThi lakha heeche?

You-poss letter writing be-pres pft-Agr “Have you (finalised ) writing the letter ?” b. Sammar lekhaa eiThi peyechi you-poss written letter got-pres pft-Agr “I have received a your-written ( written by you ) letter”.

(125) a. Osiner gooFi conlame Ektaa gh0Tmaa

Ashin-poss ear driving cue-class event “Ashim’s driving (of ) a car is an event”. b. Osimer easplasme gooFi EksiDemt kereche Ashim-poss driven ear acciden do-poss pft-Agr “Ashim’s driven car (the car driven by Ashim) not with an accident”.

3.9.3.2. In the following paragraphs I shall elaborate Dasgupta’s (1978) suggestions with much greater details. The two hypotheses would have the following structural configuration for the a-sentences as I have understood them53

Page 147: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

146

(126) a. Dasgupta – Hypothesis B. Chattopadhyay-Hypothesis To generate the b-sentences, Chattopadhyay/ Clause-Hypothesis would require that the S-mode in (126)b is embedded under another NP- mode the head noun of which would be the same as the object NP of (126)b. Thus, we have to modify (126)b in the following way:

(126)

3.9.3.3. One could now write an Equi rule that would explain as to how one would generate (124)b and (125)b from a structure such as in (126)e.

Page 148: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

147

(126) Semi-rule54, SPs [[x – [NP – VP]] –NP] NPS ProdP NP SX: X 2 3 4 ? SC: X ? 3 4 Dasgupta (1976), following the notation of Wasev (1977:333), wrote the following lexical rule that would generate the participial constructions as in (125)b from a configuration such as in (126)a: (127) Gerund/Participle Rule33: ? [[ V +Vas/as ]]A (XX) – I Here, ? is a null affix which turns some class of NPs or N- into Adjectives (or into I); and the equation on right points out that the original DO which is relationally ‘II’ (ef. Johnson, 1977a: Pullon, 1977), becomes the ‘subject of’ or SU, i..e., becaue ‘I’ when (126) applies. Thus, gooFi which was a DO in (126) a becomes SU in (129) as follows: (129)

Page 149: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

148

3.9.3.4. Dasgupta (1970) argued in favour of his proposal by showing that gerundives take definitiser/ classifier suffix –Tea56, can be inflected like a regular nominal by adding objective (-? or –ka), genitive (-(V)aar) or locative (-a or –to ) suffixes, and they can also be place before postpositions such as Sattse “despite”. 3.9.3.5. No then showed that a surface structure filter which is independently motivated would block sentences such as in (130)b automatically (whereever N in an AP-N sequence was a simple definite pro-form), whereas in the other hypothesis, one has to state it as a condition on the Equi transformation. Consider the following examples. (130) a. manaar oTea dofen ucit you (non)-poss it giving proper “your giving this is proper”. b. “aamaar devan eTea - - - “your given this is proper” 3.9.3.6. A lexicalist analysis is to be preferred to a transformational analysis because there are a number of idiosyneratic facts about this gerund/participle relation. For instance, certain compound verb and conjunct verb constructions do not have such participial counterparts in Bengali. This includes bole bRano “go saying around”, SoNgit bojhas57 “understand music”, etc. No then points out the verb-like characteristics of these gerundives, and then gives certain problem-sentences which cannot be explained by any one of these

Page 150: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

149

hypothesis. Finally, he presents some more data which could easily be analyzed by enlarging the scope of the lexical-relation rule as stated in (128). But these cannot be explained by transformations, because a T-rule formulated to capture those ‘facts’ would generate many ‘non-facts’ or unacceptable strings as well. (131) as follows can be stated by a lexical rule such as one in (134), and one would need to revise (128) as in (133) to capture the sentence-types in (132)-(133) along with (124)-(125). Consider the following sentences now. (131) a. iNreji jasmaa chole - - - English Know -ing boy “A boy who knows English----“ b. dilli thoke phire aaSan tinjun montri -- -- Delhi from turn-pte coming three ministers “Three ministers who have come back from Delhi”. Notice that there instead of DO, the SU is shifted from its original position, and retained as SU even in the gerundival phrase. The following sentences however follow (126) although they do not involve - Vas/no markers: (132) a. anmaar baaRi teiri- -- - my house building “my building (of) a house - - - -“ b. anmaar teiri baaRi “The house build by no- - --“ (Also, “My-built house”)

Page 151: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

150

(133) a. bhaareter can aandaani - - - India’s tea import “India’s importing (of) tea - - - -“ b. bhaareter aand aami eaa -- - - “Tea imported by India- - -“ (Also, “India’s imported tea”) 3.9.3.6. The lexical rule needed to explain these are as follows: (134) Gerund/Participle Rule II58 ? [N [ N –V – Vas / no] ] A (I) – I (135) Gerund/Participle Rule X ( revised from 128)59 ? N (II) –I A 3.9.4.1. There are two remarkable things about this analysis of Bengali gerundive nominals which I have been able to notice, and they are as follows: First, notice that (126)a is not an underlying configuration because here an N branches into N and V. Surely, one cannot think of a base rule where such an expansion is permitted. This must then be a change in the tree structure brought in by some lexical rule of the following kind where with an introduction of Vas/no markers a V N - - V- - structure is changed into a N N ----V ---. We can write this rule in the following way:

Page 152: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

151

(136) Gerundive Nominalisation Rule60

-Vas/no [V [N - - - -V ------] X ] X We could probably have a general condition in these rules which would automatically delete the Aux-mode wherever a verbal particle such as gerundive, infinitive or other affixes are introduced. This convention or condition would also change the level of the category V (from V to V) to one step higher. One would thus assume a configuration such as in (137) upon which (136) could apply to generate (126)a: (137) 3.9.4.2. Thus, one can see that in Bengali, one has to relate three different categories by lexical rules, and those are as different as Verb, Noun (in case of gerundive nominal) and Adjective (in case of Participle). This is therefore an instance of I-convention being used to combine61 V, N and A, as in (96)ii. Secondly, if one compares the gerundive nominal structures such as in (126)a with their derived nominal counterparts, it would turn out that (94) is completely irrelevent for Bengali. This is because a derived nominal counterpart of, say (136)a, which is

Page 153: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

152

(138)b in this case would have a configuration such as in (139): (138) a. Osim-or gaaRi gaalaa-aa- --- (Gerundive) b. Osim-or gaaRi gaalaa(aa)- --- (derived) The tree configuration for (138)b would be as follows: (139) If we compare (139) with (126)a, we would find out that the difference between gerundives and derived nominals in Bengali was very little. Sentence with derived nominals, however, do not have homophonous participial counterparts as (140) would show (140) a. * OSimer oaalon (aa) gaaRi - - - b. Osimer daaran caalite gaaRi - - - Ashim-pees by drive-p.pass-ear ptopl “A car driven by Ashim - - - : 3.9.4.3. One would, however, expect that there could be other languages tee where this differences between gerundive and derived nominals could be marginally reduced in the same way. Maithili turns out to be another language where the gerundive/participle homoymy has spread. This is due to the fact that it has two

Page 154: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

153

forms for gerundive/infinitive, and they are –ab and –ala. The latter happens to be a post-passive participle marker tee. One can thus have the following parallels: (141) a. raajaasashebak cusa daliddar jakaa2 gaaRii King sir –pass such poor manner/like cart Haan kag kana raa eon schola nah1 gola 62 Drive-ing no-te by stand-ing not go-pass-put-Agr “I could not stand the King’s driving a cart like a poor man” b. raajaasashebak onaa daliddar jakaan haankala gaaRi - - - haankab But in Maithili, the idiosynerarics are more (compared to Bengali), and I would not go into the details of those because that would not contribute much to the topic being discussed here. My main purpose of giving those examples from Maithili was to show that there could be many more languages like Bengali where Chemsky’s (1970) treatment of nominalisations of different types has to be modified even if one followed the T-convention. 3.9.5.1. There could be many more ways to argue against the T-convention in general, and the defination of ‘Subject of’ as in (96) in particular. For example, Jackendeff (1974:13-4) himself admits that there are languages where his characterisation (he calls it ‘defination’) of N as a feature complex (142) does not hold good: (142) N - + Subj - Obj + Comp

Page 155: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

154

No says that “in French, for example, nouns to not satisfy my defination of [ + subj] since there is no genitive NP possible at the beginning of an NP. Nevertheless, one would hope that Frence nouns and English nouns have the same syntactic features, and that only the way these features appear in grammar differs from one language to another” (Jackendeff, 1974:14). If we forget what Jackendeff “hopes”, for the time being, it becomes evident that this statement shows that the ‘definations’ of syntactic categories in (97) are language-particular. Those are therefore at best of typological value because there could be other languages where (97) would be deficient in other ways. For instance, most of the SOV languages are postpositional, as Greenberg’s (1965:79) ‘Universal 4’ would predict. (97) is faulty because P of Proposition is not a category in these languages. Even if we generalise (97) by redifining P as either proposition or postposition, there could be languages where none of these is an essential category. If P is not a universal category (97) falls through. Notice that the definitions of grammatical relations in (98) – (101) depend on (97), and if the latter is shown to be a language-particular phenomenon, the former cannot be taken as universal definations, and hence are also liable to be rejected. 3.9.5.2 I have, however, a more basic objection against the definations of grammatical relations as in (98)-(101). Notice that all the “Universal” syntactic categories are defined with features such as [ ? Subj] . [? Obj ], etc. ass in (97). Then,

Page 156: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

155

these syntactic categories are used say in (98) to define relational terms such as ‘Subject of’. Thus, the lexicalists seem to be comfortable of having been also to define both categories and relations in a language-independent way. But it does not take much imagination to find out the definitional parodox in this approach. You define X1 and X2 by Y1 and Y2 (by assigning some binary value to the letter), and then define Y1 and Y2 by using X1 and X2 . Such definations are circular and hence are incorrect. Moreover, in defining ‘Subject of’ Jackendeff (1974) also uses [ + Subj] like a cover symbol of abstract nature that collapses categories N and V in (96). The expansions of (98) as in (104) show this. 3.9.5.3. The last point seems to very important to me, because if someone tries to define ‘Subject of’ by employing a feature such as ? Subj, then the only conclusions that one could draw from this are that either the person using such a stratergy is trying to fool everybody by imposing a false universal defination on syntactic theory, or he is using the feature ? Subj in an arbitrary manner. Jackendeff (1974:13) seems to be well aware of these conclusions when he says: “To avoid confusion, it is worth noting that the names of the features are chosen solely for houristic purposes: nothing theoritical is intended by the names, and they could just as well be named F1, F2 and F3 “. The question that can be raised here is why use abstract arbitrary features for defining grammatical category anyway? The problem is that these 'Syntactic distinctive features’ were thought to be parallel to “Phonological distinctive features(which have

Page 157: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

156

phonetic -acoustic correlates, and which seem to be doing fairly well in Phonology),and hence they could not even be taken as ‘arbitrary’ objects. Jackendeff (1974:12-3 ) therefore NAD to define them, and for that he used SUBJBOT OF and DIRBOT OBJBOT as given syntactic-semantic notions (which unfortunately for him had ‘almost’ one-to-one correspondence with the relational notions ;Subject of’ and ‘Direct object of’) and yet he claims that there could as well be called F1 , F2 , etc. 3.9.6.1. I shall now show that Y-convention is supported to Chemsky’s (1965) earlier conception of grammar in that it overcomes some of the difficulties faces by an “Aspects”-type defination of grammatical relations. Y-convention can possibly overcome the problems which Chemsly (1965) faced with verb initial languages even if the defination in (98) is unchanged. Thus, in languages such as Barber, Nebrew, Neeri, Naya, Welsh, and Enpetee all of which have been reported (of. Greenberg, 1963a) to follow the VSO word order would probably have the following base rules in terms of Y-convention63 , (145) a. V ? (Aux) – V- N b. N ? N – N We can now identify the subject NP in the tree configuration by putting a circle around it in (144): (143) VSO-languages

Page 158: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

157

3.9.6.2. Notice that the circled mode is identified as ‘Subject of’ by (104)a – interpretation of the defination of SU in (98). However, (104)ab would identify the boxed N-mode as SU tee. May be, one can posit the following reading conventions: (144) Interpretation of the Defination of SU (as in 98) Start looking for the SU from the bottom to top of tree configuration; when you are able to identify either of the structure of (104), as abbreviated in (98), at a lower level, ignore if such a configuration is found at a level higher than that. Following (145), one can rightly identify the circles N, and eliminate the boxed N for the purpose of identification of SU. (145) would do not only be relevant for VSO languages, but it would work in OSV-languages tee, which also were problem-cases for Chemsky (1965). Probable examples of OSV language would include West Greenlandic Eskine (Rischel, 1970), Alout (Ross, 1970a), Apurina (Pickering, 1974: Berbyshire & Pullon, 1978), Dyirbal (Johnson, 1974a), Nurrian (Keeman, 1976), Iavante (Durgess, 1976), and Nadob (Weir, as quoted in Derbyshire & Pullon, 1978). The base rule for these could be written in the following way: (145) a. V ? N – V –(Aux)

b. N ? N – Y/N

Here again, (143) would help identify the circled mode as bearing the grammatical relation ‘Subject of’, as (147) would illustrate:

Page 159: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

158

(146) OSV-language: 3.9.7.0. In the paragraphs we have seen that although Chemsky (1970) and Jackendeff’s (1974) attempts to save the traditional TG characterisation of ‘Subject of’ in terms of ‘Dominance” relationship have some good points in favour, nevertheless, they are deficient in more than one way, and are therefore to be rejected. It may be mentioned at this point that some lexicalists like Wasev (1977:333) borrow the notations of Relational Grammrians such as Perlmutter & Postal (1974) – the notations like I, II, and III for SU, DO, and IO, respectively and use them in stating lexical (redundanay) rules. Wasev in particular admits the importance of RG in the following words: “….. I would like to touch upon ----- the relationship between the ideas put forward here (i.e., in Wasev, 1977: UNS) and the theory of ‘relational grammar’ currently being developed by Postal, Perlmutter, Johnson, and others. It is clear that they have mush in common: both ascribe a much greater importance to notions like “subject” and “object” than the standard theory did: both reformulate in relational terms many rules that had earlier been formulated in structural terms; and both claim that all relational rules precede all structural rules” (Wasev, 1977:333).

Page 160: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

159

3.10. Conclusion 3.10.1 In the preceding paragraphs we have seen that one can come up with a number of linguistically significant generalisations by positing grammatical relations such as ‘Subjecf of’, ‘Direct Object of’ ‘ Indirect object of’ , etc. at both underlying and superficial level, and can reformulate a number of rules (especially these that change grammatical relations) by giving them universal characterisation. At the same time, we have also seen that by taking only grammatical relations as basic tools for grammatical description relations as basic tools for grammatical description, one cannot exhause writing rules of grammar, because there are a number of (both universal and language-particular) rules which may or may not refer to grammatical relations, and which must be written in terms of grammatical categories. It was therefore found that any adequate theory of grammar must accept both grammatical relations and grammatical categories as basic units for the description of various aspects of clause structure at different levels of derivation. It was observed that although one can identify a large number of properties which are associated with subject relations of different languages of the world(- properties which may vary from language to language) it was not possible to define grammatical relations such as ‘Subject of’ Instead, one should take those relations as undefined and undefinable primitive units of grammatical description. Lexical rules were found to be more convenient than transformations in dealing with a number of problems, but it was evident

Page 161: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

160

that they could not take over all that transformations (which includes both relation-changing and non-relation-changing rules) could do. Also, lexical rules must be allowed to intersperse grammatical transformations, i.e., the former would not in all cases apply on bloc before the application of the latter rules. Among the transformations, probably those altering grammatical relations apply before others. Moreover, those transformations which change the word order or linearise the terminal elements would probably occur at the end. It was also mentioned that all types of transformations, i.e., relation-chaning (or, transrelational), non-relation-changing, and linearinsing transformations, give rise to various semantic/ pragmatic interpretation64 in discourse or oven otherwise. And, in a number of instances, these facts of the outer world so to say (or that of performance) come into play in the formulation, arrangement as well as in the selection of particular (optional) transformations. Therefore, syntactic rules would have to be characterised in a grammatical theory in such a way that they both food semantic/pragmatic interpretation rules, and be fed by the semantic/pragmatic facts. Since by now it has been established that all the logically possible order of meaningful elements are in fact attestable in human language of the world, and since there is nothing like the ‘basic word order’ of a language, one could now probably extend the notion of non-disereteness in describing the ‘favoured’ or ‘unmarked’ read order of a particular language. That is, one may now compare languages typologically to find out for instance ‘how much SVO is English (or Fulani or Quarani or Nerwogian or Thai or Yeruba

Page 162: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

161

with respect to each other)?’ All these and the assumption that grammatical relations are not to be defined suggest that one may as well do away with PS-rules (or base rules), and leave the task of predicting surface order to linearisation and late readjustment rules. That is turn implies that deep structure is a superfluous level, and that one could instead start from a deeper where only unordered grammatical relations existed. At that point, one could start by applying the Universal Relation-Changing Rules first, to be followed by language-particular rules. Such a theory of grammar would also have a number of universal (or local, and even language-specific) constraints on transformations which would serve the purpose of filtering out unwanted configurations or structures. These universal constraints are ideally model-independent and refer to universal notions like ‘grammatical relations’, ‘grammatical categories’, ‘grammatical processes’ and ‘grammatically possible word order types’ . For this purpose, one could even think of altering the existant language ‘universals’ which may be dependent on grammatical models to a varying degree and translate them into a model-independent ‘language’ if possible and if need be. It may also be pointed out that a universal may at some point of time turn out to have counter-examples, but that would not make it a non-universal, because any significant linguistic generalisation may be a universal ‘to a degree’. Similarly, a language-particular condition may block a fairly general (or universal ) transformations from applying on a particular structure. Again, this would not warrant us to declare

Page 163: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

162

such a rule to be a non-universal. Rules may also be applicable (even universal rules) ‘to a degree’. If this happens in case of a language-particular rules, lexical rules may be called for to explain or state idiosynerasies. If now we turn towards the notion of grammatical relations, and think of the attempts of some (Keenan, 1976 for example) for a cross-linguistic comparison of these notions, we can easily find out that such comparisons belonged to a level other than underlying relational level (or network). And quite often it was found out that grammatical relations. (belonging to various levels of derivation ) across different languages did not show the same characteristics. One could therefore talk about non-discreteness of (derived) grammatical relations also. That is, we could now say that in a language n, a grammatical element ORa of a language v,a dn so on and so forth. It is in this kind of comparison that one could probably make use of Keenam’s (1976) ‘Subject Properties List’ (SPL), and his ‘Promotion-to-subject Hierarchy ‘ (PSN) in a meaningful way. 3.10..2. Given the characterization of an approach to grammatical theory as above, I feel that I have to bear the burion of proof as to why in the title of this description of Maithili Syntax, the words, “A transrormational –generative approach” were used. Obviously, as the sketch of a theory of grammar given above may have indicated, I take the words “transformational-generative” more as a research strategy than as a grammatical model for ‘Language” in general (and for languages in particular). I take

Page 164: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

163

generativity of language as not only a mechanical property of recursion used in a particular framework to write language descriptions (of. Chemsky, while discussing language with Parrot, 1974:88). But also as a human faculty of creating structure (which may be trivially or interestingly infinite in number of items, structure-types or in innivativences). A “generative” approach towards language description does not give priority only to “gramaticality judgements, as opposed to judgements about what a ssentence COULD mean and under what circumstances it would be APPROPRIATE (Macawlay, while discussing language with parrot, 1974:251: Emphasis: UNS). I may mention here that I do not quite see as to why a “generative” description has to be “complete and precise” (Lukeff, while discussing language with parrot, 1974:174). The completeness- requirement is understandable, but procision or space saving need not always lead an analyst to correct results.Thus, it has long been recognised that “…although one wants the ‘simplest’ linguistic theory…, in some sense that cannot be specified in a useful way. elaborations and complications of linguistic theory are all to the good insafar as they narrow the choice of grammers and the range of admissible languages (i.e., sets of derivations )” (Chomsky, 1972b :69 ). The name “transformational” is very much applicable for the scheme of things that I sketched earlier, although different kinds of transformations (relation-changing, non-relation-changing, and linearisational) may have varying notations. However, as long as a conception of grammar “involves the motion of derivation, consisting of some kind of underlying

Page 165: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

164

structure and steps leading from it to a surface structure,… and involves a system of rules that specify how the different stages of derivation are related to each other” (Macawley, while discussing language with parrot, 1974:230),it is ‘transformational”, and in this sense my approach can surely be called so. 3.10.3. Lastly, the words ‘a’ and ‘approach’ only indicate that the way I characterised ‘Grammar’ is merely one of the ways, and that such a characterisation is not a full-fledged theory of grammar, but is, as I said in the beginning, a research strategy adopted by a language analyst (who in modern times is faced with a flood of competing theories of language all of them being at best incomplete models of grammatical description) who does not want to close all the options open to him by selecting any one particular formal approach to linguistic description, because he does not want to lose sight of linguistic facts which do not fit in any one particular model. Finally, I am of the opinion that what George Lakoff has to say in this regard is of supreme importance for any descriptivist, and I quote him ; (148) “I think the time has come for a return to the tradition of informal descriptions of exotic languages, written whenever possible in clear press rather than in formal rules, so that such descriptions will still be useful and informative when present theories are long forgotten” (lakoff, while discussing language with Parrot, 1974 : 133).

Page 166: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

165

Notes to chapter III

1. See Singh (1979 a) for an account of the developments in the theory of syntax that have taken place in recent years. 2. That “providence”, “dominance” and “command”, etc. were the basic relations in the transformational-generative grammars of different brands was admitted by Lakoff (1973 :273). 3. Notice that (6) and (7) are structurally the same, but are still kept distinct in the text because in (6), the NP in b is the subject, whereas in (7), the subject is the NP in a. That is why, their tree representations also look alike : compare (10) and (11). 4. I am not saying that (14) is the diagram suggested by all the three scholars- McGawlay (1970a), Fillners (1968) and Lakoff (1970 a). But of one takes their positions regarding the phrase structure rule of human languages seriously, one would naturally come up with tree configurations such as in (14). 5. The definitions of relational motions given in (16) have not been taken from any of the works mentioned in fn.4. These automatically follow from the tree configuration in (14). 6. Notice that if we take SOV or SVO as TNB underlying representation for all the language in the world, we have to rewrite the defintion of relational motions given in (16). For instance, in that case, we would define SU and DO in the following manners :

A. SOV-hypothesis : i. NP/ NP, V =SV ii NP/ NP ___V =DO B. SVO- hypothesis; i. NP/ V, NP =SV ii. NP/ NP V __ =DO, etc. 7. Since the dominance relation has been excluded in all these tree diagrams, the structure for VBO and VOE languages are again identical. See fn. 3 for the reasons of such an identity.

Page 167: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

166

8. See susan Sehmerling (1975) for an extreme position regarding the existence of level of surface structure. While criticising Chemsky’s (1965:16) characterization of surface structure as a level that is neccesary for a phonetic interpretation of a sentence, she said; “------- little if any serious discussion has been devoted to the question of whether a definable level of surface structure exists, despite the fact that a number of linguists have noted various problems which bear on the question of the viability of the sort of characterisation quoted above” ( meaning Chensky’s characterisation mentioned earlier; Schemerling. (1973:377). 9. In (24), we have marked the ‘moved’ NP by a subscript I,a dn the ‘affected’ NP-position by a subscript t. 10. See Keenam and Courie (1972) for a detailed discussion of this hierarchy. Also see Johnson (1977) for a modified version of the same, as noted in (71) in this chapter. 11. Courie (1974:8) calls it a ‘paradigm case’ 12. See McGawley’s (1968b) review of Chemsky (1966b), especially McGawley (1968b, repr. 1976:269), for a discussion on grammatical and semantic relation. 13. This is not to deny that this process DONE involve language-particular facts, but for someone interested in UG (Universal grammar), these facts are at best of marginal interest. 14. See Lakeff (1 973:278) where he says: “- -- -I also tentatively follow the suggestion of Postal and Perlmutter that all cyclic rules involve change of grammatical relations”. Johnson (1974a:V-80) proposed a hypothesis regarding cyclicity and relation changing rules in which he changed his earlier hypothesis that all cyclic rules alter grammatical relations, and according to what he calls a ‘revised cycle hypothesis’ cyclic rules are essentially relation-creating rule such as Passive, Tough movement, A-raising, B-raising, and Predicate-raising. 15. See section 3.6 of this chapter for a detailed discussion on an attempt within the framework of Relational Grammar to define the relational notions and the criticism thereof. 16. See Johnson (1974a), and (42) in this chapter which include a discussion of rules that convert grammatical relations of a given relational network into (grammatical) categorial labels

Page 168: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

167

17. These conditions smell of ‘derivational constraints’ of the generative semanticists. 18. See Johnson (1977b:135) for three types of promotion rules ; ‘ insertion’ (e.g. there-insertion), ‘ascention’ (e.g., subject-to-object raising,quantifier-float), and ‘advancement’ (e.g., dative-movement). 19. Trithart (1975:615) proposed a Relational Hierarchy which he characterised in the following way (quoted from Johnson, 1974a): Relational Hierarchy Subj > Do> IO> Non-term (43) appears in Johnson (1977b:123), where the sign > most ‘to have procedence over or ‘to outrank’. While the AN first appeared in Keenam & Courie (1972), AN was proposed by Perlmutter & Postal (1974) for the first time. See Johnson (1977 b: 158) for a discussion on the two different markers used by then-< in the AN, and > in the NH. 20. Notice that doubling caused by an absence of a particular relation, and the one brought about by an introduction of a causal agent are quite different things. The latter is an instance of a “derived” doubling which violates the Relational Annihilation Principle mainly, while the former nullifies the putative universals such as AN or NH. 21. See chapter V for a very long treatment of this topic. 22. There have been some more serious attempts to define relational motions, especially within the framework of categorial or X-bar grammar (of. Jackendeff, 1974) and also within the lexicalist tradition (of. Wasow, 1978). For these and some other position (such as that of male.1976), see section 3.8. to of this chapter. 23. See chapter V where it is argued that it is the generalization concerning verb agreement arrived at by Keenam, and as given in (75)f, that is truly of universal nature, and not the verb agreement law which was suggested by perlmutter & postal (1974). See Singh (1978) for details. 24. It will clearly to demonstrated in Chapter VII that semantic and pragmatic facts must be taken up while writing a syntactic description of a language, and that if one believes in the intermediate grammatical stages, semantic /pragmatic facts would surely be found to be associated with such stages.

Page 169: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

168

25. For instance, Johnson (1977 a) here takes it for granted that “Intermediate grammatical stages” existed, as would be done by the traditional TC grammar. 26. This paper was written in and around 1967, and was reprinted in Macawley (1976b). See Macawley’s editorial comments for a brief history of the origin of this work. 27. See Usha Verma (1979) for an account of this problem in Angami. 28. See Johnson (1974a: 110-6) in which he shows with the help of a quote from Fillners (1968:25) (“… none of these cases can be interpreted as matched by the surface-structure relations, subject and object, in any particualar language”.)that the latter holds that (I) the motions such as NV and DO, etc. do not have an invariant interpretation, while case relations have these, (ii) transformations are sensitive to case relations, (iii) in many instances, ‘subject’ may bear ‘agent’ or ‘patient’ case relationship, which shows that grammatical relations are peripheral to syntax, etc. Johnson’s answer to these is that the first point is not a weakness of the relational motions, but is rather a strength because of the relational motions, but is rather a strength because there are a number of universal syntactic rules such as passive where this character of grammatical relations make them more apt as points of reference while formulating these rules. Incidentally, we have already seen that a universal passive rule written in terms of case relations is faulty in many ways. Secondly, correct formulation of a number of rules demand that one assumes grammatical relations to be existing at different levels of grammar, and not only on surface structure level. Thirdly, by proposing that grammatical relations exists at both underlying and superficial levels, one does not deny that languages quite often place semantic conditions on syntactic rules. It may be interesting to note that by 1977. Fillners has changed his ideas (as presented above) completely. No now says, and I quote : “… this perspectivising corresponds… to determining the structuring of a clause in terms of the nuclear grammatical relations, a consequence of this proposal ( the proposal comes in the form of a slogan in Fillners (1977:59) that NRANINGS ARE RELATIVISED TO SCHEMES; UN;) is that it becomes necessary to recognise a level of grammatical relations, something I once thought it important to reject” (Fillners,1977;79-80).

29. See Brane (1976:xviii-xix) for a brief history of this development. 30. See Seyar (1974), and sections 3.2.1.5 to 3.2.1.7. of this chapter for arguments against the reductivion in respect of word order typologies.

Page 170: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

169

31. Lakoff for example, informs Parrot (1974:151-2) that “I would not presume to tell anyone how to do linguistics; I don’t think there is any one way… It should be borne in mind that in adopting generative semantics as a research strategy, we are not attempting to provide a complete theory of language’. See Parrot’s discussion with Macawley (Parrot, 1974:249-77) to find out the differences of opinion that exist within the generative semantics tradition. Also, none of them seen to be happy with the name ‘Generative Semantics’ for the kinds of things they have been doing, and because it (i.e., the name) evokes some undesirable connotation and thereby misleads many. 32. Note that the second point could be proven if (I) either we can show that they never took the relational motions seriously, (ii) or that they were forced to adopt a view points similar to that expressed by relatioanl grammarians only very recently. I shall try to argue in both the ways in the coming paragraphs. 33. These quotes are from Macawley’s (1976b) editorial comments on the paper. 34. Consider the following statement, for instance: “Schematically SV strikes 10 as being like O and SV reminds IO of O would have to contain a representation like: (2) IO perceive (SV similar O)’(Lakoff, 1971:271). 35. For example, Lakoff (1971:271) draws the following tree structures:

“… Subject-raising would produce (4) :”

Page 171: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

170

36. Toward the end of this chapter, I shall argue that one should take a stand on fussiness while dealing with grammatical relations and I shall point out that an assumption such as in (92) a has many other vantage points too. 37. To quote Lakoff (1973b:277), “correspondence rules are the direct descendants of so-called ‘local rules’(see Lakoff, 1969), which turned out not to be local most of the time. Such rules are well-formedness conditions on adjacent pairs of trees in a derivation”. 38. About this differences in terms of aspects, Jackendeff (1974:5) says the following : “Gerundives , being sentences in deep structure, are expected to contain aspect: since –ing substitutes for Tense and Nodal, their absence is accounted for”. 39. We shall see that these base rules (?) will have to be replaced very soon by different types of rules (for instance, see 109). 40. For a different use of COMP, see Dasgupta (In preparation) who adopts it from Brosman(1970, 1976). In Chapter IV of this dissertation, I use COMP In this sense, and not in the way it has been used by Jackendoff (1974) as shown in (97). 41. This definition of ‘subject of ‘ is actually an attempt to collapse two possible configurations which are given in (104). 42. In the subsequent paragraphs where I shall present arguments agains X-convention and in particular against (96), the definition of object relation as given in (100) will not be dealt with in detail. It is interesting to note that the lexicalists never take non-terms or oblique objects of various kinds into consideration. Moreover, they did not give the relation IO the attention that it deserved.

Page 172: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

171.

43. The relation ‘Predicate of’ seems t o me to be superfluous. 44. Taking genitive nouns was considered to be the reason why N should have this feature. 45. That is, one could do so instead of parsing this sentence into several of John’s and the rest as was done in (0143)a. 46. I shall not go into the details of Jackendeff’s (1974) defence of those base rules. I am taking it for granted that these changes are necessary if one is to use I-convention in a meaningful way. 47. See Neck (1974, 1975, 1977, 1978ab), Arora (1978) Baranikav & Baranikev (1967), Bohl (1964, 1967), Gaeffke (1967), Naekar (1958), Nasica (1976), Periska (1967-9, 1977), Shapire (1974) and others. For a discussion of Bengali Compound Verbs, see Dasgupta (1977) and Singh (1979). 48. Annamalai A(1977), Bhat (1977) Subbrao 91977) use this term. There are many others who prefer it to other names. 49. For the reasons as to why the verbs in (111)a and (111)b should have two different markers: thinka and aska, respectively, see Chapter V of this dissertation. 50. That may be due to the ‘rules’ that govern Conjunct Verb formation where a Noun and a Verb is combined in such a way that there is greater echerance between these elements than that is expected by simply juxtapposing an N and a V in languages other than Indo-Aryan. 51. There is obviously a meaning difference involved when we compare the ‘a’ and ‘b’ sentences in (117), because in course of derivation Yaktavya and the verb from it which it originated, basi-because different semantically to certain extent. Such differences are parallel to, say, to act, action and activity in English. 52. Note that I have made small changes in the data (esp. sentences 124-5) given by Dasgupta (1976). 53. The tree configurations that follow here and later have been drawned by me to work out the problem raised by Dasgupta (1976) in detail.

Page 173: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

172

54. The Equi rule was written by Dasgupta (1978:3) in the following way: (4) Equi-Noun phrase deletion (posited by the clause hypothesis) S.D : X - NP – V - NP NP S VP S.C : 1 2 3 4 ? 1, ? , 3, 4 55. The formulation given by Dasgupta (1978:3) of the lexical rule (128) was slightly different. No did not use the additional N-symbol used in (128). 56. See (121)b to get a parallel construction in Maithili whre –Tea has been used with gerundive nominals. 57. His examples include the followings A) i. Raamer baaje kOthan bole bRaane Ram;s nonsensical things say-abc go-around-for “ Ram’s going around saying nonsensical things” ii. Raamer bele bRaane baaje kOthan B) i. samaar Saastric SoNgit bejhaa My classical music understand-Ser ii. Aamaar beh\jhaa sasstric Sengit 58. (134) did not have additional N in Dasgupta’s (1976) formulation of this rule. See fn. 55 above. 59. In the original formulation, the N and I categories did not have bars on them. See Dasgupta (1978:7) 60. This rule and the general condition mentioned after it did not figure in Dasgupta (1978). 61. Notice that to justify the use of cover symbols such as X(with or without bars) or [? Subj] or [? Comp] etc., one has to show that all the combinations in (96) or atleast a majority of them are actually attested in human languages, and not only (96) ii (as shown here ) or (96) iii as shown by Chemsky (1970) exist. 62. For certain speakers of Maithili, however haa2kala in this kind of construction would not be acceptable, but for many others it was found to be quite okay.

Page 174: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

173

63. I have put Aux in parenthesis here because some of the verb initial –languages show Aux incorporated into the body of the verb in such a way that it is inseperable. The same comment applies for (146)a. 64. For a detailed discussion on this problem, see Chapter VII of this dissertation.

Page 175: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

CHAPTER IV RELATIVINATION, COMPLEMENTION, AND

INTERROGATION

4.0 In this chapter I shall discuss in detail three related and basic syntactic processes in Maithili, viz., Relative clause formation (henceforth, RCP), Sentential Complementation and Interrogation or Question Formation. 4.1.1.1. In the recent and not-very-recent literature on RCFs in various languages across different language-families, one finds linguists concerned with one or more aspects of relativization such as follows: (1) Does a language have more than one relative clause marker? If so, what are the principles of distribution of these markers? (2) In it the case that in a particular language relative and interrogative markers are the same in surface? If so, do they syntactically behave in more or less similar manner? And even if the overt forms are different, does the syntax of that language show systematic parallels in these two constructions? (3) Looking at the nature of movement rules in sentences that embed one or more relative clauses, is it worthwhile to talk about island constraints, or for that matter about a fixed ordered base on which syntactic rules would apply? (4) Should relativization be defined on a syntactic basis, or a set of “syntax-free” criteria be made basis of

Page 176: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

175

such a defination? i.e., is it the case that participialisation, nominalisation, equi, anaphera, verb-coding, gap and other related phenomena are employed in natural languages as different strategies for RCF? Or, these are different syntactic rules the input of which is somehow fed or filtered through a previous application of the relativization rule? (5) Does the relativisation rule of a particular languages adhere to the hierarchy of grammatical relations : i.e.which are the NPs that are accessible to RCP? 4.1.1.2. In this chapter, I shall first elaborate”some” of these aspects of RCP and discuss the case of Maithili in this respect, and then go on to Complementation and Interrogation rules as attested in this language. And, finally, I will come back to the various RCP strategies as attested in Maithili. 4.1.2.1 Eric Hamp (1972a:vii) noted that in Celtic languages the relative clause markers have “highly varied and elusive shapes”. However, the syntactic functions of these are the same. Similarly, for Thai, Prometasana (1978) observes that the distribution of the relative clause linkers such as uhun, thii, sin, an, den, phuntii, and phuusin is governed semanticall. All of them occur only in the following syntactic position: (6) # NP [ NP + ----------- + S]NP # What happens in such cases is that quite often forms from different dialects are assimilated into the standard dialect. One may expect to find such multiplicity of relative markers also when a language

Page 177: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

176

has a diglossic situation of some kind. Thus, in Thai an and dan are literary forms while thii is commonly used in Spoken Thai. Notice that are really or geneologically related languages may not have similarities in this respect. For instance, Basque provides a unique invariable shape for its relative marker (deRijk , 1972) which is a rare phenomenon in the ancient continent of Europe. 4.1.2.2. Although Maithili does not have different markers for the relative clause, the same form assumes varying shapes here depending upon the case marking it takes. Some other factors such as homerificity, animateness and possessiveness also come into play in this case. The paradigm for modern Maithili relative markers is as follows: (7) ANIMATE INANIMATE HONORIFIC NON-HONORIFIC DIRBOT : je je je OBLIQUE : ja(k2 raa) jam1 (kaa) jaah1 POSS: ja(kar) jami(k2)/jami (kar) jaah1 (-NP + ak). jaah1 (-kera) 4.1.2.3. In Old Maithili, however, there were more forms than these. In the Garyaapada (9th-11th e A.D)., one finds that following shapes of the relative pronoun. NOMINATIVE : je, je-je ja, je ACCUSATIVE : jaa GENITIVE : jaa (hera) INSTRUMENTAL : jo (na) LOCATIVE : johi, jasu

Page 178: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

177

4.1.2.4. The following sentences from the Garyaa are illustrative: (9) jo managoars1 so vaansa (Kashru, 7.2.2.) RNL-cognisable COREL anattached “What is cogisable, that is unattached”. (10) jo jo asila to to gelan ( Kaashru, 7.4.1) Whoever came all went”. (11) ja ehu2 jaati3 - - - - - (Saanti, 26.3.1) REL it win+pres+Agr “(No, ) who wine it - - - -“ (12) jaa lai4 achana tashera3 wha6 Na dis7 (Lui, 29.3.2) REL take+ptopl I be, of him (is) neither a direction, (nor) a purpose”. (13) jashera baane-eihna 8 remba 9 Na jaaNii, se - - - (Lui, 29.3.1/2 ) REL+gen color, form, appearance Neg know+pross + Agr, COREL “Whose color, form (and) appearance I do not know, him - - - -“. (14) kaa-back-sia 10 jasu Na samaaya 11 (Kashru, 40.2.2) “Where body, speech and heart fathoms not”. (15) johi naHa india- (pad) baNa 12 he 13 NaThan 14 ( Anjadeva, 31.1.1.1) REL +lee mind senses breath be+ pros +Agr destroyed “Where mind, senses and breath are destroyed”. 4.1.2.4. A few centuries later, the shapes of these markers changed considerably. Thus, coming to the days of Vidyaapati’s lyrics, Jyotiriswara’s “VarNaratnaakara” and Unaapati’s “PaarijaatajharaNa” 15 which date back to 14th to 17th century, the relative clause linkers assumed the following shape:

Page 179: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

178

(16) ANIMATE INANIMATE NOMINATIVE : jo (ha) jaa ACCUSATIVE : jaahi , johi, johe, janhi - GENITIVE : jaaka, jaake, jaaheri, jasu janhik(a), janikari INTRUMENTAL :jemi – LOCATIVE : - ja 4.1.2.3. In most of the cases, the relative pronoun is followed by a correlative. The correlatives can also be classified depending upon the features such as animateness, honorificity, and the grammatical relations, and they are as follows: (17) ANIMATE INANIMATE HONORIFIC NON-HONORIFIC DIRBOT : se se se OBLIQUE : ta (m1 kan) ta(k2 ran) taah1 POSS : ti (nak a ) ta (kar a) taah 1 (-NP+ak )/ taah 1(ker a ) 4.1.2.6. Apart from these relative pronouns, there are a number of relative and correlative adverbs such as : ioman ----- toman “as - - --se”, inteR ----- tataR “where ----there”. Inta has (also, latek 2 ---- takek a) “ as much ---- so much”, iakhan------- takhan “when ----then” , ishion ---tehian “which day ----- that day”, iso iso-----tee tee “by the time---- by then” and many others. 4.1.3.1 The relative and the interrogative pronouns have distinctly different shapes in almost all the Indo-Aryan languages. In these languages, the relative marker normally starts with j- whereas the question has always a k. The same is the case with Maithili (of. Singh and Bandyopadhyay, 1978:20). In English,

Page 180: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

179

however, they are strikingly similar. Both have almost an invariable shapr: wh-. But then, English is not an isolated example of this type. Howard Berman (1972) reports that even in Hittite, they are the same. It is attested also in Georgian (Aronson. 1972). Morgan (1972) points out that it is almost the same with Albanian relative and interrogative adverbs. Although this feature is absent in the relative pronouns of Albenian, the interrogative adverbs ku and kur have already entered into the relative system. 4.1.3.2. However, till recently it went unnoticed in the relevant literature that in some languages the relative pronoun and the sentential complement markers are also identical. This similarity becomes all the more significant when one notices that even the open and yes/no question markers are similar in many languages. Recently, following the suggestions originally embodied in the famous paper on English relativisation by by S-Y Kuroda (1968), and in the process improving upon the same, Probal Dasgupta (In preparation ) has made an important suggestion in this respect. The original suggestion was that since relative clauses and wh-questions showed a systematic parallel (in English) they be derived from the same source. Notice that Chemsky and Lasnik (1977) have a position that go against Kuroda’s suggestion directly. Even earlier, Kats and Postal (1964) treated those as independent phenomena. 4.1.3.3. To capture the similarities that exist between relativisation, complementation and interrogation, Dasgupta (In preparation) introduced the notion of ‘Quadrangle’. It is a common knowledge that some grammatical categories show more than

Page 181: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

180

one axis. Pronouns, for instance, show two axes in many languages- that of number and person. On the scale of person, a language may have three points: First, Second and Third. Similarly, the axis of number may also show three points: Singular, Dual and Plural . If that be the case, the pronominal quadrangle would have the following shape in such languages: (18) PERSON FIRST SECOND THIRD SINGULAR : 1 sg 2 sg 3sg DUAL : 2 du 2 du 3 du FLURAL : 1 pl 2pl 3 pl 4.1.3.4. Similarly, one can also think of some other categories that involve two axes 16, thereby allowing one to posit a few more quadrangle of this type. The relative-interrogative-complement system, according to Dasgupta (In preparation) provides us with another such example. No calls it a Q-clause quadrangle and argues that it must be incorporated into the theory of language universals. The C-Clause Quadrangle would look like the following: (19) A A X I Si X DBT-OPEN COMP-OPEN I NON-QUESTION : Relative Clause Complement Clause Si QUESTION ‘Wh’ Question “Yes-No’ Question 4.1.3.5. In terms of the markers for these constructions, the Q-Clause Quadrangle in Maithili would lack like the following 17, (20) A A X I Si X DBT-OPEN COMP-OPEN I NON-QUESTION : j- j- Si QUESTION k- k-

NU

MB

ER

Page 182: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

181

4.1.3.6. Notice that it would be ideal to work on a language where all the four points in the two axes are represented by different sets of markers. But neither Maithili nor English (nor, for that matter, Hittite, Georgian, or Albanian) is ideal in this respect. However, Maithili is different from these languages in that it does not have the same marker for the relative clause and open question. Rather it has, like in Bengali (io/io), Sanskrit (yat/yat), Russian (ehte/ehte), French (cue/cue) and many other languages, identical forms for the points on the axis1 as opposed to axisj. If it could still be shown that in Maithili (where relative and open question markers are different in form) these two constructions have the same degree of commoness from the syntactic viewpoint, it would go directly against the assumptions of the works of Chemsky 91973, 1978) and his associates, and would give support to the suggestions embodied in Kuroda (1968) and Dasgpta (in preparation). However, I would leave this task to a future date, and instead I would discuss in detail the three basic syntactic processes in the rest of this chapter. In the following paragraphs, I shall take up complementation, interrogation and relativisation in this order. 4.2.1.1. Resembaun (1967) had postulated two types of sentential complementation depending on whether the subordinate sentence is embedded in the NP mode or in the VP. No proposed two phrase structure rules for English as in (21) to capture those two types and tries to justify this distinction by applying three criteria as listed in (22)-(24) below:

Page 183: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

182

(21)a. NP? DPT N (S) b. VP? V (NP) (PP) {s} {PP}

(22) The passive test (23) The extraposition test (24) The pseude-eleft test

He then went on to describe in detail the three types of markers for complementation (that, for…to and Pass…ing ), and the rules for their introduction and related matter. Finally, he discussed the governmental nature of the rules for complementation and provided with as list of English verbs that would take complement sentences. 4.2.1.2. A year later, Kiparaky and Kiparaky (1968) showed that a number of basic semantic factors could help predict the choice of complement markers and that the most important of these was the notion of ‘Presupposition’. They brought out the distinction of [t fact ] and showed with the help of a number of systematic parallels that the true verbs and adjectives governing sentential complementation should be divided into two types : factive and non-factive. They had, in short, the following arguments in favour of their proposal; (25) a. Only factives allow the noun fact as their subject or object when it is a finite or a gerundive complementation, b.Factives take gerunds mere freely than the non-factives do, in both subject and object positions, c.Only non-factives allow the accusative and infinite constructions.

d. Only non-factives allow the subject-to subject raising, and e. With non-factives, the application of extraposition is a must in the domain of subject clause, while it is optional for factives.

Page 184: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

183

4.2.1.3.1. In the early years of this decade, studies such as Ross (1972), 1974) and Lakeff (1973b) made the common researcher more and more aware of the futility of a theoretical position based on “all-er-none” priniciples. To elaborate, they stressed that the grammatical descriptions in allowed to be “fussy” because if one looked at the way the natural languages behave, one would be almost compelled to talk only in terms of degree no matter which of the factors (such as rule application, categorisation of grammatical, syntactic “islands” grammatical environments, grammatically and informant elicitation of the data) of grammatical description one took. This “squisky phenomena” (Ross, 1972) tended to blurr the traditional distinctions between various grammatical classes and categories.Some such squishes would include: (26) Verb>Present participle > Perfect participle> Passive participle>Adjective > Preposition (?)> Adjectival noun > Noun (Ross, 1972, 316) (27) That>for-te> Q> ACC ING > POSS ING > Action nomina l > Derived nominal > Noun (Ross, 1974). 4.2.1.3.2. With the help of these squishes, they would thus talk about the “verbiness” and “nouniness” of categories. (27) would claim, for example, among other things, that “POSS-ING complements are more “nouny” that complements” (Jain, 1975 :viii). This also implies that these three complement-types “sentency” in the reverse order. One can find out similar interpretations for (26) or other squishes.

Page 185: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

184

4.2.14.1. In the field of syntactic descriptions of natural languages the latter half of this decade say linguists engaging in long argumentations regarding the nature of the phrase structure rules of particular languages. We have already seen (ef.Chapter III) that McGawley (1970a) and Bach (1970) were one of the first people to join in such wild speculative game as whether English and Anharie could be considered as VSO and SOV languages, respectively. The same exercise was repeated by several development lied in the “myth” that somehow sneaked into linguistics by the misunderstanding of the first universal of Greenberg (1963a : 77): (28) “universal 1. In declarative sentences with nominal subject and object, the dominant order is almost always one in which the subject precedes the object”. (Under-score :UNS) 4.2.1.4.2. Greenberg (1963a: 103) was himself aware of the fact that (26) was at best a statistical or “tendency” generalization and he did not conceal the fact that languages like Siuslaw, Gees and Coour d’ Alond were genuine exceptions to this tendency. But somehow the theoreticians of this period worked with the assumption that Greenberg’s elimination of three mathematically possible combinations was not enough and they were doing good to the theory of language universals by trying to reduce further the three linguistically possible sentence-type into two types depending on whether the verb is in the initial or final position of the declarative sentence. All they had to do was to posit a few additional movement rules that would take care of the verb-medial order of the surface structure.

Page 186: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

185

4.2.1.5.1. As an extension of the above studies on the “basics” word order of grammatical elements came the argumentation as to what should be the order of embodied sentential complements and the nominal (be it lexical or dummy) that functioned as the “head” of such complements in different languages. An example near home will be the debate on the PS rule for NP in Hindi. The hypothesis implicit in the works of Kachru (1965, 1966) and Verma (1971) that the PS rule for the noun phrase complements in Hindi was NP ? (S)N was first challenged by Sinha (1970). Later a number or arguments based mainly on the notions of analytical simiplicity, economy of symbols and on Greenberg’s language universals were presented by Subbarao (1974b) who tried to show that Sinha’s (1970) hypothesis that the PS rule had an N (S) structure was not tamable. The debate continued when Sinha (1976) tried to answer Subbaroa’s (1974ab) criticisms. 4.2.1.5.2. If one views this or similar debates from a different angle, to a great extent, they would now seem to be exercises that were probably not of much necessity. I shall not repeat here the arguments against the theoretical assumptions regarding the components of grammar (especially regarding the phrase structure component) that underlie most of the works mentioned in the earlier paragraph. 4.2.2.1 I would now discuss different types of complementation and complementisers in Maithili. Maithili has two main complementisers i.e. (comparable to Hindi ki, and Bengali and Oriya ie.e ) and ak-ab (comparable to Hindi, Kee-nam and Bengali ar-aa). The former is a finite complementiser and in that sense similar to

Page 187: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

186

that- complementiser in English, while the latter is an infinite complement marker. Maithili, like the most of the other Indo-Aryan languages, does not distinguish between the infinitive and the gerundive complements, and therefore it is natural that many of the equations that are true for English do not hold good for Maithili. Traditionally, one can think of different types of complement constructions depending on two factors-the grammatical relation that the NP governing the embedded sentence holds in the higher S-mode, and the status of the verb that governs the constructions vis-à-vis the notion of transitivity. If one adds the notion of finiteness to the ones above, there emerge seven possible types of complement structures in Maithili. I shall enumerate them below ( in the way Rosenbaun (1967) does for English) along with their structural descriptions in the traditional way for the convenience of the discussions that follow. Notice that I have taken it for granted that the “favoured” word-order in Maithili will be (S) N (within the framework of traditional TG grammar too, the PS rule for Maithili will have an order such as (S) N, and not N (S). The seven possible types of structures are as follows

(29) a. Transitive subject NP complements with a finite complementiser when the structural description is

S[ NP[ je + S [N] ]NP VP[ NP V + AUX ]VP ]S b. Transitive subject NP complements with an infinite complementiser – S[ NP[ S [NP +ak VP +ab ]S [N]NP VP[ NP V + AUX ]VP ]S

d. Intransitive subject NP complements with a finite complementiser. S [NP[ je + S[N]NP ] VP [ V + AUX ]VP ]S

Page 188: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

187 d. Intransitive subject NP complements with an infinite complementiser---

S[NP[ S[ NP + ak VP +ab ]S [N] ]NP VP[ V +AUX ]VP ]S e. Objects NP complements with a finite complementiser- S[ NP VP[ NP[ je +S [N] ]NP V + AUX ]VP ]S f. Object NP complements with an infinite complementiser S [ NP VP[ NP[ S[ NP +ak VP + ab ]S [N] ]NP V + AUX ]VP ]S g. Oblique NP complements (which always have the infinite complementiser marking) S[ X –ADVP [ NP [ S[NP subj +ak NPobj +KPn V +AUX +ab ]S ]NT ADV] ADVP – Y]S 4.2.2.2. Now, an example of each of these types could be given :

(30) a. behin daay (TT jo ) kaalh1 animahatyaa kaaliiha sister elder COMPL yesterday suicide do-pattoAgr ii samnad humkaa dukh pehun cosal a b. behin daayak kaalih1 aatnahatyan kar2 baak samaad sister elder +yesterday suicide de+INF +Gen move + POSS humkaa dukh pohun camela to him sorrow carry+pet+Agr “The news of the elder sister’s death shocked him”.

c. (T je ) aab1 rahal a chath 1 ii khabar 1 pokkan aah 1 No COMPL come+ptopl Prog +pst beufo + this news definite Ptopl Agr be+ pr +Agr “It is a confirmed news that he is coming”.

Page 189: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

188

d. hunak asbaak khabar1 pakkaa ach1 Ne +POSS come +Inf news definite be + pros +Agr “The news of his coming is confirment”.

e. ham a aab 1 rahal a chath 1 ii khabar 1 jamait 2 chii I he come +ptopl preg + be + pros + this move know + be +pr + p.ptepl Agr ptopl Agr “ I know the move that he is coming”. f. e kich u kinback lel a dukaan jan rahal a chath 1

“No is going to shop to buy something”. 4.2.2.3.1. This is a common knowledge that not all sentences take complements. Both the finite and infinite complement-types have to depend on the nature of the verb in the matrix sentences. There does not seem to be any grammatical label for the verbs that allow such embeddings as shown above but some kind of semantic feature seems to guide as to which verbs should take complement sentences which should not. Out of well over thirteen hundred verbs that I have come across in Maithili the following verbs allow sentential complementation. I have classified these into two types depending on whether they presuppose the “factivitity” of the sentences embedded. It must be borne in mind that the following list is incomplete

Page 190: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

189

(31) FACTIVE ( + COMPLEMENT) VERBS Udes “ tell a secret” Upech/upokh “ignore” Ghesar/bisar “ forget” Jaan “know” Dekh “see” Pataa-laag “come to know” Bujh “understand” (not when it means “think”) Makaa-lang “amuse” Aj a guta –lang “seem strange” KaBt-de “ pain” Afacea – bha “regret” SpaSt-kO-de “make it explicit” AaT-pahuN e “be shocked” Einh “recognise” Oot “war,” Bhuul i – jan “forget” Thee-laag “be hurt”

(32) NON-FACTIVE (+COMPLEMENT) VERBS

Uear “seems” Khag “appear” Pkur “occur” Caah “occur” Kah “speak” Baaj “say” Khoj I – mikaal “say” Par1 ooar “figure out” Parn ooar “spread rumour” Phusian “tell a lie” Bicaar “conclude” Bujh “think, guess” Laag “seem likely” Sun “hear” 4.2.2.3.2. Note that apart from these verbs there are a number of adjectives in Maithili that also take complement sentences and these could again be divided into factives and non-factives. More it must be borne in mind that not all the sentences that show i.e. before an embedded sentence are cases of complementation. For instance ie in (33) marks an adverbial clause only.

Page 191: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

190

(33) a. o jo etae gela chalaaha takar kii bhel a ? He ADV there go-pts have+pst + that+gen what happen-pst-Agr “What happened to that-to his going there?” b. hum hunkaa etae paThalisah1 jo e hamar I he + Obj there send +pst +Agr so that he I – gen Baalak kNn khoj1 ammath1 Sem +Obj mark search +abs bring +opt +Agr “I sent him there so that he finds out my son and brings (him back)” 4.2.2.3.3. Maithili ie, like Bengali complementiser ie, marks many other things also and some of these include the followings

(34) i. Ie as a relative clause marker, ii. ie as a marker for conditionality (as in 33a), iii. ie which marks a purpose (as in 33b), iv. ie that is used as a pronoun in the constructions such as ie kyee “anybody” and ie kichu “anything”. (35) (kintu) aami je a Sob kichui jaami naa But I EMPH this all nothing know-pr-Agr not “(But) I don’t know all these”. 4.2.2.3.5. In Maithili, ie also marks the direct speech in certain types of constructions, such as in (36) (36) a. o haaran rokula thiiuha je nah1 jan No no stop +pst +Agr that not go+imp +Agr “No stopped me (saying ) ‘ Don’t go”. b. o haaraa ii kanj saen pa lashi jo shaan kO dicha He no this work assign+pst +Agr that you de + abc give + imp +Agr “ He assigned no this work (saying) ‘ you do ( it )’

Page 192: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

191 In (36), ie has been used as a quotative and hence is not really a complement marker. This ie is, however, a different type of quotative compared to bels (Kachru, 1978) in Bengali and bali in Oriya. Bali in Oriya has been taken as a full complementiser by Patnaik (1976; 1978) whereas in Bengali its occurrence and distribution does not seem to be so wide (of. Singh, 1978a). It is also worth noticing that the embedded sentences in the Maithili examples are not in the reported speech. Rather they are in the direct speech. This is not the case with bale in Bengali and Bali in Oriya. 4.2.2.4.1. A few more points emerge out of the examples given earlier. Notice that the verbs that permit sentence embeddings in these cases are essentially transitives. But this does not automatically rule out the possibility of an “instrasitive” complementation because in the latter case the structure of the matrix clause is

invariably the following: 4.2.2.4.2. There are several things to be noted in respect to the structure in (37). The presence of an adjective of a special type is a must for the complement sentences to occur. That is, just like the verbs in (31) and (32), these adjectives also “govern” complementation and those are also classifiable into two types-factive and non-factive. The matrix verb is always the verb UN

Page 193: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

192

does not matter which of the different forms of NN appear in the surface. I could be cha, bha or rah. Then the head noun can be a simple pre-form such as ii or it can have optionally the nouns such as heat “fact”, khabar1 “move”, samaad “information”, haa “sense”, gyaan “knowledge”, dhiiraj “patience” and budh 1 “sense”. It is comparable to Hindi vah baat and Bengali e-kO than constructions where like in Maithili, vab and oTan (and not simply o as one would expect by looking at Hindi and Maithili data) can stay certain cases, the lexical head noun may be retained while the dummy ii, yah or eTea is dropped. 4.2.2.4.3. Also notice that in the examples given earlier, the oblique NP complementation emerges as a very different type of construction in that it does not require to have a particular type of verb in the matrix sentence. Thus it does not seem to be a “governed” structure. Moreover, a class look at (29)g would allow the complement, it does not matter whether the ADV P concerned comes in the initial, medial or final position of a sentence. The adverb comes at the end of the adverbial phrase, i.e. immediately after the embedded complement is normally one of the following: laR or lel a “for”, aO n “from, with” hot u “purpose” and nar “en” (used rarely). It is to cover those cases that Sinha (1970) had proposed a third complementists –naa ke live (of Subbarao, 1976 for arguments against this claim). One also finds hand “after” and kaarann “because” in place of this adverb. There is nothing to disagree with Subbarao’s (1974ab) arguments against Sinha’s (1970) claim and I shall not repeat these arguments here.

Page 194: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

193

4.2.2.5.1. It must, however, be mentioned here that out of the four possible tests (to find out whether the adverb laR or lal a is purposive or it is a semantically null post-position)- question, deletion, translatability into the finite complementation and the identical subject-test, not all of then hold good for Maithili. Consider the following sentences the first one of which has a purposive lel a, while the other one has a post-positive (37) a. o shaa n aOn beat kar- a b- aak – lel a jane No you with talk de+Inf +purpose go +inf Caahait a chath1 Wish + ptopl be+ pros + Agr “No wants to go to talk to you”. b. o hamran sO n sup rah-a b – aaklel a kahal a thiiah a No me to silent remain +Inf+ ? say +pst+Agr “no asked no to remain silent”. 4.2.2.5.2. One can now question both the le l a by kiylk “what for” which will show that the lel a in the b-sentence was non-purposive. In the same way, one can also shift lel a to the next higher sentence and “translate” the embedded sentence into a finite complement-type and show that the b-sentence does not have a purposive interpretation: (30) a. o kiylk jaae caahait a chath1 No why go + inf wish + ptopl be +pres +Agr “Why does he want to go?” b. * o kiylk kahal a thiiuh a No why say +pst +Agr “Why did he say?”

Page 195: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

194

(40) a. o eh1 lel a jane caahait a chath 1 jo ahaa n eOn He this for go+inf wish +ptopl be+pres + CPL you with Baat kar a taah a Talk do + fut +Agr “ He wants to go for this that he will talk to you”. b. o eh1 lel a kahal a thiinh a je han cup rahii He this for say + pst + Agr CPL I silent be+Opt +Agr “He asked me for this that I remain silent”. 4.2.2.3.5. However, the deletion test, if applied to these Maithili sentences, would give bad results as is evident from the ungrammaticality of both the sentence in (41): (41) a . * o shaan sOn heat kar-(ab) jane caahait2 chath1 b. * o hamran cup rah – (ab) kahala thiiuha The deletion of aak – lela in those cases is required to be followed by an –0 addition to the right of the verbs in the embedded sentence. The resultant sentences are acceptable: (42) a. o shann sO heat kar-O jana caahait a chath1 b. o hamran cup rah-O kahala thiinha 4.2.2.5.4. It is thus evident that in Maithili no clear line of distinction can be drawn between the purposive and non-purposive (ask) – lela on the basis of the deletion-test. But since the other tests are applicable I do not see why Subbarao’s (1974ab) arguments do not hold good. Of course, one may say that -) is comparable to Hindi –ke and that it is just another post-position the application of which has no relation with the deletion of aak-lela . Under this interpretation, one is forced to say that the deletion of aak-lela is simply not permitted in Maithili. A third interpretation seems to be as follows: The element aak-lela consists

Page 196: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

195

of –O-ak-lel a, and when the formet is deleted –ab also goes leaving behind –0. It may be noted here that alternatively one can also delete –ab and –ak and form grammatical sentences with only –O lel a. Thus, (38) and (42) have the following alternative: (43) a. o shaan sO n baat kar-O lel2 jana caahait a chath1 b. o hamran cup rah – Olela kahal2 thiiaha 4.2.3.1.1. Maithili presents a very interesting case if one looks into the oblique complementation more closely. While it is true that Subbarao’s (1974ab, 1976) arguments against a naa-ke-live complementiser are equally valid in case of Maithili ab-aak-lel a, some other types of oblique complements seem to show yet another marker – ak-laa, Consider the following sentences: (44) shaa n ka ii kaaj keelan par o asauntuST a bhelash a your this work doing by he unhappy be+pst +Agr “He was unhappy by your doing this work”. (45) lalitaak ang a re zii sukhlaa sO n hamran kii laabh? Lalita’s English learning from no what gain “what is my gain in Lalita’s learning English” (46) ang2 resii sukhlaa sO n Lalitaa kB n laakh haatamh1 “ Lalita will be benefited by learning English”. 4.2.3.1.2. Notice that the embedded sentence does not have to be in the sentence-initial position to be able to take this marker. One can, for instance, shuffle it around in the matrix clause as may be made evident in (47) where the elements in the sentences (44) through (46) have been reordered.

Page 197: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

196

(47) a. o asantuSTa bholasha [ ahaan k ii kanj keelan par] b. haaraa [ lalitaak anga resii sikhlaa sOn ] kii labh? c.Lalita kBn laabh [ anga rezii sukhlaa sOn ] haataahi 4.2.3.2. These ak-lan construction apparantly look somewhat like ek-ab (Hindi kee- nan and Bengali or-as ) complements, but if one replaces the former with the latter in these cases the result is ungrammatical strings. This explains why (48) a through (48) c are starred. (48) a. ? shaan k ii kaaj karn ban par o asantuSTa bholasha b. * lalitaak ang a resii sukhbaa sOn hamraa kii laabh? c. * angn rezii sukhbaa sOn lalitaa kBn laabh haataaha 4.2.3.3. The marker ek-laa, however, must be further segmented into ek + el a +aa in the same way as one would divide ak-ban into ak + ab + aa where aa is an adverbial marker comparable to –e in Hindi which is suffixed to naa to give the resultant n o in the oblique complements. Since ek-ab and ek-ela have so many things in common, one wonders whether one can lump them together as alternants. If ak-abv is not allowed to occur where ak-ela , therefoe, seems to be limited to only certain types of adverbial or oblique complements. But all of these adverbs also cannote a [+past ] sense so far as the verbs in the embedded sentences are concerned. This is also supported by the language history. In early stages of Maithili, -ala was tagged to the root to denote the past participle base, the proof of which still exists in Modern Maithili conjugations; e.g., gela ash1 “has gone”. Similarly – b in both Bengali and

Page 198: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

197

Maithili originated from a future participle base. That is why in the Saadhu style of Bengali the verbs bear –baa as infinitive marker as against the Gelit –aa; e.g. Saadhu. Dekhibaa (-mantre) “(the moment one) sees”, Gelibaa (-kaale)” (While) moving”. Notice that the adverbials that occur with ak-al a carry a [ + past ] feature in their meaning entry, and these include so, nar, ek, haad, ek up a raant a , etc. All these arguments point out that although formally ak-ab and ak-al a are two markers for infinite complementation, at some point of derivation they may be taken to be coming from the same sources. 4.2.4.1. The question that I shall now discuss is: Why was it taken for granted that a grammar based on the principles of deep word order would take (S) N as the initial PS rule for an NP in Maithili. Notice that in the “normal” declarative Maithili sentences the “favoured” word order is SOV. It is only under special conditions of emphasis that Maithili, like other sister Indi-Aryan languages, change (masion, 1974) this basic order of elements. Thus like these languages, Maithili permits adjectives, genitives, demonstratives and numerals to occur only before nouns. As in Hindi surface structures evidences do not show decisivelly as to which of the two orders is better or is preferrable, so in Maithili and Bengali those evidences are not decisive. In Bengali, at times, the relative length of the complement sentences tilt the balance of judgement in favour of one or the other order. Thus while the S-N is adjudged better in (49), the N-S order gives a better result in (50) a:

Page 199: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

198

(49) a. eTan enr bOlan ucit hOe ni [ je acni bokaa] This his say + Inf proper be+pros not that I feel “it was not proper for him to have said that I am a fool” b. [acni bokaa] eTan e n r bOlan ucit hOe ni (50 a. eTan e n r bOlaa ucit hOe no [je acni o kaaj This his say +Inf proper be+pros – Agr not that I this work Kerte paarge naa] De+Inf able +fut –Agr not “ It was not proper for him to have said that I shall not be able to do this work”. b. [acmi o kanj kerte paarbe naa] eTan e n r bOlaa ucit hOe ni 4.2.4.2. In comparison to that, Bengali er-aa (i.e. infinitive) complementiser seems to favour only an S-N order. In Maithili, however, both the orders are permitted and are current even in –ak-ab complementation. Consider the following sentences: (51) han [agnoipuSpak paas haabaak] samaacaar {suna lisnhi / suna lahun } I Agniyashpa’s pass be+Inf +ef news hear +past+Agr (52) a. han samaacaar { suna lisnhi / suna lahun

} [ agnipuSpak pass haabaak] b. haa { suna lisnhi / suna lahun } samacaar [ agnipuSpak pass haabaak] c. haa sammacaar [ agnipuSpak pass haabaak] { suna lisnhi / suna lahun } 4.2.4.3. In case of ie-complement constructions in Maithili all the orders such as N-X-S, N-S-X and S-N-X generate grammatical strings, as the following examples will illustrate:

Page 200: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

199

(53) a. ii kahab Thiik nahi haft a [jo mithilaa aP] This say + Inf right not be+fut + Agr that Mithila in sab kyee kavi chath i ] Everu one [poet be + pres+ Agr b. ii [ jo mithilaa nF sab kavi chathi ] kahab Thiik najo haFt a c. [mithilaa nF sab kyee kavi chath I] ii kahab Thiik nahi haFta 4.2.4.4.1. All these sentences point out is that in Maithili the two types of complement sentences do not show different preferences for the order of elements such as S and N. In comparison to this, Hindi evidences showed (Subbarao, 1974a: 9-10) that the finite complementiser had a preference for N (S) order whereas the infinite complementiser showed an (S) N order invariably. Thus, Maithili examples make the choice between the two even more difficult at least in this case. 4.2.4.4.2. In the same way, even the modifiers originated from the relative clause structure do not provide any strong evidence in favour of a particular structure. In a number of examples the Noun-Modifier order is tolerated and even more frequently used. Consider (54) and (55) for example (54) a. baraita diyaa baRa sundar lagaita achi burning lamp very nice seem+pros – ptopl be +pres+Agr “The burning lamp looks very nice” b. bhiina son dundhbaalaa aela chala Bhimsen milkaan come+p.ptopl be+ pst +Agr “Bhimsen, the milkman, had come”.

Page 201: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

200 c. git a haari mangli-laken I kataN geliiha ? singer woman +pl where go + pst+Agr “where did the singer-women go?” d. raanak baalak hamraa to niike lagal a Ram+gen son no ENPN good +ENPH soon +pst+Agr “Ram’s son seemed good to me”. e. tehar a dimaag a unTo chOh a Your brain opposite be+pres+Agr “Your brain is dull”. f. kunnalak kavitaa niike chalaahi kunal+gen poem good +ENPH be + pst +Agr “kunal’s poem was nice, indeed”. 4.2.4.4.3. The order of elements in sentence (54)a through (54) f is reversed in (55) below: (55) a. diyaa baraita baRa sundar lagaita aah1 b. dundhbaalaa bhiimsen ancla chala c. mangii-lekani gitn haari kataN geliiha ? d. baalak raamak han2 raa ta miike laagala e. dimaaga tohara unTo chOha f. kavitaa kunaalak niike chalashi 4.2.4.3.1. It would soon that the answer to the problem of word order could not be found even if one considered the reordering rules as applied to the complement sentential structure. Although the reordering rules will be discussed at length in the Chapter VII, a few examples of what is known as extraposition could be presented here to illustrate this point. If one considers the structure of three sentences listed in (56), it would be evident that bothe S-N and N-S orders are rated equally if the reordering rules do not change their formation. Also notice that whenever the head noun occurs to the right of the embedded sentences, the complementiser

Page 202: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

201

ie has eithere to be deleted or it should be shifted to the right of the initial element of the embedded sentence: (56) a . o ii [ jo kunaal eh1 dis1 aab1 rahul a chath1] No this that Kunal this direction come+pte prog+ be + pr + p.ptc Agr Jamait a chal a thiinha “ he know that Kunal was coming in this direction”. b. * o [je kunaal eh1 dis1 aab1 rahula chath1] ii jamaita chala thiiuha c. o [kunaal (jo) eh1 dis1 sab1 rahul a chathi] ii jamaita chal1 thiinha 4.2.4.5.2. Now, the embedded sentence can optionally be moved in such a way that it can either go to the initial position of the matrix sentence or it can be transposed to the extreme right of the matrix sentence. The sentence below would thus suggest that Maithili needs both a rightward and a leftward movement rule in case of jo-complementation; (57) a. o (ii) jamait a chala thiinha [jo kunaal eh1 dis1 sab1 rahul a

chath1] b. [kunaal (jo) eh1 dis1 aab1 rahula chath1 ] {o /ii} {ii/o } jamaita chala thiinha 4.2.4.5.3. Notice that Bengali is very similar to Maithili in this respect in that here too the S-movement rule (er, extraposition) must be allowed to take both directions. The following sentences show a striking parallel. (58) a. tini (eTan) bujhlen [(jo) e oer nOe ] No (hem ) it understand+ that he +AN thief not + be +pres + Agr

“He (hon ) understand it that he (non-homerific ) was not a thief”.

Page 203: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

202

b. *tini eTan [ jo o oer nOe ] bujalan c.[ o (je) eer nOe] {eTan/tea } tini bualan ? d. tini [e (jo) eer nOe ] {eTan tan} bujalan The juxtaposed N (S) structure gives a bad result in Bengali as may be seen in (58)b. In this case the movement becomes a must. It may also be noted that the dummy pronoun may not be dropped from the structures such as (58)c and d. But I had already pointed out that in certain cases length seems to play a crucial role. It is true of these constructions too. If the verb of the matrix sentence is lengthened by adding near “be able to”, these pre-forms can be deleted : e.g., (59) a. [o (jo) oer nOe ] tini bajate perechilon b. tini [o (jo) oer nOe ] bujhte perechilon 4.2.4.6.1. Since surface structure, syntactic rules and other evidences do not show as to which of the “deep” word order is the right choice for Maithili grammar, the only way to push Maithili to the (S) N order languages is to argue that since other areas of Maithili syntax show proofs that it is a verb-final language, and since the SOV languages have normally an (S) N order (in TG terms), Maithili cannot be an exception. If one took such a position, it would based on a few ‘folk-beliefs. The first of these is that since a “just” theory of syntactic structure should automatically assign phrase structures to all acceptable and possible sentences in a language, and since it is not possible to do so if both the orders of S and N are taken to be basic, one of them must somehow be selected as “the” underlying word order. If an analyst still persists that both of them are equally good candidates for

Page 204: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

203

the phrase structure derivations, then one is forced to draw either of the two conclusions: (60) i. The claim made by the Transformational Grammar that syntactic elements are ordered in the deep structure in false, ii. The analysis has a lacuna. The normal line of argument that is taken is this: (60) iii. Since to think of (I) to be true is absurd, (ii) must be correct. 4.2.4.6.2. In the case of Maithili ( and Bengali), one can also argue that since this language needs both rightward and leftward movement rules (such as extraposition), one of these can be shown to be deriving from one of the two possible orders- (S) N and N (S), whichever is taken as non-basic. One may argue that if by such minor manipulations in specifying the scope of a rule one can save the claim of the TG grammar referred to in (60)I, it must be allowed by the rule of the game. It would seem all the more favourable if is also confirms to the pattern of SOV languages showing a (S) N structure. Then one can also bring statistical and “intuition” arguments hard to disprove. But all such arguments seem to be futile because they are based on the assumption that the theoritical position of the TG grammarians regarding the nature of the “deep structure” ( or for that matter, the notion of deep structure itself) is not vulnerable, and that it cannot even be questioned. But the arguments in the chapter III and this particular case show that it is not so vulnerable as it was thought to be. The nature of movement rules and of sentential complementation will be discussed in greater detail in chapter VII and therefore we ahall move on

Page 205: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

204

to a discussion on the interrogation in Maithili. 4.3.1.1.1. The open question sentences are marked by K-question words in Maithili and the yes-no question marker is ki in both Maithili and Bengali (op. Hindi Kyaa. Assamese iame/no). However, like all other Indo-Aryan languages of today, one can use a rising intemation in yes-no question sentences of Maithili8 instead of using the marker ki. The k-question words include a set of markers depending on which elements is being questioned and what is its grammatical status. The rising intenation is to be found occurring once in K-question sentences but the question markers are never deleted from these sentences. Thus depending on what they replace, these K-questions can be classified into the following types; (61)

4.3.1.1.2. These can be further divided into various sub-types such as follows:

Page 206: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

205

4.3.1.2.1.Before I explain some of the terms used in the classification presented in (62) which may seen unfamiliar, it would not be a bad idea to list the question markers that form the basis of such a typification as above. This will also remove a let of unclarity otherwise associated with the terms. For the sake of comparison, parallels have been drawn from the other Eastern Indo-Aryan speeches such as Bengali, Assamese and Oriya. The Hindi counterparts of these interregative pronouns have also been included here (;in the following list, B.A.O and N stand for Bengali, Assamese, oriya and Hindi, respectively ): (65) a. ke “who” (B. ke. A.kem /kome, o.kis,n.koem);used as human active subjective simple nominal K-Q. b. ki “what” (B,A. ki, O.komo, N.kyan);used for many of the categories mentioned in (62) a through

Page 207: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

206

(62) c, which include : complemental nominal K-q, non-human active and passive subjective simple nominal K-Q, and as non-human transitive and direct di-transitive objective simple nominal K-Q. c.ken :which” (B,A. ken,O. ke n m, H. kOom): used as a demonstrative adjectival K-Q. e. kehan :Now (quality)” (B.kD en, A.kene/ kene kuaa O.ki pori, N. kBsan); used as qualitative adjectival K-Q. f. katac/kal “Now (quantity )” (B.kOta/kOe/kO,A.kinan/koi. O.ketta, N. kiyanaa0; used as quantitative adjectival K-Q. g. kenan/konan kaP/ken tar n h v n “Now (manner)” (B.ki kera/ ki khaabe, A.keme koi, O. kemiti, N.Kbse); used as a manner adverbial K-Q. h. kiy Nk /kathil lel a /ken kaar a b E “why” (B.kBne/ki jumpe / ki kaarepe/ ken kaarane. A.kele /kio, O.kias / kaahi ki, N.kta n/ kis jliya / kis kaaral ) : used as reason adverbial K-Q. i. kokhom “when “ (B.kOkhem,A.ketian, O. kebe, N.kab); used as an unspecified time adverbial K-Q. j. kahivaa/ ken din “Which day” (B.koba / ken din, A.kahani, O. kette beRe, N.kis din ); used as a specified time adverbial K-Q. k. kayaB “where” (B.keythaas, A.keyt. O. knnBe. N. kahpa n ); used as an unspecified adessive space adverbial K-Q. l. ken Thaam “which place “ (B.kenkhaama / ken jaaegaae. O. ke n urthii, H.kis jagha ) used as a specified adessive space adverbial K-Q. m. kinkaa “when” (B. kaa n ke. N. kinka / kinbe ); used as honerific human transitive and honerific active indirect di-transitive objective simple nominal K-Q. n. kakraa “When “ (B.kaake, A. kaak, O. kashaaku, N.kiske );used as non-honerific human transitive and non-honerific active indirect di-transitive K-Q. o. kinak “whose” (B.kaa n r, N. kinkaa ) : used as honerific genitive simple nominal K-Q.

Page 208: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

207

p. kakar “whose” (B.kaar. A.kaar, O. kashaarO, N. kiskaa); used as non-honerific genitive simple nominal K-Q. q. kinkaa dwaaraa / kinkaa so n “By whom “ (B. kaa n r daaraa/kaapr/ sashaajje/kaa nke dive, N. kinke dyaaran); used as honerific passive subjective simple nominal and hinerific human instrumental adjectival K-Q. r. kakraa dwaaran / kakran op n “By whom “ (B. kaar daaras/ kaar saahaajje/ kaake dive. A. Naare /kemera/kaar betuaai/ kaak di, O. kahaaddyaara. N. kiake dyaaran ); used as non-honerific passive indirect di-transitive objective simple nominal and non-honerific human instrumental adverbial K-Q. s. kii so n / kathil so n “From what “ (B. ki theya /ki heys, A. kibor poram, O.ke n uthn, N. kis so ); used as inanimate ablative space adverbial K-Q. t. kinkan so n “From whom “ (B. kaa n r (kaach ) theke,kothan hots, A. kor poram, O. ke n uthu/ ke n uthaaru, N. kshaa n ne ); used as inanimate ablative space adverbial K-Q. u. kii dhar I /kathii dhar I “Upto what “ (b.ki (/ken) ebdi (/parjents ), A. ki ioika, O.ke n n perjants/ kepparjanyts, N.kyaa tak / kabaa n tak ): used as non –human animate allative space adverbial K-Q. v. kinkaa dhar I “Upto whom” (B. kaa n r kaache / kaa n r kaach ebdi, N. kin tak);used as honerific human animate allative space adverbial K-Q. w. kakraa dhar I “Upto when “ (B.kaar kaacha /kaar kaach ebdi, A. kaar loike, O.kaahaaku, N.kis tak ); used as non-honerific human animate allative space adverbial K-Q x. kataB dhar I “Upto where “ (B.kethan ebdi / kethan porjents, A. ker loika, C. ke n uthii, N. kahaa n jek ); used as inanimate allative space adverbial K-Q.

Page 209: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

208

4.3.1.2.2. Notice that there are several features that recur in the diagramme as well as in the list given in (63) a through y. These include : Active –passive, Human-non-human, Animate-Inanimate, honerific-Non-honerific, and specified-unspecified. There si nothing much to talk about these because barring one, vis., [t specified ] all others have already been discussed in Chapter II. As the examples in (63) h,I,j, and k show, in case of time adverbials they are distinct words, while in case of space adverbials the specified form is a compound word. Although an Assamese speaker utilises the concept of specificity in time, he does not do so in case of space adverbials. Assamese, however, leaves some more gaps in the diagrams in (62). For instance, Assamese does not use honerificity in K-question words. Similarly, other Eastern Indo-Aryan languages seen to use the same form for a number of categories. From this, however, one should not draw a conclusion that the typology presented in (62) is too wide, and hence must be rejected. That this is not true is proven by the fact that some of the unusual distinctions exist in other areas of grammar of the same language, or it did exist there at some point of time. For instance, although Assamese does not have any honerific question word, it does have honerific and non-honerific pronouns and even the verb shows honerific agreement pattern in case the subject pronoun is in the second person. Again, in Maithili and other related languages, the pronouns in passive utilise the feature of honerificity, while in active they do not. But in case of active sentences this distinction is maintained in the selection of verbal affixes. Thus it follows from the above discussion that although in some instances a few

Page 210: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

209

distinctions are lost leaving gaps in the sub-classification, it does not mean that these are not expressed by some other mechanism in another area of grammar. 4,3.2.1. Maithili, Bengali, Assamese and oriya involve no obligatory fronting of K-question words as in English. But if one pairs the interrogative sentences with their declarative counterparts, another type of problem energes out of the agrteement pattern of these languages. The person- agreementpattern, which is obligatory for all the Indo-Aryan languages (of. Singh and Bandyopadhyay, 1978; also see chapter V), poses a problem in the formulation of the question rule. Notice that when a subject, object or an oblige pre-form is questioned, in some languages the personal terminations are the only signs left in the body of the sentences that can help in recovering the identity of the pronown that has been questioned. But there are many languages where they are also deleted from the surface of the sentence. Thus, (64) would be ungrammatical in English; (64) *Who am going there ? 4.3.2.2.1. Similar examples could be found in Maithili and other Indo-Aryan languages : (65) a. “han kakraa maar a ban ? I when kill+ fut+Agr (1psu ?2p NH Obj) “when shall I kill ?” b. *o kakraa maaral a thuumh a? He when kill+pst+Agr 93p SU?2p nod Obj) “When did he kill ?”

Page 211: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

210

4.3.2.2.2. These Maithili question-sentences are parallel to the declarative sentences that follow: (66) a. han teraa maara ban I you (NN) kill +fut + Agr “ I shall kill you” b. teraa maar a l a thuunh a No you (Med ) kill +pst+Agr “No killed you”. 4.3.2.2.3. These suggest that one may have the following condition on the I-question rule: (67) The I-question rule is blocked if the NP to be questioned in not in the third person in the corresponding declarative sentence. Such a condition may be used in case of other IA languages also. For instance, in Bengali one finds similar things happening. The ungrammaticality of the following Bengali sentences is a proof of it: (68) a. * ke ekhaane jaabe? Who there go+fut+lp Agr “Who will go there?” b. * ke ekhaane geehile ? Who there go+pst+lp Agr “Who went there?” 4.3.2.2.4. But there are examples where (67) seems to be apparently violated. Consider the following sentences from Bengali and Hinid”: Bengali : (69) a. tumi ekhame jaabe You (Ned) there go+fut +Ip NRD Agr “you will go there” b. ke ekhaane jaabe ? “ Who will go there?”

Page 212: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

211

Hindi : (70) a. nB2 wahan n gayaa than I there go + past +pft +lp mase Agr “ I went there” b. kOen wahaa n gayan than “Who went there?” 4.3.2.2.5. If one looks into these counter-examples more carefully, the reason for the apparent failure of (67) will become obvious. Whenever due to some sort of “terminational syneretion” ( whateverr be the reason for such developments), the personal terminations of third and non-third persons are identical, the subject can be questioned not withstanding the condition (67). This condition, however, works all right in Maithili and Assamese, because these do not show such terminational syneretions. Still, to take care of the Bengali and Hindi counter-examples, one can modify (67) in the following way:

(68) The K-question rule is blocked if the NPs to be questioned agree with the verb in such a way that the personsl terminations in the verb disclose the disclose the identity of the “ marked” pronouns.

By “marked”, I noun the non-third person pronouns. 4.3.2.3. In Maithili as well as in other Eastern NIA languages, the question rule simply replaces the element to be questioned by an appropriate question-word. This replacement, as noted earlier in (62) and (63), is dependent upon the grammatical status of that Particular element. The condition (67)’ can be incorporated into a K-Question Replacement rule as in (71) common to all the IA languages.

Page 213: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

212

(71) N-QUESTION REPLACEMENT RULE SB:V-S[X –Q+{NP/ABJP/ADVP} Pers –Y –V (+Aff) + inf Pers]S –E SX:1 2 3 4 5 6 7 => SG:1 2 X+3 4 5 6 +UNMARKED Once this rule applied thereby introducing the various K-equation words, further rules can move the K-Q optionally to other syntactic positions. Such movement rules will be discussed in detail in Chapter VII, and hence we shall leave the question of movement of question words for the time being. 4.3.3.1. Although the K-question replacement rule as formulate in (71) captures most of the instances of interrogation of nominal, adjectival and adverbial phrases, it leaves out those examples from Maithili, Assamese and Bengali, where the verb is questioned : Bengali : (72) a. aaSan ghunooche Asha sleep+pros +prog+Agr “Asha is sleeping”. b. aaSan ki korche? Asha what de+pres+prog+Agr “What is Asha doing?” Assamese (75) a. assame fotak dikhOn kitaab dise Asha +SU Lata+BO a book +IO give+pres+prog+Agr “Asha is giving a book in Lata b. Aasaao lOtaak ki kOrise? Asha +Su Lata+Bo what de+pres +prog+Agr “What is Asha doing to Lata?”

Page 214: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

213

Maithili (74) a. aaSaa glit a gaab 1 rahul a chath I Asha song sing+ptopl prog + be+pres +Agr “Asha is singing a song”. b. aaSan kii kO rahal a chath I? Asha what de+ptopl prog+ be +pres + Agr “What is Asha doing?” 4.3.3.2.1. In all the examples given above the verb phrases are being questioned. Although it is true that there must be a rule to account for these constructions, such a rule has to be restricted in such a way as to eliminate generating Maithili sentences such as in the b-sentences that follow:

(75) a. han ghare aN {chii/chalahu a / rahab}

I home +EMPH lee {be+pres+Agr/be +pst+Agr/ be+fut+Agr} “I am /was/will be at home ( and nowhere else)”. b. * haa (ghare nB) kii {karait a chii / keelahu n / karab} ? What do/did/will I do (at home)?

(76) * o etaR kii {kaRl a / kaanB} chath I ? “What is he doing here?”

Page 215: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

214

(77) a. shaa n ii beat jaamait a chalahu n jo o sundari You this fact know+ptopl be+pst + Agr that she beautiful clathi Ve+pr +Agr “You know the fact that she was beaufiful”. b. *shann n (ii best ) kii karait n chalahu n jo - - - - ? “what were you doing to the fact that she was beautiful ?”

4.3.3. 2.2. The fact that the b-sentences are unacceptable in (73) through (77) forces one to put a contraint on the rule that would otherwise account for (72) though (74). It seems from the ungrammtical sentences that for a verb to be questioned in Maithili, the string concerned must involve a verb which is [ -static]. Notice that the verbs in these sentences, viz., aeh- “be”, rah- “be” bais- “sit” jann “know”, are not [-static]. Similar example can be found in other Eastern Indo-Aryan languages also. 4.3.3.2.3. However, the same verbs as above can in certain constructions denote non-stationess and in such cases one finds the interrogative equivalents to be acceptable. Consider the following constructions: (76) a. haa sai sO n mass dhar I roj ghare mE Fahab I today from one month upto everyday home+EMPH lee be+ fut +Agr “I would be at home from new onwards till a month’s time everyday” b. haa aai sO n ek mass dhar I roj ghare mP kii karab? “What shall I do at home from now onwards till a month’s time everyday?

Page 216: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

215

(78) a. o eTan prat I tiin din par bais a taah a he here every three day after sit+fut+Agr “He will be sitting here once in every three days” b. o etaN prati tiin din par kii kar a taah a? “ What will he be doing here once in every three days?”

(80). a. o ii heat dhiire dhire jaani jastaah a No this fact slowly slowly know+phopl go+fut +Agr “He will gradually learn these things” b. o (ii best ) dhiire dhiire kii kar a taah a ? “ What will be gradually do (with these things)? 4.3.3.3. The b-sentence in the above mentioned constructions involve the same verbs as in (75) through (77). But there is a difference in meaning if one compares the two instances of their use carefully. In (78) through (80), these verbs denote “the act of being (-living)”, “the act of sitting”, and “the act of knowing”. Thus, in those examples they have become “process” verbs. One can therefore put a condition on the rule that would question the verbal element that the verbs to be questioned must involve “process (or, may be some kind of action). 4.3.3.4.1. One faces some more problems when one actually sits to write the rule that would explain the sentences (74 – (80). One such problem lies in the application of this rule on the verbal sequence. In some Indo-Aryan languages such as Bengali and Assamese different types of verbal sequences are not overtly differentiated. Maithili is more like Hindi in this respect. In Maithili, compound verbs have normally a POLAR + VBOTOR order and the first number of the compound receives a marker : -1. In case of BI-POLAR compounds (see Singh and Bandypadhyay, 1978), the structure

Page 217: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

216 is basically the same with a little difference in that here the two verbal elements are formally seperated by kaa. In Bengali, both types of verbal sequence show –o after the first number of the compound. Meaningwise these two are of course, different in all these languages. The compound verbs show greater cohesiveness and the vector here only modifies the meaning of the polar. Moreover, the compound verbs are generally replaceble by single verb forms of the same language with a slight or no difference in meaning. Finally, only a few verbs can function as vectors in a particular language, whereas any two verbs can be combined to form a bi-polar compound as long as their sequenctial order is semantically compatible. An example each of these two types of constructions would make the above points clear (; for a detailed discussion, see Chapter VI of this dissertation); (81) a. Shaa n k elij a biit o dekh I letaah a Your (non ) things he (Non ) see+ptopl take+fut +Agr “He will look after your things” b. shaa nk eiij a biit o dekh I kaa letash a Your (non ) things he (non ) acc+ptopl BO +pte take +fut +Agr “he will take your things scrutinisingly” 4.3.3.4.2. When one applies the verb question rule on these two types of structures, the results are different, and any formulation of this rule must take a note of that. In case of bi-polar compounds, the first, second, or the whole sequence can be questioned. For instance:

Page 218: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

217

(82) a. shaa n k ciij a biit o kii kO (kee) lataah a “What will he do with your things and take?” b. shaa n ka ciij a biit o dekh I kae kii kar n taah a? “ What will he seen your things and do?” c. shaa n k eiij a biit o kii kar n taah a ? “What will he do with your things?” 4.3.3.4.3. Note that in Maithili, like in Hindi, two successive KAR elements are not preferred and to get a better result the bi-polar compound marker kee should be deleted. This can be seen in (82)a. In case of simple compound verbs, all such possibilities are not found. A parallel to (82)c is alwaysa generable in case of compound verbs, but structures such as (82) can never be found. A parallel to (82)c is always generable in case of compound verbs, but structures such as (82)b can never be found. As Singh and Bandyopathyay (1978) point out (also see chapter VI) this fact can actually be used as a test to find out whether a particular construction is a true compound verb or a bi-polar compound. This test will, however, be of more relevance in Bengali and Assamese where the two verbal sequences do not differ formally. 4.3.3.4.4. Consider the following examples from Bengali: (83) a. chobigule dekhe no picture+pl acc +ptopl take +Imp +Agr “Have a look at the pictures”. b. * shobigule dekho ki kOr? c. Shobigule ki kere me? (84) a. chobigule dekho (ekhame ) aao picture+pl see +ptopl here come +Im+Agr “Look at the pictures (and ) come (here)”. b. shobigule dekhe (ekhaane ) ki kOr? c. shobigule ki kere (dekhanne ) aae? The first set of sentences here has a true compound verb: dikhana:

Page 219: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

218

hence the ungrammatical sentence in (83)b, Notice that (83)b does exist in Bengali. But in that cse dhikane will have to be interpreted as a bi-polar compound. Again, the bi-polar compound dukh+aa in (84) can also be used alternatively as a true compound where it would mean “see +perfect”, Under this interpretation, (84)b would not be acceptable. It may be pointed out that ekhaane “here” has been used in (84) only to show that the sequence of verbs is a bi-polar compound, and that is why it can easily be intervened by an adverb. Such intrusions are not allowed in case of true compound verbs. 4.3.3.4.3. The parallels of these sentence can easily be found in Maithili. For instance: (83)a. kaa n kir n baa Tabul par sO n murtii khasam dot a Child table upon from status fall +cause+ give+ fut+Agr “The child will make the status fall from upon the table”. b. *kan a kir a baa Tobul par sO n murtii khasan kaR kii kart a? c. kan a kir a baa Tobul par sO n murii kii kO do a?

e. kan a kir a baa Tobul par sO n murtii (kB n ) kii karat a? In (85)c, only polar has been questioned and in (85)d, the whole compound has been questioned. But questioning of the vector in (85)b gave an ungrammatical string in the intended sense. 4.3.3.5.1. At this stage, however, it is not known as to why a certain set of compound verbs require that the verb question rule would replace the whole compound and only the whole of it by the interrogative marker plus a dummy verb: kar. In other words, there are some compound verbs which do not allow the polar verbs to be

Page 220: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

219

optionally questioned. It may be worthwhile to examine different vector sequences of Maithili and other Eastern NIA languages carefully to find out whether one can come out with a classification of the polar member of compound verbs on the basis of their sensitivity to Verb Interrogation rule. One may guess that the combinations that allowed this rule to replace only the polar members by retaining their vector counterparts involve these vectors which somehow underwent more ‘lexical emptying’ then the others. The reason may also lie in the greater cohesiveness of the polar-vector combinations, but nothing could be said for sure at this point. Consider the following examples from Bengali whre khaa- “eat” allows the verb interrogation rule to replace itself whereas eOl- “move” of cale-inn “go away” does not do so: (86) a. So miSTi kheya philchile he sweet eat+abs throw +pst +prog+Agr “He was (almost) eating up (all ) the sweets” b. So (miSTi) ki kerchile? c. so (miSTi) ki kar –e pholechile? (87) a. so cele jaaochile he go+abs go+pst+prog+Agr “He/she was going away” b. So ki kerchile? c. * so ki kar-e jaachile? 4.3.3.5.2. Similar examples could be found in Maithili and Assamese. In these languages, structures such as (86) are perfectly acceptable while (87)c would be completely ill-formed in the intended sense. Although at this stage we do not have any

Page 221: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

220

answer as to why (87)e or similar structures are unaccetable, but this much is clear that a Verb Interrogation rule would partly be governed rule - governend by the nature of the polar+ vector sequences involved. 4.3.3.6. To explain all these facts, therefore, one has to have a rule of interrogation to question the verbal complex, and such a rule would be different from (71). It is also clear that this rule could be properly framed only after a careful consideration of different types of verbal complexes. That is, one also has to consider the conjunct verbs along with compound verbs to be able to write the rule properly. However, this rule could tentatively be written in the following way: (86) VERB INTERROGATION RULE SD: V – V[X [VB[ Y – V +Dos]VB }V X SX : 1 2 3 4 5 6 SC : (a) 1,2, ki, KAR + 5, 6 , er (b) 1, 2, ki, KAR+ 4, 5, 6 4.4.1.1. In this section, we shall look into the RCF strategies that are employed in Maithili. It has already been noted at the beginning of this chaper that recently attempts have been made by a number of syntacticians (ef. Keenam of a “WH + S2” type of relative clause was not a must for all languages, but that different languages employed different strategies for relative clause formation and ‘Relative Pronoun (-WM) strategy was one of them. The other strategies included participialisation, or nominalisation, equi-case,

Page 222: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

221

anaphera, gap, non-reduction, verb-coding and word order strategies. Also, there is no reason to believe that all languages would show at least one particular type of strategy viz. Such as the WE-strategy. There are languages like Tankhul Naga (Sachideva, 1977) where WN-strategy is never used. In Tankhul Naga, the most prevalent strategy is that of participialisation. What is more important to understand is that languages may even use a ‘mixture’ of differentg strategies (Given, 1975:61) at a time. Thus, in defining (restrictive ) relative clauses, we follow the “syntax-free” defination of RCF given in Keenam and Courie (1977:63-4) such as follows: (89) “We consider any syntactic object to be an RC if it specifies a set of Objects (perhaps a one-member set) in two steps : a larger set is specified, called the domain of relativesation, and then restricted to some subset of which a certain sentences, the restricting sentence, is tru”. In the following paragraphs,we would at first try to find put as to how many positioned RCP – strategies are employed by Maithili. 4.4.1.2.1 In terms of relative position of the head NP vis-à-vis the restricting clause, Keenam and Courie (1977: 65-5) talked about three possible strategies : (1) Postnominal RCP (head NP +S2). (2) Prenominal RCP (S2+ head NP),a nd (3) Internal RCP, where the head occurs within the restricting clause. We may add to this list a fourth strategy which could be called Displacement strategy where the head NP and the restricting clause are intervened by different kinds of sentential elements. Maithili has, apparantly, all the four strategies, the examples of which follow

Page 223: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

222

(90)a. ha pane-sab-Taa-kB n niik jakaa n jamait a chaligah I I panch +pl +dot +DO good way know+pte aux+pst+Agr “I know all the panch very well”. b. mame-sab [jan I kaa –sab – kB n han miik jakaa n - - - -] Panch + pl Rel +pl +DO I good way- - - - “All the Panch when I know very well” (91)a. o mukhiaa hamraa khuub Daa n Tal a thiiah a that village chief + I + DO much reprimand+pst +Agr “That village-chief reprimanded me much”. b. [jo hamraa khumb Daa n Tal a thiink a ] o mukhian Rel I + DO much reprimand + pst + Agr that village chief “That village-chief who reprimand no much”. (92)a. o ekTan beRad kiimal a thiiuh a he(hon ) one+dot ex buy + pst +Agr “He bought an ox”. b. o ekTan is beRed kiinal a thiiuh a hamran possin he (hon ) one+dot Rel on buy +pst+Agr I +obj like hah 1 raRal a not fall + pst + Agr “ The ox that he bought was not likable to me”. (93) a o baalak kaalh1 etao khasala ehal a that boy yesterday here fall+p.te aux+pst+Agr “that boy fell off here yesterday” b. [jo kaalh I etam khasala ehal a Rel yesterday here fall+p.pte aux+pst+Agr I Eh baalakak bhasi tjutahu a That+ebl boy+gem bother be+pres +Agr “I am the brother of that boy who feel off here yesterday”.

Page 224: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

223

4.4.1.2.2. One can find out from the above examples that (90)b, (91)b are undisputable instances of Postnominal, prenominal and displacement RCP strategies, respectively. There may be some doubt whether (92)b really shows as, for instance, Banbara or Diogene would show Notice that in (92)b, the matrix sentence does not have a head noun. Therefore, one can argue htat (92)b has undergone some kind of deletion-strategy and that it is not relevant for a discussion on Internal RCF here. Also notice that one can, of course, alternately use the noun balmi with the correlative so before it in the matrix sentence, or one can simply use the pronoun as there, as the following examples would show: (94) I ekTan jo baRai kunala thiinka so (baRed) hamran pasiin nahI paRala. 4.4.1.2.3. But interestingly (94) could be altered in such a way that the relative and correlative markers interchanged their position, and in that case such a sentence may be described as an instance of an Interval RCF. Consider the following sentence: (94) o ekTan no baRed kkinala thiinka jo (baRed) hamran pasiin nahI paRala It may be mentioned here that (94)’ is meaningless not parallel to (95) which follows (95) [jo (baRed) hamran passiin nah 1 poral a ] o {*ekTan} so baRed kiinal a thiink a “He bought that or which (or) was not likable to me” 4.4.1.3.1 The presence of positional (or) strategies of RCF in a particular language, however, does not indicate that it does not

Page 225: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

224

involve any of the case strategies. Rather, quite ofter it is the case that they go head-in-hand. In this and the subsequent paragraphs, we would look into the various case- strategies employed by Maithili. The first among these is the ‘non-reduction’ strategy, examples of which would include the following sentences: 90) o [jaah1 pane –sab –KBn ham niik jakana jamaita chaliaahI ] o pane-sab . . . . 91) o [ jo mukkian hamran khuub Daa r Tal thiinha] o mukkian 92) o [jo baalak kaalh1 etae khasala ehala ] ham ej1 baalakak bhaai thikahun In all these examples, which are comparable to (90)b, (91)b and (93)b respectively, the nominals being relativised (some, mulhian and baalak) are occuring twice. The typical examples of non-reduction strategy (given from Hittite and Bombare; of. Given, 1975:62) show that the nominal concerned in atleast present withing the restricting clause, although it may or may not be presen in the matrix clause. Here, both of them are present, but one can if one wants to, delete the second occurance of name-sab, mukhian and baalak in (90)c. (91)c and (93)c, respectively. Another fact about the non-reduction strategy is that it is found to be occuring mainly with a subsequent shift of the restricting clause to a Topic position (which may be the sentence-initial position as in Hittite and Maithili or the sentence-final position as in Bambare). This is borne out by the fact that, for instance, (0)c would have been a little clumsy (although it would be acceptable had it retained the order of elements as in (90)b. The construction, in that case, would have been as follows:] (90)d (e) pane-sab [ jaaha pane – sab – kBn ham niik jokann jamaita chalisahI ]….

Page 226: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

225

It may also be mentioned here that the selection of a particular shape of markers i.e./ jaah I and a/ eh I as in the sentence given earlier depended on whether the following nominal (which the marker is attached to ) is in subject or non-subject relation. 4.4.1.3.2. The ‘gap’ – strategy is typically shown by Japanese where the noun being relativised is not present in the restricting clause nor is the relative marker or pronoun found there in its place. In restricted number of cases, Maithili like in other IA-languages employs this strategy. Consider the following sentences: (96)a. [ham {jaah I murant-kB n / jan I kan} eiTThii likkaB I. {which guy +obj / when} letter write+ptopl chalisahI {o / no} murrent kyee dos a re rahathI be+pst +Agr that guy someone other +Emph be +pst +Agr “The man I wrote the letter to was a different guy. “ b. [han ? ciTThii likhaP chaliaah1 ] {o oo} murent kyee dosa re rahathI Here, in (96)b the ? indicated the place of gapping, and if one compares a- and b-sentences as above, one easily find out that not only the nominal (current), but also the pronoun (ie-) could be gapped optionally. Aa seems to make the resultant sentence better probably because it is assiciated with the relativisation process as a correlative and hence makes the recoverability easier. 4.4.1.3.3. Obviously, the ‘relative-pronoun’- strategy is the commonest of all, and is widely attested in the Indo-European family of languages. (90)b, (91)b, (93)b and many other examples given in the

Page 227: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

226

earlier paragraphs show this strategy which involves at least (I) replacement of the relativised nominal in the restricting clause by a ia- pronoun, and optionally (ii) attachment of a corresponding e-/e- (correlative) pronoun or marker to the head nominal in the matrix sentence. These markers, as it has already been observed (of 4.4.1.3.2.), also carry the case-marking of the deleted nominal. (96)a shows another interesting thing in this respect. If the nominal to which io-/e-/e- is attached is present on the surface, these markers would still be inflected to show whether the noun concerned holds, SU or non-SU relation in the sentence, but when this noun is deleted the pronominal markers not only take up the case-marking of the deleted noun (aa –kan in jam I kan which is a variant of –kB in muruut –kB n ), it also takes up the appropriate honorific/number marking. For instance, in (90)b, the plural marker –sab as well as the case marker – kBn were attached to in-pronoun. But both (90)b and (96)a show that the pronoun in- has selected a –a- shape (as in ie -u 1 kaa ) rather than taking up a –k- marking (as in, say, in-k-a kan ) which is used mainly for the non-honorific human nominal. This points out that when the relative pronoun RCF strategy is used to replace the affected nominal concerned with a pronoun, the latter also copies mentioned above, same (in 90b) and murrut (in 96a) were marked [ +honorific], and hence –a- forms were selected in place of –k- forms. 4.4.1.3.4. ‘Anaphoric pronoun’ – strategy occurs commonly in Arabic and Hebrew in case of object-position RCP, the restricting

Page 228: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

227

clause shows an anaphoric pronoun in place of the deleted nominal apart from a relative marker attached to the embedded clause. If one trics to do such things in English, one would come up with the following constructions: (97) a. Ramesh gave a kiss to Rakhi b. The kiss that Ramesh gave to Rakhi c.?? The kiss that Ramesh gave it to Rakhi Maithili employs this strategy only in a limited number of cases mainly for the purpose of disambiguation. Consider (98) where ‘b’ is the simple relative equivalent (in the object nominal position) of ‘a’ (98) a. ham eh I laRkaa-kR n jamait a challiaj a I that +obl boy +Obj know +pte be +pst +Agr “I know that boy” b. {jaah I laRkan-kB n / jak a raa} han janait a chalaik a (o /o laRkan} --- “The boy when I know, he - - - -/ “When I know, that boy - - - - -“ Now, if (98)a is embedded as a complement clause of another sentence as in (99)a, the resultant RCF operation may give rise to an anaphoric pronoun placement as (99)b shows: (99)a shan n accait a chalah n [ jo han eh 1 ---- chaliaij a ] You (han) think +pte be +pst +Agr that “you thought that I know the boy” b. jaah I laRkan – kB n shan n secait a chalahu a [jo ham {ek a raa/o } jamait chaliaik a ]- “The boy when you thought that I know him - - -“

Page 229: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

228

At present, I do not know of any situation in Maithili other than the one above where anaphoric pronoun strategy could be used. 4.4.1.2.5. ‘Verb-coding’ strategy is very common in the Philippino and a few other Malayo- Polynesian languages and in some Bantu languages, as given (1975:66-7) tells us. His examples of a parallel of the use of this strategy in English include the followings (100)a. I worked with the boy. b. The boy I worked with. This is, of course, only an approximation as to what happens in the languages where verb-coding is a productive strategy. But Maithili does not employ this strategy at all. 4.4.1.3.6. If the syntactic rule of the head noun is the same in the main and subordinate clauses, in certain languages, one needs case-mark only one of them and once only. There are languages which use the case-marking twice. An example of such a double case-marking could be given from English (Bernard Courie, personal communication): (101) John is sitting on the table on which there is a vase. Modern Hebrew (Given, 1975) and Tamil are languages where in such cases of double case-marking, one of the markers may be dropped. This strategy, known as the equi-case’ –strategy is only abbreviatory in nature, and is rarly found. Even if it is found to be production in a language, it is always matched with other RCF strategies. It is inoperative in Maithili.

Page 230: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

229

4.4.1.3.7. The ‘participialization’ – or ‘nominalization’- strategy is a very common method of RCF in the Indian languages, and we have already noted that in some Tibete-Burman languages in India, this is the commonest strategy of all, In such cases, the verb in the restricting clause appears in a non-finite nominalised shape, and ‘normally’ the subject nominal receives a genitive case-marking. Consider the following sentences given earlier where nominalization-strategy could be applied now: (90)a. han pane-sab-Tan-kB n niik jokan n janait a challisah I “I know all the Punch very well”. b. ham-ar pane-sab-Tan-kB n niik jokan n jaan-ab I + gen know + inf “My knowing the Punch very well….” 4.4.1.3.8. Given (1975:64) reports that a ‘word order’ – strategy is used “for only subject and direct object relativisation, in a language in which the position of those to argument before and after the verb is rigidly grammaticalised”. SVO vs OBV – order in the Subject and Direct Object positon relativisation (respectively) in English was given as an example of this strategy. But we have already seen (4.4.1.3.1) that Maithili ( and many other NIA languages ) use all positional – RCP strategies, and it is an acknowledged fact that the IA languages have relatively “free”- word order. This strategy is thus not utilised by Maithili strictly. Notice that normally, gap and word-order strategies go together in a language so far as the recoverability is concerned. When one of the affected NPs is gapped, the clue for recoverability is left behind either in the word order or in the case-marking of the

Page 231: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

230

relative pronoun. This is what happens in English or even in Maithili (of 4.4.1.3.1. for discusstion on possibility of applying such gapping on 90c, 91c and 93 c ) . But in some Altaic languages and in Japanese, the use of gapping strategy is stretched to an extreme in that the coreferene noun in the restricting clause is deleted without leaving any trace or clue. It is thus an exactly opposite to the non-reduction strategy. 4.5. In this chapter, we have discussed the nature of relative clause marking in Maithili and a number of other languages. In the beginning, the relative markers in Old and Middle Maithili were given with examples from the Garrandaada and VarNaratnaakara. Then the Modern Maithili relative and correlative markers were briefly discussed. This was followed by a report on some recent and ongoing studies to find out whether the marking-similarities between open interrogation and relativisation rules could be arranged in a quadrangle where complementation and yes/no –questions could also be placed on a different axis. After this discussion, the history of studies in the field of sentiential complementation was outlined which was followed by a detailed study of different types of complement constructions in Maithili. The governmental nature of the process of complementation and its relation vis- a – vis ‘factivity’ were also touched upon . At this point, the question of deep word-order and the debate on the phrase structural expansion of Nps in Indian languages were also dealt with in the light of Maithili. Then the process of interoggation was taken up and a classification of K-question words in Maithili, Bengali, Oriya and assames was presented. This was succeeded gy a different discussion

Page 232: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

231

on two types of K-question – one that involves the verbal complex,a dn the other that involves the non-verbal categories- and the rules for both the types were formulated. Some unresolved problems are also pointed out in this respect. Finally, a number of RCF strategies were defined and it was shown that except a few rare types, Maithili employed all the other strategies, which included the positional-strategies of four types and the case strategies such as non-reduction, gapping, relative pronoun and anaphoric pronoun.

Page 233: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

232

Notes to Chaper IV 1. Nilratan Son (1973): io-meNa go-ara. 2. Shastri (1916) : inaku 3. Shastri (1916): inati; inati; Son (1948) : juti: Bagehi and Shastri (1956): tugati ;

Mukerji (19643); tuati. 4. Shastri (1916) : inalai 5. Shastri (1916) ; naahamatan hara : Son (1948), Bengali and Shastri (1956);

naaham,a taahara ; Mukerji (963) ; anaaha na taahara, 6. Son (1948) : who (I) 7. Shastri (1916) ; dis ; Son (1948), Bengali and Shastri (1956); Diss; Mukerji (1963)

; diing : Nilratan Son (1973) agrees with Mukherji (1963) in this respect. 8. Son (1948); baaN-eihna-ramha ; Mukerji (1963) adopts the same reading but

deletes the hyphen marks, and has shastri (1916); baamasihaa ramha. 9. Bagati and Shastri (1956) ; Kynya. 10. Shastri (1916) ; kaabackeia Son (1948) l kan back ein l Bagathi and Shastri (1956)

; kaa vaak eial Nilratan son (1975) ; kan-a book-a cia. 11. Original, Bagahi and Shastri (1956), Nilratan Son (1975); Nanaa-a: Shastri

(1916)l samagya. 12. Shastri (1916) ; india (med)nonnaha, because the commentary had Rayandriyakan

l Sahidullah (1940) had the same readings ; Mukerji (1968) ; indiabaNa; Bagehi and Shastri (1956) India navalla.

13. Bagehi (1938) l hei 14. Shastri (1916); Na Than ; others had NaThaa. 15. YarNaratnaakara is socio-cultural discourse and Paariiantaraha is a play 16. Dasgupta (in preparation) gives the Bar-Structure (involving N, A, P, and V as

another instance of such a bi-axes quadrangle.

Page 234: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

233

17. In some of the neighboring languages, the Q-clause quadrangle show more similarities in the markers. For instance, the quadrangle in Marathi (Hock, personal communication) would look like as follows:

The Q-clause Quadrangle in Marathi But Onam Comp Onam Non-Question doo, io, etc. kii (+S2), or nhaNum (S2 + ) Question: kon, kaay, etc k an Notice that nhaNum may be excluded from the quadrangle as it is parallel to Oriya bali and Bengali bole-complementiser, all of whic h are basically quatative in nature (of. Patnaik, 1976, 1978; Kachru, 1979; Singh, 1978). The quadrangle for Hindi would be very much similar to that of Marathi. Bengali would, however, show a quadrangle which is identical with Maithili.

Page 235: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

CHAPTER V VERB AGREEMENT

5.0. In this chapter I propose to examine some generalizations concerning verb agreement rule first and then discuss in detail the nature of agreement rule first and then discuss in detail the nature of agreement rule in Maithili. I shall attempt to show that a large number of languages of the worlk “tend” to show some regularities in respect of verb agreement systems attested in them. I shall then argue for and against such regularities and show that those can, at best be called “tendency universals” I could be concluded from this discussion that such tendency universals have nothing to apologise for, and that those ought to be encouraged and accepted as working hypothesis if the research in the field of language universal and language typology has to progress. 5.1.1 A verb is said to be in agreement with a nominal argument when some or all of the grammatical categories of the nominal is reflected in the conjunctional pattern of the verb concerned. Verb Agreement is not a universal phenomenon because there are languages such as Igbe, Tankhul Naga and Thai which do not show any sign of agreement. The categories that may show up in the verb include person, number, gender, and honorificity. In languages like Sanksrit, Latin and Serbe-Croation the nominal arguments triggring agreement in verbs may “optionally” be deleted. Again in certain other types of languages such as in some Austro-Asiatic tongues (especially in Santhali, Nahle, Birhar and Dhangar), the verb seems to copy the overt shape of triggering pronouns which are

Page 236: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

235

deleted by an “obligatory” pronoun deletion rule later. It must also be mentioned here that not all languages which have verb agreement show all the four categories in the verbal complex. Similarly, a language may not show rigid agreement pattern rhroughout its grammar. Thus, in Sanksrit, participles show gender distinctions, but finite verbs do no. In kannada, verbs do not show agreement in negative sentences (Hiremath, 1961). In Maithili too, the gender distinction is now confined to certain types of intransistive constructions. The nature of gender-distinction may also differ from language to language. Thus, Hindi has a masculine/feminine distinction whereas Gujarati and Marathi have the nouter in addition (Mistry, 1978). Sometimes, the same marker is employed in a language to show more than one category. For example, in Sanskrit (Jha 1958) and Chichewa (Watkins; 1937), the markers denoting plurality are used to show honour or politeness. This is the case in present-day Hindi, too. Sometimes, markers showing more that one category are fused in such a way that they become inseperable. However, if a language does not show a particular category in the verbal complex, it does not indicate that the distinction is not made in other areas of grammar. Finally, although in most of the languages, subjects tend to trigger and control verb agreement, there are instances of non-subjects such as DO, IO and even OO controlling the same (See Woodbury, 1975; Smith-Stark, 1976; Bani and Kickeid, Unpub; Singh, 1976b). It may be mentioned at this point that Verb Agreement, in my characterization, does not include those cases where a language selects different sets of verbs to agree with the

Page 237: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

236

honirific levels of the argument. Instances of such cases are found in Thai (Nuangail, Personal Communication) and Tibetan I (Grieroon, 1906c). Morever, there are also instances where the nominals, especially the pronouns, agree with the categories of verbs. Aginsky (1935) reports that in Hondo, one finds different sets of pronouns to be used when the verb is in any one of the following categories; Simple Present, Present Continuous, Perfective, Imperative and Nortative. 5.1.2. I shall now discuss different types of agreement systems found in human languages across the world. It is interesting to note that geneologically related languages do not have the same agreement pattern in all cases. There does not seem to be a satisfactory answer as to why, how and at what point two or more sister dialects or languages develop different patterns of verb agreement. For instance, if we look into the Non-Banta languages of North-East Africa (Tucker and Bryan, 1966), it would be evident that while the Neru – Nangbetu and Londu) and Bonge-Bagirai group ( including Barirai, Sara, Kara, Central Krosjh and East Benga) mark numbers in the verbal complex by overt markers, the Kba languages (in particular, Nba, Edunga’, Denge and Na) make the singular-plural distinction by the height of tone mainly. On the other hand, Sare-Nundu2 and Benda-Obaya3 languages do not make this distinction at all. If one turns towards the gender-marking in those tongue, an altogether different pattern emerges. While Noru-Mangebutu, Sande, Sere-Nundu and Banda-Obya show no gender-agreement, the Kba languages (except in Edunge) do show such

Page 238: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

237

agreements, Again, among the Benge-Bagirai group, only Benge has gender-marking in verb. However, these languages show different gender-patterns such as Masculine/Feminie (Cushitic, Bari-Lothuhe, Kba), Masculine/Feminie/ Nouter (Kaduli, Kronge), Masculine/Femine/Place (Nasai), Masculine/Feminine/ Common4 (tese) and Animate/Inanimate (Na). In some of those, gender can be distinguished only in the singular. 5.1.3. Now, if we look into the Indo-Iranian languages, the same types of differences can be found. Thus, while Hindi and Punjabi show all the four possible categories of agreement, Gujarti, Marathi and Sinhalese mark only Person, number and Gender, and Oriya has a Person-Number-Honorific system. Other Eastern NIA languages do not show any similarity with Oriya, as Maithili has a Person-Gender-Honorific pattern and Bengali and Assamese have only Person-Honorific marking. Similarly, in Old Persian the gender-distinction is missing whereas in Sanskrit it is present at least in the participials. Greek is nearer to Sanskrit in this respect, but Hittite does not show gender distinction in nouns. In Gypar (of. Bhatia, 1963) again, all the three categories are marked. In some languages different styles have different pronominal patterns. Chao (1968:639) notes that in Chinese, especially “in traditional opistelary style, there is a whole system of humble and honorific forms consisting mostly terms of address, instead of the usual personal pronouns. In actual speech- - -- the forms are limited to a very few.” Such differences in the pronominal system may be reflected in the verb agreement pattern also.

Page 239: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

238

5.1.4. The copying strategy seems to be productive in many language to mark the category of person. It however remains to be decided as to at what point would one say that the “copied” element is a person-marker and not merely an allomorph of a personal pronoun. Consider the case of Austro-Asiatic languages. Here, the personal pronouns and their “shadow” do not differ very much. In most of these languages, the difference is restricted to the first person singular and third person singular pronouns and their reflections. In most of these, the difference is noticeable in the latter instances only. These shadow pronouns are preceded by a marker : -aa ~ -at ~-cd. But while in Santhali, Mahle, Birhar and Dhangar, the order of elements is Verb-Tense Particle – “Shadow”, in Mundari, Kerva, Kerku and in some other Austre-Asiatic tongues, the order is “Shadow”-Verb-Tense-Particle5. In some such as in Mahali, Kharia and Savara, these four elements are fused in such a way that it is almost impossible to separate them6. 5.2.1. Inspite of the difference in the nature of verb agreement, if we look into the agreement systems of the languages belonging to different language-families of the world, one point of similarity does show up. This similarity (of Singh. 1978c) has to do with the selection of a particular type of agreement pattern from among the four possible types: Person-Agreement, Number-Agreement, Gender-Agreement, and Honorificity-Agreement. We found that those four types form some sort of hierarchy where the Person-Agreement pattern is on top, all other types being clustered under it. This generalization could be expressed in the following way:

Page 240: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

239

(1) THE VERB AGREEMENT HIERARCHY (VAN): If a language has verb agreement, it will ideally show, at least, a Person-Agreement pattern, and it may, in addition, show any, some, or all of the other agreement-patterns that involve the categories of number, gender and honorificity. If VAN as formulated in (1) is absolutely true, then we would expect that there is no language where one can find only a Number, Gender-, or Honorific-agreement pattern. Similarly, we should not expect that there are languages that show only Number-Gender or Number-Honorific or Gender-Honorific or Number-Gender-Honorific agreement-systems. A search for the confirmation of (1) through 94 languages belonging to different language-families showed that there are the following types of agreement patterns. (2) i. No agreement : Igbe, Tankhul Naga, Thai, Tibetan, Kolekuma (I jo), Bai, Biri, Bende-Gbya7, etc ii. P(erson) : Diogueno, Tiddish8, etc iii N(umber) : (Toda)9 iv. G(ender) : - v. H(onorific) : (Japanese)10 vi. P,N : Bruhui, Bagirai, Baka, ‘Boli, Central Krosh, Chuvash, Butanian, English, Fanti, Finnish, French, Greek, Gheri, Hittite, Hungarian, Karek, Kabba, Kara, Lappish, Londu, Hangbutu-Efe, Mangbetu, Avukaya, Kolike, Logo, Lugbara, Luluba, Modi, Leri, Nerokode, Hondo, Nordvin, Navahe, Nos Perce, Nde, Edunge, Eyanusa, Old church, Slavanic, Old Irish, Old Persian, Ostyak, Selkup, Shona, Spanish, Sara, Tajik, Vegul, Yurak, Siryone,

Page 241: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

240

etc.11 vii. P,G12 : - viii. P,N : Assamese, Bengali13 ix. N. G : (Modern Mongolian)14 x. N.N : - xii. P,N,G : Arabic, Benge, Chicheva, ‘Denge, Gujarati, Gypay, Hebrew, Irula, Kannada, Kod, Kota, Na, Malayalam, Naltese, Marathi, Nba, Nomonini, Mid. Mongelism, Ndwaga, Penga, Sinhalese, Tamil, Tulu, Gande 15 xiii. P,N,N ̀ : Burmese, Nahuati, Oriya16 xiv P,G,N : Maithili17 xvi. P, N, G,N : Hindi, Punjabi18 5.2.3 Notice that the Austrie languages of India would fall in the sixth column in the typological index given above. It may be pointed out here that there are certain counter-examples to this “tendency” universal in (1). These have been yet within parenthesis in the table in (2). At least two of them had at one point of time a system that confirms the statement quoted earlier. These are Toda and Mangolian. There are ample evidences to show that at the Prote-Toda stage, the agreement- Pattern included Person also (Krishnamurti, 1975). And, the column twelve shows that Middle Mongolian had a person-number-gender pattern as against Modern Mongelian’s number-gender system. This does not seem to be the case with Japanese honorific marking (Pridcaur, 1970) But

Page 242: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

241

since VAN is not an absolute generalisation, falsification in the sense of Popper (1965), the comments such as “every putative exception to a generalisation is an admission of failure” (Green, 1974a:4) assumes special importance. Such a position implies that universals must not be falsified by the actual diversity of languages which, in turn, demands that language universals be absolute. The belief behind this is that only then these statements can account for the “rapidly and uniformity of language learning” (Chemsky, 1965:28). Although it was considered for a long period to be the ideal sort of characterisation of ‘ Language Universal’ , there has been all along serious studies in this field where the findings have been of “statistical” nature. The studies done by Greenberg 1963a) and Keenan and Courie (1972:1977) are good examples of such a probabilistic approach. These and some other universals have been claimed to be “paradigm cases” which do not become falsified in the event of a counter-example or two, because the latter are incorporated within the theory as deviations (Courie, 1974). Moreover, many of these universals are further weakened to be made “implicational” which take the form such as “if X is found to be occuring in a language, I would also occur”. According to Greenberg (1963a:73), “ there is nothing to apologise for in such results”, because “the lowest level laws as described in manuals of scientific method take precisely this form”, and even the “non-implicational universals----- are implied by definitional characteristics of language”. Moreover, the assumptions that the proponents of the absolutive nature of universals take for granted have no empirical consequences

Page 243: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

242

and hence must be rejected. This assumption, namely “that language vary little in their deep structure, though there may be wide variability in surface manifestations” (Chemsky, 1966b :54) is again based on another assumption that the existence of deep structure is a “fact”. The arguments against such claims have already been presented in chapter III. It may, however, be mentioned here that Chemsky here makes the same mistakes as Ross (1970) is to make a few years later in formulating the rule of “gapping” (ef. Bekman, 1976:39). Now, note that Keeman and Courie has to say about the strategy they employed while working on relative clause formation: “We are attempting to determine the universal properties of relative clauses (RCs) by comparing their syntactic form in a large number of languages. To do this it is necessary to have a largely syntax-free way of identifying RCs in an arbitrary language” (Keeman & Courie, 1977:63). I interpret this “freedom” as a way to get rid of the Chemskyan direction of research in the field of language universals 19. Evidently, in this case they used the so-called surface manifestations and yet came up with one of the most important generalisations of the recent time. Notice that the “tendency” universals, as against the absolute universal statements have a weaker falsifiability criterion. VAN is, therefore, no exception to this rule. 5.2.4. One can argue that the “ tendency” universals somehow make the statements non-falsifiable in the sense that even if a good number of languages are found to be violating these tendencies, the universals are not falsified. Thus, the argument goes, that these statements would be abandoned only if at least 51% of the

Page 244: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

243

world’s languages could be shown to be going clearly against them. This is evidently and impossible task given the number of undescribed languages of the world. Therefore, the question is : should the tendency universals be rejected because they do not meet the falsifiability criterion in the sense of Popper (12963) ? To my mind, the answer is yes and no at the same time. Yes, because a non-falsifiable generalization may be called a superfluous generalization. No, because if those universals are not accepted as working hypothesis, it would become very difficult, if not impossible to carry out research in the field of language universals and language typology, given the present state of knowledge. Moreover, cross-linguistic surveys allow one to make only “tendency” statements. A “strict” falsifiability criterion would lead one to frustration in this case because “much more often than not, the generalisations proved to be as elusive as butterflies. It seems that for every reasonably well-justified hypotheses there would always be some forms it failed to predict, or it would predict forms which never occurred, or both” (Green, 1974a:1 ). It would, therefore, only be fair if the axe of falsifiability is applied sparingly on the tendency universals such as Verb Agreement Hierarchy (VAN). 5.2.5. One final question that is generally asked regarding the kind of statement that is made while formulating a universal such as VAN is : What does a statement such as VAN prove? Or, how much does one gain in achieving the goal of constructing a Universal Grammar by such statements? The justification of a condition, constraint of a universal in terms of case of language

Page 245: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

244

learning provides an answer to this question. Thus, after describing the impact the universal of Relational Hierarchy would have on the Cereferential Deletion, Johnson (1977b) at one point claims that there would be 726 possible languages types if the statements of relational hierarchy, double chaining constraint and cereferential deletion law20 are taken together as against 1,64,432 possible types without them (see chapter VI for a similar claim about “main verb-vector” combination in Maithili). No then goes on to argue that “from the standpoint of language acquisition, on can see why such constraints might exist. Without strong assumptions such as those regarding what particular grammars can be like, a child, on the basis of a limited amount of primary data, would have a hard time constructing a grammar of the language he is attempting to learn since there would be a tremendously large number of possible grammars compatible with the primary data” (Johnson, 1977b : 171). 5.3.1. Another recent generalization concerning verb agreement that has already been mentioned in brief in chapter III is known as “The verb Agreement Law”. Johnson (1977b) informs us very briefly that the various universal principles that Perlmutter and Postal (1974) posit include a generalisation such as follows: (2) The Verb Agreement Law: Only terms can trigger verbal agreement. A little more elaborate exposition of their claim can be found in Lawler (1977:225) and I quote him:

Page 246: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

245

(3) “ Perlmutter and Postal’s original claim, which seems to be supported by evidence from most languages with verb agreement, was that such agreement could only be triggered by terms (i.e., by subjects (Su’s) direct objects (Dos), or indirect objects (Ios), and further, that those must be cyclic terms of the clause in which the agreement takes place rules have applied within the relevant clause).” 5.3.2. Since it was shown by Lawler (1975, 1977) that at least in one language, Assamese, verbs tended to agree with the pre-cyclic or “logical” (as opposed to “grammatical”) terms, the law-makers seem to have retreated from their original position and “the new position they take is that agreement must be with either underlying or cyclic terms” (Lawler, 1977:223). In what follows it will be shown that even this amendment is no good because in Maithili even non-terms can trigger and control verbal agreement as much as the terms can do. But before going into the details of agreement system of Maithili, we may recall Keeman’s (1976) principle regarding Verb Agreement which is relatively weaker compared to Perlmutter and Postal’s law, and which found place in his Subject Properties List (SPL) discussed in chapter III. Keeman (1976) observed that (4) “The NPs which control verb agreement, if any, include b-subjects” (Keenam,

1976:316). The claim that non-terms cannot control the selection of the agreement markers is, obviously, much stronger than to say that even if they do, it will always be the case that terms will also have control over the same. Notice that the primary of subjects over the non-subjects has parallel in the claims of Keenam (1976) regarding the primary of sons subject-properties

Page 247: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

246

over the others. The general principles that relates those properties in a special way is that they follow a hierarchy, which has been called the promotion-to-subject hierarchy (of Chapter III). 5.4.1. Maithili has a person-gender-honorificity agreement pattern, as it has already been shown in (2) Except in Oriya. The decay of the old number agreement pattern is a common feature of the Eastern NIA languages. According to Chatterji (1926:935-6), in these language the plural forms gave birth to honorifics. But there does not seem to be any special reason as to why the plural markers should lose their original function and start marking the honorific levels of the arguments only. This is more so because in Hindi and Punjabi the markers for those two categories are the same and yet both number and honorificity are preserved. 5.4.2 Subhadra Jha (1958) informs us that the honorific system is not attested in the “Caryaa” songs (9th –12th C A.D). However he seems to overlook the fact that in this text the same markers were used for plurals as well as for honorifics, although he had himself noted that such examples were very common in Classical Sanskrit texts. The following examples from the Caryanapada would show that Jha’s observations were not well-founded (5) bhaNanti biruaa21 thira kari caala (Biruaa, 3.5.2.)

say+3p pl Biruaa quite dc+abc gc +cause +imp “Biruaa says, make (it ) go quietly”.

(6) kamal-kulisa ghaaNb 22 karahan25 biaalii 24 (Gubbrii, 4.1.2)

lotus-bolt stri+abc. Make+imp+hon eternal “By stirring the lotus-bolt, make no eternal”.

Page 248: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

247

(8) (a) bidyaa-kari25 dama-kun26 akilase n (kahNu, 9.5.2) ignorance-elephant resist+do+ imp +hon easily “Resist the ignorance-elephant easily”

(9) bhaNanti mahittan mai othu buRante kimpi27 na diThan (Mahidhara, 16.5.2) say +3p +pl mahittan by me here drown+ptply anything -not seen thing “Mahittan says, (even after having been) drowned, by no nothing has been noticed”.

The verb-forms bhaNanti, karahu n, and dama-ku n show that like in Sanskrit, in this earliest Eastern NIA text too the plural marker was employed to denote the honorific. 5.4.3. In VarNaratnaakara of Jyotiriiswara, the earliest prose in Maithili (dated (about 1350 A.D. ) the honorific distinction is attested only in certain tenses and modes-mainly in the present and past perfective and also in the optative. Consider the following sentences from VatNaratnakara28 : (10) I ye aaThae naayikaa athikaha sehao nandi hothi (Vr. 5) these RNL eight+emp

heoines, they also become pale” (11) caanda nlalna bholaaha (Vr. 27) noon dim become +past perform “The noon faded” (11) ghaTavaaranhi jalaasaya aarahala (Vr. 15)

boatman tank go+past perf+non – hon “The boatman went to the tank”

Coming down to the period of Vidyaapati 29 and his contemperaries the use of Honorific –Agreement pattern was extended to several

Page 249: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

248

other tenses, and its frequence of occurrence increased, too. However, even Vidyapati did not use this distinction in a rigid manner. But in Modern Maithili, this rule is strictly adhered to. This is true for both written and spoken standard colloquial varieties. 5.4.4. Modern Maithili has composite agreement affirms showing person, honorificity and gender in the verb at the same time. The gender, however, plays an unimportant role here. This is mainly because the gender distinction in Modern Maithili is confined to intrasitive sentences that are in a tense other than present. Again, the distinction is maintained in future tense only if the subject nominals of these sentences are otherwise high in the honorific-hierarchy. In case of non-honorifics, the distinction is made only in the past tense. Consider the following examples “ (13) raan caudah a barkhak lola ayodhyaa choR 1 kaa Ram fourteen years for Ayodhya leave-ptopl BO cal1 gallah a

proceed-abc go-pst-Agr “Ram left Ayodhya and went away for fourteen years”.

(14) hym a kah 1 sang a silltaa so-he cal 1 galiih a him(hon –Haph with Sita she –Raph proceed-abc go-pst-Agr “With him only, Sita also went away”. (15) homan kan 2 kir a baa gaan cal 1 gel a my son village proceed-abc go-pst-Agr. “My son went away to village”.

Page 250: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

249

(16) iamaadharin a cal I gel I sweeper-fon proceed-abc ge-pst Agr “The sweeper-woman went away”. Notice that (13)-(14) have subjects that are [+honorific] in terms of feature. Hence, we would expect that this distinction of gender would be maintained, even if the tense is future. That our expectation is true is proven by the sentence in (13’)-(14’) that follow: (13’ ) raam cauiah a barkhak lel a ayodhyaa choR1 kaa cal I inotaah a “Ram will go away from Ayodhya for fourteen years”. (14’) hum a kah 1 sang a siitan so-he cal 1 inatiih a “With him only, Sita will also go away”. This gender-distinction in future tense would be absent if the subject nominals are [-honorific ] which kaa a kir a haa “son” in (15) and innadaaria “sweeper-woman” in (16) are. This is the reason the sentences in (195’) and (16’) below show the same agreement marking in spite of the fact that the subject in (15’) is [+masculine] and that in (16’) in [+feminine]; (15’) hamar kaa a kir 2 baa gaan cal 1 inact a “My son will go away the village”. (16’) innadaaria cal I inact a “The sweeper-woman will go away”. 5.4.5. Thus for all practical purposes, Modern Maithili has mainly a Person-Honorificity agreement pattern. Now, in terms

Page 251: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

250

of selections restrictions on the verb agreement rule, the sentences in Maithili could be divided into two types that follows: (17) a. Sentences which have only one [ +human ] nominal argument which will also be the subject. b. Sentences which have more than one [+human ] noun phrase. The first type of sentences will include all intransitive sentences and these transitives which have non-human nouns as objects. These sentences show relatively simple agreement markers which are as follows: (18) 1st Person 2nd person 3rd person Honorifics -hu n -hu n {-aaha (use)/ -iih a (fon) }aah1 Medium -hu n - h a {-? (use) / -1 (fon)} –ka Non-honorifics –hua -Bn/-hii {- ? (use) / -1 (fon) }-ka Non-honorifics -hu a -Bn / -hii {-? (use) / -1 (fem )}-ka 5.4.6. Notice that in the first person the marker remains the same because the three-level distinction does not exist in this case. Also note that the notion of transitivity is important only in case of the third person. The markers for the first and the second person honorific are the same. Finally, in the third are blurred 30 and that is why the suffixes are the same here. The following sentences demonstrate the type of agreement marking listed in (18)

Page 252: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

251

(19) ham/ ahaa n sut-a la – hu n I / You ( hen) sleep+past+Agr

?? I / You (hen) slept”. (20) to n kha-ela-h a You(nod) ear+past+Agr “You (nod) ate”. (21) jhaatil ghar g-el-aah a g-el-aah I Mr. Jha home ge+past+Agr(intr) Agr(tran) “Mr. Jha went home”. (23) jhaa-jii pothii paRh – a I – ank1 Mr. Jha book read +past +Agr (tran) Agr(intr) “Mr. Jha read the book”. 5.4.7 It may be mentioned at this point that the optatives seem to fall within this type of pattern then in the type b. Thus, in such cases, even if the sentence has two human nominals where only one of them is the sentential subject, the markers are selected in accordance with the subject nominal from the list given in (24) below:

Page 253: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

252

(24) 1st person 2nd person 3rd person Honorific : -ii -ii -ath1 Medium - aha -ae Non-honorifics - -Bn -ae Consider the following examples: (25) han cash-ait a eh-ii je ahan n hama raa naar-ii I want+ptopy AUI +pres that You (hon) no kill+Opt+Agr “I want that yoy (hon) kill me”. (26) e teran maar-athii He(hon) you+obj kill +Opt +Agr “Let him (hon) kill you”. 5.5.1 In all other cases, the agreement markers are selected depending upon the peron-honorificity level of both subject and non-subject nominal arguments present in the surface of a sentence. But if both of these refer to the same person(s), the reflexivisation rule would apply and the agreement-pattern would then become the same as in case of (17)a . In that case, the markers are selected from the list of suffixes that follows: (27) Subject NON-SUBJNOT MARKERS 1P 2p. hon -ahun 1P 2p.nod -iahu 1P 2p, non0hon -iau(ka) 1P 3p, hon -iauk1 1P 3p, non-hon -iani (ka) 2P, hon 1p, -ahun 2p, hon 3p, hon -iaauh1

Page 254: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

253

2p, hon 3p, non-hon -iai(ka) 2p, mod 1p -aha 2p, mod 3p, hon -ahuuha 2p,nod 3p,non-hon -akaka 2p, non-hon 1p -Pn 2p,non-hon 3p,hon -ajhunha 2p, non-hon 3p,non-hon -hii(ka) 3p, hon 2p,hon -anh1 3p, hon 2p, mod -thuunha 3p, hon 2p, non-hon -thuunha 3p, hon 3p, hon -thiinka 3p, hon 3p, non-hon -thiinka 3p, non-hon 1p -aka 3p, non-hon 2p, hon -aka 3p, non-hon 2p, nod -kahu 3p, non-hon 2p, non-hon -kau(ka) 3p, non-hon 3p, hon -kaiuh1 3p, non-hon 3p, non-hon -(k) ai(ka) 5.5.2. Notice that it is extremely difficult, if not impossible, to segment these affixes into two neat pieces in each case so that one could identify the markers for person and honorificity. A close scrutiny of the list would also show that in many instances the markers are the same. Thus, -ahun, -iaamh ,̀ -iai (ka ), -ahunha, -aah1, -thuuaha, -thiinka, and –aka re potentially ambiguous in a sentence which does not have the triggering nominals in the

Page 255: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

254

surface. This ambiguity would be due to an optional pronoun deletion rule which may delete such triggering pro-forms after the appropriate agreement suffix is selected. Notice that this rule would give rise to recoverability problem if the agreement marker happens to be either of the ones mentioned above. For instance, (28) may be understood as (29) or (30) depending on how one would like to interpret it: (28) phuddii jhaak phula bagiaa nN dekh a lahua Phuddi Jha-poss flower-gardent lee see+past+Agr “Saw in the flower-garden of Phuddi jha”. (29) han ahaan kBn phuddii jhaak- - - - - - I you +obj Phuddi jha+poss “I saw you in the flower-garden of Phuddi jha”. (30) ahaan ham a ram phuddii jhaak- - - -- you I + obj Phuddii jha+poss “You saw me in the flower-garden of Phuddi jha”. Thus, although one would ideally expect about twenty-six markers to be used in the sentence type (17)b, it is a matter of accident that is reality there are only eighteen of them. 5.5.3. Let us now consider the cases where the non-subject nominal is either a BO or IO. It will be clear from the following sentences that the agreement markers in the verb keep changing with the change in the person-honorific level of the DO or IO; (31) o {ahaan – kNn/toran/huna kaa} {naar-a l-anh1/ naar-ala –thuuah a / naar-ala – thiinka} He (hon) {you (hon)+DO /you (med)+DO/ he(hon)+DO} hit+past+Agr “He (hon) hit you (hon)/ you (mod)/ him (hon)”

Page 256: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

255

(32) ham {shaan –kBn / terraa1 /teraa2} pasi {do-1- ahu n /do- l – iah u / do-l-iau(ka)} I {you (hon) +IO/ you (med)+IO/ you (NN) +IO } money give +past +Agr “I gave you (hon/mod/NN) money”. 5.5.4. There are, however, instances where one can argue that the oblique case markings attached to an argument are the results of some well-formedeness conditions on the surface structure of the sentences. Thus, one can, on the basis of the newer position taken by Perlmutter and Postal (1974), explain the apparent violation of the verb agreement law in the following sentences by showing those Oos to be pre-cyclic Dos and Ios: (33) ham {hunak /ekar} khuun {kar-2b- ahn1/ kar-ab-ai(ka)} I he (hon) +poss murder de+fut+gr He (NN) + poss “I will do his (hon/non-hon)murder”. (34) e {teran sOn/ eka ran Son} pron kar-aita {cha-thuuha/ cha-thiinka } He (hon) {you(mod) +comit/ he(NN)+comit} love de+pross ptopl AUX +Agr. “He (hon) leves you (mod)/ her(mon-hon)” Such constructions as above are common in Maithili and some other related languages when the sentence has a conjunct verb the first member of which is a nominal 5.5.5. However, if the verb agreement law is to be taken as an absolute universal generalisation which is empirically verifiable,

Page 257: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

256

it becomes difficult to explain as to why both deep and surface Oos in the following sentences should control and trigger the selection of verb agreement markees: (35) a. ham ahaank gaan nB gela challahn I your (hon) village+les ge+past ptopl AUX +past +Agr “I went to your (hon) village:. b. ham tehar gaan nB gela chaiah u “I went to your (mod) village”. c. ham hama kar gaan nB gela challiah1 “I went to his (hon) village”. d. ham oKar gaan nB gela chaliai (ka) “I went to his (non-hon) village”. 36. a. c ii vastujant hama ran aOn lekala No (NN) these things me+from take+past +Agr “No (non-hon) took these things from me”. b. c ii vastujaat teran aOn . lel2 kahu “He (non-hon) took these things from you (mod)”. e. e ii vastujaat huma kaa aOn lela kaiauha “He (non-hon) took these things from him (hon)” (37) a. ahaah n honar atitha chalahun You (hon) my guest be+past+Agr “You(hon were my guest”. b. ahhan huma kar athith a chaliauh 1 “you (hon) were his(hon) guest” c. ahaan ekar atith a chalinika. “you (hon) were his (non-hon) guest”. 5.5.6. Notice that I have not included in the above examples sentences where a change in the subject nominal brings about a change in the agreement suffix in the verb. This is not because

Page 258: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

257

I want to give an impression that in Maithili it is only the non-subjects that control verb agreement. Rather, the aim was to show that Perlmutter-Postal generalisation cannot be accepted as an absolutely true universal statement just as the Verb Agreement Hierarchy proposed in the first half of this chapter is not an absolute generalisation. If we take the a-sentence from the examples in (35) –(37), and change the subjects the agreement markers would also be found to be different. (35) c. e ahaan k gaa, nB gel a challaha He +hon your +hon village +lee go+past ptopl AUX +past +Agr “He (hon ) went to your (hon) village” (36) d. dam ii vastujaat teraa aOn leliah u I these things you(mod) from take+past+Agr “I took those things from you (mod)” (37) e. ton hamar atitha chalaha You (mod) my guest be+past+Agr “you (mod ) were my guest.” It may be mentioned here that the agreement rule in Maithili does not seem to be without exception 32 and the first of the three sentences given above provide us with one such case. According to the list of markers presented in (35)c , a third person honorific subject and a second person honorific non-subject (here, an oblique object) should trigger –aaho. But the marker that has actually been selected is –aaha which is normally used in the sentence type (17)a, i.e. in the intransitive sentence and in these sentences where the non-object is a non-human nominal.

Page 259: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

257

I want to give an impression that in Maithili it is only the non-subjects that control verb agreement. Rather, the aim was to show that Perlmutter-Postal generalisation cannot be accepted as an absolutely true universal statement just as the Verb Agreement Hierarchy proposed in the first half of this chapter is not an absolute generalisation. If we take the a-sentences from the examples in (35-37), and change the subjects the agreement markers would also be found to be different: (35)e. e ahhan k gaan nE gela chalaah a No +hen your+hem village +lee go+past ptoply AUX + past +Agr “He (hon) went to your (hon) village”. (36) d. ham ii vastujaat terun son leliahsu I these things you (mod) from take+past+Agr “I took these things from you (mod)” (37) d . ton hamar atith a chalaha You (mod) my guest be+past+Agr “Your (mod) were my guest”. It may be mentioned here that the agreement rule in Maithili does not seem to be without exception 32 and the first of the three sentences given above provide us with one such case. According to the list of markers presented in (35)e, a third person honorific subject and second person honorific non-subject(here, and oblique object) should trigger –aah1. But the marker that has actually been selected is –aah a which is normally used in the sentence type-(17)a, i.e. in the intransitive sentence and in these sentence where the non-subject where the non-subject is a non-human nominal.

Page 260: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

258

5.5.7. It must be obvious by now that Perlmutter-Postal Law is at best a tendency universal and that this relatively strong hypothesis is to be replaced by (5), proposed by Keenam (1976), because it does not matter as to what kind of nominal we select to fill in the SV, DO, IO and OO-slots, it is always the case that b-subjects are co-controllers of verb agreement. It seems that in the languages where the agreement. It seems that the b-subjects are co-controllers of verb agreement. It seems that in the languages where the agreement markers are triggered and controlled by the non-subjects (including the non non-terms) either the markers for both subjects and non-subjects would be found in the verbal complex in a clearly identifiable manner or they would be found in a syneritized form as in Maithili 33.

Page 261: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

259

Note to Chapter V. 1. Gricreem (1906c:30) gives the following examples where depending on who one

talks to, one selects entirely different sets of verbs, nouns, adjectives and markers: A) i. Khyed –kyi rta yag-pe (non-honorific) ii. nyid-rang-gi chhibe-pa bsang-pe (honorific) “You have a fine horse”. Thus, some verbs have different shapes (of. Grierson, 1906c:31) depending upon the honorific level the addressee belongs to. For instance: B) NON HONORIFIC HONORIFIC GLOSS i. nthing-ba guigs-pa see ii. cnam-pa changs-enon-pa smell iii. sa-ba bshee-pa eat iv. go-ba/thes-pa gsam-pa hear 2. The Sore-kunda group includes the following sub-grup of languages: Ndega-Sere,

Ferage, and Nundu-Ngbaka (Tucker & Bryan, 1966:85-107). 3. The Banda-Ghaya group of languages have the following sub-divisions; Benda,

Ghaya, and Ngbandi (Tucker & Bryan, 1966: 85-107). 4. It includes diminutive and pejorative, tee 5. An example of each of these could be from Santhali and Nundari (of. Greirson,

1906a): SANTHALI MUNDARI PERSON & NUMBER Pronoun Conjugation Pronoun Conjugation Lin v+ete + aa+a sing ling+V+1+aa 1st p, sg aa-bee-n v+et’+aa+bee aa-bu aa-bu+V+1+aa 1st p, pl(incl) aa-laa v+st’+aa+laa aa-le+ aa-le+V+1+aa 1st p, pl(excl) aan v+st’+aa+a asm asm+b+1+aa 2nd p, sg aa-paa v+st’+aa+paa aa-pe aa-pe+V+1+aa 2nd p, pl un-ii v+st’+aa+ee limii limii+V+1+aa 3rd p, sg en-kee/na-kee v+st’+aa+kee con-kee/aa-kee askee+V+1+aa 3rd p, pl Note that the “V” in this chart indicates the “verb”.

Page 262: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

260

6. For instance, the personal pronouns in Mahali include ie “X” hingan “We”, No “You, ag” naakum “you, pl”, and haa “No/They”, but the verbs are conjugated only for first and second person singular on the one hand and the same form is used for the rest of the categories. Thus, koTTe “to beat” has only two forms; koTTange “I/you beat” and koTTagan “You (pl) /No/We/They beat”. In savara, similarly, tiDa “to beat” has three forms: Pronoun+tittai “I/We beat” Pronoun +tiitel “You (sg)/you(pl/No beat”, and Pronoun +tittleji “They beat”, Notice that one may explain the problem of these “shadow” pronominal agreement in Austre-Asiatic verbs by a rule of pronoun-flipping of some kind, thereby taking a very different position, viz., that these language do not have verb agreement at all.

7. In Nee-Nelamesian (Nihalic, 1957:22-3), there does not seem to be any agreement

marker, but occasionally one finds an affix I- which most of the time “appears after a third person subject”. Although Nihalic (1957) calls it a “prodicate” marker, the examples he gives shows it to be a topic marker. In such cases, I suppose, the languages of other languages in this column, see Green & Iqwe (1963), Prometasana (under proparation) Grierson (1906), Willianson (1965), and Tucker & Bryan (1966)

8. For Yiddish and Dioguene, see R.N.R Hall (1967) and N. Langenden (1970),

respectively. For Diogusac, also see Nurue Acki (1970). 9. See Krishnamurti (1975) for gender and number in the Dravidian Languages. 10. For a detailed study on the Japanese honorific system, see Prideaux (1970). 11. Some languages, such a Roal Achemenid Elamite (Paper, 1955) will fall in this

category just because only in one of the persons the number distinction is blurred in only one of the persons. For the language in this column, see Denys Eray (1909), Tu7cker & Bryan (1966), Kruger (1961), Collinder (1957), O’leary (1969), Welmers (1946), Gall (1966), gall (1966), Jammaris (1968), Mista (1969), R.A Hall (1966), Bright (1957), Tucker (1967), Samarin (1966), Aginsky (1933), Sapit & Heijer (1967), Lunt (1955), Thurneyason (1970), Kont (1935), Hudson-Williams (1935), Fertune (1955), Puente (1959), Rastergueva (1963) and Robins (1958).

Page 263: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

261

12. If, for a moment, one excludes the exceptions such as Toda, Japanese and Modern Mongolian, the VAN can be said to predict that there could theoretically be mine types of agreement-patterns. It is an accident that no example of a language with a person-gender (P-G) agreement was found. The fact that both P-N-G and P-G-N patterns are attested in human languages suggests that if the corpus is enlarged, the P-G. pattern can also be found.

13. See Chatterji (1926) for Assamese and Bengali agreement system and their development

14. See Nicholas Poppe (1955) for a brief history of the Mongolian agreement system. 15. Among the Indo-Iranian frontier languages, again, one finds instances where one

of the languages, viz. Fidgha-Nunji, there exixts a P-N-G system whereas the related tongues such as Banglechi-Ishkashmi and Wakhi do not have gender-agreemtn (of. Norgonstirerne, 1938). For the details of languages included here, see Goveli (1964), Watkins (1937), Aquilina (1959), Blocj (1970), Bleemefield (1962), Poppe (1955), Gair (1970), Bhat (1967), Greirson (1906ab), Hiramath (1961), Krishnamurti (1975), Subramanyam (1968), and Tucker (1967).

16. See Geraya (1944), Fittman (1954) and Chatterji (1926) for Burmese, Nahualt and Oriya, respectively.

17. See Jha (1958) and Singh (1978b) for the agreement pattern in Maithili. 18. Broj Kachru (1969:71) notes that like in Hindi and Punjabi in Kashmiri also, “a

finite verb is inflected for tense, gender, number, status, person”. Hence, it can also be included in this list. See Yamuna Kachura (1965, 1966) and Gill & Gleason (1969) for Hindi and Punjabi.

19. These two lines of research have been noticed by some more scholars and almost all of them would agree to what Navy Sanches and Linda Slobin (1975) have to say about this development. They held that

“In discovering the particular formal and substantive features of a universal grammar, which in turn reflect biologically determined underlying human cognitive capacities, at least two approaches are possible. One approach is to determine the rationally necessary features which all language systems must have in order to function. An example of this kind of approach can be soon in Fillmore’s study (1968), according to which a description of any of the world’s language must include as part of the base component, a case system. The

Page 264: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

262

question than becomes one of investigating how such a feature is realised in the various languages of the world. Statements of this kind of a ‘universal’ are very close to what we might call the cognitive capacities particular to humans who speak natural languages. One can also proceed from the other end of the theoretical spectrum and observe regularities in the distribution of certain features in the languages of the world. This approach is exemplified in the work of Berlin and Kay (1969); Ferguson (1963) and Greenberg (1966). It usually results, though not always, in statistical or implicational statements” (Sanehos & Slobin, 1973 :3) Note that one need not agree with Sanches and Globin’s (1973) linking the absolutive universal with cognitive capacities. However, their general observation regarding the two directions of research in the field of language universals is shared by many 20. See Ross (1974), Keenam and Courie (1972, 1977) and Johnson (1977b) for the

history and development of the ideas behind using the notion of grammatical relations on predicting certain generalisations about language. For Double Chaining Constraint, see Johnson (1977b). The cereferential deletion law was proposed by Perlmutter and Postal (1974)

21. Viruua in Bagchi and Shastri (1956). The names of the saint-poets who are given credit for a particular verse appear in the parenthesis. The numerals after their name (eg. 3.5.2, in this case) indicate the serial number of the song, couplet number and whether first or second line of the couplet, respectively. Those numbers refer to Naraprasad Shastri’s 91916) work where the Caryaa-songs appeared for the first time.

22. In the original text and Naraprasad Shastri (1916); kamalakulinghaaTTa ; in Tarapada Mukerji (1963); kammla kulisa chasTre Rd. Sahidullah (1940) ; gthanTi.

23. In Sukumar Son (1948):karahu 24. Cemmentary :bikaala 25. Original, Shastri (1916); kidyaakari: Mueherji (1963); kidyaa kari; Son (1948);

abidyaakari, Bagchi and Shastri (1956); evidyaakarikun, 26. Original, Shastri (1916), Mukerji (1963); damakun ; son Sahidullah (1940): dama

Page 265: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

263

27. Bagehi and Shastri (1956) : kinvi. 28. See Chatterju and Mishra (1940) for those and many other examples. 29. A good number of examples from the poems of Vidyapati and others have been

given in Jha (1958: 469) to prove this point. 30. This is because the third person pronouns have a two way distinction only. One

may guess that while there are language where pronouns show more categories than the agreement affixes in the verb do, the vice versa is not true. This may be a universally true statement.

31. Ton functions in Maithili as second person pronouns of two different levels. See the pronominal chart in chapter II.

32. Talking about exception, in a personal communication, Bernard Courie posed in Maithili. The question has to do with the multiple human nominals in the surface; Does the verb in such cases agree only with the subject or even here the object(s) play an equally important role in deciding the exact shape of the marker? I think it would be more interesting to see whether the relational hierarchy is maintained in case all the non-subject positions (i.e. DO, IO, and various kinds of OO) are human nominals. Consider the following constructions:

a. han shaan kBn kunak pachuaR1 nB dekh a lahu n 7 deck a lianh 1 I you +DO ho(hon)+ backyard +lee see+past +Agr “I saw you (hon ) is his (hon) backyarda. b. ham huma kaa chaan k packnaaR1 nB dekha lahua dekha liauh1 “I say him (hon) in your (hon) backyard”. Notice that in the first set of sentences when the OO (hunak) controls the agreement along with SV, the result is an odd construction. But in the b-sentences, both DO and OO may cenral agreement. It is not known as to why such a difference shall exist in this case. However, in no case the relational hierarchy seems to be violated. But if SU, DO, IO and OO are all [besides agreeing with the SU]. But there is no instance where the verb agreeing with SU and IO gives a better sentence than when it agrees with, say, SU and DO.

Page 266: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

264

33. Eduth A. Maravasik’s (1971, 1978) work on Agreement gives rise to about nine generalisations and these are based on a survey of a large number of languages. These are again of implicational type and she gives a series of hierarchies among various grammatical notions depending on their triggering properties as follows: (1) Intransitive Subject will have higher or equal control over agreement than any other relation; hence, +Intr SU ? -Intr SU; (2) Do complement ? Dative complement; (3) –Adverb ?+ adverb; (4) Definite ? Indefinate (if both are attested in a language); (5) animate ? - inanimate (if both are attested in a language ); (6) +topical ? -topical (if both are attested); (7) Pre-verbal elements ? Post-verbal elements ; (9) (At the point of the application of verb agreement rule, ) the IOs of the verb. These generalisations have not been discussed in detail in the present chapter because Maithili does not seem to violate any of them.

Page 267: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

CHAPTER VI COMPOUND VERBS IN MAITHILI

6.0.1 The problem of defining and identifying the set of Vector verbs (or call them ‘aspectual’ , auxilaries; ‘ operators’ or ‘ explicators’) in a language which has compound verbs as a productive grammatical category has been tackled on both formal and functional grounds. The problem revolves around, to my mind, a few questions which are as follows: (1) a . What is it that allows certain Verb+Verb sequences to be called “Compound

verbs” as against others? I.e. what are the formal criteria to define such a word-class, and what are the functional tests that help an analyst in identifying this category?

b.What is the normal word order within a compound verb? And , can this order be reversed in such a way that the two verbs concerned exchange not only their positions but also their respective markers, and if at all that can be done, is or isn’t the reversal restricted to a few combinations? c. How many of the verbs that could perform the function of vectors can themselves

be used as polars with all other possible vector verbs? And, how many cases of the same verb being used as both polars and vectors does a language have?

d. Can the main verbs be classified depending on whether they can occur with any of the vector verbs or with a partcicular set of vectors, and not with those outside the set?

Page 268: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

266

e. What are the semantic shades that each or all of the vectors bring in when they are used with different polars? Are they desirable only as “aspects” in true sense of the term1?

f. What are the pragmatic considerations in selecting a compound verb as against a simple verb in a particular senrtence, and what makes one select a normal or a reversed order of a polar and a vector verb in different sentences?

6.02. It is with an aim to provide at least a partial answer to those questions with a special reference to Maithili that this chapter has been prepared. Note, however, that one of these questions, namely (1)f would be dealt with in the following chapter. Also, for the sake of convenience, I shall on many occasions be dealing with more than one question at a time and not neccessarity in the order as in (1) 6.1.1. Instead of advancing a definition of what a Compound Verb is at this point. I would start with a discusstion of the nature of what has been called a Compound Verb in the ever-increasing literature on this topic. Thus, I would, to start with, concentric on what a Compound Verb “does”. In one of the recent studies (Dasgupta, 1977), the nearest equivalent to a compound verb in English has been described a verb plus an adverbial modifier (or a preposition) construction. What is significant in this statement is the realisation that “Compound” verb is a misnomer for the construction it refers to because it does not involve any conjunction. Rather, meaningwise it is

Page 269: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

267

simply a modification of what is signified by the polar2. That is, when used in Compound verbs, the verbs that function as vectors lose their original meaning. 6.1.2.1. In some of the Indian languages, notably in Bengali and also to an extent in Maithili, this seems to be an important criterion that helps one differentiate between Compound Verbs ( a Polar + Vector constuction ) and Verbal Conjucnts ( a main verb + main verb combination ) when by chance the same verbs are selected for those two different constructions and are placed in identical environments. Let us take a Bengali example to make the point clear: (2) a . Chobigule dekho no picture-pl see-ptopl take – imp – Agr “Take the pictures seeingly (-scrutinisingly)” b. chobigule dekho no picture -pl see – abs VECT (benefactive ) – imp – Agr “See the pictures (for yourself)” 6.1.2.2. (2) is a clearly derivable from two sentential sources such as follows, and is therefore an instance of Verbal Conjunct: (3) a. chobigule dikh picture-pl see-imp-Agr “Look at the pictures”. b. chobigule no picture – pl take-imp-Agr “Take the pictures”.

Page 270: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

268

The order of actions denoted by the two verbs in (3)a and b could be changed, thereby generating a sentence parallel (2) a : (4) chobigule nie dikh picture-pl take-ptopl see-imp-Agr “Take the pictures (and ) look at (them)” 6.1.2.3. However, (2) b cannot have a parallel such as in (4) because the verb dikh – “see” cannot function as a vector, and also because Modern Bengali does not allow a reversal of Polar-Vector combination in such a way that they interchange their markers too. Notice that if these were allowed, it would have been more difficult to differentiate between (2)a and (b). But even in that case, the meaning-difference between the two clearly show that no in (2)a and b have different functions . 6.1.3.1. The situation in Bengali is complicated mainly because here in both Compound Verbs and Verbal Conjuncts, the first member receives the same marker –a. Also, Maithili and Hindi would show a dummy verb DO inserted between the two members of a Verbal Conjunct, whereas Bengali would never show it. In most of the cases, Maithili however shows a preference for doing away with this dummy DO. Which in Maithili is Kao “having done” (< kar “do”). Thus, there are examples in Maithili parallel to the Bengali sentences in (2). (5) a. ahann apan pasand a aOa dekh I livOh a “You take it seeingly, according to (your) own choice.”

Page 271: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

269

b. chan n apan pasand a sOn dekh I livOh a you (hon) self-pass choice from see-abc VNCT-imp-Agr “You see (-check) it, according to (your ) own choice”. If we insert the dummy kao, (5) a would remain grammatical while (5)b would become unaccpetable in the intended sense as the following sentences would show: (6) a’ chaan apan pasanda sOn dekhi kao liyOha b’ shaan apan pasand sOn dekh1 kao liyOha 6.1.3.2. These sentences give us one criterion that differentiates the Verbal Conjuncts from the Compound verbs. We can test a Verb-Verb sequence by inserting a kee-element between the verbs, and can find out in this way whether a particular combination is a Compound Verb. We shall call it the kao- insertion test (parallel to Hindi kar-insertion test; of. Hock, 1974). Other characteristics of the Compound Verb include the following: (I) the favoured word in Maithili in this respect is Polar-Vector, and (ii) the absolutive marker that is attached to the first member of compound verb is, in most of the the cases, -1, i.e., a half-short I 6.1.4. I think I should explain why I said “in most of the cases” The marker –I could be seem almost invariably if the first verb of a compound (which is usually the polar member, too) ended in a consonant. There are two exceptions to this rule: kar- “do”, and dhar- “keep”. These two verbs would behave like vowel-ending stems in this case. These tow –( r ) ending stems could be grouped with the –o and –e ending

Page 272: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

270

stems because all of them change to –O before the marker –i. Once this change takes place, the original marker is dropped. Thus, one would get kO, dhO, do, IO and bhO from ka( r ) “do”, dha ( r ) “keep”, do “ give” , lo “take, and (b) ho “be”, respectively. As against this, there are a few verb stems which appear in most of the time as vowel-ending take a consonant –b- when placed before the marker –i. This, however, suggests that these be taken as –v ending stems where –v before –I because –b. These verbs included aa(v) “come”, gaa(v) “sing”, vaa(v) “get”, kaa(v) “open (one’e mouth)”, enu (v) “lock”, and chuu(v) “touch”. All other vowel-ending verb stems would remain unchanged and the marker –I would simply be dropped4 . 6.1.5.1. If one tries to expand the category of Verb, one can come up with at least the following structure-types: (7) a. [V-(aff-AUX) – Desinense] b. [N/Adj/Adv } – V- (aff-Aux ) – Desinence] c. [V- {aita/anF} – {V/Aux} – Desinence d. {V+1 – (Kao) –V – (Aff- Aux) – Desinence] e. [V+I – V – (aff-Aux) – Desinence] An example of each of these structure-types could be given here: (6)’ a . hon [go-1a cha-1-ahun] I go-aff aux-pst-Agr “I had gone”.

Page 273: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

271

b. e [cuppii naar-ala –thiinka ] hc silence kill – pst –Agr “he became silent” c. e manch [ {paka n – aita / pakaN-bB} rah- a 1- aaha] he fish catch –{ habit.ptc/ prog.ptc} aux-pst-agr “He went on catching fish/No remained holding (a) fish”. d. han niioaan sOn kursii par [uTh1 (kao) I floor from chair on risc+pte DO Bais- a l-ahu n] Sit-pst-Agr “I get up from the floor and sat on the chair”. e. taavat ham hunkaa sab-Taa Taakan by then I him all-dot money [dO bais-al1 cha-l-ianh1] give +abc sit –off aux – pst- Agr “By then, I had given away all the money to him”. 6.1.5.2. The last of the types in (6) is the structure for true compound verbs. (6)’a is an example of Simple verb, while (6)’ b has been calling Verbal Conjuncts in the preceding paragraphs. (6)’ c is also similar to (6)’ d, the only difference being that the main verb here is always in progressive/habitual aspect. If Kai is deleted from the structure-type (6)d, the “formal” distinction between (6)d and c is lost. We have already seen that they can still be differentiated by their meaning, and by reinserting kao in the sentence to find out whether a particular sentence shows a true compound verb or a verbal conjunct. Thus,

Page 274: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

272

while one can have optionally kao in (6)’d, its insertion would make (6)’ c ungrammatical. One can think of some other tests 6 as well 6.1.6.1. One such test is given in Singh and Bandopadhyay (1978), which can be called the Verb Interrogation Test7. This test has already briefly been discussed earlier (see chapter IV). We have seen that one has to posit a Verb-interrogation transformation to capture those instances where the verb or the whole V-mode is under interrogation. Examples of this kind of interrogation included the following sentences from Bengali and Assamese, the neighbouring languages: (8) assanac 10taak EhhOn kitaab disc (Assamese) Asha –Erg Lata IO one book –DO give-pres-prog –Agr “Asha is giving a book to Lata”. b. aasaaca Iotaak ki kOrise? Asha-Erg Lata-DO what Do-pres-Prog-Agr “What is Asha doing to Lata?” (9) a. aaSaa ghunooche Asha-mon sleep-pres-prog-Agr “Asha is sleeping” b. aaSaa ki kerche? Asha-non What DO-pres-prog-Agr “What is Asha doing? 6.1.6.2. This kind of interrogation is not possible with certain verbs, especially with the existentials. For instance,

Page 275: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

273

consider the following sentences from Bengali: (10) a. aami baaRite {aachi / chillan / thaakbe} I house-lee be-pres/pst/fut-Agr “I am/was/will be at home” b. I house-lce what IO- pres/pst/fut-Agr “What do/did/will I do at home?” 6.1.6.3. Moreover , when the static verbs are questioned in this way, thy acquire a sense of ‘an act of doing’ something. Thus, the verbs that are traditionally known to be static in Maithili, such as iaan “know” (of. Bengali iaan; Assamese iOn) or bais “sit” (ef. Bengali bOS; Assamese bOh) , would denote “the act of learning” or the act of sitting” when this rule is applied. This suggests that the verbs to be questioned in this way must involve “process”. 6.1.6.4. In Singh and Bandyopadhyay (1978), the construction in (6)d was described as an instance of “Bi-Polar Compound” which was different from (6)e, the “simple compound verbs” in that there would be a greater “cohesiveneses” (whatever that may mean) between the two members of the compound in the latter case than that in the former. That is, semantically, true compound verbs would behave as single units, and the Bi-polar case would not. When one applies the verb interrogation rule on these two constructions, the difference between them becomes evident. We

Page 276: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

274

have already noticed that this rule replaces the V-mode by a dummy kar- “do”, keeping the desinence of the original verb intact, and it also introduces a ki “what “ question-word before the dummy verb. This rule when applied on Bi-polar compounds give rise to three possibilities. Consider the following examples from Maithili 10) ahhaa k ciijbiit e dekh1 kee lataah a You-poss things he see-ptopy BO take –fut – Agr “He will take your things seemingly (scrutinisingly)” 11) a. ahaan k ciijbiit e kii kar 2 taah a? you poss things he what DO-fut –Agr “What will he do with your things?” b. shaank eiijbiit o dekh1 kaa kii kar1 taaha ? you-poss things he see-ptopl DO what DO-fut-Agr “What will he do with your things after seeing?” c. Shaan k diijbiit o kii kO (kae) latasha? You-poss things he what DO-ptopy (DO) take-fut-Agr “What will he do to your things and take?” 6.1.6.5. Notice that in (11)b and c, the second and first member of the “Bi-Polar compound” have been questioned respectively, whereas in (11)a the whole Bi-Polar compound (or the verbal conjunct ) has been questioned. Not all such possibilities are found in the case of true compound verbs. Consider the following examples

Page 277: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

275

(11) aai bhore-bher e jala khae khan lela thiinha today morning –Hmph – he breakfast eat-abc take-past-Agr “Today, he took breakfast early in the morning”. (12) a. aai bhore-bher e kii kaal a thiiuha What DO- pst-Agr “What did he do today early in the morning?” b. aai bhore-bhor e kii kO lela thiinha What DO-abc take pst-Agr “What did he do today early in the morning?” c. *aai bhore-bher o jala khaa khan kii kaal a thiinha? 6.1.6.6. Similar examples are found also in Bengali: (14)So his khoye pholechile He poison eat-abs throw-pst-prog-Agr “He was (almost drinking up the poison”. (15) B. So ki kerchile? What DO-pst-prog-Agr “What was he doing?” c. *So biS khoye ki kerchile? 6.1.7.1. There are three points that become clear from these examples and they are : First, in case of Compound Verbs, only the Polar number may be replace by a dummy DO by a Verb interrogation Rule, and not the Vector. Alternatively, one

Page 278: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

276

can, of course, replace the whole Compound Verb. But Verbal Conjuncts allow both the verbs to be replaced by DO, and of course, in this case too, the whole combination can be replaced by the same. 6.1.7.2. Secondly, in case of Compound Verbs, when the Polar member is replaced by DO, the verbs that perform the role of vector and remain in the surface, do not show their original meaning. Thus, in both (13) and (14), the b-sentences did not differ from a-sentences in meaning. That is, the vectors in (13)b and (14)b – lo “take” and phul “threw” were semantically almost empty or null. This is in confirmation with the definitions of Compound Verb where the vector verb is characterized as a slotfilled in a restricted set of verbs that “lose” their original meaning. 6.1.7.3. Thirdly, we are aware of the fact that not all Compound verbs allow the polar member to be replaced by DO. In such cases, Verb Interrogation Rule can affect only the whole of the Compound Verb, and not its part. The following Bengali examples have already been given in Singh and Bandyopadhyay (1978:30) to prove this point: (16)so cole jaaschile he proceed-abc go-past-prog-Agr “He was going away”. (17)a. So ki kerchile ? What DO-past-prog-Agr “What was he doing?”

Page 279: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

277

b. *No ki kere janochile ? e. *so cele ki kerchile? 6.1.7.4. What (12)-(17) prove can be summarised in the following words: (18)a. The Verb interrogation Rule can replace the whole Compound Verb by a

dummy kar “DO, b. The Verb Interrogation Rule may replace the Polar member of a Compound Verb by a dummy kar “DO, c. The verb Interrogation Rule can never replace the Vector member of a Compound Verb by a dummy kar “DO. If (18) is a generalisation applicable to all the languages that have compound verb ( and have a problem of distinguishing between compound verb and verbal conjunct), one can try to apply the verb interrogation Rule on the vector verb and see whether (18) c is maintained. If it is maintained, the Verb+verb combination concerned in a compound verb, and if it is violated, the combination is not a compound verb (i.e., in that case, it may be an example of 6c or d, but not of 6c) 6.1.8.1. Another test that has found place in a number of works on Compound Verbs has to do with the fact that “compound verbs do not occur in most types of negative expressions” (Hook, 1974:201), and that “the element deleted under negation is the vector verb” (Hook, 1978:130). Most of the scholars who noticed it tried to show negative expressions as environments which permit a “Simple Verb” as oppose to a Compound verb. Various explanations have been advanced for this phenomenon. For example, it has been

Page 280: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

278

suggested that the “idea of finality and completion” (Pahwa, 1936:245) associated with Compound Verb and negative environment are semantically conflicting things (see Hacker, 1978; Gaeffke, 1967). Assuming that Compound Verbs developed from Verbal conjuncts at some stage, some (Pray, unpublished) explained the ungrammatically of a Negative Compound Verb in terms of a “formal gap” that exists otherwise because of the absence of a negative Verbal Conjunct expression in Hindi. The “Semantic Conflict” explanation has been criticized on the ground that Russian and many other languages have examples where negation of a perfective act is possible. Similarly, the “Formal Gap” theory has been shown to be inadequate (see Hook, 1974:204-6) because Hindi does have negative Verbal Conjunct of various types, and because on the one hand one may find some languages which do not have conjunctive participle at all and still lack negative Compound verb, and on the other hand, language like Tajiki have both. 6.1.8.2. The explanation that Hook (1974) gives is that “Relative Completion” as a function of Compound Verb is incomparable with the non-completion of an action denoted by the rule of negation, and hence the absence of a negative compound verb in Hindi. Hook (1974) who went beyond Gaeffke (1967) to find out the nature of the environments in which negation and compound manifestations may co-occur, later concluded that in all such cases of negative compound verb, they “are either semantically affirmative. . . . or emphatic denials of either a spoken or pre-supposed affirmative statement in which a compound verb is

Page 281: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

279

present” (Hook, 1978: 134). Thus, it was claimed that a sentence such as in (19) was semantically an affirmative sentence, and that (20) was in emphatic denial. The following Hindi sentences are from Hook (1978:133-4): (19)yahaan ruke rahnaa jab-tak we tuahe phso na do do here struck remain until he to-you money not give give –Due “Stick have until he gives you the money” (20)nBn no use phse nahiin do dive I ENG him money not give five DVS “I didn’t give him the money (although you seem to think I did)” 8.1.8.3. Although it needs to be explored whether a co-occurance of Compound verb and negation in Hindi would always have the interpretation suggested by Hook (19874, 1978), one can find examples that go against his explanation in other languages that have compound verbs. For instance, although the sentences in (21)-(22) in Bengali confirm to what he says, those in (23) clearly go against his claim: (21)uni Sos muhurte eSe pORon he(hon) last moment-in come abs fall-pres-pft-Agr “No came at the last moment”. b. ? uni SoS muhurte oSe pORon ni not-Agr “He did not come at the last moment”. c. uni Sos muhurte-e oSe PoRon ni moment-Raph “He did not even come at the last moment”.

Page 282: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

280

(22)uni Set muhurte oSe pORon ni, aagai Osechilan before-Huph come-pst-Agr “He did not come at the last moment, (he) came (much) before (that)” (23)a. uni kaal raate kan de phRlon he (hon) last might-in weep-abs throw-pst-Agr “He wept last night”. b. uni kaal raate kan de phElon ni “he did not weep last night”. c. uni kaal raate kaan don ni “He did not weep last night”. (21)b is odd because it gives an impression that it is incomplete. Thus, when annai eSechilkan is added to it in (22), it is perfectly acceptable. That the co-existence of negation and Compound Verb results in emphatic denial is borne out best by (21)c which is the same as (21)b the only difference being that an emphatic –e has been added in (21)c which makes it grammatical. But (25) tells a different story. In this case, a compound verb with a negation marker is completely acceptable. Notice that a simple verb plus negation would carry a stronger denial than a compound verb plus negation as is shown in both (23)c and (24)c: (24)a uni nijor stri-ke more phElon he(hon ) self-poss wife-DO kill-abs throw-pst-Agr “He killed his own wife”. b. uni nijer stri-ke more phElon ni “He did not kill his own wife” c. uni nijor stri-ke maaram ni “He did not kill his own wife”.

Page 283: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

281

6.1.8.5. What is strange here is that (24)c seems to be incomplete, and not (24)b. Thus, (24)c would sound better if it could be as follows: (25)uni mijor stri-ke maaram ni, cune kichu kerechilen else something do-pst-Agr “he did not kill his own wife, he did something else (with her)”. (26)uni nijor stri-ke maaran ni, onner stri-ke morechilon someone-else’s wife DO kill-pst-Agr “He did not kill his own wife, (he) killed someone else’s wife”. Notice that the emphasis in (25) is on the “act of killing” and in (26) on “wife”, and hence the subsequent denials in the following clause refer to these in (25) and (26), respectively. It would be interesting to find out whether the situation in Tajiki and other non-IA languages that have compound verbs is the same as in Hindi, or they are nearer to Bengali in this respect. Maithili seems to be closer to Hindi in this regards. 6.1.9.1. There has at least been one attempt to my knowledge to draw a line of distinction between Compound and Conjunct Verbs with respect to their behaviour under negation. Harbir Arora (1978) claims that negated Compound verbs in declarative sentences result in ungrammatical sequences, whereas with conjunct verbs, negation is possible. She gives the following examples from Hindi: (27)*wah nahii aa gavaa he not come – ptopl go-pst-Agr “He did not come”

Page 284: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

282

(28)us-no koli bhual nahii n kii he-krg any mistake not de-pst-Agr “He did not make any mistake”. 6.1.9.2 Again, one can present examples to show that negation provides no test to distinguish between Compound and conjunct verbs at least in Bengali. Consider the following sentences from Bengali: (29)a. uni naaran jaan (Conjunct verb) he (hon) death go+press+pft-agr “he has died” b. uni naa naaran jaan not “Lest he should die”. c. uni naaran naa jaan “lest he should die”. d. uni naaran jaan ni not –Agr “He did not die”. (30)uni eSe pORon (Compound verb) he (hon) come-abs threw-press-pft –Agr “He has come”. b. uni naa oSe pORan “Lest he should come”. c. uni oSe naa pORan “Lest he should come”. d. uni oSe pORan ni “He did not come”.

Page 285: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

283

Bengali, is most of the cases, does not allow NRG to be place in the proverbial position8. The ideal position for the negative particle is post verbal. In that position, both Conjunct Verb in (29)d and Compound Verb in (30)d give grammatical results. Similarly in both the cases, a proverbal and intervening position of NRG result in optative interpretation9. 6.1.10.1. Apart from kar-insertion, Verb interrogation and Negation tests to determine whether a particular combination is Compound Verb, another test has been suggested by Hook (1974: 182-95), and it was called ‘until’-test or a jab tak - - --tab tak- test in hindi. In this kind of construction, hindi requires that the verb in the first, i.e. jab tak clause, be a simple verb as against that in tab tak-clause which prefers to take a compound Verb. This remains true even if the contents are reversed. Hook (1978:134) gives the following examples from Hindi: (31)a jab tak us no mujhe puse diva, tab tak paan e by-the-time he Erg no money give-pst-agr by then five bai gave the ring go-pst-pft –agr “By the time he gave me the money, it had already struck five”. (31)a jab tak paan e baje tab tak us-me mujhe phsi do five the “By the time it struck five, he had already given me the money”.

Page 286: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

284

6.1.10.2 What is implied here is that if this balance is not maintained in the jab tak ---- tab tak clauses, the result would be either ungrammatical strings, or very odd constructions. One can try to do it and show how the test works. Notice that in the sentence given below. (32)b seems to be worse than (31)b. (31)b ?* jab tak us-me mujhe phse do dive, tab tak paan. Baje the (32)b. * jab tak paann e baj gave the, tab tak us-no mujhe phse dive. 6.1.10.3 It may however be noted that this test does not indicate that jab tak- clause in general allows only simple verbs, because if both the clauses have Compound verbvs the construction seems to be perfectly acceptable. For instance: (31)c. jab tak us-no mujhe phse do dive, tab tak pann e baj gave the (32)c ? jab tak paan e baj gave the, tab tak us-me mujhe phe do dive. This would mean that a tab tak-clause can never have a simple verb. Consider the following sentences (31)d. * jab tak us-no mujhe phse dive, tab tak paane baje the. (32)d * jab tak paan e baje, tab tak us-no mujhe phse dive the. 6.1.10.4. It may be mentioned here that jab-tak test (or, ‘until’-test) is very similar to another test called ‘By the time’- test which in hindi involves the same adverb ‘jab-tak’ overtly.

Page 287: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

285

The difference between the two uses of this adverb lies in the fact that under’By-the-time’ –interpretation, jab-tak takes simple verb and allows the tab-tak clause to take a non-simple as well as a compound verb. Consider the following examples (32)jab tak us-no mujhe phse dive. By the time he-Erg no money give-pst-Agr Tab tak merra bhaai jhar se By then my brother home from Khaa (kar) aayan thaa Cat-ptopy DO come pst-pft-agr “By the time he gave me the money, my brother had already come back from home after eating”. (33)jab tak wah mujhe phse dogaa, tab tak by the time ho no money give-fut-agr by then kyann nN us paan e rupaye ke-liya What I that five rupee for {?ro dunn / rotaa rahumn} ? {weep-abc give-press-Agr/ weep-pte prog-pres-Agr} “should I weep/go on weeping for the five rupee till he gave no the money?” Normally, in such sentences the compound verb or the verbal conjunct sounds better with negation as in the followings (34)jab tak us-no mujhe phse diya, tab tak meraa bhaai ghar so khaa (kar) nahiin aayaa than “My brother did not come back from home after eating when he gave me the money back”.

Page 288: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

286

(35)jab tak wah mujhe phse degan, tab tak kyan nE us paane rupaye ke-liye {ro na dun n / na rotan rahun n } ? “Should I weep/go on weeping for the five rupee till he gave no the money ?” 6.1.11.1. While trying to make a distinction between Compound and Conjunct Verbs, Marbir Arora (1978) at one point says that a conjunct verb may be followed by a verb as in (37) that follows, but that a compound verb plus a simple verb combination is unacceptable, as in (38); (37)nBn – no uskii maini kar dii i-erg he-obl-poss help do give-pst-Agr “I helped him”. (38) * wah khaanan khaa la gayan he food eat take go-pst-Agr “No ate food”. 6.1.11.2. But the same construction is acceptable in Bengali as the following example shows: (39) anni eke Saahaaije kere gholaahi I him help do threw-press-pft-Agr “I helped him”. What is important to understand here is that there are various co-occurrence restrictions (which are language-specific, i.e. they may be different in Hindi and Bengali as a comparison of 37-38 and 39 would show) that make a particular CONJ V +V or COMP V +V combination acceptable or unacceptable in a language.

Page 289: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

287

6.2.1.1. This discussion brings one to a more interesting question. What are the possible Verbal environment that Compound Verbs can enter into, and what could be possible arrangement of elements, I believe, are at least of the following types: (40) i. V[ i. V- (kao) – CV ]V 10 ii. V[CV – (kao) – V]V iii. V [ ConjV – (kao) – CV]V iv. V[ CV – (kao) – ConjV]V 6.2.1.2. Notice that there is a fifth possibility which is not grammatical because the last verb has to be finite by nature. This logical possibility is as follows: (40) v. * V[CV – V – (kao) ]V 6.2.1.3. Within the structure of Compound Verb, there could be the following types of arrangements: (41) i. CV [Vpol – Vvee ] CV ii. CV [Vpol – ConjVvee]CV iii. CV[ConjVpol – Vvee]CV If one agree with Hook (1974) that at least for Hindi, choR do and rakh do (which are themselves Compound Verbs) can function as vectors, and if some of the Verbal Conjuncts such as le jan (see Hook, 1974, 1978) are allowed to function as vectors, one can did two more types to (41) above:

Page 290: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

288

(41) iv. CV [Vpol – Cvvee]CV v. CV [Vpol – Vn – Vn vee ]CV 6.2.1.4. It may, however, be mentioned here that such combinations as (41)iv and v are unacceptable in Maithili as Compound Verbs. Also note that if we reverse the elements in (41)iv, it will be the same as in (40)ii and will lose its compound Verb character. I shall now present Maithili examples of the construction-types (40) and (41) I-iii in that orders (42) aatna hatyaak anak saadhan nE son Dumb1 nar1 jannab suicide-poss many ways in from drown-ptopl die- abs go-inf sab sOn besil kaSTkar hoich a (40 I) all of most painful be-pres-Agr “Of the different ways of commiting suicide, to die by drowning (oneself ) is the most painful of all”. (43) o jaldii jaldii kaari IO kao paRaelash a he (hon ) quickly eat-abs take ptoply DO run-pst-Agr apan ghOr dis1 (40 ii) self-poss home towards “Having eaten quickly (like a gourmet) he ran towards (his own ) house”. (44) e eh1Thaan dum-caari din majan lumT1 uR1 gela (40 iii) he this-plaed two-four days fun lest-pte fly-abc go-pst-Agr “No, having had fun there for a few days, went away”.

Page 291: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

(45) o iikSyan sOn sadha bhae ekar baat par he maline from blind be-pte he-pass words on jal1 uTh kao chalaanga naaran rahath1 chat sOn (40iv) burn-abs rise-pte rise-pte DO leap kill-pft be-Agr roof from “Being blind with envy, (and ) heated (to hear) his words, he leapt out from roof-top”. (46) ghar nB Dakaita kBn dekh1 tah 1 o bhae sOn room in dacoit DO see-the moment-Emph he fear from kaanv1 vTala (41 I) shiver-abs rise-pst-Agr “The moment (he) saw the decoit in the room, he started trembling in fear”. (47) ek goTan kBn apa naa dis1 abaita dekha dejh1 e takha nah1 one person-DO self-to side come-prog see-ptopy he immediately naRaaao ThaaRa bholaaha (41 ii) run-abs stand be-pst-Agr “seeing a person coming towards him, he immediately run away”. (48) huma kan ta han eh1 bor 1 sang a lao khapa baah1 (41 iii) him (hon ) PART I this-time along take-abs spend-fut-Agr “This time, I shall ruin him along (with myself)”. Notice that one could insert kao in (41)-(44) optionally, thereby generating forms like Duubi kae mari jaseb, naari khaa kao gollah a, and maina luuri kao uRi gela. In the following paragraphs, I will be concentrating mainly on the construction-type (41)I, i.e. the sentence-type such as in (46).

Page 292: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

290

6.3.1.1. Before we start exploring as to which of the verbs function as vectors in Maithili, let us at the outset make it clear as to which verbs may not be considered as vectors. First, of all, verbs of being such as he, aah, rah are used as auxiliaries (quite often to denote tense, mode and aspect, and at least in the present tense their use is obligatory; thus, kaalaaha “did” , and kara taaha “will do”, but karaita chatti “does”. Rah has been considered as vector verb in Hindi by a number of grammarians, but most of them did not make any distinction between two types of use of rah, examples of which are given from Hindi; (49) aur yah kahte kuo motii - - - -apne bandhum ko pass and this say-ing own friend of near aa kar so rahan (Hook, 1974:38) come DO sleep be-pst-Agr “And saying this Noti- - - -coming near to his friend went to sleep”. (50) raam ke he r th gumkh rabo the (Hook, 1974:96) Ram poss lips parch Prog past-agr “Ram’s lips were parching (/had become parched)”. 6.3.1.2. These who made such a distinction on the other hand describe rah in (4) as a vector and the same in (5)) as a progressive aspect market. But in all the examples comparable to rah in (49) what we find is that it occurs only with stative verbs, and has an effect of prolongation of such a state. Hook (1974:142) did note this quality of rah and commented ; “Every case we have seen of the use of vector rah involves a change

Page 293: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

291

with an ensuing fixity of state - - --rah –sequences are semantically very much akin to conjunct sequence”. But somehow he then compared these examples with rah (he had leT-rah, he-rah, baiTh-rah, etc) with their counterpart in Verbal Conjunct by inserting kar in between those verbs, and concluded that since they gave “quite different meaning”, rah in the examples without kar was used as any other vector verb. 6.3.1.3. This is a very suspicious argument because in most of the cases if one adds –a hhua in between the two elements, the (progressive) aspectual nature of rah becomes clear. Also, there is no significant change in meaning between so raho and see hua rahu (both meaning “ve asleep of a while”), or between loT rahive and loTa hue rahive (meaning “keep lying down”). The only difference is that –e hue makes it more explicit, and presupposes that the addresse has been lying down for sometime. 6.3.1.4. Again, kar can be used in some cases without any change in meaning in Hook’s (1974) own Hindi examples. Consider the following sentence: (51) nBn jab yah baat soctan hunn to hat buddhii I when this matter think-aux-pres then puzzled He rahtaan huma Be be –aux – pres-Agr “When I consider this matter I lose my wits”. The meaning of (51) remains the same even if one inserts kar hore: (51’) nBn jab yah baat sectan hum n to hat buddii he kar rahtaa hum n / he kar rah innatan hum n

Page 294: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

292

6.3.2. Secondly, it has already been mentioned that Maithili does not have compound verb constructions such as in (41) iv and v. Thus, in Maithili, chOR1 do, raakh1 do and IO jaa cannot be used as vectors. The vectors in Maithili are all otherwise simple verbs. 6.3.3.1. Thirdly, there are a few verbs in Maithili which function as models, and those can be used with almost any verb, and thus no selectional restrictions need be stated for their use. These are sak, uaa, and naar 11 all of which are ‘abilitative’ in meaning, and out which is a ‘completive’ marker. Aak is more commonly found that naa, and naar is found only in certain dialects. Naa, however, is used frequently in conditional, infinitive and similar constructions, e.g., (52) jOn ham etae jaa pab1 tahu n if I there go could “if I could go there - - --“ (53) ahaan k otae jane paayab kaThin-e kanj thiika you-poss there go able-inf difficult-emph work be-pr “It is difficult indeed to go to your place”. 6.3.3.2. Maithili has similarly with Bengali in that unlike in Hindi, the abilitative markers here could be used as main verbs. For instance: (54) ii kaaj ham nah1 sakab/paarab this work I not able -fut-agr “I shall not be able (to do) this work”.

Page 295: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

293

6.3.3.3. Compare the following example from Bengali with (54): (55) aani ei kaahTan paarbe naa I this work-class able-fut not “I shall not be able to do this work”. 5.3.4.1. Fourthly, I do not include what has been called the ‘Unit Idioms’ or ‘Constelled Main Verbs” (Hook, 1973:72-5) in the list of Compound Verbs here because these are virtually frozen of closed constuctions. Some examples of such constructions in Maithili include: (56) ghur 1 phir1 waRh mougii waRh giit gane turn-pte invert-pte same woman same song sing-pros-Agr “The same women sings the same song again and again”. (57) o hulak1 bulak1 kaa cal 1 goliih a she poop-look (-furtively)- ptopl DO walk-ptopl go-pst-Agr “Se walked away having peeped in standing on one’s tee”. (58) kumaR1 ahummaR1 kao baadal barsala make –sound –ptopl DI clound rain –pst-Agr “The clouds rained having mae tremendous noise”. (59) hama raa sOn sain tab usaarab nah1 host a I-obl by arrange-inf put-into-not be – pass-fut-agr order-inf “Puttin things into order is not possible for me”. (60) aae bhar1 aan th1 raaj I iE, kaalh1 sOn ta up2 basse today only dress make-up-ptopl take-imp tomorrow from Part fast – Emph “Do your dress and make up till today, tomorrow onwards (you have to ) fast”.

Page 296: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

294

6.3.4.2. This kind of constuction is not now to Modern Maithili, because even in the Garyaapada (9th – Iith C. A.D) such combinations could be found. Consider the following example from the Garyaa: (61) aahere12 ahini mali acchahu kiisa (Bhusuku, 6.1.1. ) who-Do take leave be-press-1st p how “What:to take, (what to ) leave? How to live?” 6.4.1.1. While talking about Compound Verbs, Nasica (19876: 148) had suggested that “these usages are perhaps espicially characteristic of colloquial languages and have found their way into literary language ( and then to the notice of gramarians ) very unevenly, though more in India than in Europe"” This is confirmed when we look into the kind of Compound Verb constructions found in the oldest texts of Eastern NIA languages 6.4.1.2. I shall now take up the Garyapada, the oldest text found in any of the Eastern NIA tongues written at a point of time when this branch of Indo-Aryan was about to split into what we now know as Bengali, Assamese, Oriya and Maithili, etc. It is a collection of about fifty songs each of which had an average five 13 couplets. These mystic songs were written by a number of saints belonging to the ‘Sahajiyaa’ sect of Tantric Buddhists between 9th to 11th Century A.D. The text has a large number of Conjunct Verbs and also many instances of Verbal Conjuncts 14. 6.4.2.1 Here are a few examples of Verbal Conjuncts:

Page 297: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

295

(62) gaabare n 15 teliaa 16 popaa jana gholiu 17 (Krishne, 12.3.2.) elephant DO move-ptopl five-pawns destroy-pose-pst-Agr “Five pawns were destroyed by moving the bishop” (63) causaThi 18 koThan aanisa lahu n (Krishna, 12.5.2.) sixty-four squares count-ptopl take-pros-Agr (1st p) “I take, having counted sixty four squares”. (64) aaila garashak apoNe bahiyaa (Biruaa, 3.3.2.) come-pst-Agr customer self by row -ptopl “The customer came rowing by himself”. (65) suma 19 laina 20 apaNaa caTaarian (Saanti, 26.3.2.) void take-ptopl self scrap-pass-pst-Agr “By taking the void the self is scrapped”. 6.4.2.2. Certain things become very clear when we look into these examples carefully. First, almost anything can intervene the two members of a verbal conjunct. In (62), it was a direct object, in (64) it was the subject with an oblique pseudo-reflexive, and in (65) it was again an object that came in between two parts. Secondly, a reversal of the verb with conjunctive participle and the main verb was allowed in the text. The instances are so many that it is difficult to believe the explanation that this feature is only due to the fact that the text is in verse. The number cannot be so high unless they were current in the spoken languages too. Finally, the commonest form of the conjunctive participel seems to be –I(v)aa (which is the same in the Saadhu’ or ‘High’ variety of Modern Bengali). 6.4.3. As expected, the Conjunct Verbs in this text also allow different things to intervene, and here too both the

Page 298: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

296 orders are current. Here are some of the Conjunct verbs that are extremely found in the text with their word-to-word gloss;- (66)a. Verb-final Orders. Tikhe-karia 21 “firm-doing”, thira-kari22 “fix doing”, nibhara-tarai23 “fearlessly-aresses”, abara-karia immovable-doing”, nide-mala25 “sleep-went”, haaka-nalaa 26 “call-fall”, daava-dena “stake-give”, haara-innix28 “path-ge”, kaabi-kiya 29 “beat-doing”, hara-kiri 30 “crossing-do”, etc., b. Verb initial orders: “lahu-(re-)paasa31 “take-near”, haiba-raarageamii32 “will be going across stream”, habu-bhaante33 “be-mistaken”, de-ankabaali34 “give embrace”, kariba-mibaasa “ will-do staying”, karibha-(ma)samba “will do intercourse”, kin-sabharahe37 “did-ornamentation”, kia-keduaala38 “did-care’, bhaila -uchaaha39 “became delayed”, etc., 6.4.4.1 The most clear and convincing cases of occurrence of compound Verb in the Cary an text are the following sentences as noted in Singh (1979b:59-60): (67) ndi gala40 candraa rabi astaanga (daarika, 51.4.1.) rise-abs ge-yat-Agr moon

sun eight-limbo-lee “the moon has arisen, (and) the sun (is in ) eight limbs”.

Page 299: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

297.

(68) panea-naale nTha41 gala paaNii Crunjarii/Ihaama, five-chamnele-through lift up- go-pet-Agr water “water went up through the five chaneles”.

69) taa mahaamuderii42 tuti gali Kankhad43 (TaaRaka, 37.1.2.) that Mahaamudraa- break-abs ge-pst-Agr desire “The desire of that Mahaamadraa was lost”.

6.4.4.2. Notice that in all these examples, the vector is a form of Quinn ‘ge”. The commonest motion verb is so, and it is not quite unexpected that out of the languages that are now known to have Compound Verb, i.e., Indy-Aryan languages, the Dravidian group, Tajik, Uzbek, Kirghis, Turkish, Mongolian, Korean Japanese and Dharmese – all have at least 60 as a vector. The only exceptions seem to be Santhali and Singhalese which have only two or three vectors (Mastic, 1976:146-7) all of which are comparatively uncommon I others. Secondly, although the vector was the same in (67)-(69), these sentences had three different verbs acting as pillars which proves that 60 was a very common vector and that these are not instances of itemized verbal combination. Thirdly, there are two different markers tagged t the polar verbs here: -1 and –2. The first one is attested in modern Nathalie as noted earlier, while the second one is found in present –day Denali. The reason for these being two marker, I guess, could be that the saints who used –1 came from a dialect-area closer to the western branch of Eastern NIA, whereas these who used –2 came from the more eastern region . Or, it is possible that both were in use at that point of time.

Page 300: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

298

Fourthly, gale in (69) agrees in gender with the feminine noun kapkhaa ‘desire” which supports the claim that some of the composers must have originated from an area where gender-agreement was still prevalent. Modern Nathalie and Oriya still have traces of it while Bengali and Assamese have lost it. Finally, what is remarkable is that at least I these instances nothing has been allowed to intervene the polar and the vector.

6.4.5.1. There are some more sentences in the Cary an which I have argued elsewhere (Singh, 1979b: 62-3) show Compound Verbs with reversed order. Consider the following sentences: (70) baajule44 dila maha-kakhu45 bhaNias (Khaade, 33.4.1.) Baajula-Erg gie-pet-Agr ignorance –sphere say-abs “Baajula showed me the abode of ignorance”. (71) Na jaanani46 apa kani47 gai paiThaa (aajadeva, 31.1.2.) Not know-pres-Agr self where ge-abs sit –pet-Agr “I do not know where the self has entered”. 6.4.5.2. In (70), bhaN- “say” followed de- “give”. If just for this reason one describes the verb SAY as a vector, it would be very peculiar because no known language seems to use this particular very as a vector, In terms of meaning, the nearest to is it found in Burmese nya “Show” which us used a vector in this language. Moreover, the overall meaning of dila… bhaNias is translatable as “showed” or “exposed” (Sam, 1948:114) which shares more with SAY than with GIVE. And since it is the polar that bears the major burden of meaning while the vector is only a modifier, meaning-wise, bhaN- seems to be the closest to the

Page 301: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

299

conception of polar verb. Finally, bhaNias has a marker –iaa which is some for the modern Eastern NIA tongues is one of the markers48 for polar verb. Also, dile is carrying the dissidence. All these prove that for some reason (be it figurative or material) the Compound Verb expression bhaniaa dila has been reversed. Interestingly, this is the lone convincing example of compound Verb where something (here an object) has been allowed to intervene two numbers of the Compound.

6.4.3.3. (71) has a reversal of a different kind. This kind of reversal is very common in Modern Hindi. Thus, gai paiThaa “entered’ is comparable with Hindi jaa gusaa or jaa baiThaa (Hindi also has a verb –naiTh “enter , sit” used rarely,) where what happens is that the polar and the vector interchange their positions as well as their markers. Unless jaa baiThaa is used as a Verbal Conjunct in which case it would men “sat, having gene” (or, “went and sat”), it would be a reversal of baiTh gayaa “”sat down” Similarly gai naiThaa should have a normal order equivalent naiThi gale (or , paiThaa glaa / naiTha gale: the vector could alo be gan or gai , etc.,) Moreover, although COME-OUT is used inb Usbee as a vector, so far we know of no language where ENTER has been used as a vector. Finally, although it is true that different linguists translated this combination in different ways and some have even taken it as a Verbal Conjunct (Sam, 1948 for instance), but in all such translations the meaning “movement into” or “enter” has found place, which shows that ENTER is the focal point of this construction in terms of meaning.

Page 302: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

300

6.4.6.1.The following combinations may also be interpreted as examples of Compound Verbs: (72) pahila49 takiaa bahiaa naraariu 50 9Krma, 12.3.1.) first lift –ptepl pawas kill-pase-pat-Agr “First, pawas were erushed (by no)”. (73) eia bikaraNo51 tahiTali naisai52 9Aajadeva, 31.3.2.) heart senseless-Brg there fall-ptrpl enter-press-Agr “The senseless heart enters there”. (74) Ga(a)Nasanude53 Talina naiThaa (khaade, 35.2.2.) sky-sea-lee fall-petal enter-pet-Agr “Ne) entered into the sky-sea”. 6.4.6.2. The combination naKiaa nar-(Cause-Desinonse) has parallels in Modern Hindi tak naar- , but nar-/naar- may not be a vector in all the modern “Eastern” NIA languages. It functions as a vector in Maithili only in limited cases. One could, therefore, take it as a Verbal conjunct and translate it as “lifted and (get) killed ‘ as some commentator’s have indeed done. But the other two combinations pose a different problems altogether. The first thing that strikes one is the fact that both involve the same verbs. But the markers are different in (73) and (74) . In the first case, the polar marker is –1, and in the other it is –ian. Similarly, nais- “to enter” in (73) is in th present tense, and hense has an usual third person singular marker –ai (akt. –ati) But in (74), it has assumed a past participial base UpaiTha – and has the usual declensional ending – ns (paiThaa < skt. NravissTash) . Both of these constructions are translations are translatable into Hindi as jaa Susana “entered”.

Page 303: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

301

Now, most of the linguists translated these as “fall and (be) absorbed” or “fell and entered”. The reason for this to my mind lies in their inability to interpret this construction as a reversed compound Verb where the original form could have been something like paisi Thalai (for 73) and Paisi Talias or the root cause of this inability is in the fact that a great number of these commentators are from Bengali, and Modern Bengali, their mother tongue, does into have these reversed Compound Verbs where the polar and vector exchange their positions as well as their markers. Thus, while Thuke ias – in Modern Bengali would mean “enter”, gie Thak – would always mena “go and enter”. The verb nOB- “fall” which is the same as the Caryaa Tal- “fall”, and which is in use in Modern Bengali as a vector can combine with Thak- “enter” and form a Compound Dhuke nOB “enter’ , but I am sure all would agree that a reversal of this in Modern Bengali, *poBe Bhek -, would sound nonsensical. I would therefore think that there are nough reasons to suspect that Tali paisaiin (73) and Talina paiThaa in (74) are reversed Compound Verbs.

6.5.1. The following is a list of “probable” Compound Verb explicators or vectors in Maithili. (The arrangement has been made more or less impressionistically at this point and it would be revised in the subsequent sections:

Page 304: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

302

(75) a. Very frequent: iaa- “go”, aa- “come”, de- “give”, la- “take”, nTh- “rise”, naB- “fall”, rakh- “put”. b. Lees frequent: Hais- “sit”, cal-U “walk/go”, dhar- “put/catch”, naThaa- “send”, nikal- “emerge”, nikaal- “release”. c. Infrequent and rre: Nan- “bring”, nar- “die”, naar- “kill”, khasnam- “bring”, UnarU- “die”, naar- “kill’, khas- “fall”, dhak “see”, vhenk “threw”. 6.5.2.1.. Before I present examples of each of these vectors where hey combine with various verbs, I would like to talk about those verbs that can fill the slot of polar and vectors at a time. In the neighboring la nguages, this kind of contractions is limited to de-de, la-la in Hindi and Bengali (of course, Bengali has an additional bKe-bKE combination 54, where it means “wander around”). In Maithili, the verbs such as cal- “walk/go”, ias0 “go”, DaB- “fall”, mar- “die”, dhar- “put/hold” can be used twice in a Compound Verb combination apart from de- “give” and “le- “take”. 6.5.2.2.I would now present examples that support this combustion: 6.5..2..2.1. iaa-iaa” (76) aaikhum baaraat jaae jaetaah today-time marriage-party ge-abe ge-fut-Agr “The marriage-party would go today”.

Page 305: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

303

(77) e-sab hamara ghar so jaae jaetaah he-pl my house from ge-abs go fut-Agr “They will go away from my house”. 6.5.2.2.2. de—de: (78) humkaa e bast do dianh ne hin (hun) that thing give-abs give-imp-Agr Bnph “why don’t’t you give him that thing”. (79) O hamara aai anh ta kaalh pani He(hun) no today not part tomorrow money do-e deaah Give-give-abs fut-Agr “If not today, he would give me the money tomorrow”. 6.5.2..2.3. le—le : (80) je hinchaa bhelanh se Ulo lal thinkU what wish be-put-agr that take-abs take-put-Agr. “He took away whatever he wished”.

(81) ekhan ii 10 liyah , baa kii baad now his take- imp rest labor nl deb elm give fut. “Take this much now, (I) shall give (you) the rest later on “.

6.5.2..2.4.paB---paB: (82) biFhniiga Riyaak yhalna-baalum ayan Birhnigaria -pess X – Mr. self-press be Taak naukrik lel jhaa-jiik son-press job-pass for jha-Mr. pes upar Upal paH laah On fall-fall-put-Agr “Mr .X from Birhaigaria almost thrust himself upon Mr .Jha for his son’s job”.

Page 306: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

304 6.5.2..25. cal—cal : (83) yKh bumjhum je aab bhaii-jii this-emyh understand –imp that now brother cal cal lach walk-abs walk-pet-Agr “you may take it now that (big) brother has almost started the journey for himself (m. about to die)’’

(84) etas so cali-ye calm, sKh niik here from wlk-abs emph walk-Imp that good haet be-fut-emph “Let us walk away from here, that will be good”.

6.5.2.2.6. dhar---dhar( 85) hamara lag gaar pakae astaak ta ne near fall-inf come-fut- Part ham nareti-b Taa dhar dhar hamp ki I meer- emph class catch –abs catch-fut-Agr what “If he comes near me and calls name, I shall catch him by his neck”.

86) banbas-jiik kamal Dul-e taa dhar dhar sadhu-Mr.-pess bowl-emph class catch-abs banh ta e jaetaah kamaa? Catch-subj-Agr them he (hum) ge-fut-Agr has ‘How can the sadhu go if (I) catch hold of his bowl?

Page 307: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

305 6.5.2..2.7. nar----nar : (87) Biihak jamiinak lel e niseas homestead – land – for he(hon) surely pess pess. Nar nar teah Die-abs die-fut-Agr.

“He will go to the extent of sacrificing his life for the land of the honestead”.

(88) nah kuF ta kar narum not be-pres then die-abs die-Tup “if you cannot bear it, then give up life”

6.5.2.3. Notice that if one includes gak- “to be able to” also (which I have argued against) . one can get constructions with gak- which are “similar” to the ones above. Here are a few examples: Gak –Gak: (89) e kii ii kanj ongare gak gak teah ? he what this work alone emph able –abe able-fut-Agr “would he be able to do this work alone?”

(90) e pahile ber nE nah gak gak leah ta he first- emph cxhance in not ablew-abs able-parts pat- Agr. Desar berak gappe ken? Second chance-pess talk-emph why “we could not do it in the first chance; so why talk about a second chance?

6.5.3.1. I shall now innumerable examples of the ways the probable vectors listed in (75) “may be “ used in order to establish which are the “real” vectors, and them discuss about the semantic modifications they bring in when combined with different verbs as polars. (This, it is not claimed here that the examples in 91-127 are all instances of true compound verb. ) This will be followed by an account by an accounts of reversibility of elements in a compound.

Page 308: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

306

Verb in Maithili and a speculation about the select ional restrictions of various kinds that tare imposed on the Compound Verb formation in Maithili. 6.5.3.2. The following sentences55 show how the verbs listed as vectors in (75) function: 6.5.3.2.1.1. iaa : (91) e bcaare gariibii eo nay gelash he (hon) poor-man poverty from die ge-pet-Agr “That poor man dies out of poverty”.

(92) eh Daag daraiik prayaas eo e ek-ber that lady-doctor-pess effort form she(hon) emetine.

Phor jii geliih Again live ge-per-Agr “she was once again born (-got a new lese of life) because of that lady –doctor’s efforts”. 6.5.3.2.2. em: (93) hamaras-sabhak gaam ml ii prathaa ve-pess-pl village in this ouster bahut din nO cal bab rahal ach many days form be-current come-preg-yet-Agr “This ourstem has been current in our village for a long time”.

(94) eh baat par kunak aankh noraa aelanh this talk upon his/her eyes water come-pet Agr “His/her eyes started shedding water upon (hearing ) these words”.

Page 309: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

307 6.5.3.2.3. de : (95) nantrii –maheday afsar-sab kN ekTaa minister –esquire officer-pl-Do one-class tuech khumlak lel niik jaka ordinary mistake-pess for good manner humth delkach. Rebuke give-pet -agr “The reversed minister chilled the officers in a befitting manner for an ordinary mistake”. phalnaa-baabunk akhab par ham eilaa I-Mr.-pess say-inf on I Y-baabum Ky tenaa kae ne ach delivered Mr.DO in such –a-way Naph-pool give pat-Agr

je aab duureduurahait chath tht new far-huyh fr remain sux-pres –Agr. “On Mr. X’s request, I taught Mr.Y a lesson in such a way that (he) does not dare

come nearer now” 6.5.3.2.1.la : (97) raanm sab-Taa pirakiyaa eeraa kae Ramu all-claass sweatmeat and ate (them) up”. (98) khaaeb ta niik jakan khaeu no, eat -fut-Agr then good way eat-fut-Agr Part hunr lebaak ken kaaj? Stuff-take-inf-pess what was “Hat if you want to, (but) in a sober way, what is the necessity of stuffing (your belly) in s great haste?”.

Page 310: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

308 6.5.3.2.5. uTh: (99) ghar nN takh-a-nah Dakait –a kN-n home in still dacoit DO achait -a- dekh-I kae e bhae sO-n be-ing see-ing DO he(hem) fear from kaa-n-p-I- uTh-a-leah tremble rise-pet-Agr “Seeing the dacoit still present in the room, he started ttrembling with fear”. dui-ye din mN ii lahans ta two-Huyh days in this corpse part saR-I- uThal am?

SaR-I yThat am? Decompose rise Vee-pet-Agr. “Hey, this corpse got decomposed in only two days”.

-n kari-Taa eiKhaa diyanh-u no, you (hon) little-class tease –give-part e ap-apnah-I ukhar-I- saR-a-task-a he (hon) self Bmph flare/upgreet fall-fut-agr “why don’t you tease him a little, he will flare up himself”. humka jakhan non hoit-a ehauh-I katch-u him-to whenever wish be –pts aux-pres somewhere sO-n aab-I Kno huul-I paRait-n chath-I from come-ing DO enter flal-pte-aux-pres-Agr” whenever he feels like, he comes from somewhere and dashes in”. 6.5.3.2.7 rakh: (103) paakum lel-a ham haath-mum-n-h-a pakhaar-a baak guest for I hand-face wash-inf-pess jOI dho rakhaliauh-I duaar-I par water keep put-pet-Agr outhouse on “I kept water at the outhouse for the guest to was (his hand and ) face.

Page 311: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

309

(104) phucaai jhaa bhain daaik lel-a kumbh-a phucai Jha sister elder-pose for Kumbh nelaa so-n samal taabiij ekh-anah-u fair form brough arnlst still raakh-I rakh-anR chath-I keep put-pte-aux-pres-Agr. “phueai Jha has still preserved the armlet brought form the fire of Kumbh for the older sister”. 6.5.3.2.8.haisa: (103) jakhan ahaa-n ii kaaj Uko bais-a lahu-n when you (hon) this work do sit-pet-Agr takhan aaleemnaak dar-akaare kon? Then discussion-pess use what “what is the use of discussion when you have (already) done this work? . (106) hunak nitr-a ta humko lo bais-a lank-I he (hon) –pess friend part he(hon ) take sit-pat- Agr “His friend rained him, too”.

6.5.3.2.9. cals : (107) khaaeb ta jhatpar khea liyeh-a, eat-fut-Agr part quickly eat-ptepl take-imp grahan ta laag-I calat-a ecliyee part hit walk-fut-Agr.

“but quickly if you want to, otherwise it is almost going to eclipsed”.

(108) ahar-n ta bahut din so-n humkar sab you (hon) part many days from he(hon) –pess all kich-u nather-I calaliauh-I, aabe kich-u things acquire walk-pet-Agr still more baa-n kii ach-I ? left be-pres-Agr “You have been grubbing his all possessions for quite some time: is there still some thing left?

Page 312: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

310

6.5.3.2.10 dhar: (109) kaangresiyaa sabtaa kN -u kaalh-I Congress-(leaders) all class DO tomorrow

Pearliyaamam-a nK bachaal dhar-u henh-I Parliament in press held-fut-Agr Taken pataa cal-atamh-I Then information ge-fut-Agr. “When I (will) change the congress-party am in the parliament . only then they would know”.

(109) kattah-u kichuo nukaa kae raaka-I dharma wherever whatever hide-ing DO keep keep-Imp be mamkarnau KH-u Thiik [atan ca;-I that sever ant- te surely information jaait-a ach-I walk-ptepl go-fut-Aun-Agr “Does not matter what you hide and where, the servant would surely come to know about it”.

6.5.3.2.11. pathaa : (111) cartiilaal kH -u has aaiye haisa nathaa Cartilal DO I today-Emph call send-fut- Hank-I kii Agr what “I shall summon cardinal today only”.

(112) hum-akan lol-n han tiin kap case he)hum) – ebj for I three cup tea nanaa nathaaliah-I order send-pet-Agr “I ordered for him three cups of tea”.

6.5.3.2.12. nikal: (113) taka-anan-I ektaa saa-u-pak yumaa nun-nh-a then –Emph one-class snake –pess child mouth baab-I nik-a lal-a open emerge –pet-Agr “Only then a young snake opened its mouth”.

Page 313: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

311

(113) betaa ta dekhait-a chiaik-a baape so-u son part see-pts-aux-pres-Agr father-from Emph hakh-I nik-a lal-a eress emerge-pet-Agr “The son, I see, has (almost) ever taken his father”.

6.5.3.2.13. nikaal : (115) hamar kam-a kirbaa ehen no baat my son(-dear) such Part matter see-I nikaal-a-lak je ham-araa think release-pet-Agr that no-te yhur-abe nah-I kar-a tihaik-a strike-Dayh not de-subjn-Agr “My dear son has thought out such a plan which would have never struck me”.

(116) antaegatwaa e sab – Taa baat jal so-u inally-going he (hon) all-class matter root from ukhaaRi-ye nikaal-a leah-a Uproot –Emph release –put-Agr “Finally, he rooted out/unraveled the mystery from the very beginning”.

6.5.3.4.14 aan : (117) Tikar ta 10 anna, jaseb kii ticket part take bring Imp go-fut-what nah-I se baad na dekhal-a jaeutaik-a not that later on see-p.ptepl go-fut-Agr. “why don’t you purchase the ticket: (we) will see later whether to go or not “. 6.5.3.2.15 nar?118) apuaa ta dhwra-a bholaahe, des-a-re self part destroyed be-yet-Agr- others- kN-a 10 nar-a-leah-a DO take die-pet-Agr “Not only did he destroy himself, (he) ruined others, too”.

312 (119) shaa n to tohan no packhan dn chii you (hon) part such not regue be-pres-Agr jo ahaan Hagi-ye-merab to ham that you (hon) deficate-enyn die- part I fut-Agr

Page 314: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

nah1 aseb net come-fut-Agr “You are such a rague that even if you are vulnerable, I would not come”. (120) ge dase, e ta kan si-ye- par lach” Vee Sister, he (han) part laugh-emph die-pst- Agr “ky ged, be laughed his head off”. 6.5.3.2.16. Saar(121) buckuan-kaa harn bashaa sabran knn rakeep euchkum-male tiller-pl-class-DO scold naarla a thiiun kill-prt-Agr. “Buchkum-uncle scolded the tillers heavity”.

(122) lagnsea-balea Mejn ND O Ok dannil lagnea-of feast in he(hom) one pet bhaat1 naarala kasha rice swallow kill-pst-Agr “He devoured a full plate of ries in the lagna-feast”.

6.5.3.2.17.Khas (123) hmar heath son gilass fuur my hand from glass for ins ktasa” go fall-pst-Agr “The glass flow off at a distance from my hand’

(124) ii sab kD 10khasab salea this all No take fall-taf(fxt0-of matter hanges pasin that yaal mete like not fall-pst-Agr

“I don’t quite like the idea of involving everybody in this matter”.

Page 315: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

313 6.5.3.2.18 dekh : (125) ham am kaa tharese kiyik rahuu, I other –obl dependence-in why be-press-Agr teach I so 2 ap2ash1 kich u that from self emph something ? ? De see-couse-fut-Agr “Why should I depend on others? I would rather do something myself”.

(126) etas kaalh I ias dekhah to puraa there tomorrow

go see-fut-Agr part fall pataa calat’ know go-gut-Agr ‘I will come to know about it all when I go there

tomorrow”.

6.5.3.2.19 who2 k (127) lek a nanyak jayprakaas naaraayan indiraa people here J.P narayan Indira grandhik sar kaar kn n ? Gandhi-press government DO uproot

who2 kal a thiuk a throw-pst-Agr “J.P Narayan, the people’s leader, brought the downfall

fo the government of Indira Gandhi”. 6.5.4.1. We shall now look back into the examples given for each of the ‘probable” vectors in (76)-(127) to find out as to which of these confirmed to the functional criteria of the vector verbs discussed earlier in the section. It would seem evident if we look into these sentences carefully that some of these sentences may look like examples of the use of compound verb, but are actually not so. ‘What is more interesting is that these sentences may involve verbs which are otherwise demonstrable as real vectors. Let us take the case of (93) involving the verb.

Page 316: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

314 In “some “ as a vector. The use of Is indicates that the aspect involved is progressive. This auxiliary can be inserted into other compound verbs too. For instance, one can have the following sentences from (91), (97) and (99), respectively: (91) a became gariibii son That peer man was almost distingue of poverty” (97) ras,, sab-Tea pirakiyas sered kas khas 10 raheal dhal “Ramu was stealing all the sweated and was teaching (them all)”. (99) ghar ne takh nah dekait kn n achait n dekh kae thae so n kes n p I “Seeing the dacoit still present in the room, he was trembling in fear”. Now, compare these sentences with (93) which I reproduce here for the sake of conveniences; (93) hamraa-sakhak gaan nl ii prathaa bekut ida “This custom has been current in our village for a long time”. 6.5.4.2. Out of these sentences, only (91) and (93) have Counter parts such as follows;

(91) a. o becare garibiii son narai jaa rahala chalaaha

(92) a hamaras gaa nE ii prabha bahut din son calal saba rahaa achi

Page 317: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

315 This introduction of past tense marker-all’s does not alter the meaning of (91) and (93). But notice that although (97) and (99) are in the past tense as the markers in the auxiliary show, these cannot have such sentential alternations. Consider the following ungrammatical sequences that result if one tries to do so; (97’) a.*raaman sab-Tan pirakiyaa ceran kaO khaasela IO rahala chalaaha (99’) a. * ghar nE takha nahu Dakait kBn achait a dekha kae e bhao sOn kaan nal a uTh 1 rahal a chal a 6.5.4.3. If we check with all the verbs that have been stipulated as vectors in the examples given earlier, we would find that such constructions are possible with one more verb, ? ? which of course does not always give acceptable sentences. It may, however, be noted here that all the three verbs concerned are verbs of movement of a particular kind” ‘going’, ‘coming’, and ‘walking’. It turns out that these verbs have different kinds of functions. The ? / / that combine with (Where V-a particular act of verbs) are thus not the same as these that are used as vectors. The differences of these two can be shown by the following sentences.

(128) a . e mar2 laah a he (hon ) die-pst-Agr “He died”.

Page 318: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

316

b. o mar-I gelaah-a ho(hon) die -abs ge-put-Agr “He died (-in perfective)”.

e. O naral-a gelaah-a He(hon) die-pft-pte ge-put-Agr “He went on dieing 9-figuratively)”

It may also be noted that –al-a attached to nar- here in (138) is not the perceptive –al-a, but that it signifies “imminence” (which is marked by –2 in Bengali). 6.5.4..4 In (91’)a, and in (132) e, the function for ias and as has been that of a progressive (or iterative) marker, In the first two, they function as progressive indicators in spite of then presence of the traditional progressive marker rah – Here , the presence of “double” progressive markers make the a special reading for the same strong. This , therefore, suggests as vectors. This does not rule out the use of ias in (91) as that of a vector. Nor does it imply that na in (94) is nto a vector. The same comment is applicable for jaa in (92). More so, because (92) and (94) do not allow the progressive use of ias or na : (92’) *eh-I Daag-adarniik prayaas so-u e ek-bar-I pher jiiyal-a geliih-a / iii ina rahal-a chalith-a. (94) * eh-I bat par humak aa-ukh-I naraaal-a aslamh-i/ noras nab-I rahal-a ehalambh-i. 6.5.4.5. Thus, to conclude, ias is possible to use as a vector as in (91) and (92) , but not all uses of ias in similar positions.

Page 319: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

317

Are that of vector, similarly, aa may be a vector in, say (94), but certainly not so in (93).

6.5.5.1. one may be critical about the conclusive statement regarding the discussion that proceed on a different ground altogether, to say that a particular combination of verbs is Compound Verbs and not an instance of other types of verbal combinations discussed earlier, one must have stronger arguments. Such arguments may come in form of tests that can identify Compound Verbs ? uniquely. These tests should preferably be based on the behavior of Compound Verb constructions in larger synthetic environments, and on their behavior vis-à-vis the selection of various transformations to apply on the sentimental structures which have Compound Verbs. We have already discussed at length about some such tests, viz., Insertion of Conjunctive kaa, Negation, verb Interrogation, and jab tak …. Tab tak - (or, UNTIL), and By-the-time tests. We have already seen that while those tests are fairly general, not all of them help us identify various types of Compound Verb combinations that are found in Maithilli and other neighboring languages. 6.5.5.2. Now, if we agrees with heek (1978b) that the perspective/non-perfective contrast is shown in Modern NIA languages (he takes only Hindi, but we can generalize the case) with the help of polar-vector combinations as against simple verbs (of the type 6a). we would assume that one of the most important functions of the vectors in these languages would be to “mark” that apart from doing this, the vectors may in addition show a lot.

Page 320: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

318

Of other things (see Kachru, 1978b for some of those). These have nicely been summarized by Masica (1976: 143) who says: “From one perspective the function of V2 is varied and manifold: it connects completion, giddiness, directionality, benefaction, intensity, violence, stubbornness, reluctances, egret, forethought, thoroughness, etc., depending not items involved and on the circumstances.”. If it is true that along with those different shades of meaning, the V2 or the vector shows “persfctivity” them we may add a few more tests sch as the phasal Verb test (periaka), 1967: Neak , 1978b),Apprehnsion-Test, (Farayth, 1970” Verma, 1975: Neak, 1978b) Compound Verb Education Test (Neck, 1978b), Vasudeva/contradiction-Test (Heak, 1974, 1978b) and the Incompletion-test. Let us first see how these tests work in out languages, and then we shall find our as to how many of the probable vectors listed in (75) would pass them.

6.6.1.1. The first of these, the phasal Verb Test, involves the verbs of “phase” , i.e., the verbs that donate the beginning, progression and termination of an event or an action. It has been noticed that the languages showing perfective aspects (by verb-marking o by a vector sequence) do not allow the perfective or combine with any of the phasal verb. The example given by perika (1969:30), and quested in Kock (1978b:9), is from Hindi, and it is as follows:

(129) A. pardan giraa (Non perfective)

“Curtain fell”. b.pardaa gir gayaa (perfective) “curtain fell”.

Page 321: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

319 (130) a. pardaa girna lagaa (Non perfective) curtain fall-inf start-pst-Agr

“The curtain began to fall” b. “pardaa gir jnane legaa (perceptive)

“The curtain began to fall”. Here, Lag- “start” (which otherwise means “seem”) is a phasal verb and hence the combination of a compound verb gir-jea “fall-ge” (literally) and lag produced ungrammatical results. 6.6.1.2. In Maithili, a variety of verb forms could be used as phrasal verbs. We do not mean to say that compound verbs cannot occur with any of these phrasal verbs in Maithili, but they can certainly provide a good test. These phrasal verbs included simple verbs, conjunct verbs as well as what we have been calling compound verbs. And, different verbs allow different kinds of infinitives before them. The following is a list of some of the phrasal verbs in Maithili:

(131) START CONTINUE END -lag/laag -rah cvk {-Suran-} kar/ho -rah1 jaa {SoSa/-amta/-aapaapta/-argan} -kar/ho -jaa (rah) -usar -aa (rah) -naar laar -cal (rah) -bho jaa -kO lo

?

Page 322: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

320

6.6.1.3.These phrasal verbs require that the proceeding verbs have the following shapes “ (132) PHASAL VERB : SHAPE OF THE PROCEEDING VERBS V-ao/-O V-aba V-ala V-aita V-I I II III IV V Lag : ? x x x x Laag : ? x x x x Suran kar/ho : ? ? x x x Aaranbh2 kar/ho : x ? x x x Rah : x x x ? ? Jaa/aa/cal (rah) : x x ? ? ? Ouk : x x x x ? Usar : x ? ? x ? Paar laag : ? ? ? x x BhO-jaa/kO-le : x ? ? x x 6.6.2.1. The apprehension-test or the Fear-test works in a way that is opposite to the phrasal test. In an appropriate context where someone apprehends about an action or an event, the verb to be selected is a compound verb, and not a simple verb, Consider the following examples from Bengali; (133) a. ammi bHOe paaochillan e mere naa jase I fear get-prog-pst-Agr he die -abs neg go-pres-Agr “( I was afraid) he might die”. b. aami bhOs pasachillan o {*mOre naa/ ? naa mOre} * ( I was afraid( he might die”.

Page 323: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

321 6.6.2.2. Similar examples could be found in Maithili, too, it may be noted that the word for ‘apprehension” or “fear” need not be present in the surface of the sentence concerned. It is enough if these sentences contain some emphatic particles in appropriate positions: (134) a. chan aavaaj sOn regii kaan p1 no uTath1 such noise from patient trouble Neg rise-Fut-Agr “( I am afraid) the patient might get shaken up by such a noise”. b. chan aavaaj sOn regii {* kaa2 path 1 ne / ? * ne kiaan path1} “(I am afraid) the patient might get shaken up by such a noise”. 6.6.2.3. There are, however, counter-examples to this test which make it a little weak. The following Maithili counter example in a parallel to the one quoted from guru (1952: 577) by pavika (1977:72) (133) a, ?? Daraita chii je kyee dekh1 no leta hoth1 feat-pte-Aux-pres that someone see-abs Neg take-be-pr- pte Agr “ (I) am afraid that someone may be watching A Similar distinction between a and b sentences about could be seen if chii “be” is replaced by Chalahun “was”. 6.6.3.1. The Conjunction-Reduction test emerges from a few examples given by hook (1972 a b 98, fn 31). It could be developed into a full-fledged test to identify compound verbs. Hook (1978b:98)

Page 324: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

322 Suggests that “in a series of actions Hindi often reduces all (-compound Verbs: UNS) But the last to non-perfective forms’, and illustrates this point with the following sentence from hindi where apparently the first occurrence of compound verb (which one night guess could be ? ? has been reduced leaving the second intact; (136) us no gilaas uThaavan aur ok hii ghunn R non

he-Erg glass lift-pst-Agr and one only swallow in saaran paanii pii Daalaa

all water drink-abs threw-pst-Agr “He raised the glass and in a single swallow effort break up all the water”. 6.6.3.2. This kind of “neutralization”, he noted, was not possible in some other languages like Russian. We can now extend Hook’s idea in the following way. It could now be predicted that although one may have an alternative to the sentential structure of (136) as in (136) where nothing has been reduced, a structure such as in (136) would be either very odd or unacceptable: (136’) us no gilass uthas lived aur ek hii ghm me saaraa passii nii Daalaa (136’) *us no gilass uthas lived aur ek hii ghm me saaraa pannii nivan Notice that this test does not say a word about the possibility of reducing both the compound verbs from a structure such as (136’) This would give a perfectly grammatical sentence: (137) us no gilaas uThaavaa aur ek hii ghm r me saaraa paanil ?

Page 325: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

323 6.6.3.3. However, compared with (136) this sentence sounds a little worse. But

the acceptability of (136) and (137) and the oddity of (136”) shows that we have to make alternations in the way Heek (1970b) had described the rule of reduction of Compound Verbs (quested earlier) . We would instead say that

(138) Compound Verb-Reduction:

When two or more coordinated clauses have a series compound Verbs (preferably one in each clause: there must, however, be at least two of them in total), an optional rule of reduction may educe one or more of the compound verbs (‘reduce’ – replace the Compound Verb by its polar), but if the reduction takes place only once, it cannot affect the last of the series of Compound Verb alone thereby leaving the other occurrences of the same intact.

6.6.3.4. In the light of this characterization of a constraint on the rule of compound Verb reduction, we can now formulate the test easily. To test whether a particular verbal combination is compound verbs in coordinated clauses, and them try to apply the rule of reduction. If the application of this rule gives an ungrammatical sentence, the combination involved is a compound Verb. But by now it must also have been clear that to apply this test successfully, one has to establish at least a few “genuine” compound verbs which could be tested. Therefore, this test can only be taken as a supplementary to other independently.

Page 326: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

224

established tests. More over, this test is negative in nature as it does not say anything about what happens when a combination of verbs remain grammatical after being reduced in the position “marked’ by (198). What it says is that a combination is a compound verb if the sentence is UNGRAMATICAL after a reduction rule has applied. This does not mean that a particular combination is not a compound verb if it does not less the grammaticality. 6.6.5.5. To make this test meaningful, one must point out that there indeed could be a situation where (138) is violated but still a particular combination has to be defined as a compound verb, because it fulfill other criteria. This shows the further weakness of this condition. The following sentence from Hindi would illustrate this point:

(139) a. us no khud to cain kii niin d so lii, par unpar he-Erg self Part comfort of sleep sleep-take-pst but up aa kar, baooe n ke apne bistare par aaraam so committee-ing DO children-Obj self-poss bods on comfort with sets dekh kar, gusse so sag babuulan he kar, sleep-ing see ptopl Do anger from fire be-ing DO unke uThaavan (/uThan divaa) them wake-up-pst-Agr (Wake-up-abs give-pst-Agr) “He had himself had a comfortable sleep, but (when he) went up and

saw the children sleeping delightfully on their own bods, he flared up with rags and woke them up”.

Page 327: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

325

b. us no khud to cain kii niin d so lii, par unpar aa kar, baceesn ke apne bistare par aaraam se sete dekh kar, unke uThaavan “He had himself had a comfortable sleep, but (when he) went up and saw the children sleeping deligtfully on their own beds, he wake them ip”

c. us no khud to cain kii nii n d so lii, par unpar aa kar, bacceen ke sete dekh kar unke uThaavan “He had himself had a comfortable sleep, but (when he) went up and saw children sleeping, he woke them up”. d. us na khud to cain kii nii n d so lii par vaces n ke sete dekh kar unke uTaavan “he had himself had a comfortable sleep, but seeing the children sleeping, he woke them up” e. ? us no khud to cain kii nii n d ne lii, par baces n ke sete dekh kar uThaavan “He had himself had a comfortable sleep, but he woke the children up seeing (them) sleeping”. f. ? us no khud to cain kii niin d so lii, par unpar as kar bases n ke uThaavan “He had himself had a comfortable sleep, but (he) having gone up, woke the children up” g. * us no khud to cain kii ni n d so lii, par bases n ke uThaayaa “He had himself had a comfortable sleep, but woke the children up”.

Page 328: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

326

6.6.3.6 The sentences in (139) a through indicate that the length of the clause where the last of the series of compound verbs is placed IS one of the fasten that decide whether the reduction of compound verbs by violating the constraint in (13a) gives grammatical results. It may be the ‘Length’ of the clause or it may be the ‘distance’ between the two assurances of compound verbs in such an environment as in (139). Whatever be the reason, this much is clear that although (139) a to d are all acceptable sentences in Hindi, the lengthier the clause, the higher is the degree of grammaticality. That is, (139)a is better than ab, and b is in turn better that e, and so on and as forth. (139)e and f are equally add, but the sentence in (139)g is the vest of the let, because the distance between ae-1ii and uTan(-di)yas is the minimum in this sentence. These again prove that this test is relatively weaker than either

6.6.4.1. The Vasudeva-test or the contradiction-test as we would call it originated from Hook’s (1974) dissertation where he showed that in a sentence such as in (140) in Hindi, the first clause cannot have a compound Verb if the following clause ‘sentradicts’ the action described in the first clause:

(140) a. taswlir decision ku´’, taswiir sahit” bentil picture makes-press- axe-agr- (-X) picture net make-pre-Agr-(Intr) “(I) try to paint (her, but) the picture does not become”. B “Taswiir makes leter kum”, taswiir mahil” bent ill make-abs take-pft-gun-press-agr “(X) paint (her, but) the picture does not become”.

Page 329: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

327

Far as its first part is concerned: “The trumpet (declaring) an end of the sitting blew up (just because the ) king gesticulated ( I may be it wouldn’t have, had it been anyone else)”. This surely not implied by (81). I did not mark (81) as odd because the second part here does not centralist the first. It is thus evident that the ending sentences of the medium length which have coordinate structure show a preference for retaining the normal word order unless such a retention gives an altogether different meaning. In two other cases where the reordering took place in Manipedma (1972: 74& 112), what happened was that in course of joining two actions which took place one after another or simultaneously, the event that assured first or whichever is unimportant was placed in an adverbial phrase before the matrix sentences. They are thus met cases of simple adverbs being proposed for rhetorical purposes. Only in one important a reordering was done with this purposes. This sentence is as follows:

(81) Staidhas, prem an sauhayisak Vastasvaral” 31

as respect, leve and good wish-pess environment in naikas as dev”pedma agman dal sahit ratnasen so naika and Devapadma self -poss group with natmasema from vida length”32.

Leave takpet-Agr “In an environment of (mutual) respect, lave and good wishes, Maika and Deevapadma with thirty en took leave of ratuasena”.

Page 330: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

328

Yuddyan wiswa-saanti-k lel” runs an am rlikas although world-peace-poss for Russia and Anerica salt-Tum ki” ( ‘magn”lelak” magn”lak” agree-abs take-pet-Agr Agree-pet-Agr) tathaap” wiswa-samti Salt-2 obj still world-peace Nah’ Asel” Not come-pet-Agr ‘Although Russia and America agreed to SALT-2 in order to (achieve) world-peace, the world-peace did not come”. It may be mentioned here that the effect of using the beginning of a sentence is that it expresses a doubt in respect of the festivity of the context of what follows- Thus, in (141) with the introduction of , a doubt as to whether “all the members of the family died” is expressed. This is the reason why a translation of such as “although’ is only an approximation, and it does not tell everything. What these sentences show is that one can use the contradiction-test with only a limited degree of success. Alternatively, one can select only these contexts, which give clear verdict in respect of grammaticality, and keep the other contexts away.

6.6.5.1. The last of the tests mentioned earlier is called the incompletion test. This has to do with the fact that in ”most of the cases, the motion of perfectivity overlaps with that of completion. If this is true, than one can always find out contexts in particular languages where the first part of a sentence would have a compound verb and the subsequent joined

Page 331: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

329

Sentences would indicate that the action denoted by the compound verb in the previous sentences was not completed. Take the following Mayhill sentences for example:

(143) thsahit per atthasais-Tea rasgullas raskhal” dhal”

plate upon twenty-eight-diet Rasgulla keep-p-pte be-part-Agr as khesaarii between thashil parak rasgulla nad khesari Mr. Plate upon- poss Rasgulla

Kheal’think” Eat-put-Agr

“There were twenty-eight Rasgullas kept on the plate, and Khesari Bathu ate Rasgullas (kept) on the plate”.

(144) Thsakil far aTThsais-Taa Rasgulla raskhal” ehal” an kheseariii beabum thsaili parak rasgullas khas

eat-abs ge-pst-Agr

‘There were twenty-eight rasgullas kept on the plate and Khesari Bathu ate up rasgullas (kept) on the plate”.

Notice that (145) a has a simple verb Khas- “eat” whereas (145) b has a compound verb Khas-ias “eat up”. How, to test whether Khas-ias is a compound verb or not, we can extend the sentences in (143)-(144) in such a way that the extension indicates incompletion of Kheasari Babu’s task. Consider the following sentences:

Page 332: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

330

(145) thsarii par aTTasis-Tsa rasgulla rsakhal” ahal”, aa khasaari bathu thgaill parak rasgullas khasl think , mudaa tiin-taa rahi-ye gelash but three-dot remain-abu- ge-pst-agr “There were twenty eight rasgullas kept on the plate, and Khesari Babu ate rasgullas (kept) on the plate, but three of them were still left behind.

(146) thsarii par aTTasais-Taa rasgullah raakhal” ahal” as khesaarii baabun thasrii parak rasgullah mudaa tiin-taa rahi-ye gelash.

6.6.5.2. Notice that one can keep an changing the context of (146) as the following sentences would show, but in all such cases the constructions would sound very odd because of this perceptive/ in completive contradictions:

(147) thsarii per………..khas gel think”, mudaa (sab-tea puuraa) nah

but (all-dot complete) not

“There were twenty-eight rasgullas kept on the plate, and Khesari Babu ate up rasgullas (kept) on the plate, but not all of them”.

One can think of more and more contexts such as these if one takes up other candidates for Vestor-ship. Here is one more example of this type:

(148) A .pardaa Khasal” mudaa adhe dxur dhar curtain fall-pst-Agr but half-mph distance up to “The curtain fell, but only up to half the distance

Page 333: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

331

(148) b. pardaa khas pakal”, mudaa adhe dwar dhar fall-abs fall-pst-agr

“The curtain fell down, but only up to half the distance”.

6.7.1. We have seen while discussing different tests for establishing a Verb-Verb sequence as a instance of compound verb that almost all the tests have some weak points or the other. Someone working on Indian languages would them naturally ask: Does it mean that there is no way to find out whether a language has compound verbs or whether a particular verb is or is not a vaster? It appears that these “aspectual” verbs are not a category of meaning in the same way as other regular aspects-indicators (such as ask-“able”, “Completive” or “Progressive”, etc., are, and that is why these semantive and syntactic tests do not seem to be fool-proof. These verbs mark “perfective” and took (1976: 102) rightly says that “perfective aspect. Like the color yellow or the tests of lemonade, is a category not of meaning, but of experiences, incomprehensible to these who have not developed a sense for it”. But at the same time, one would like to provide a elute to an investigator working on language that have compound verbs as to how to recognize vectors (without really having to know the intradoses of this system until at an advanced stage of research). This, I think, is possible to do if we take up the different tests for identifying compound verbs described earlier, and try to find out as to how many of the “probable” candidates for vector verbs pass these tests. We would not expect that all such “probable” would

Page 334: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

332

Pass all the tests. But they are expected to obtain high score and fare well in these tests. Notice that I purposely leave this statement a little vague in that I do not specify as to what score would qualify a very to be described as a “true” vector. But this statement assumes that not all the vectors are same in nature. This is borne out by the realization that “in some cases, this lexical emptying (of the vectors- UMS) is almost complete: in other cases enough of the literal meaning is retained to render a literal OP+V interpretation 56 plausible: most are somewhere between, with the semantic contribution of the V2 rather transparent but a literal translation not quite admissible” (Masica, 1976:141). The scholars who believe that compound verb expressions as a category generated historically from verbal compounds also note that the “progression” from verbal compound to compound verb has not been “even” and “regular-57 in case of all verbs (of. Masica, 1976:142). This also explain why certain vectors are very frequent, while certain others are relatively elapsed. Therefore, if some of the probable vaster pass in all or most for these tests, and some score quite low, the differences could perhaps be attributed to their slow rate of progression from verbal compound to compound verb, and also to their difference in terms of lexical emptying and such other characteristics.

6.7.2.1. With this limitation of applying mechanical tests to identify vector verbs in mind, one could take up all the “probable” vectors (nineteen of them as noted in 75) and apply the tests on a few of them to prepare a sample score-sheet and analyze the results of this test on these “probable” vectors. A sample score-sheet is presented below:

Page 335: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

333

(149) A sample score-sheet of a multiplex compound verb testsu

C O M P O U N D V E R B T E S T S

Page 336: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

334 For each verb, one can find ‘+’, ‘-‘ or ‘O’ in the nine columns in (149), where the ‘plus’ indicates that a particular tests okays a particular verb, and ‘minus’ means the opposite, whereas ‘score’ indicates that although a particular tests works for certain combinations that a probable vector may enter into with a host of other verbs, there are many instances where this tests does not work. The scores for these values are as follows: 2 for a plus, I for a score and O for a minus mark. That means a verb can get at the most 15 points in which case it has to be okayed by all of these tests. Notice that one can, if one wants to, devise more and more tests and can include them in the multiplex test, and that in such cases the maximum point would no longer be 18. For instance, we have fast out the until-test in preparing (149), and one can always include this and many others tests in here. 6.7.3.1. We can take up some of these verbs and find out as to why a particular score was given in the sample score-sheet in (149) against it under a particular column. Let us take the case of one of the most frequently occurring probable vector is “go” which scores only 11 points as the table in (149) shows. We shall now onwards refer to the serial number of the tests as noted in the vertical columns instead of repeatedly mentioning their names. Notice that it passes through the tests 1,2,2,6,7, and 9. As an example of how the test-1 works, we can take up the sentences in (76) and (91), and show that Kar-insertion gives wrong results in this cases: (76’) “saikhum bearast jaa kas jaetash” (91’) “O becare gariibil so” Mar

Page 337: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

335 6.7.3.2. It may be mentioned here that a plus score in the test-1 does not mean that there can be no sentence in Maithili where Y-1-ias-pasineses would not fail it. That there are such cases is proven by the fast that (150) b and d are possible sentences in Maithili, and are parallel to these in (1500 a and e , respectively. It has still been given a plus because cases such as in (150) are rare 58:

159) a. e hamraa apan sab kich” 10_________ he (hum) me own every thing give ge-pst-Agr “He gave me away everything”.

b. e jamraa a[am sab locj” 10 kas gallak” c. o hamraa so apan sab kich: 10 from

take ge-pst-Agr He took away everything from me”. d. e hamraa so apan sab kich 10 kab gallak

6.7.3.3. Notice that not only (76) and (91), but even (150)a and e pass in the test-2 as

the following unacceptable sentences would show: (150)a. Saikhum bearest jaa (kae) kii kar tash” ?

b. e Because gariibii so mar (kae) kii kar lash”? c. e hamraa apan sab kich do (kae) kii (karlash,

Karal think )

What is strange is that (151) e seems to be perfectly acceptable if it is taken as an interrogation of (150) b, but not of (150)a, 6.7.3.4.1. The third test fails because one can have the following sentences with negative ‘compound verb’ where the ‘vector’ is Let us take up the same sentences as in (76) and

Page 338: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

336 (91) here:

(152)a. saikhum beareast nah jase jstaah” The marriage-party would not go today”. b. e becaare gariibii so nah mar gelash ‘That poor man did not die out of poverty:

6.7.3.4.2. Although test-4 was found to be working in certain cases where ias was involved, there are a number of sentences such as follows where the test did not seem to work,

(153)a. jat 2 baa dor nE o raakSas kBn maar a tihath1 By-the-time in he (hon) demon to kill-Cond-pst-Agr Tat2baa der nE raakSas ap a ne mar1 gel a By-then in demon self-Emph die -abs go-pst-Agr “By the time he would kill the demon, (by then) the demon died himself”. b. jat 2 baa der nE o raakSas kBn maar 1 detihath 1, tat2

ban der nE raasSas ap a ne { uniil a / maral / mar2 gel2} (153) jaa jaa o hamraa kaagaj dO {jastihath 1/ detihath

1? Taa taa by- the – time he (hon ) no paper {give go-cond-pst-Agr/ give-send-pst-Agr} by them

ma-bii a huukaa pED { delianh1 / dO anitaah1} I-Emph him (hon) ;ped {give-pst-Agr/ give go-pst-Agr} “By the time he gave no (some) paper, I (myself) gave him a pad”

6.7.3.5. The phasal verb such as lag, Surukarm ouk, and paar lang mentioned earlier in 6.6.1.2.3. can easily combine with

Page 339: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

337 Y-ins verb forms, and hence ina does not pass in the fifth test. Thus, for example. One can have constructions such as dO-jass lag-a lach-a started giving”, hais-I jaeeb near (nah-I) leagal-a Has gotton up’, haia-I jaasb near (nah-I) leagal-a “was not possible to sit down”, 10-jaaeb suru kal-a thinh-a started taking away”, and so on and so forth. 6.7.3.6. The last but one test does not work in case of because there are a number of examples where the clauses before the contradicting statement can take Y-ias combinations. Consider the following sentences:

(155) a. e 10 iaait”chaath-i Mudsa lel nah jaait” chanh he(hum) take ge-pres-Agr but take-inf neg ge-press-Agr He tries to carry it along, but (for him) to carry is not possible”.

b. O nar-I jane eaahait-a chath-I, nudan natait-a he(hon) die ge-inf want –ous-press-Agr but die nah-I chath-i.

nah-I chath-I nag sun-pres-Agr “He wants to die, but he cannot die”.

6.7.3.7.1. But the ninth test seems to work all right for most of the probable vector sequences with. So do the tests 6 and 7. It may also be mentioned here in passing that the selection restrictions of the verbs that are established, as vectors by this multiplex test have to be stated differentially because they vary from verb to verb. For example. Without bringing in other verbs, if we take up all the nineteen verbs from (149) and try to make

Page 340: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

338 Polars out of them by keeping as the vector in each case, only eight combinations score near about 11 points. They are:

156) a. Usab-I-jaa ‘ come up”, hais-I-jaa “sit down” cal-I-jaa “go away” nikal-I-jaa

‘come out”. Nar-I-jaa “die”, jase-jaa “proceed”,

do-jaa “give” 10-jaa “take away’.

7.3.7.2. We have already seen that among these, do-jaa and 10-jaa score lower than the others because of their verbal compound-like behavior. The following combinations tutrn out to be impossible, not because they cannot lexically combine with jaa (because other related languages do have some of these combinations). I think that it is a matter of accident that we do not have combinations such as follows.

(156) b. *paR-I-jaa, *rakh-I-jaa, *san-I-jaa, khas-I-jaa 6.7.3.7.3. Note that poB-e-jaa ‘fall off’ is a perfectly acceptable compound verb in Bengali. There is a third set of verbs which can combine with only to form verbal compound or sequences fo main verbs. They are the following:

(156) c. raath-I-jaa “Keep and go “, dho-jaa

“keep and go”, nikaal-I-jaa “turn out and go”, naar-I-

jaa “hit/kill/cheat and go”, dekh-I-jaa “(come) see and go”, rha-nk-I-jaa ‘throw and go”.

6.7.3.7.4. Here too, there is nothing inherent in the verbs such as dekh- see that it can combine with factors as it is only to result in a verbal compound. This becomes evident if we combine dekh-i with le to got dekh-I-le” see” which is a

Page 341: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

339 True compound verb, although it is homophonous with dekh-I-la, which means “wee and take” or “take scrutinisingly” as noted in (5) a earlier. 6.7.4.1. How, if we look at the score of as “come” as a vector, we can find out that it has one of the lowest score: 6 In fast, of these nineteen verbs, it can combine with only the following:

(157) dO-en, 10-aa, uTh-I-aa, rakh-i/reach-I-aa, cal-I-aa, nikal-I-aa, near—I-aa, dekh-I-aa, rha-n-k-I-aa

And, except for cal-I-aa, ‘ Come back/ever, all the either combinations give a parallel verbal compound reading. If we exclude combinations such as naree-aa“become tearful” For the time being, the score would naturally abe low. But if we consider these, the score becomes as high as 10. This is because they pass in the first two tests in that cases 59. Consider the following sentences, where (93) shown that insertion works, and (94) proves what any attempt to question as in this sentences is futile: (93’) *hamaraa sabhak gaan me ii puthaa bahut din so-n cal-I kaa sab-I rahal-a ach-I (94’) *ch-I bant par kumak aa-n-kh-I neraa (kaa) kii kaelanh-I? 6.7.4.2.1. Thus, combinations such as cal-I-aa and naras-aa get additional four points which make the total for as 10 for certain a elasted combinations. The normal of this discussion would be that while rating a particular probable vector, one should select especially these vector-sequences that can score maximum in this multiplex test. Even if one finds one or two combinations with a particular probable vector to be scoring very high, it should be

Page 342: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

340 Accepted as a vector on its own merit. In that case, one can describe it only as a rather “closed” vector (-closed in a sense that it has a limited combinatory possibility). 6.7.4.2.2. If we turn towards do and its rating as a vector, the total score shown that it is as common as but the score does not say many things. For instance, many of the verbs which can combine with, cannot occur as polars with as a vector. If one compares the following list of impossible combinations with (156)a, the differences would become quite evident:

(158) a * aab1-do, * bais –do, * nikal1 – do, * mar1 – do, * iaa-do,

*iTh1 – do, * naR-do *khan 1 – do, etc do as a vector can combine with a greater number of verbs than can combine with, and this becomes clear if we look into some of the combinations that can tner into with the verbs listed in (149): (158) b. do-do “give away”, rakh1 / raakh I – do “keep” aal 1 – do “start”, dhO -do “place”, naThaa-de “send”, nikaal 1 – do “turn out”, nan I- do “bring (benefactive)” unaar I – do “kill/hit/beat” , etc. 6.7.4.2.3. What makes it more different from is that when it is placed after the causative verb forms, the combinations are perfectly acceptable as compound verb in most of the cases without becoming verbal compound. Thus, compare (156) e with the following: (158) e. raakh1 – do nikaal 1 – do, naar I –do (already noted in 158b), iThavaa-do “get someone up” baises-do “make someone sit” naThvaa-do “send through someone”, khasan-do “case away, etc.

Page 343: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

341 6.7.5. The problems with the sample score-sheet in (149) that we discussed as far suggests the following ramifications in the procedure of identifying vector verbs: First, one should also select these combinations with a particular probable vector which can pass in the maximum number of tests, i.e., can obtain the maximum score along with those that may score minimum points. Secondly, to select such combinations, the least one can do is to find out which are the verbs that can be used as polars with a particular probable vector which can pass in at least the second test. The test would more or less 60 filter out the main verb sequences or verbal compounds from among them. How, one can select all the verbs listed as probable in (149) to get polar out of them, or one can also take up a good number of verbs which cannot function as vectors themselves, but which are otherwise very frequent as the first members for verbal sequences. Thus, after a list of combinatory possibilities is prepared. It would not be very difficult to find out which of the possible combinations pass in the verb interrogation test and can be taken as ideal samples of a vector sequences which may fetch the maximum score possible. Thirdly, the overall score for a particular verb would not show its frequency or openness in terms of combinatory possibilities as a vector 61. These have to be found out separately. The multiplex test would thus be only a confirmatory test 62 for establishing a particular verb as a vector. Therefore, one has to make a survey of combinatory possibilities separately where one could possibly select a representative sample of verbs for the language concerned and pair each of them with all the probable.

Page 344: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

342 Vectors to find out the following facts about the vector sequences: ) I) which vectors take how many verbs as polars? (ii) whether there are verbs which can not occur as polar of any vector? (iii) whether the polars show at least some sort of (partially) complementary distribution as regards to selection of vectors? (iv) what percentage of verbs can occur as polar with a particular probable vectors? (v) which combinations have become more or less idiomised ? (iv) whether one can classify the main verbs into several classes depending on which vaster they can take, and also on how many they can occur with? While trying to answer these questions, one may also find out more about various types of verbal compounds. 6.8.1.1. Out of the three takes that one has to do to make the multiplex test more meaningful, the third task seems to me to be most important. Because once one can answer all the questions enumerated in the last paragraph, it becomes easy to find out these polar-vector combinations which would score the maximum points that the vector concerned may obtain. Therefore, we shall start with the third task first. 6.8.1.2. To find out the answer to the six questions noted earlier, we took up in all 819 verbs of Maithili including the verbs that can otherwise occur in the vested position (as shown in 149). We found that at least two verbs- cash- “want” and ask “bring” cannot occur with any of the vectors 63. It is difficult to place cash in the first position of even a verbal compound, and would always give a verbal compound reading.

Page 345: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

343 All the other verbs were found to be occuring as polars with one or more of the probable vectors. But it was found out very soon that there were no combinations with the “probable” vectors who n k – “throw” and khas – “fall” which could pass in the kar-insertion and verb interrogation tests. The frequency of these probable vectors could be given at this point: (159) Freqency of Occurance of the 19 probable vectors with 217 verbs used as polars: VECTOR NUMBER OF OCCURRENCE PERCENTAGE64

Le 128 58.99% Do 119 54.84% Jaa 104 47.93% Mar 66 30.41% UTa 55 25.24% Bais 44 20.20% Aa 26 11.90% PoR 23 10.6% Rakh 18 8.29% Cal 13 5.99% PaThan 11 5.07 Dekh 8 3.69% Maar 6 2.76% Nikal 5 2.3% Dhar 5 2.3% Nikaal 3 2.3% Aan 2 0.92% Phon k 0 0.00% Khas 0 0.00% The table above clearly shows that phn k- and khas- should be dropped from the final list of vector verbs 6.8.1.3. The table in (159) also shows 65 that the major vector verbs in Maithili are : le, do, jaa, mar uTh, bais, aa and naR. These are the verbs which occur with more than 10% of the polars. Now, there are about 38 verbs (out of 217) which can occur with only one of the vector verbs. If these eight verbs are the commonest vectors in Maithili, one would expect that all the

Page 346: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

344 As main verbs should select any one of the vectors from this list. The expectation comes true when we see that the same-Vector polars select only the following: (160) VECTORS NUMBERS POLARS le 2 ai a naa- “test”, lo- “take”, do 2 ugar- “vomit, diselese”, thaTak- “throw”; mar N2 dama- “ram”, chaan o – “trap}; naR 1 khas- “fall” jaa NN3 an Tak “get struk”, an t- “be accomodated”, aR “stiffor”, aa- “come, unai- “grow”, utal- “descend”, unaT/nlat- “turn upside down”, kaT- “out”, kanen “get tightened”, shaT- “be loss”, shasak- “go off” chiTak- “splash up” chummu- “get post” inRaa- “be glued”, thul- “hang”,Tunt- “break”, Imp- “drown” dhak- “cover”, dhoh- “crumble”, naa- “be digested”, Raak- “be cooked, be scorched”, nasar “spread”, naab- “got”, chap- “be lit up”, han t- “be distributed”, ban taa- be distributed”. 6.8.1.4. We may new revise the “frequency- based” classification of probable vectors that was presented earlier in (750 which was more or less impressionistic, and had no statistical basis. We can, however, retain the three-way classification by accepting the vectorn that are high in frequency in (159), and which at the same time occur also in (160) as “Very frequent” as against these that are above in (159) but do not figure in (160) as “Mere or less frequent”. The rest of them are, of course,

Page 347: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

345 “Infrequent” vectors. The revised classes would look like the following:

(161) a. Very Frequent Vectors: la, do, jaa, mar, naR

b. More or less Frequent Vectors; uTh, bais, aa , rakh, cal, paThan:

c. Infrequent Vectors: dekh, maar, nikal, dhar, nikaal and aan. 6.8.2.1. The table in (160) shows that about 17.5% of the verbs taken to find out the various combinatory possibilities in Maithili can occur only with one vector. While these verbs became a “class” of main or polar verbs in themselves, some of the combinations in (160) look like ”idiomised” to some extent. For instance, it is more common to say ai 2 naa – le, khas 1 – naR, ugar1 – do then to use these main verbs alone. Another interesting thing is that in (160) seems to take three types of verbs as polars: (162) a. Polars that can occur with only [ + human] subjects: b. Polars that can occur with both [ ? human] subjects: A n Tak, an T-, aab-, utar- unaT-/ulaT, ghanak-, duuh-, naak- uha n- hao-, ha T- and baTaa-:

c. Polars that can occur with only [? human] subjects: Unai, kaT, kasam-, ghaT- chiTak, chuuT-, inFaa-, thul-, TunT-, Ihak- ihak-, naa-, basaT-, rhat-, uhaaT-, uHuuT,a nd baraa-.

Page 348: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

346 6.8.2.2. The following combinations with the infrequent vector verbs are the ideal ones and can later be taken to find out whether they pass at least in the verb interrogation test: (165) a. With dekh: khaa-dekh “taste (self-benefactive)”, kenan-dekh “earn (by oneself)”, gaab1-dekh “sing (for oneself)” ko-dekh “do (for oneself; by oneself)” eiikh1 – dekh “taste (yourself)”, cahaa-dekh “chew (yourself)” baaj1 – dekh “speak off”, dekh 1 – dekh “lick (yourself)”; b. With naar: ag n taa- naar “do hurry-scurry”, anThan – maar “be languid”, auma-naar “annoy/vox”, kah1 – naar “utter out” khii n e1 – naar “make text/drag”, Daraa – naar “frighten”; c. With nikal: ng 1 – nikal “rise, rhar 1 anikal “bear fruit”, bhang 1 – nikal “run away”, tar 1- nikal “cross over” chaar1 – nikal “be published”;

d. With dhars Arh 1 dhar “cover”/ envelope”, raT1 – dhar “mug up”,

Nasaar 1 – dhar “spread away”, dhar 1 udhar “catch up/ Ensnare”, adsar1 – dhar “extirpate” e. With nikaal:

Chaap1 – nikaal “publish”, aa n k1 – nikaal “draw up” Taak 1 – nikaal “search out”;

f. With aan:

GaPh1 – nan “concost:, kumaa-aam “case”

Page 349: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

347 6.8.2.3. It may be mentioned more that by giving these examples here, we do not claim that these are the only cases of the use of these six infrequent vectors. These are the ones that were found out to the occurring when we searched out of the 217-odd main verbs. Moreover, some of them may have both compound verb and verbal compound interpretations. In such cases, we could take up the former interpretations to test them against insertion and verb interrogation. Finally, it is possible that for some native speakers of Maithili, a few combinations are set acceptable, but some is likely to reject all of these as cases of verbal compound. 6.8.3.1 The only serious attempts to classify the polar or main verbs of a particular Indo-Aryan language seems to me to be that of heed (1976:144) where he summarizes the findings of heed (1974). In this paper, he has suggested the following classification for polar verbs in Hindi” (164) a. kuvaadi (intrans) Vectors iaa : eg., he- “become” uTh- “sit”, Uth- “get up”, etc.

b. ravandi (intrans) Vectors do ; e.g. re-“cry” ha 2 a- “laugh”, cal- “set out”, muskuran “smile”, cillao “scream”, etc c. santhaddi (trans) Vectors iaa, la: e.g. samajh “understand” jasu- “find out”, maan – “accept/admit”, etc. d. liyaadi (trans; Vectors; le : e.g. le- “take” khao “eat” dekh “see”, chin “grab”, etc.

Page 350: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

348

e. Divaadi (trans) Vectors do ; e.g. do- “give”, nikal- “object”, rho n k- “throw away”, chod – “let go off”, etc. f. karaadi (trans) Vectors lo, do: e.g. kar- “do” karaa- “have ( someone ) don rakh – “put/set”, etc g. jaadi (exeptional items) : e.g., jaa- “go” bhuul (jaa) “forget” laa/la- (aa) “bring” bajaa (laa) “carry out”, etc., h. cahandi (stative) Vectors : e.g., chaah “want”, rakh – “have/possess”, kahraa “be called”, etc.

6.8.3.2. While accounting on this typology, Hook (1976: 144) had noted himself that “this classification of the main verbs of Hindi is highly approximate if not actually misleading”. My impression is that it is both appreciative and misleading. It gives an impression that Hook somehow tries to force the paavinion classification of Sanskrit verbs into Hindi polar verbs. There is nothing objectionable if this classical typologination works for Hindi: but the fast is that it does not. Firstly, one would expect that since Hook (1974, 1976a) has 22 vector verbs, he would use all of them to find out as to which polar verbs show similarity with each other in selecting the some vectors or vectors. But he concentrates only on three vectors jaa, do and lo, which are the most frequent in both Hindi and Maithili, Secondly, Hindi in (164) is a store-house for exceptions and irregularities which is not a desirable thing. Thirdly, if one takes up the logically possible patterns of occurrence of vector verbs in a particular language, the higher the number of probable.

Page 351: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

349 Vector verbs, the greater is the number of such possibilities. In that case, an eight-way classification of main verbs is bound to leak very simplistic and appreciative. 6.8.3.3.Let me elaborate this point in a little more detailed manner which would lead us to a classification of polar or main verbs of Maithili that can overcome these defects. If, for the time being, we take it for granted that Maithili has 17 vector verbs as listed in (161) earlier, there seem to be a huge number of logical possibilities in case all the possible sets are utilized by all the 17 vectors. Notice that theoretically there could be 17 types of verbs which could occur with only one particular vector ( and not with any other vector) out of a total number of 17. In the same way, if one tried to find out the number of main verbs which could occur with only two the 17 vectors, one would come up with 136 logical types of patterning. We could go on calculating the logical possibilities for “some-vector” verbs to these hypothetical polar verbs that can occur with all the 17 vectors, for instance. 6.8.3.4. A competition of such logical possibilities would give us the following results for a language that has 17 vectors verbs. (165) POLARS OCCURRING WITH LOGICAL POSSIBILITIES A Number of Victors (Out of a Total of 17) at a time a. Name- vector combinations: 17 b. Two-Vector combinations : 136 c. Three-vector combinations: 600 d. Four-Vector combinations: 2580

Page 352: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

350 e. Five-Vector Combinations : 6188 f. Six-Vector combinations : 12376 g. Seven-Vector combinations ; 19448 h. Eight-Vector combinations : 24310 i. Nine-vector combinations : 24310 j. Ten-Vector combinations : 19448 k. Eleven-Vector combination : 12376 l. Twelve-Vector combination : 6188 m. Thirteen-Vector combination : 2380 n. Fourteen-Vector combination : 680 o. Fifteen-Vector combination : 136 p. Sixteen-Vector combinations : 17 q. Seventeen-Vector combination 1 Total number of Verbs-types : 131071 (by addition) 6.8.3.5. Logically, on the basis of combinatory possibilities with 17 vector verbs, Maithili can have a total of the following verb-classes: (166) 17x136x600x1280x6188x12376x19440x24310x X 24310x19448x12376x6188x2380x600x136x17x = 2133118895435450650139314120202865055368811104000000 types of main verbs. Instead of multiplying each of (165) a to q with each other, If we simply add the 17 separate logical possibilities as in (165) the total becomes 131071 6.8.3.6. The total would naturally increase if a language had 22 vector verbs. In that case, it is too much to expect a handy eight-way “grammatical” possibilities, as hook (1978:144) did. By keeping the verbs which do not occur in compound verb construction aside, we took up all the 217 verbs of Maithili to find out as to how many of these logical possibilities are grammatically possible, and the result was as follows:

Page 353: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

351 a. None-Vector Combinations : 5 b. Two-Vector Combinations : 15 c. Three-Vector Combinations : 20 d. Four- Vector Combinations : 19 e. Five- Vector Combinations : 11 f. Six- Vector Combinations : 7 g. Seven- Vector Combinations : 3 h. Eight- Vector Combinations : 4 i. Nine- Vector Combinations : None j. Ten- Vector Combinations : None k. Eleven- Vector Combinations : None l. Twelve- Vector Combinations: None m. Thirteen- Vector Combinations: None n. Fourteen- Vector Combinations: None o. Fifteen- Vector Combinations : None p. Sixteen- Vector Combinations : None q. Seventeen- Vector Combinations None Total number of Verb-types : 48 (by addition) 6.8.3.7. I would propose that there are 84 “types” of polar or main verbs in Maithili. However, it would emerge from the enumeration of these 84 types of combinations very soon that these verb-classes could be further collapsed into fewer “generalized” class of polars. We have already seen that 38 out of 217 verbs of Maithili taken in the sample survey occur with 5 (out of 17) vectors: The possible combinations have already been shown in (160). We can new look into the combinations noted in (167)b to h. 6.9.1.1. The following are the possible two-vector combinations where the main verbs mentioned in the right in each column can occur with only two of the 17 vectors noted against each class on the left.

Page 354: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

352 (168) Two-Vector Combinations: A.le & de : ann-“smear”, ykheal- “upreet”, ? “Cover (someone)”. Kaat-“out (something)”, katar- “scisser up”, kuut-“mince”, khat-“work for”, khx“serape”, khah-“rew”, shai kaa- “disledge”, sharp yan- “get (something) published”, Iiitea- “help (someone) win”, tyaas- “relinguish”, jhang- ‘suppert/held” jker-“ley upon”, dhme-“seern/beat”, polat- “turn back”, phansag- “trap (someone)” khal kas- “unserve”, bhmi- fry” ? “prepare bed”, Taar- “Avoid”, tiin- “knead/wink’, thak- “Cheat”, ? “Save”, ? “Keep”, and near- ‘Kill”, b. le & jea : suxt- “sleep”, uth- “got up”, bais- “sit”. c. le & rakh : dekh “sit”, d. le & dhar : odh-“cover (with something)”, e. le & jna : it rea- “Vulgarise/traduce’. f. de & bais : pethas- “sound”. h.ja & nar : ashae-“be content/ be bered”, tiit- “Get drenched”. i. jaa & del : suw- “leak/eese out”, dhak- orumble down”, pal- “fall”. j.jaa & uth : gal-“prick’, jaga-“be awakened”, ? “sjpw I[. Dar ‘fear”, del ‘swing jhask- “be tired’, ? “be unserved”. k.jaa & aa : aneth- “be obstinate”, dab-“be under”, l.jna & sel : edar- “be shavan off”, kun-“be less”, ? “open up, khar- fit in” be glned”. m.jaa & oel : bai-‘ring”. n.jaa & nikal : shap-“be printed”, tar- “crees”, dhar- “bear fruit”. e. as & rakh : sher-“eneirole”.

Page 355: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

353 6.9.1.2. A close look at (168) would reveal that it could be broadened up by collapsing (168) and into one category where is a must: Similarly, it is a must in (168)e-g, and is invariable in (168)h-n. Finally, (168)e is very different from the others. We would therefore collapse the combinations into the followings:

(168) i. Le and another vector ii de and another vector

iii jaa and another vector iv an and rakh.

6.9.1.3. Notice that some of the combinations in (168) would fall into more than one category as shown in (168). For example, (186) a would fall in both (168) I and ii, and (168)e would come under both (168) ii and iii. Similarly, (168) b would be in (168)I and iii. The frequency of occurrence of different vectors in (168) is as follows 67: in (168) is as follows 67: (168) VECTORS NUMBERS % VECTORS NUMBERS % Le 4 13.33 bais 2 6.66 De 4 13.33 aa 2 6.66 Jaa 9 30.00 rakh 2 6.66 Mar 2 6.66 eal 1 3.33 Pak 1 3.33 dhar 1 3.33 Uth 1 3.33 nikal 1 3.33 6.9.2.1. Three vectors combinations have the following twenty grammatical possibilities:

(169) a. le,de &jaa : ktaar- “bring down”, untea-“reverse”. Shilige “setter”, ?

“reast/parch”, sai-“dress up”, jhal kae- “be shown up”, ivn-

“ evertale”, dajaa- “come under/suppress’, ? “be saved”.

Page 356: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

354 6.9.1.2. A close look at (168) would reveal that it could be broadened up by collapsing (168) and into one category where is a must: Similarly, it is a must in (168)e-g, and is invariable in (168)h-n. Finally, (168)e is very different from the others. We would therefore collapse the combinations into the followings:

(168) i. Le and another vector ii de and another vector

iii jaa and another vector iv an and rakh.

6.9.1.3. Notice that some of the combinations in (168) would fall into more than one category as shown in (168). For example, (186) a would fall in both (168) I and ii, and (168)e would come under both (168) ii and iii. Similarly, (168) b would be in (168)I and iii. The frequency of occurrence of different vectors in (168) is as follows 67: (168) VECTORS NUMBERS % VECTORS NUMBERS % Le 4 13.33 bais 2 6.66 De 4 13.33 aa 2 6.66 Jaa 9 30.00 rakh 2 6.66 Mar 2 6.66 eal 1 3.33 Pak 1 3.33 dhar 1 3.33 Uth 1 3.33 nikal 1 3.33 6.9.2.1. Three vectors combinations have the following twenty grammatical possibilities:

(169) a. le,de &jaa : ktaar- “bring down”, untea-“reverse”. Shilige “setter”, ?

“reast/parch”, sai-“dress up”, jhal kae- “be shown up”, ivn-

“ evertale”, dajaa- “come under/suppress’, ? “be saved”.

355

b. le, de & nar : khadh-“dig out”, khal-“open”, gnap-‘bury under”, gan- "

Page 357: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

“bury under”, gan- “count’, ? “wreathe”, chea- “mart out/ select”, jap- “pray”, lag-t-“message”, jap “join”, teak-“stich/tack”, ? “fill/insert”, thal-“push”, / “tap/beat’, the-“carry”, dpak “trample under”, dush-“milk”. Near”measure”. c.le, do &uTh : sang-“light a stove”, kaan-“weep”, ? “be adement/take responsibility ? “abuse/scoff at”, dhep-“chide, prate’, ? “light”. d. le, do & bais : ? “melt (something)” shatne- “sunteast”, ghah-“leave”, nikea “take out/bring out”. e. le,do & rakh : beth- “read/study”. f. le,do & pathas : nang yea- “tet (someone) bring”. g.le,do & dekh : elikh-‘taste”, eehb- “lick”. h.le, do 7 near : barea-“frighten”, nethae-“avoid”.

i.le, do & dhar : ? “spread out “. j.le, do & nikael ; taek-“leak for/search”. k.le, jaa & mar : nar-‘die’. l.le, jaa & dalk : let-“lie down”. n.jaa nar & dat : runk-“step”. n.jaa,nar & uTh : jharak- “he burut”, long-“touch at /feel/ adhre to”. o.jaa, uTh & pak : ? “shoudder/ be startled”, bar “be lit up”. p.jaa, uTh & beis: : skhap-“seem improper”. q.jaa, uTh & ad : kag- “be tightened”, lag- “scerea/ freege”, ? “strike (in mind_”. Sah-“ret”. r.jaa, dek & ad : khar-“be uprected”, gal-“melt/deem- pose”, jhuk- “bond down, nikal- “come out” s.uTh, bais & as : near-“seem/saw”.

t.uTh par & naar: : ? “be vexed/be ameyed”, ? ‘stir/stub”.

Page 358: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

356 6.9.2.2. Compared to two-vector combinations, participation of vectors in three-vector combinations seems to be more because in the former only 12 vectors participated as (168) shows, while in the latter, 14 of them showed up. The vectors and occur in the former but not in the latter, while which are found in (169) are not present in (168). The frequency of these 14 vectors in (169) is as follows 68: (169’) VECTORS NUMBER ? VECTORS NUMBER ? LE 12 20.00 aa 3 5.00 Do 10 16.66 rakh 1 1.66 Jaa 9 15.00 paThan 1 1.66 Nar 5 8.33 dekh 1 1.66 PoR 4 6.66 naar 1 3.33 UTh 7 11.66 dhar 1 1.33 Bais 3 5.00 nikaal 1 1.33 6.9.2.3. I shall not try to collapse (169) at into smaller blocks at this point, because I would like to take up the matter later. Later, I shall also try to find out as to what percentage of the verbs belonged to a particular vector-combination, For instance, 59 out of 217 verbs were found to be occurring with twe vectors, and 65 occurred with three vectors. It is thus evident that the lower you go down the combinations listed in (165) or (167), (i.e., the greater is the number of vectors) the lesser will the number of main verbs be for a particular vector combination. This can be established after we found out about other vector combinations. 6.9.3.1. The following 27 verbs would occur in four-vector combinations:

Page 359: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

357

(169) a. le, do, inn & mar : uB- “fly” b.le, do, jaa & bais : jiit – “win” c. le, do, mar & uTh: nano- “dance”, niin- “annoint/plaster”, haan k – “call/ shout”. d. le, do, mar & bais : thankne- ‘crush/grind”, pakaa- “cock/ roast”, uhanB- “tear off” phoR- “break/ dissolve”, kiin – “buy”, gili- “mix/spoil”. e. la, do, ,mar & dekh : gaab- “sing”, cahaa- “chew” f. la, do, mar & naa : gaph- “concect/ build”

g. la, do, uTh & naThaa : bajaa – “call” h. la, do, bais & ghar : odaar – “uproot/extirpate”, i. la, do, aa & nikaal : chaap- “publish”. j. La, iaa, mar, bais : glip – “swallow”. k. Lo, do, bais & naar : khiin a – “drag” l. Do, mar, bais & rakh : khuaa – “food/invite to eat”. m. Do, bais, rakh & naThan : do- “give”. n. Do, iaa, mar & na : aguar – “ come/go forward”. o. Jaa, mar, uTh & aa : jar- “burn”, khisiaan – “be angry upon/ scold”. p. Jaa, mar, cal & nikal bhaag-“run away” q. Jaa, bais, rakh & naThaa jaa – “go”. r. Mar, uTh, rakh & cal ; cikar – “shout”.

6.9.3.2. Notice that except baR, all other probable vectors as listed in (161) participated in four-vector combinations, and their frequency of occurrence in (171) is as follows69 : (171) VECTORS NUMBER ? VECTORS NUMBER ? lo 11 15.28 cal 2 2.78 do 13 18.06 paThan 3 4.17 jaa 7 9.72 dekh 1 1.39 mar 11 15.28 dekh 1 1.39 uTh 4 5.56 nikal 1 1.39 bais 8 11.11 dhar 1 1.39 aa 3 4.17 nikaal 1 1.39 rakh 4 5.56 aan 1 1.39

Page 360: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

358 6.9.4.1. The grammatically realised five-vector combinations are eleven in number, and they are as follows: (172) a. lo, do, jaa, uTh & aa : bhar- “fill” b. lo, do, jaa, bais & dekh : khan= “eat” c. lo, do, jaa, mar & bais : jap- “annoint/plaster”. Juut – “loot”.

d. la, do, mar, uTh & bais : chaa12k – “throw eatables into one’s own mouth”, haan T- “distribute”, naa- “scratch”, rak- “stop (someone)” e. lo, do, mar, uTh & rakh : Tak- “interrupt/ butt in “ f. lo, do, mar, bais & rakh : jet- “yoke ( enon)/plough (land)” g. lo, do, mar, bais & naThan : maane-“want/demand”. h. lo, do, bais, dekh & aan : kaman- “earn”. i. lo, do, uTh, bais & cal : taR- “break”. j. lo, do, uTh, naThaa & nikaal : aan k – “draw/ paint”. k. lo, jaa, mar, naR & cal : han T- “go off/ withdraw”.

6.9.4.2. Five-vector combinations involve 14 vectors out of 17 and 15 main verbs out of a sample of 217. The frequency of occurance of these 14 vectos in (172) is as follows 70 (175) VECTORS NUMBER ? VECTORS NUMBER ? lo 11 20.00 ca 1 1.82 do 10 18.18 rakh 2 3.64 jaa 4 7.27 cak 2 3.64 mar 6 10.90 paThan 2 3.64 uTh 5 9.09 nikaal 1 1.82 bais 7 12.75 aan 1 1.82 6.9.51. As noted in (167), there are about 7 types of grammatically possible combinations involving six vectors at a

Page 361: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

359 Time. Only eight main verbs were found to occur with six vectors, and (174) involves only 12 of 17 vectors: (174) a. lo, do, jaa, uTh, bais & dekh : kO- “do”. b. lo, do, mar, naR, uTh & bais : jhagaD- “quarrlel” uhaan-“jump” c. lo, do, mar, vaR, Uth * cal : han s- “laugh” d. lo, do, mar, naR, uTh & dekh : baaj- “speak out” e. lo, do, mar, bais, rakh & dhar : dhO/dhar- “keep” f. lo, jaa, mar, naR, uTh & bais : rah- “remain”. g. do, jaa, mar, uTh, bais & maar : agu tan “hasten”. 6.9.5.2. The frequency distribution of 12 vectors is as follows71. (175) VECTORS NUMBER ? VECTORS NUMBER ? LO 6 15.29 bais 5 11.90 Do 6 14.29 rakh 1 2.38 Jaa 3 7.14 cal 1 2.38 Mar 6 14.29 dekh 2 4.76 Par 4 9.52 naar 1 2.38 UTh 6 14.29 dhar 1 2.38 6.9.6.1. The vectors involved in seven-vector combinations are lesser than these shown in (174)-(175). Grammatically possible combinations are three which involve 11 vectors and only three main verbs. These combinations are as follows: (176) a. lo, do, jaa, mar, uTh, bais & rakh : piih- “drink” b. lo, do, jaa, naR, bais, aa & rakh : jaan “know”. c. jaa, mar, uTh, bais, aa, cal & nikal : no- “rise”.

Page 362: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

360 6.9.6.2. The frequency distribution of II Vectors involved in (176) is as follows 72. (177) VECTORS NUMBER ? VECTORS NUMBER ? lo 2 9.52 bais 3 14.29 do 2 9.52 aa 2 9.52 jaa 3 14.29 rakh 2 9.52 mar 2 9.52 cal 1 4.76 paR 1 4.76 nikal 1 4.76 uTh 2 9.52 6.9.7.1.Eight-vector combinations involve 13 vectors and 4 main verbs. The grammatical possibilities are also four such as follows: (178) a. lo, do, jaa, mar, naR, uTh, bais & cal : rama- “be piqued/ be conceited”. b. lo, do, jaa, mar, bais, rakh, cal & dhar : raT- “monorise” c. lo, do, mar, uTh, bais, rakh, naThaa & naar : bol “say” d. do, jaa, mar naR, uTh, bais, aa & cal : cal- “proceed/ move”. 6.9.7.2. The frequency distribution of 13 vectors is as follows75. (179) VECTORS NUMBER ? VECTORS NUMBER ? lo 3 9.38 aa 1 3.15 do 4 12.5 rakh 2 6.25 jaa 3 9.38 cal 3 9.38 mar 4 12.5 paThaa 1 3.13 paR 2 6.25 maar 1 3.13 uTh 3 9.38 dhar 1 3.13 bais 4 12.5 6.10.1.1. If we now compare the frequency distribution of 17 vectors as noted in eight grammatically possible vector-combination as noted in eight grammatically possible vector-combinations (of, 168”, 169’, 171, 175, 177 and 179), and add to this the vectors (5 if them) that can occur in “some vectors” combinations, we would get a table of frequency of occurrences of 17 vectors which would be a little different from that in (159):

Page 363: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

361

(180) frequency of occurrences of 17 probable vectors in name-vector combinations to eight-victor combinations ( in total 84), where they occur 317 times.

VECTORS NUMBER OF PERCENTAGE (% AS SHOWN IN 159) OCCURRENCE Lo 50 15.77% 54.84% Do 50 15.77% 58.99% Mar 39 12.30% 47.99% PaR 14 4.42% 10.6% UTh 28 8.83% 23.34% Bais 32 10.09% 20.28% Aa 12 3.79% 11.90% Rakh 14 4.42% 8.29% Cal 10 3.19% 11.90% PaThaa 7 2.19% 5.07% Dekh 6 1.09% 3.69% Naar 5 1.58% 2..76% Nikal 2 0.69% 2.30% Dhar 5 1.58% 2.30% Nikaal 4 1.26% 2.30% Aan 2 0.69% 0.92% 6.10.1.2. Recall that the percentage of occurrences shown in (159) for each of these vectors was overlapping by nature as at that point we were interested in finding out as to how many main verbs these vectors could occur with. Naturally, many of these vectors occurred with the same main verbs, and in the preceding paragraphs we found out these instances of co-assurances. The third and the fourth columns are therefore not strictly comparable 74. However, the purpose of giving the percentages noted in (159) in (180) again was that one could easily see that some of the “more or less frequent verbs” such as uTh, bais, and rakh are “more frequent” in vector combinations than some of the “very frequent verbs” such as naR The table in (180). However, above that the three-way classification of IT vector verbs in Maithili is more or less justified. This is because except naR, all the other verbs in (161)a are more frequent.

Page 364: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

362 Than these in (161), and the latter, in turn, are much more frequent than these in (161)a. 6.10.1.3. It follows from the above discussion that the main verbs of Maithili enter into eight major combinations-types which involve IT vectors. A survey of 217 main verbs shows that the polars in Maithili will occur more with one to four vector-combinations than with combinations that involve more than four vectors. The following table that emerges out of the detailed discussions on which polar can take how many vectors would give a clear idea of what would happen if the corpus is further enlarged:

(181) Selection of the number of vectors in respect of the main verbs of Maithili: VECTOR-COMBINATIONS NUMBER OF POLARS ? OPTING FOR IT

a. None-Vector Combinations: 38 17.51 b. Two-Vector Combinations : 99 27.19 c. Three-Vector Combinations : 65 29.85 d. Four- Vector Combinations : 27 12.44 e. Five- Vector Combinations : 15 6.91 f. Six- Vector Combinations : 8 5.69 g. Seven- Vector Combinations : 3 1.58 h. Eight- Vector Combinations 4 1.84 6.10.1.4. Under each of these eight types of select ional patterns, there are various combinatory possibilities which have been mentioned in (167) earlier, and which have been described in (160) and in (168)-(80). We would here tentatively suggest that these 84 combination-types would be broadened in the following way:

Page 365: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

363 (182) a. None –vector combinations:

(i) with lo; (ii) with d0; (iii) with jaa: (iv) with dar; and (v) with naR.

b. Two-vector combinations: (i) with lo and another vector; (ii) with do and another vector; (iii) with jaa and

another vector, and (iv) with aa and rakh. c. Three-vector combinations: (i) with lo, do and another vector (ii) with lo, jaa and another vector ; (iii) with

jaa, naR and (iv) with jaa, uTh and another vector; (v) with uTh, bais and aa ; and (vi) with uTh, nar and maar.

d. Four-vector combinations: (i) with lo, do, mar and another vector (which would include ???? e-f ), (ii) with

lo, do, bais and another vector (which would include 1766b h and k ), (iii) with jaa, mar and two other vectors (which would include 170 j, n , o and p), (iv) with do, bais, rakh and another vector (which include 170 l and m) (v) lo, do and two other vectors (vi) jaa/mar and three verbs of the second category (i..e. of 161 b).

e. Five-vector combinations: (i) lo, do, jaa and two other vectors (ii) lo, do, nar and two other vectors, (iii) lo,

do uTh and two other vectors (iv) lo, do bais and the other vectors and (v) four very frequent vectors (i.e., these belonging to 161a) and another vector.

f. Six-vector combinations: (i) with lo, do, mar, naR, uTh and another vectors (ii) with jaa, mar, uTh, bais and

two other vectors; and (iii) lo, do, bais and three other vectors. g. Seven-vector combinations (i) with lo, do, jaa, bais, rakh and two other vectors; and (ii) with jaa, nar, uTh,

bais, aa , cal and uthal h. Eight-vectors combinations: (i) ? (iv) as in (178) a-d.

Page 366: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

363 6.10.2.1. It must have become clear by new that the probable vectors belonging to (161)a and b are likely to score quite high in the multiples best described earlier. I would therefore concentrate more on the infrequent probable vectors to find out how they fare in am multiple test such as in (149).The table of score that I would come up with new will be slightly different from (149) in that for each of these vectors, there would be two horizontal score-sheets instead of one. The top line would show the minimum score and the bottom the maximum with respect to a particular test. The following is the score-sheet of the nine test applied on the probable vectors . dekh, naar, nikal, dhar, nikaal and naars (183) MULTIPLE TEST APPLIED ON THE INFREQUENT VECTORS

Page 367: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

364 6.10.2.2.1. By locking at the results of the tests applied on . One notices a remarkable thing. It does not pass in the first two tests, but still scores very high. I call 8/9 a high score in this case because if it had passed in the first two tests which all “tree” vectors are supposed to do, the score would have been 12/13, which would be as good as that of naar and nikal. The question that can now be raised is whether dekh should be counted as a vector if it fails in the first two tests my impression is that dekh in Maithili is in the process of becoming a regular vector, and the lexical emptying has not taken place fully in this case, and hence Test-1 and 2 became ineffective in this case dekh and naThaaa Seen to me to be parallel to lo and do in respect fo a feature complex such as follows: (184) [ + Benefactive] [+Self] [- Self] 1. lo do 2. dekh naThaa 6.10.2.2.2. Apart from the fast that the pair dekh/nathha is much less frequent compared to lo/do there are either differences too. For instance, the former seem to be in some sort of complementary distribution, as the ungrammatical forms in (185) would show: (185) eiikh I – dekh but * eiikhi – paThaal; dekh, but ?? cabh I – nathaa gaab I – dekh , but * saab I – natthan; ?? khan-dekh, but *khaa-naThaa, etc.

Page 368: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

365 6.10.2.2.3. There are, however, forms like ko-dekh and ko-uthas where they select, the same polar. Moreover, the differences shown in (185) do not indicate that dekh should be rejected as a vector, because there are polar which can occur with do and not with lo or vice versa. Consider the following combinations, for example: (186) With lo with do close aj a nmek - test eth- - cover sunt - sleep uth - get up baie - sit dekh - see let - lie down mar- - die (rare comb) giin _ swallow ha nt - go off/withdraw rah - remain - ugar vomit - jhatah throw - it/n ras vulgarize - pathan send - kkuae food - ag/nae move forward - ag/nae hasten - cal proceed/move 6.10.2.3 In spite of the difference in selection of polar verbs as shown above, de/le jointly take 106 out of 217 verbs that I have chocked with. Therefore, what (183) shows is that dekh is in the process of becoming a regular vector But as Kasica (1976) pointed out that there are some verbs which do lexical emptying in a greater speed that the others which take a

Page 369: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

366 Little longer, time in the history for a particular language to do so. I believe, dekh is one of these slow -moving vectors. Another vector which is in between dekh and r egular/frequent vectors such as le in this matter is nikal. Thus, while come combinations with nikal would resist a kar in between them, some would optionally allow then; for instance, ng-I kae-nikal, ghari/kae-nikal, and bhari-kae-nikal would nto be different in meaning them compared to nki-nikal chapi-nikal and phaki-nikal but that is not ture for bhaag I-kae-nikal and bhangi- nikal or for texi-kae-nikal and tar I-nikal. This shows that although nikal underwent lexical emptying to a considerable extent, it has not completed the process. Interestingly, even nginikal chaw I-nikal and phar I-nikal cannot be allowed to be “verb interrogated”. That is, one cannot have sentential parallels such as follows:

(187) a. Surunj ug/2 nik/nlel/n ach/1 gun rise-abs come-out-pft. Pte be-press-Agr “The sun has arisen”. b. *surunj ug/1 jae jukk jeek/a acg/1 ?

(188) a. ah/1 bor samae co/n pahinah/1 phal phari nth-

alala this time season/time before furit bear come-out “This time fruite riponed before the

season.”. b. cj/1 bor/1 some no/n pch/1nah/2 phal phar/1 kee kii

keel.a?

6.10.2.4. Notice that some combinations involving near and nikal do not allow relegation marked before them while some do. That is why the minimum score is shown as ‘e’ for both under this test.

Page 370: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

367

To elaborate, one cannot have constructions such as nahi ugi nikal, nahi uhari-nikale, nahi kahi- pear, and “nchi uthas-marg. But at the same time, combinations such as khasi nikal. Tari-nikal. As/ntae-near, and khii/ne/I near do allow a NDS before them. The apprehension test works for neither a Y dekh compound verb, now for corresponding simple verbs, and hence it is neutral so far as dekh is concerned. This explains ‘o’ for both minimum and maximum score under this column for dekh. By-the-time-test posses some of the combinations with nikal such as ugi-nikal as the ungrammaticality of (189)b would show. But at the same it does not block khane- as against khagi-nikal in (189)d.

(189) a. jaa jaa e bhangi nikal/nlaah/n, tea tea surunj

by- the-time he(hun) run-abs come-out by-then sun bhag/n naam ugi-nikal/nlash/n

god rise-abs come-out-pst-Agr “By the time he run away, the sun god had

already arisen”. b.”jaa jaa e bhangi nikal /nlash/n, taa taa surunj bhagn maan ug/nlash/n/ugal/nthink” c.jaa jaa e bhaagi nikal/nlaash/n, taa taa has

bhangi nikal/n liaahi “By the time he run away, I run away”. d.jaa jaa e bhangi nikal/nlash/n, han

bhang/nliashi Similarly, while tari-nikal and as/ntaa-near pass in the compound verb reduction test, ngi-nikal nd khii/ne/I-near do not pass, and hence the o/- variation under this column for near and nikal.

Page 371: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

368 6.10.2.5. Let us now turn towards other vectors. Nikeal is very much similar to nickel in that it allows a kar insertion in tecki-nikeal without any significant change in meaning. While as/nk/I-nikeal would not allow it to be changed to as/nki-kae-nikeal in the instead sense. While both tecki-nikeal and as/nk/I-nikeal would reject a NDS, combinations with and dhar and run are divided into two groups in this respect. It would give a verbal compound reading in case of odeeri-dhar, For other combinations, consider the following sentences, where (190)a and b do not allow NDS, but (190)e is perfectly okay with it.

(190) a. ii sab nahi rari dharah, so nahi ? this pl not memorise-abs keep-fut-Agr that not

“In not it possible that (we) do not memories these ?”

b. “han e candar nahi edhi dharaliahi I that robe not cover-abs keep-put-Agr ‘I did nto cover (myself) with that robe”.

c. kunkah onaa nahi dhari dhariahu him that-way not held-abs keep-imp-Agr “Do nto hold him that way ?”

The sentences in (1900 a and b become acceptable if we remove the vector dhar from their surface. (190)e explain as to why there is a o/+ marking for dhar under the column of negation test. 6.10.2.6. In case of combinations with sun, dhari- snn would reject a NDS element, while kamrann would not. The former would again allow a kar to be inserted between the polar and vector, while the latter would not. This is the reason why sun scores o/+

Page 372: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

369 for these two tests. These two combinations are perfectly acceptable even with the phasal verbs like sarnkar “begin” or Long “continue”, etc. Among the combinations with dhar, rari-dhar and odeeri-dhat May occur with such phase verbs, while dhari-dhar and ashi-dhar produce ungrammatical results with such verbs. The apprehension test does not seem to work for nickel at all, because here after Barait/n ohii jo.,. “I am afraid that …”, “I am afraid that ……..”. Neither Y-nikeal near their simple verb counterparts would be acceptable. Hence, it goats o/o score there. While Kaman-man would fall in this test because it can be replaced by its simple verbal counterpart kaman in “apprehensive” sentences, gathi-reen would not allow that. Hence, scores o/+ in this test. What is remarkable is gathi-sen seems to be t he only combination which would score in this test for these six infrequent vector sequences as the score-sheet would show. Again, while this particular combination would pass in a contradiction test.kadhar-man Would not, and therefore scores a o/+ score there. This distinction between two combinations with is maintained in both and compound verb reduction-tests. Gathi-raan Passes in both whereas Kaman-aen fails in both. But for this differences, both minimum and maximum score and new would have been 16, which would have been quite high by any standard. Compared to that, nicely seems to be the vector which shows the least differences between the minimum and the maximum score. The overall score for dhar seems to be quite low although it is more frequent than aan and nikaal and as frequent as naar.

Page 373: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

370 6.10.2.7. At this stage, I think I must mention that the vector-sequences which are less frequent are expected to obtain higher minimum scores than the very frequent ones because the letter combine with various types of verbs and in the process “explicate” more than one semen tie shade. Therefore, if one runs these different types of combinations through these nine tests, one would naturally expect more ‘O’ scores in the minimum line, although their overall maximum score may be quite high. However, this needs to be worked out in different language 76 that have compound verbs to find out whether the more frequent vectors would always score so differently from the less frequent ones. 6.11. In conclusion, we have seen that although compound verbs in Maithili do not show any one formal criterion that could differentiate them from all other types of verbal combinations, there are a number of operational tests that could help in identifying as to which verbs can function as vectors. We have also seem that many of these tests do not work for certain types of compound verbs, but almost all of them are sure to be identified by a verb interrogation test, unless the vector concerned has undergone very little lexical emptying. The test of insertion can also point out to how far as particular vector has progressed in lexical emptying. We found that while very frequent vectors such as le and do have become semantically null, nickel, nickeal and have not progressed so far, and that is much behind them. We also tried to find out as to what are the select ional restrictions on the main verbs in terms of taking a particular vector, and came

Page 374: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

371 out with eight major classes of main verbs which could be divided into 84 sub-types. We also attempted to reduce the 84-combinatory possibilities into 34 real categories. We found out the relative frequency of 17 vectors in Maithili with a sample list of 217 main verbs and found out that there were certain verbs that would not take any of these vectors. Finally, we demonstrated as to how a multiplex test invaliding nine compound verb tests (i.e., kaa-insertion, verb interrogation, negation, by-the-time, phrasal verbs, apprehension, compound verb reduction, contradiction and incompletion) could be used to find out the similarities and differences in the behavior of different vectors is a particular language.

Page 375: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

372

Notes to chapter VI

1. See Peter Hook’s (1976) paper on “perfecting a test for the perfective-: Aspectual parallels in Russian, Lithuanian, Modern Greek, Hindi and pashte for details.

2. The first member of a compound verb in IA Languages the meaning of

which is modified by the following verb is called the polar. What follows polar has been called the vector verb in the following paragraphs.

3. The marker found in the Garyanpada (9th-11th e. A.D) included-I and –e both. In modern times,-I is found in Maithili and Oriya, whereas-o is attested in Bengali. Note that the absolute marker in modern Hindi is-o(Hook, 9174).

4. If one goes into a little more detail of these vowel ending roots plus compound verb marker-I, one may come up with an underlying-(X) for the verbs listed in (X) and (IV) below (Notice that this-X shows up when these verbs receive the infinitive suffix- e.g. nii-k-ah “be drink”, aamaan-ab * to proceed, but not do-raab to give”):

BASE FORM POLAR FORM CLOSS

(x) pii pii drink jii jii live sii sii new khan khan eat dhe dhe wash,etc,

(II) ae(y) aabi come

gua(w) gaabi sing paa(w) paabi got baa(W) baabi open (one’s mouth) own(y) coubi leak chun(w) chumbi touch, etc.

(III) de do give

le le take he the be ka® ko de dha ® dhe keep,etc.

(IV) aggan aggan proceed ag/ntea ag/ntea hasten

aghan aghan be content

Page 376: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

373

5. The conjunct verb comes from different sources. First, OIA “particle verb designs” constructions such as npo –vas-ax-arti “takes seat” are made noun to which kar de or similar verbs are added. Secondly, sometimes simple verbs are nominalised and then kar is added to them, eg., khal-karab “to play around” ; maar-karab “be importunate”. Thirdly, a number of contraceptive words are added to verbs to form conjunct verbs, eg bakkah-karab “be talk message”. Bhankhan-karab “to mutter”. Lastly of course, simple noun can be added to various verbs to form conjunct verbs. Depending on the frequency of occurrences, the verbs that participate as the second number of conjunct verbs in Maithili can be classified into three groups.

(X) Very frequent verbs:

laa- “some”, do-“give”, pag- “die”, uTh- “get up”, dekh- “see”, mar- “die”, kat- “cut”, pass, nag-“spread”, tod- “break, khan- “eat”, khel- “play (music)” cheek “shout”, jaa- “go”, lath- “load”, bhar-“fill”, he- “be”, lo- “take”, bah- “flow”, etc.

(XX) Infrequent verbs; rhag “get down”, kaaph- “try to change a state” jaath- “look”, jab- “add’, gil- “swallow”, nath- “road”, jee – “plough”, phul – “be fat”, That- “strike”, cal – ‘proceed”, walk, naan – “assume”. Bait – “sit” tho2k – “throw”, cavel- “rice”, the2k “steep”, loan”, haan2 “tie”, unur- “fill in”, uttap – “get uprooted”, can- “flow, han- “set as”, etc. (XXX) Verbs that never occur independive ( and hence, have no independent meaning); anraaD-, nivap- biihuk-, khisaT, kaath-, manar, miRaar- Kith, etc.

6. One of the trivial tests be differentiate a compound verb from a verbal conjunct is

the coordinated paraphrase test. Under this test both of these are expanded in fuller coordinated phrases, in which case one can find out in no time that the resultant sentence that has three compound verb is ungrammatical. Consider the following sentences which will make the point clear”. (A) o baRda saabhanii nOn dekh1 (kao) lel a thinha

he(hon) very caution with see-ptopl DO take-put-Agr “he took it having looked at it with great caution”.

(B) o raastan par oOr calait a calait a nar 1 golaah a

he(hon) road on through walk-ing walk-ing dic- pte go-pst-Agr “He died while walking along the road”.

Page 377: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

374

This test would give us the following sentences from (A)-(B).

This test would give us the following sentences from (A)- (B). (a) o barDa saabhaanii sOn dekhal a thinha aa (o baRda saabdhaamii sOn) lel a thinh

a (b) * o raastaa par sOn calaita calait a mar a laah a aa (o raastan par sOn calait a calait a ) golaah a

7. Karbir Arora (1978) independently arrived at the same test which she had applied

to establish whether a particular order of elements in a compound verb is normal or reversed. If the first member is not dilatable, while the seemed member is deleted, one can immediately tell that it was a case of reversal of polar-vector order. See (18) to appreciate this position.

8. See Singh (1976b) for exceptions. 9. Compare (29)b and (29)c with (30)b and (30)c. 10. Here V around the verbal constructions could be used for the sake of conversions.

It would denote the ‘verbal complex’ in that case. 11. Naar is used mostly in the non-standard variety. It is a common feature of the

dialect of Bhagalpur 12. It is Kashmiri in the original test, while the commentary shows Bhaadha Shastri

(1914) had also noted this. Jayadhari Singh (1969) transited it as kash 1 in Maithili.

13. The minimum was three complete, and the maximum seven. 14. Kasica (1976) calls it ‘or…..V’.

15. Bagehi and Shastri (1956): gaavare a 16. Son (1948): taPiaan Bagehi and Shastri (1956): Todian Shahidullah (1948) and

Mukherji (1963) tolias. 17. Son (1948, 1936) : shaalin 18. Shastri (1916) : camTatthai 19. Bagehi and shastri (1996): unNa son (1948) anna mukherji (1963)” anna 20. Bagehi and Shastri (1956): loian l others; laiaa n

Page 378: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

375

21. Cited from lui , 1.2.1. Original, commentary, and Shastri (1916): diTa keria. 22. Cited from biruaa, 3.2.1. 23. Cited from CaaTilla, 5.2.2. 24. Cited from KrSNA, 12.4.2. Begahi and shastri (1956) avase.. 25. Cited from Kukkuri 2.3.1. 26. Cited from Khusuka, 6.1.2 original, Nilratan sen (1975): Backi shastri (1916). Shahidullah (1940), and Bagehi and Shastri (1956): Baska. 27. Cited from KrSNA. 12.3.1. Original, Shastri (1916), Sun (1948):daahi : Bagehi and Shastri (1956): damma : Mukherji (1963): dekha- 28. Cited from Banti, 15.5.2. 29. Cited from KrSNA (caarya), 13.1.1 Mukkerji (1965): nahii (printing mistake ? ) Bagehi and Shastri (1956):Naavii 30. Cited from Dumbii, 14.1.2.& 14.3.1. For the second occurrence, Sen (1948): naara karaii: Bagehi and Shastri (1956): naara karai. 31. Cited from lmi, 1.4.2. Shastri (1916) and Mukherji (1963): laaku 32. cited from GaaTilla, 5.5.1 33. Cited from bhusuka, 6.4.2. 34. Cited from OuNBarii, 4.1.1 Bagehi and Shastri (1956): adkavaali 35. Cited from NaahNu, 7.2.1. 36. Cited from NaahNu, 10.2.1. Shahidullah (1940): kariba: Bagehi and Shastri (1956): karibe. Shashidullah (1940), son (1948) and Bagehi and Shastri(1956), do instead of na. 37. Cited from KaahNu, 11.3.2. Original, Shastri (1916), and Mukkerji (1963) Kimu ; Sen (1948) and Bagehi and Shastri (1956): kiu 38. Cited from KrSNaa (caarya), 13.3.1. Sen (1948, 1956) Shahidullah (1940) and Bagehi and Shastri (1956): kaRuaala 39. Cited from dumbii, 14.2.1. Sen (1948): uahaaraa 40. Maraprasad Shastri (1916) nora in place of gola.

Page 379: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

376

41. Sen (1948) and Mukherji (1963): uThi 42. Bagehi and Shastri (1956) nahaa nuderii instead of mahanedarii 43. Shastri (1916): Kanthaa: Begahi and Shastri (1956): kadkhaa. 44. Sen (1948): raakula n. 45. Bagehi (1938) maho-lekbhu 46. sen (1948): jaaNami 47. Bagehi and Shastri (1956): kahin. 48. This is attested in the N-Variety in the diglossic Bengali which is known as

“Saadhu” Bengali as against the L-or ‘Calit’ Bengali. See Singh (1976a) for this distinction.

49. Bagehi and Shastri (1956) nahile. 50. Shastri (1916) nermaRiius Mukherji (1963) Bagehi (1938) & Shahidullah (1940)

nornaRius Nilraten and son (1973):naaraRius sen (1948)naadarius 51. Bagehi and Shastri (1956) vikanadha. 52. Begahi and Shastri (1956) vaison 53. Shastri (1916) gaVasamda; son (1948) ga(a) Na-samudro 54. As an example of this combination, consider the following sentence from Bengali, uni to Sukaar thoke-I baRic bHraasehan he (hon) part morning” from –Emph wonder-abs wander-pros pft-Agr :He has been wandering around right from morning:. 55. For a detailed discussion of these vectors, see-6.7.1.onward. 56. “CP-V interpretation” is what I have been calling “Verbal compound” it has also

been called “Main verb sequence )Hock, 1974), and “ni-polar compounds” (Singh and Bandhopathyay. 1978)

57. We shall take this up later towards the end of this chapter. See (183) and the

subsequent discussion. 58. we would later modify this kind of marking and give “e” to such vectors in these

instances, because of the probable vectors which occur lost frequently are themselves “rare”.

Page 380: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

377

59. Coming to a score-sheet such as in (183), we would see that one has to find out both “minimum” and “maximum” scores for each of the probable vectors to get an idea of such differences between the combinations involving the same vectors. 60. We would finally see that at least one vector-dekh fails even in this test, but still Scores quite high. 61. In fact, some of the less frequent vectors score much higher than the more Frequent ones. For instance, non scores a maximum of 16 points as (183) would show, whereas dh can obtain a maximum of 10. But (159) shows that DS occurs with 11. and of a sample of 217 Maithili verbs, while nag would occur with a more 0.92 of them. Even in combinations, as occurs in 3.47 of the instances, whereas one occurs only in 0.6% of the cases. 62. Notice that some of these tests do not work in either languages showing Compound verbs. In such cases, the multiplex tests in these languages may involve only theme tests out of (149) that are relevant, unless of course this particular language has some more indigenous tests of its own. For instances, while kar-dissertation works fine in Hindi, and its parallel does a good job in Maithili, it is irrelevant for Bengali because the markers for both polar of compound verbs and the first number of a verbal compound are- e. 63. Hook (1976a: 144) gives the following Hindi verbs which do not take vectors, e.g . caah- “went”, rakh- “have”, possess and kahlaa “be called”. 64. These percentages would not total 100 because these have been calculated Separately for each vector to find out as to how many out of 217 main verbs they can combine with. Many of them would naturally occur with more than one main verb. 65. It may be mentioned have that all the 217 verbs have been listed in (160), (168) (170), (172), (174), (176) and in (178) under various types of combinations. 66. If we multiply the grammatically possible combinations with each other, we Would find out that 26334000 combinations out of a logical possibility of a huge number as calculated in (166) are acceptable in Maithili. 67. The frequency is calculated here out of a total of 30 which one gets from Multiplication of the length of the vector-combinations (which is 2 here) and the number of grammatical possibilities (which is 15 here, as (168a-e show).

Page 381: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

378 68. Here, again, the frequency is calculated out of a total of 60 occurrences of these 14 vectors. 69. The total is 72 here. 70. The total is 55 here. 71. the total is 42 here. 72. the total is 21 here. 73. The total is 32 in this case. 74. Another frequency-table can be prepared on the basis of a number of combinations (the total being 84 as 16V shows) a particular vector can enter into. This table would be exactly parallel to that in (139) and hence is given here: VECTOR NUMBER % % AS IN159 Le 50 59.52% 58.99% De 50 59.52% 54.84% Jan 39 46.43% 47.93% Nar 37 44.05% 30.41% PaR 14 16.67% 10.00% UTh 20 33.33% 25.34% Bais 32 38.10% 20.28% Aa 12 14.29% 11.98% Rakh 14 16.67% 8.29% Cal 10 11.90% 5.99% PaThan 7 8.33% 5.07 Dekh 6 7.14 3.60 Naar 5 5.35 2.76 Nikal 2 2.38 2.38 Dhar 5 5.95 2.30 Nikaal 4 4.75 2.38 Nan 2 2.38 0.92 75. Notice that some of the score may change if one searches through the whole verb lexicon of Maithili, and comes out with some more types of combinations involving these vectors. 76. see Singh, Subbarao and Bandyopadhay (In preparation) for an account of compound verbs in seven South Asian languages.

Page 382: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

379

CHAPTER VII

7.0 In this chapter, I wish to demonstrate the ways in which pragmatic fasts influence the semantic interpretation of particular sentences in Maithili. Barring a few rules such as raising, equip, adverb-creating rule, unspecified agent deletion and comparative deduction, most of the either transformations reorder and read just various grammatical elements (-both categories and relations). The liberalization rules that I would discuss here include rules that are well-known for their emphatic and other import such as Negative adverb not too well known for their implicational interpretation. The letter type of rules include compound verb beer eying, extra position- to- the-eight, extra position-to-the-left, heavy shift, sentence-shift, sifting, adverb movement, right dislocation, topic alisasion, etc. I shall show that in a number of cases, selection of a particular set of rules as against another depend on the speaks to whom, whom, where and why. To exemplify this aspects of Maithili grammar, discourse structures from novels, short stories, and plays would be quested as examples. 7.1.1. Quite often, it happens that one uses certain types of movement rules for the purposes of emphasis. While some syntactic rules are marked by “emphatic” rules, some others also serve the same purpose. Two typically emphatic rules are Negative (Adverb)

Page 383: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

380 Proposing and verb proposing. Consider the following sentences from Maithilis: (1)a ham humkar kaaj raj-I kar-bath-I

I hie (kan) week nog do-fut-Agr. “I shall not do his work”.

b. ne ham humkar kaaj kar-nbah-I (,an ne kumkan nag I his (hon) work de-fut-Agr and Neg him (hon) kare deband-I) kare debunh-I) de-Inf allow-fub-Agr “neither I shal do his work (,nor shll I allow him to do the same )

2 O kahie errak bhas Unah-I bhgol chalaah-a

Ne (hon) ever so much happy Nag be –pet ann-pet-Agr. “He was never so happy “.

b. O Unah-I akhie ettck khus bhol challah 3.a. O gam gelach-a He (Hon) village ge-pet-Agr He went to village”. b. O jaan Upah-I gelaah-a He (hon) village nog ge-pet-Agr. “He did not go to village”. c. O gam gelaah-a Uuah-I d. O.Unah-I gaan gelaah-a

7.1.2. A comparison of a- and b- version in (i) and (2) reveals some important fasts. First, nag- proposing rule in Maithili is actually net allowed to transport the negative marker from within the sentence to on E-initial position, Thus, nah-i in place of no in (i)b would generate very odd sentences. In potpie usages and in certain other places

nah-i.

Page 384: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

381 does occur the initial position, however, for example, one can say the following: (4) nah-I ham kar-anakh humkar kaaj But notice that in (4), the verb has also been shifted along with the negative marker. If the verb is not proposed, the result is a very odd construction, such as (5); (5) *nah-I han hum-a kar kaaj karhanh-I The nag-initial constructions are allowed in another set of case, where the surface subject has already been deleted by some other rule such as imperative subject deflection or an optimal subject plenums deletion. For instance

(6) nah-I deban toran kii kar-abihii-n ? Neg give-pres-you, what de-fut-Agr

“(X) shall not give you, what can you do? What is implied here is the structure in (6) was altered by an optimal subject from deletion (which operates when the subject pre-form has already left a trace in the verbal agreement marking, so that there is no desirability problem). (6) would thus be identical with (6) that follows: has nah-I deban toraa, kii kar-abihii-u ? 7.1.3. All these prove that the nag-proposing rule is Maithili has a limitation in t hat at the most it can bring the marker to a yeast-subjective position, as in (2)b. Notice, however, that for emphasize purposes (2) can have some other surface orderings even when this rule is applied. They are as follows:

Page 385: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

382 (1) e. O kahie nah-I ettak khus beol-a chalah-a d. kahie ettak nah-I khus beol-a chalah-a

Also note that the ‘ang’ can be allowed to intervene the nata Auxiliary verb combination. (2) e. O kahie ettak khus bhol-a rah-I chalaaah-a if we ew compare (2) a-e with each other, we find out that different placement of ‘nag’ has resulted jin emphasizing different sentimental element. Thus in (2) a khus ‘happy’ was and in (2)b he as stressed similarly, in (2) e, it was kahia ‘ever’ in (2) ettck ‘so much’ and in (2)e, it is the whole verbal combination.

7.2.1. The sentences in 92)a-s and (3) b-d are instances jl of an application of an k”emphatic” transformation which we have called Non-proposing transformation. Similarly one can also find examples of some other emphatic rules in Maithili, such as Verb proposing. What happens in this case in is that the verbal construction in the matrix sentence, following a conditional adverb phrase, is allowed to shuffle its position around. Consider the following examples.

(7) A. jo ahaa anbo cashit chii

if you (hon) come –inf want aun yree-agr tra hamraa etae nab sakait chii then I-obj place-Ise come ypopl anaus-press-Agr “If you want to come, you can come ever to my place’.

Page 386: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

383 c…….. nab-I ta sakait-a chii hamaraa etee d………nah-I sakait-a ta chii hamraa etae. e…….. nab-I akaita chii tahamran etae.

7..2.2. In these sentences, one can either say that the verb proposing has taken place in various ways, or maintain that it has applied first uniformly on (7)b through e, and then an optimal movement has given rise to structure such as (7)a and d. It may be noted at this point that it can not be moved across the verbal construction here, whereas in other places, is allowed to move around relatively freely, e.g., (8) a. O te kaalh-I ael-a chalaah-a No PART yesterday come-yet Aun-pet-Agr “He came yesterday”. b. O. kaalh-I ta anol-a chalaah-a O. kaalh-I aael-a ta chalaah-a O. kaalh-I anol-a chalaah-a ta. It may be noted here that to movement (or partial-movement) has a similar effect to that of non-proposing or nag-displacement. But even in this case, to-movement is restricted in the same way as in (7)e, especially when the verbal construction is preceded by a negative marker, e.g., (9) a. o ta kaalh-I nah-I aael-a chalaah-a “He did not come yesteraday”

b. O. kaalh-I ta nah-I aael-a chlaah-a

Page 387: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

384 c. *O kaalh-I nah-I ja aael-a chalaah-a d. *O kaalh-I nah-I aael-a ta chalaah-a e. *O kaalh-I nah-I aael-a chalaah-a ta. 7.2.3. It does not matter as to whichever analysis one prefers for the sentences in (7)b through e, the point remains the same, namely, that both (2)-(3) and (7) are instances where a change in emphasis (or a change in the position of emphatic particles) is being brought by certain transformations which can well be named as “emphatic” transformations. But there are many other mechanisms by which the same purpose can be served, although such rules are not known to be linked with the function of emphasis in any direct manner. 7.3.1. As an example of one such mechanism, we would take up the case of compound verb beer eying transformation 4, whereby the normal order of polar and vector or main and explicator) verbs is reserved. In Indo-Aryan language, the normal order is POLAR+ VECTOR, and the reversal of this order is grammatical only in some 5. While Hindi and Maithili allow such a reordering, Bengali6 and a few others do not permit it. Examples from Hindi and Maithili could be given at this point to show how this mechanism serves the purpose of emphasizing a particular element. Mechanism serves the purposes of emphasizing a particular element.

(10) a. raaj ghar se baahar niakl aayaa (kachru, 1979) Raj house from out emerge came. “Raj come out of house”. b. raaj ghar se baahar an niklaa. c. Raaj ghar se baahar saraa nikal.

Page 388: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

385 (11) A. ii baat humjar non ull shun-I gal-a chaa-I This matter his(hon) mind in enter go-abe aur –pre-Agr. “This matter has entered into his need” b. . ii baat humkar non nel jaa shunal-a-a chaa-I c. ii baat humjar non ull sal-a chaah-I abus-i.

7.3.2. The Hindi sentences in (10) and the Maithili ones in (11) show that the reordering of polar and vector verbs can take place in two ways. In the b-sentences, not only did the polar and vector verbs inter change their positions, they had also exchanged their markers. In comparison to that, in the e-sentences, the verbs have only exchanged fir respective positions in the sentence without affecting the markers. In case of (11)e. the polar verb has not only erased the vector base, but also want across the auxiliary or supporting verb He one who has even a working knowledge of these languages would like the maintain that the optimal compound verb reordering rule did not “change” the meaning of these sentences. In both the cases, the b-sentences can never be tagged to following clauses: (12) a. raaj ghar se baahar aa niklaa, kye-nki use Raj house form out comes emerge past came him Nikkaal diyaa gayan Ouster give-plus-pet be –yet “Raj came out of house, because he was evicted”.

b. * raj dhire ghar se baahar aa aiklaa. Slowly slowly. “Raj came out of house very slowly.”

Page 389: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

386 (12) a. ii baat kumkar nan ne jaaghusal-a ehamh-I this mater his(hon) mind in go enter-pst ptepl be-pres agr. Unida taiye oshan-nk kahab baab-I letah-a But still he(hon) you(hon) say-Inf agree take –fut agr. ‘This matter has entered into his head, but still he would agree to (what) you say”. b. * ii baat hum-akar nan nE eh-I dui barkh nii this two years in c. jaa ghusal-a chah-i. d. “This matter has in the last two yeas. e. Entered into his head”.

7.3.3. Notice that the some sentences are perfectly acceptable when either the polar-vector has been changed without a change in the markers or when the order is normal. Compared to the sentence in (12) and (13) in the latter came, the degree separately seems to increase in Maithili. Consider the following examples.

(14) a. raj ghar se baahar (nikal nayan) kye-nki use . nikaal diyan gaya

b. raaj dhire dhire ghar se bahar (nikal sayan

(15) a. ii baat humkar sannn nf (ghus-I gal chauh undan taiye o chann-nk kahab naan letah-a ghus gel-u chandh-I

gal-a chandh-I chaus-i.

Page 390: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

387 47.3.4. The examples in (12) through (13) clearly show that a linear reordering in compound verbs change the meaning. The unacceptability of (12)-(13) as against the complete acceptability of the polar vector order in (14)-(15) indicate that a reversal accompanied by a marker-shift make the resultant compound verb construction somewhat “volitional” and “mutative” thereby bringing a degree of “guide ness” in the action being described. In all such instance’s they Laos serve an empties function. Such a reordering has a parallel in the directional adverb proposing rule of English (green, 1978): (16) a. The ballon went up

Up went the ballon. 7.4.1. Some other types of movement transformations have different kinds of motivations associated with their selection. The traditional rule of extra position, for example, brings changes in the degree of festivity as well as introduces a feature of incoherence in the complement sentence at times. Let us take up the following sentences as examples: (17) a. O murut (je shan-n so-u pa-nis some rahath-I That clown who you form loan take sum-pet-Agr. Etas sael-a chalah-a Here come-pet sun-pst-Ar “The man who borrowed money form you had come here”. b. o murut etas sall-a chalaah (je chaa-nso-n pa-n-ie some rahath-I a. (je (murut) aha-n se-n pa-nie name ruhath-I) o (murut) etas saal-n chalaah-a.

.

Page 391: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

388 (18) a. ham (je hum-akaa lakhE eer thikianh-I) ii ta I GPL him-to in-the-eye-of thief be-pres-Agr it-PiuL. Spaete ah-i/ach-i. “That I am a thief in his eyes is obvious”. b. ii in spaffe ah-I/ach-I (jo ham kum-akan lakh out thikian-I) (19) a. (MuTTe niik jakan-a see-I Meaar-I kae anthir Bhutte good way-in think-debate-de-ptepl cold. Eitt-abhae kak-arak-u khum kar-abame chath-a Heart be-ptepl someone’s murder de-cuse aur-pet-Agr. Ii ta asraje bealan gappy-a bhel-a It PTCL strange like matter be-pres-Agr. “That Bhutte had got someone murder in cold blood, having debated and thought ever. Fit is a strange theory”.

c. ii ta ae-araje baalaa gap-a bhel-a (je bhhuffe) niik jakan-n see-I bicaar-I kae asthir eith-a bhae kak-a rah-a kkum kar-a bame chath-I) 7.4.2. These sentences have at lest one thing in common. In al of these the embedded

sentence is optionally movable to different positions. In (17) the optional structure were as follows.

(19) a. NP(Det-N)NP-(je-S)-adv-VB b. NP (Bet-N)NP-ADV-VB – (je-S) c. (je-S)NP (Det-NP-ADV-VB.

i.

Page 392: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

389 The movements in (18) result in similar placements of s2 although structurally the constrictions may be a little different. The same in the case with (19). Their structural arrangements could be schematized in the following way: (20) a.(je-S)2 – PRO-PART-ADJ-VB

b. PRO-Part-adj-Vb – (je-S)2

(21) a.(je-S)2 – PRO-PART-ADJ-P-VB

b. PRO-Part-adj-P-Vb – (je-S)2

7.4.3. It need not be mentioned that the major differences between the two structures above lies in the “heaviness” of the sentential subject IP which is strikingly similar to (20)e and (20)b, respectively. The commences is more than structural. In the b- sentences (in 17 through 19), the effect of a rightward sentence-movement is that of “de-factorization”. Thus, while (17)a, (18)a and (19)a do not permit “assertion” of the contents of the embedded sentences, their b-versions seem to be an opposite effect. I would assign the “rightness site of a sentences the reason behind such an interpretation. Thus, I would not simply say (as give, 1973, 1976 would do ) that the cause lies in the more selections of extra position transformation. This is because although 917)a has also undergone a similar operation (with a differences that here this transformation has moved the embedded sentence to the “leftmost” position of the higher shade), there was no such effect. Rather, in (17)a, there is a remarkable “increase” in the presupposition strength of the contents of the embedded s.

Page 393: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

390 7.4.4. This can be made clearer if we take up another structure where s2 originates a

little more towards the right corner such as in (23): (23)a.ham ch-I murt kB-n (je chaa-n se-n pa-nle I tht-obj elewn to who you form loan. None rahath-I khumb dhopalianh-I Take aun-pet-Agr very beat/scold –pet-Agr. “I gave that man who had borrowed money’ from you a good boating”. b. han ch-I murat kB-u khumb dhopalianh-I c. (je ahaa-u so-n pa-none rathath-I d. je ahaa-u so-n pa-none rathath-I han ch-I murut kB-u khumb dhopaliamph.

In this case, s2 appears in the subject IP position in (23)a. Notice that here too, the rightward extra position an applied in (23)b and the leftward extra position as in (23) e polarize the situation. (23)e makes the content of s2 more festive than it is evening (23)a. On the other hand, a deactivating effects is clearly visible in (23)b. give (1975, 1976) while discussing some of these structures attributed the change in interpretation to extraposition in general due to a reason that becomes more and more evident now. Give was dealing with language structures where a leftward movement was not possible. It would be interesting to see whether an SVO language that permits a leftward extra position shows similarity with Maithili and other Indian languages in this respect. A similar search can be made among

Page 394: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

391 The languages with the kinds of word order tee. I would however, tentatively assume (24) to be true for Maithili (if into for SOV Languages in general): (24) a. The “eastern site” of an embedding sentence is more prone to defactivization than any other parts of it. B. The “Western site” of an embedding sentence increases the degree of the presupposition associated with the content of the embedded 7.4.5. A final test for (240 can be carried out with the help of those sentences which are “traditionally” thrown in the garbage of non-factice predicates, because they involve verbs such a buth- “think”, aee-“consider’ etc., If it can be shown that (24) is valid even if one applies extra position on structures which have those predicates, it would only confirm the explanations given for different interpretations of (17)b and (17)a. Consider the following sentences. (25) a. han (o niik log chath-I) buthait chalaku I he (kon) good man be-pres-Agr think petpl sun-pet-Agr. I thought he was a good man”.

b. ham buthait’ chalaku-n(o niik log chath-I) c. (o niik log chath-I ) han buthait-a chalahu-n

26. a.ham-a raa laagal-a (je o miik lag chath-I) I-obj seem –pet-agr. “It seemed to me that he was a good man”.

Page 395: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

392

b. ham-a raa o niik log lag-a laah-a I-obj be good man seem –pet-agr (with o) “To me , he a Goodman seemed”. c. o ham-araa niik lge lag-ulaah-a “He seemed to me (and I was almost sure about it) to be a good man”. d. ham-a raa lag-a laah-a o niik lag (mudan…………) “He seemed to me to be a good man (but actually……..)

Here, (25) shows extra position (where the completive is has optionally been deleted) and (26) shows raising and related phenomena. What is worth noticing in these two sentences is that while in (25) think a non-factive predicate (traditionally), does not “change” its meaning altogether, log “seems” undergoes a total “change” with respect to factivity. Since does not “yield” from its original non-factive meaning, when embedded sentence is brought forward to the western size as in (25)e, the result is disastrous. This particular sentence is not only odd, but one also feels that it has an effect of erasing such a “psychological” gap between the two parts of the sentence that they seem to be two unrelated sentences. That is why (25)e seems to be verse that (25)a, where gain a movement of s2 towards the western site has made the sentence a little odd. Notice that (25) a is odd despite a transformational assumption that it is the nearest to the deep structure order of elements in this particular case. (25) b is undoubtedly the best of the three, because the predicate concerned is already a non-factice and the movement of s2 towards the eastern-most corner of (25) has only increased in the degree of non-festivity, as expected. The situation in (26) is very

Page 396: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

393 different. (26)b is the result of an application of raising rule. Prier to this, the subject IP of the main sentence has already been denoted to do position, because the verb last long in the sense of “seem” requires that the do be advanced and subsequently SUBJ be denoted. Thus, (27)a is acceptable while (27)b is not: (27) a. han-a aran ii laagla-a

me it sems-yet-Agr(with ii) “It seemed/occurred to me”.

b. *ham (ii ek-a ran) lag-n lahu-u long lahu-n) it-(obj) sem-pet-Agr (with ham) “It seemed/occurred to me”.

7.4.6. In (26)b, the sentential complement structure has been destroyed by lifting the subject of s2 to the subject of s1 position. But this change of relations was, in this case, accompanied by a change in position of the derived subject “he” along with its attribute niik lag “good man”.They moved almost to the western most site. According to (24). It should have an effect on the context of original s2 vis-a-vis presupposition interpretation. This is exactly what has happened. The result has been something more than that was expected, i.e., the verb lag-long here meant “surely seem”. One may argue the (26)a and (26)b have verbs which are accidentally homophonous. But then one could fail to explain as to why the same thing should happen when lag-long is replaced by another non-factive verb such a buth “think” as in the

Page 397: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

394 following sentences: (28) a. ham bujhait-a ehalahu-u je o niik log chath-I I think-ptepl sum-pet-Agr Cfi he good man be-ye-Agr. “I thought he was a good man”.

b. han hum-akaa niik log bujhait” chalianh-I “I was almost sure that he was a good man’.

7.4.7. It is also worth-mentioning that in traditional terms, (26)b is likely to be described as a result of something like tough-movement or subject-to-subject raising, whereas (26)b as a case of a simple subject-to-subject raising. The effect of both is, apart from a change in relations, a movement towards the west. This, like the westward extra position, follows the principle enumerated in (24)b. I may also mention here that a similar thing happens even in English sentences which I quote from rests (19670 and Forte (1971) here: (29) a. It is hard to play sonates on this violin. b.Sonates are hard to play on this violin. c.This violin is hard to play sonates on. (ye) a. The crud on my glasses makes it difficult to read my notes. b. The crud on my glasses makes my notes difficult to read. Green’s (1978) observation for these and similar sentences is that tough-movement attributes tough predicate to the inherent nature of the derived subject. I feel that in 9190b,e and in (30)b, the degree of ‘difficulty” in doing certain things increases.

Page 398: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

395 7.4.6. These similarities between Maithili and English arise curiosity as to what happens in Hindi so far as raising is concerned. Subbarao (1974a-104-b) supports the claim of Kipareky and Kearney (1968) that raising is not permitted by the festive complements. In Hindi, naan “Consider”samahth “think”, understood and las “appear” frequently take raising, but not the non-festive conjunct verbs such as ViSvas he believe” (Subbarao, 1974a-104-5). This implies that the festive predicates jaam “knew” dukh he “regret” and paresaan he“worry” would not permit subject raising. But Anjani Kumar Sinha (1975) argues that festive completes in Hindi do permit raising, and he gives examples involving predicates such as noon- “great” and samath “realize”. His examples included the following. (31) a. ve dhekevaas jaataa hai he evindler is-granted “He is granted to be a seindle” b.yungaii acchii nahii sajhii sayii Susie good not was realized. “Susie was realize to be no good’.

Subbarao (1974a: 112-3, far 16) objected to this analysis mainly on two counts: First, he did not find Sinha’s translation of saahaa jaastee hai “is greeted” as proper; and, secondly, (31)a could be followed by a clause that denies the so-called factice assertion. Now, Subbarao (1974) also tried to explain a problem raised by Kasha (1972) regarding estimate of raising rue application for lag- while it was obligatory for jaa-na- (when both of these near “appear”), by taking a position.

Page 399: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

396 that lag had many interpretations in Hindi. He claimed that under the interpretation of “seem”, raising was optional whereas it was to be applied obligatorily where the meaning was “find”. If we take (24) to be “mere or less” true for Hindi too (one has to work in detail to find out whether it really is true for Hindi), the reason for the debate and confusions noted above became clear. Subbarao (1974a) is quite correct in translating naan and samath as “consider” and “think/understand” as these pre non-festive verbs. However, these verbs while permitting raising de got a relatively festive-like interpretation (of.24b), and this is why one can understand why Sinha (1975) translates the same verbs as ‘great” and “realize”, respectively in (31)a and b. This also explains why the assertions in (31) a and b. This also explains why the assertions in sentences in (31) or similar constructions could be denied by another clause starting with “but”. If this is correct, one would also like to imagine that the two interpretations of results from (24)b. Thus, a number of Hindi speakers would find (32)a to be acceptable, and (32)b is surely grammatical if lag is taken to mean “soon” (32)e above that the “find” interpretation is not completely “festive” here as Subbarao (1974a-108) thought it to be. Consider the following sentences: (32)a. raan ke vak kakaanii saeeii legii, lakia ram obj that story tree appeared but mere ke nahil others obj not. “The story appeared to be true to Ram but not to others”. Or, “Ram found the story to be true but others did not”.

Page 400: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

397 b. readhikaa ke lagaa ki am miithe hai’ Radhika obj seem that mangoes sweet are “Radhika seemed to have found the mangoes sweet”. c. Raadhikaa ke vah lakas asehass legaa, radhika obj that bey good found lekin dar-asal vah or behut hii but in actuality to one very emph khataraek baskum thaa dangerous decit was “Radhika found the bey to be gee, but in actuality he was a very dangerous deceit”. Like Hindi lag- “seem” , seem to find, find”, in Maithili even jaan- “knew” is capable of two interpretations; ‘new” (factive) and “seem to know” (non-factive), and in the latter interpretation jaam- may permit raising as the following Maithili sentences would show: (33) a. ham janait-a ehalahu-n je raam das-a-rathak I know be-pet-Agr that Ram dasharath-gam Bhaal ehalaah-a Brother be-yet-agr “I know that Ram was Dasharath’s brother b. ham raam-kB-n das-arathak bhaai jamait-a I Ram-obj Dasharath –gum brother know Ehaliamh-I (madan baad mB pataa- Be-pat-Agr but later in finding Calal-a je e humak baalak ehalaah-a He-pst-Agr that he he-gum son be-pet-Agr “I know Ram to be Dasharath’s brother (but later on I came to know that he was his son)

7.5.1. Recall that we had started with a discussion of T-rules

Which are not known as emphatic but which actually are meaning.

Page 401: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

398

changing transformations, and as examples we discussed the nature of compound Verb Recording, extra position, and Subj-te-object as well as Subject-to-Subject Raising rules. The same point could be further stressed by giving examples of Sentence shift from Maithili. (34) a.T-ham apan eh-I nitrak (jam-I-kaa so-n aai so-n I myself-gen that friend-gen REL with today from Das barkh yumrav-a paricay bhel-a chalgah-I Ten years before introduction be-gptopl aur-pet-Agr Khwn kaeliaah-I Murder de-pet-Agr “I murdered that friend of mine when I was introduced to ten years age since new”. b. ( jan-I-kan so-n aai so-n das barkh yurv-a paricay bhai-a ehalaah-I )ham apan eh-I mitrak khum kaelianh-I c. ham apan eh-I nitrak khwan kaeliaah-I (jam –I-kaa so-n aai so-n das barkh uuurv-a paricay bhel-a chalanh-I)

Here , the “weight “ of the MP taking a relative clause has increased so much that the S2 has to be shifted either way, depending upon which part of the sentence is to be emphasis than the one that follows. Interestingly, (34) a, b, and c, do not follow (24) , but they do not violate this generalization either,

7.5.2. Another rule that allows intentional exaggeration, and would thus emphasis on the action being described by the predicate is comparative reduction. In the sentences that follow, the a-sentence derives form its counterpart, in b, and they differ from each other in that the former is hyperbola by nature:

Page 402: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

399

(35) a. taarini baahu garaj rahal chath janaa baagh heth tarini mr. Shout-ptepl pre axe-pr-Agr as if tiger be “ Tarini baabu is shouting as (if he) were a tiger “.

(36) B. . taarini baahu garaj rahal chath janaa baagh garaj rahal he

“Tarani baabu is shouting as the tiger shouts” 7.5.3. with this, we may add another construction which also comes from a source closer to (35)b. After the comparative reduction, another rule may delete even the dummy verb he – “be” from the surface, replace-9 jaman “as if” by jakan “like, and reverse the order of the words jakaan baack and finally advance this now -formed 00. (or some kind of adverb, if we may say so) immediately after the higher subject NP. The results would be as follows: 36. taarini baabu baach jakaan garaji rahala chathi tarini mr. Tiger like shout-ptepl prog sum-yr Agr “tarini baabu is shouting like a tiger “

This adverb-creating rule when applied on structures such as (35) b ahs an effect of “exaggeration” that supposes (35)a.

7.6.1. Under the emphatic function, I would discuss two more rules;

Sifting (see Ross, 1973) and Unspecified Agent Deletion. In the first case, the net result is that the otherwise unimportant matrix sentence becomes suddenly “visible”’ and announces its existence, as it were. But all these happen when the matrix sentence is lifted towards the rightist corner of the original sentence. Following are the examples:

Page 403: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

400

a. bujhaai yE (aab humraa jaeback eachii) seem-pass- axe-press now me go-inf-gon must pst-ptopl. “I guess, I better leave new” b. (aab hamara jaeback caehii) bujai yE

“I better leave now, I guess”. (37) a. hamraa lagai yE (o aab nahi actech to meseem –press aun –press he now nag come-fut-agr. “I think , he would not come now”. b. (o aab nahi kecta ) hamaran lagai yE “He would not come now, I think”.

7.6.2. One can possibly add to this kind of “lifting” these sentences which follow a sentimental adverb such as the following chari-nak ‘possibly’ kina-hum ‘perhaps” aakira to finally”, laga-bhag “almost” bari-bari “again and again ytha-To ‘ in the contray”, etc., let us take a few comparable examples. (39) a. bhari-nak aab hawran jaabaak oachii “Possibly, I should go now”. b. aab hamara jasback caachii char—tak (40) a. laga-bhag aadhaakaaj baskii achi almost half the work remain be-pres-Agr. “Approximately half of the work remains’.

b. aadhaa kaaj baakhi achi lagabhag

41. a. unu-ta o hamaara kesa laga-bhag on the contrary he(Hon) me occur keep-on-pet-Agr “On the contrary , he kept on cursing me”. b. o hamara kesae laga-lasha una-te.

Page 404: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

401

In all these instances, the force of the adverb considerably increased when they occurred in the final position rather than in the normal sentence-initial position. In the b-sentences, the adverbs are taken almost literally, whereas in a-sentences, they occur merely as a way of speaking. Thus in the a-instances, their presence is negligible. In fact, the short matrix sentences in (37)-(38) too behave like (sentimental) adverbs.

7.7.1. Unspecified Agent Deletion also serves the function of emphasis, but in a different way. It can be described as rule which de-emphasizes the element(s) which would not be of much relevance in undertaking the message conveyed. The de-emphaisization takes a form of deletion of the unknown or un important agent. In effect, it can also be used to show the neutrality of the speaker describing a particular situation. Or, it is possible that both speaker and hearer know the person (unnamed agent) concerned, whose name should not be uttered publicly because of fear, shame, or some kind of taboo. Alternatively, the speaker may use this devise to arouse interest in the kind of listener(s) and to sustain such an interest. Finally, one could also use it to avoid fixing one’s responsibility of an even as someone. Notice we that some of these could be achieved when one passivised an active sentence. Thus, by relegating something form subject to an X or N relation is itself a way to minimize the importance and relevance of the same in speaker’s view. Similarly, the passive rule does also have a lot of other functions (see Sinha, 1973, 1974).

Page 405: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

402

7.7.2. The examples of unspecified agent deletion which would show the above-mentioned intentions of the speaker could be given here under different heads where some of the examples appear in the form of dialogue: (A) Accnt Inaumita (nt Inaumita (Setting: A news paper report) (A2) Kaalhi raati kura-saulda bala salak par yesterday night kursaela about rad on dui vyakti 10rik dhakkaa son mearala gelaaha two persons truck-gun hit from kill-pass ge-pet-Agr. “last night, two persons were killed having been hit by a truck on the road to Eurseelaa”.

(B) Uninpertant agent (setting : A market) (43) desar desar dukaan par ta ii barton ekar other other shop at PART this utencil it-gun sadhaa damn nE biki rahala achi half price in sell preg-prs-pet-ptepl aur-yr-Agr. In all other shops, this utencil is being sold at halfthe (-your0 price”

( c ) Speaker’s impartiality (Setting converstion between N&S) (44) N (carer) antn nf bhuTTe kii bhelanhi yan? End at Butte to what be-pet-Agr, Ch “Ney ! what happened to Bhutte, finally ?”

(45) S (peaker) : haetanhi kii? Phaan sikn sajaa dela gelanhi Be-fut-Agr what? Hanging penalty give ge-pst ‘what (do you think) will happen? (He was) given the death penalty”.

Page 406: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

403 Speaker-Neareer’s introduction/Tabes (Setting : Pest emergency period in India) :chi-cab din nF heita kii chalika ? These days in be-ptepl what aur-pet-Agr “what happened these days’. (47) S : gaach-birich repaa gelaika, pakaRi pakaRi tree-lant saw-pass ge-pst-Agr cath-redupl-ptepl ake nasbadii kanela gelaila,an ekarae de-ptepl sterilization de-pase ge-pet, and him(this) ekaraa sonjhiiri kae paai jamaa kael gelaila him(that from cheat de-ptepl money collect-de aux-pet-Agr ‘Trees were planted , sterilistion was done by feree, and a (huge) sun was accumulated by cheating every body”. (B) Interest sustainer(Setting: story telling by grandma) (48) Child: taker baad kii bhola ? It-gen after what be-pet-Agr. “what happened after that ? (49) c-na : taker baad, raajanyaak biaah bho gelanhi princess-gen marriage be-pres Aux-pet-Agr “After that, the princess was married’. (50) child: nudaa, bholanhi kinakan son? But, be-pass-pet-Agr-when with “But, when (was she married) to”. (F)Non-innutaion of responsibility: (settings: asling for pardon)

(51) nonayan nF bhiul baiaye jaai chai:

childhood in PART mistake be-emph –go-ptepl sun-ept-agr.ettak geTan nE son kaka0ra paka Bub?

These many in from when catch –fut-Agr.

Page 407: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

404

ekTaa bhuul bha gle achi cheRi dianhu One mistake be-pass aen-pet-Agr leve give-imp Hima kae-sab kBu Them(Hon) obj “Mistakes are commonly done in childhood, when should I catch out of so many? A mistake was made. Let them go”.

7.8.1. Having discussed various types of emphatic functions performed by transformations such as Negative Adverb proposing, Verb Proposing, compound verb reordering, extra position, subject-to-object Raising, subject-to-subject Raising, heavy NP-shift, sentence-shift, comparative reduction, adverb creating rule, sifting, adverb movement, and unspecified agent delectation, now we can turn to “discourse-structural rhetorical” functions performed by a few rules which involve inversions, movements, dislocations and some sort of elating.

7.8.2.1 We shall start with some rules that create topic comment structure. Topicalisation and right dislocation are the ones we would discuss here. Notice that the functions they perform are similar to that of emphasis. Examples of topicalisation rule are as follows: (32) a. ham skill nahi thikahun I pleadernog be-prs-Agr. “I am not a lawyer”.

b. ekiil ham nahi thikshun

‘Lawyer I am not”.

The word skill is the topic in (32)b, and a change of the position makes the following distinctions between the two sentences: In the second sentence, what is implied is that

Page 408: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

405

Although the speaker is not a lawyer, he is no less than one in his knowledge of legal matters: but no such meaning is implied in (52)a. This can be tested in the following extensions of these sentences: (53) a. han ekill nahi thiakhun, tBn hamraa kanunak so I-obj law-gun gyaan nahi achi knowledge nog be-pres-Agr. “I am not a lawyer, hence I do not know abot legal matters”

c. * ekill han nahi thikahun , tru hamara kaanwanak gyaan anhi achi

The second part of (53)b contradicts what is implied in the first part, and hence this unacceptability. The second part has a negative particle in (53)a because the first part has one. For this test to be successful in this case, the two parts must have exact parallels in the matter of negation, otherwise, both of them will be equally unacceptable, as in (54) which follows: (54) a.* han ekill nahi thikahu , tBn hamraa kanunak so I-obj law-gun gyaan nahi achi knowledge nog be-pres-Agr.7.8.2.2. Notice that if the first part is affirmative, the test still works. That is, the b-sentence becomes unacceptable even in this case. Let us take the same sets of sentences and prove the points.

Page 409: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

406 (35) a.ham ekill thikahun, tBu hamraa kaanumakgyaan achi

“I am a lawyer, hence I know about legal matters”. b. * ekill ham thikarun, tFn hamraa kaanuank gyaan achi

In other words, what (52)b implies is not (53)b or (55)b. To make the latter sentence acceptable, one has to introduce tatve “still” in place of X “hence”. The taive-test has a condition that the two parts of a sentence must have a difference in terms of affirmation-negation. In this matter, it is exactly opposite to the Xth-test. That is why, in this case, the sentence in (53) would be ungrammatical, while these in (54) would be acceptable. However, in that case, (57)a (parallel to 52a). consider the following parallels of (53) and (54): (56) a. a.ham ekill thikahun, tBu hamraa kaanumakgyaan achi “I am not a lawyer, still, I don’t know about legal matters”. b. * ekill ham thikarun, tFn hamraa kaanuank gyaan achi

“I am not a lawyer , still I know about legal matters. ekill ham thikarun, tFn hamraa kaanuank gyaan achi

7.8.2.3. what tN-n and taive-tests show is that there is a systematic difference in meaning between a normal and a topicalised sentence. It also shows that the topicalised sentence has certain implications, which its counterpart does not necessarily have.

Page 410: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

407

This is the reason why (52)b seems to be an incomplete sentence to the native speaker of Maithili, while (52)a does not.

7.8.3.1. Right dislocation in Maithili occurs in the sentence such as follows: (58) a. hamar nB;chi bhikamganian KBn bhagan deln thinha I-pres nother that beggar-woman obj drive out give-pet-Agr. “My mother drove that beggar-women out”.

b. Hamar nBo ckras bhagan dela think chi, bhikhanganian kpn “My mother drove her out, the beggar-women”. (39) a. O hamara inglBaD nB bhenTala chalacha Ne(Hon) I-obj England leefind-pess be-pet Agr “I met him in England”.

b. O hamar etas bhen Tala chalaahn , inglBaD nB “I met him there, in England”.

(60) a. chenn pacha beRiyas-balaa gaechii humkaa “You (Hon) backyard-one garden be(Hon)-obj do delianhi ? give-ptepl give-pet-Agr. “Did you give away the garden at the backyard to him?”.

b. ahaan o humkaa do delianhi, pachabaRiya-balaa haachii?

7.8.3.2. In all the b-sentences, one of the object NPs is being displaced form its original position to the eastern end of the sentence, and depending upon what relation it held in the original.

.

Page 411: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

408

clause, a pre-form is introduced in its place. The pre-form include ekras “her” in (58)b, eted “there” in (59)b, and a “that” in (60)b, because the dislocated NPs were DS, OO (locative) and DO of di-transitive, respectively, Notice that if the “subject” is dislocated (to the right), this PRO may not show up in the surface of these sentences. For instance, consider the following: (58) e. chi bhikhamangnian KBn bhagaa delathink , kamar nBo “(The) drove the beggar-woman away, my mother”. (59) e. hamran inglNaB nN bhan Taln chalana “(He) met me in England, he”. (60) e. pacha haRiyan-balaaa ganchii humkaa do delianhi “(You) gave away the garden at the backyard to him, you ?”

7.8.5.3. What is worth-noticing here is that in case of (58)e, one can, if one wants to, use a pre-form in the subject position, but cannot do so in case of (59)e and (60) e. This is because the latter already have proximal subjects which are being displaced, e.g., (58) d. O chi bhikhnangniyan kBn bhagaa deathinka, hamar nBe. (59) d. *O hamra inglBaB nB bhenTala chaacha o. (60)d (*2) pachabaRiyan-balan gaachii kumkan do delianhi, chaan? Delianhi, chaan?

For example, (59) o can be made grammatical by introducing a subject such as raanandar balm “ Mr.Ramchandra”, e.g.,

Page 412: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

409

(59)d’. o hamara InglBaD bE chenTal chalacha

raameandar –baabu

7.8.3.4. Also notice that there is no pause-mark before the dislocated o in (59) e and d, but (59)d’ has the same. This, I believe, has to do with the relative weight of the NP being dislocated. The general rule seems to be: the heavier the NP d (-the ‘d’ is located NP) the better is the sentence with a pause before it. We can take (59)d’ and exemplify this rule by changing the NP d with gradually lengthier (and hence heavier) elements:

(64) a *o hamra inglBaD bhenTala chalaaha, o “He met me in England, he”.

b. o hamraa inglBaD nB bhenTala chalacha, raamandarhaabu “He met me in England, Ram chandra”.

c. o hamraa inglBaD nE bhenTala chalaaha, shaanku bhasisaahab “He met me in England ,your (elder)brother”.

d. o hamraa ingBaD bhenTala chalacha, bhaachii balaa dulhan-iii “He met me in England, the brother-in-law from Barachii”.

e. o hamaraa inglBaD nB bhoeTala chaacha , barachii bala dulhaa iiik nisiaxta bhaai

Here, (61)e is the best of the let, while (61)d is better than (61)b and e, and between the last two, (61)e sounds better than

Page 413: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

410

(61)b, And, finally, (61)a which is nothing but (59)d is not acceptable at all. Similarly, ()f would be better than even (60)e:

(61) f. o hamran inglBaD nB bhonTala chalaacha NPF2 (ahaank ofis nE kichudin puurva kanj karaita Your (Hon) office Ice cometimes back work de-ptepl chalcha)n2, So)NP. Aur-pet-Agr he “He met me in England he (who) worked in your office till recently.”.

7.8.3.5. (61)f is, of course, a little different from (61)a-e in that the “dislocated” NP here has a whole sentence under it. The embedded sentence in this case is a relative clause without a relative pronoun is (which night have been deleted by an optional relative pronoun delectation rule). As it is, (61)f does not seem to be a convincing case for right dislocation, as one night argue that here the relative clause has been extra posed. The question that has to be answered then is how is it that the correlative pronoun as has also been shifted from its normal (i.e ., matrix or higher s-initial) position along with the s2. One may again say that it is a case of shift of the sentential subject NP. But even this analysis would run into serious problems, because if at all the pronoun so is moved along with s2 , how can one explain the presence of a dummy pronoun o in the s-initial position. (61)f would have the following parallels where so and e can occur

Page 414: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

411

Optionally but cannot occur simultaneously; (62) a.(jo ahaank ofis all kichndin purva kaaj karaita chalacha (se) hamraa inglBaD mE bhonRala chalaaha b. (se0 hamraa inglBaD mE bhonTala chalaaha (je chaank ofis NE kichudin purva kaaj karaita chalacha) e. hamran inglBaD bhonTala chalaha (je ahank ofis nB kichudin puravaa kaaj karaita chalacha (se)

(62)a has an order of elements which may optionally be changed in such a way as (Ns-PRO)np is changed to (PRO-S2)np It hardly matters as to which order is taken as the “basic” by the traditional transformation lists, the point is that one of the two structures would have to be described by them as a case of sentence-shift. Similarly, S2 has been extra posed to the right in (62)b and (62)c, and this is an instance of (complex0 NP-Shift. In all such versions, 2 and so are to be found in the same position. Therefore, (61)f is clearly different from the sentences in (62), and the argument that (61)f has resulted form an application of two different rules-extra position and subject NP (here a “pronoun”)-shift, is clearly faulty. The only explanation that seems to be convincing to me is that (61)f is parallel to the sentence in (61)a-c, where a rule of right dislocation has applied.

7.8.3.6. Left Dislocation does not seem to exist in Maithili.

Page 415: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

412

(and in some other neighboring languages), unless one takes sentences such a (63)b as instances of the name: (63) a. raamchandar-baabu is hamar saathum thikash” Ramachandra Mr. Part I-poss Sister-in-law’s husband be-free-Agr “Mr. Ramachandra is my Sister-in-law’s husband:.

b. ramacnandra-basha (?). e ta hamar saathum thikash: “(are you talking about) Mr. Ramachandra? He is my Sister-in-law’s husband:.

7.9.1.1. A language that shows Negative Adverb Proposing as well as Verb proposing rules to have emphatic connotations, is also expected to show movements of other kinds of adverbs for rhetorical purposes. Maithili is no exception to this. Movement of adverbials is a very common device is Maithili. Consider the alternative structures in (64), where adverbs could be moved singly or it could be coupled with another reordering transformation such as inversion of subject and object, or that of subject and verb. That is, if for a moment we assume that Maithili has a “favored” order of elements such as in (64)a, one can expect (64)b to (64)h as possible alternative orders that could be selected by a person speaking or writing in Maithili for various rhetorical purposes: (64) a. SU-IO-DO-Adv-V

b. Adv-Su-10-D—V c. Adv-V-Su-10-Do d. Adv-v-Do-10-Su e. Adv-V-Su-10-Do f. Adv-V-Su-Do-10 g. Adv-V-10-Do-Su h. Adv-V-Do-Su-10, etc.

Page 416: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

413

7.9.1.2. The adverb phrase can also be moved to the Left without bringing it to S-initial position. Similarly, it can be moved to the right of sentence, too. The verb may also have a potential to be moved to any of the four sleuths other than its original position as in (64)a. that would make the possible number of order of elements quiet lager, and it would be futile to give examples of each of the orders that are actually “realized” I would also leave out the question of that “constraints” or “blocks” the structures that tare not generable. I would refer restrict my a attention here to examples of a few of these orders to show her reordering rules carry out discourse-structural rhetorical functions. I shall quite from the creative writings of a few reckoned authors of modern maithili for this purposes. Most of the sentences that follow have been taken from literally works of Rajkamal Chuadary, Yatri, Lalit, Vyathit, Nanipadma, Narichan Jha, and Prabhodh Narayan Sing.

7.9.1.3. I may also mention here that while looking into though ways the “normal” or “favored” word order is violated in the creative literature. I shall also take u these instances where there is not adverb, but still for bringing in certain special defects, the SU-10-DO-V order is altered in some way or the other Similarly, I shall also include intransitives and simple transitive, and would not limit the discussion to dis-transitive sentence only.

7.9.1.4. A sentence can have, as we know, one or more simple adverbs or a longer phrase , or even a whole adverbial.

Page 417: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

414

Clause. If there are a number of adverbs or adverb phrases, it is possible that one or some or all of them are moved out of their “original” position. Some sentences such as the following show that the adverb to be emphasized has been brought forward in each cases: (65) rnaffree dar iekham sankar unasthit beit” ashi. Nation-emph on when trouble present be-ptepl sun-pree-Agr tan no kie raanil ashi an no raajaa11.Then neither anyone been be-pree-Agr and nor king “when the nation faces a crisis, there is no queen or king”. (66) askaal die tak8isk sarmp. Aankhi ap prithvii-putrak aky towards put-swarup. Eye in earth-sen-poss yaneanaa rahaik: 12 desire be-pst-Agr “Swarup locked at the sky. In his eye was the desire of a sun of the earth”.

(67) baarthp-sanglitak rune nT. ‘aakaafvaath’ asn side unsie-gun from in Radio from ravisankarak sitear baaj1 rahala ashi 13 Ravishankar-poss siter play-ptepl preg aan-pree-Agr “As a background must. Ravishankar’s sister was being played in the (All India) Radio”.

(68) paarak swar Np garmil sab1 rahala chalaika 14 pare-gen voice in heat come-ptopl can-put-Agr “The anger was showing in pare’s voice”

Page 418: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

415 7.9.1.5. One could increase this list to ad Infiniti on. Notice that in (65)-(66), the order of elements has been as follows: (65’) ( ADVP – PRE(ret) –nP-v) 22, Preeerel…. (66’) ( ADVP – P-NOP. Advp ( Adj-NP-pess NP)NP Veux.…. (67’) ( ADVP – ADVP- ( –nP-pess NP) 22, Vaux…. (68’) ( ADVP – NP-Vaux.

7.9.1.6. In all these sentences there has been a proposing operation. Now, the question is: why has the adverb phrases been moved? The purposes in (65) is easier to understand, because here the author has not only brought the adverb to the sentence- initial position to make it sure that realty-a nar is emphasized, but he also put an emphatic marker –2. the novel “Prithviiption” which is the source of (66) starts with a detailed description of the sky, the dark unruly cloud, and the thirsty farming land of Bwarup, the here, Although Bwarup si the conventional here of this work, the nature here has an equally important role to play possibly move important than that of Bwarup in this chapter. This is reflected in the order f elements in (66) it may be moved here that the first chapter of this novel where (66) appears has 361 sentences distributed ever 14 pages, and Swarma occurs only in 33 places (and once in every second page) Out of 35, only 8 time does it appear sentence-initially. But it appears in the initial position as a subject NP (or within the noun phrase functioning as a subject) only thrice. Thus, it proves that the author is consciously trying either to avoid referring to Swarmap who is otherwise present throughout this chapter, or

Page 419: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

416

He is reordering the sentential elements (mostly by bringing the adverbials in the initial position) in such a way as to make the background of the novel familiar to the readers. (67), I believe , has the same motivation, it appears in the fifth section of a short story as its third sentence. (67) is actually the first full sentence in this section, the previous “sentences” being a more description of the setting of an event. The proceeding sentences (15) were:

(69) a. dinaanka bils jaanawarii unnaia saF chisasathi day 20 january 19 hundred 66 “20th day of January, 1966”.

b. samay saayankaal saat bajR time evening 7 be-pres “(It ) was seven (p.m) in the evening”.

7.9.1.7. when we lock at (69)a and b which appear before (67) in the beginning of one of the sections in this story, the placement of two adverb phrases in the sentence-initial position becomes justifiable. In case of (68) , Rearej swear nl occurs first for a different reason. Although in surface the nominalised element garmii “heat” (coming from garam “hot” ) is the subject , for all practical purposes, Bearek gvar “Pares voice” (which is the dative subject) retains the meaning properties of these subject even after a demotion. 7.9.1.8. The sentence in (68) has a parallel such as follows:

Page 420: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

417

(70)paarak swar garan cho rahala chalaika pare-pres voice het ge-ptopl preg aux-pet-Agr. “pare’s voice was being angry”.

But (70) when compared with (68) does not seem to be an expression that could have been used by the writer concerned in the context where it occurs in the novel. this sentence is an intervening narration by the another who is himself the here. It intervenes what Pearle had to say in answer to the letter’s allegation that she was hiding something serious from him when she was not supposed to hold them back (because of their intimate relation). 7.9.2.1. I have an impression that novelists and short story writers tend to violate the normal order of elements in the sentences that conclude or open a seems in most of the cases. While going through 17 short stories of Professor Narimohan Jha (1949) , one of the best story-tellers in modern Maithili, I found that out of 17 stories, 13 open with sentence of various length and complexity which have some or the other kind of adverbial phrases. And, 8 out of 13 sentences show adverbs in the sentence-initial position. Most of these 8 stories were centered around one or more events, whereas in the remaining five, the emphasis was more on painting a particular character. The adverbs moved in these 8 sentences included time, place and manner adverbs, conditionals, source and goal as well as simple NP+ Pest position constructions. The following are some of these sentences:

Page 421: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

418

(17) chi din jahinaa haath aB kakan laitn chii that day the cement hand in pen take ptopl aue-pres-Agr ki ekTan sajjan pahunsi golaaha that one-det gentleman reach-ptopl ge-pet-Agr. “the other day, the moment I took the pen (to write) a gentlemen entered. (72) aai son praayaha biis vearkh pahilak bat kahaita chii today from about twenty years agre-pers story tell ptopl sum-pres-agr :I am telling you a story of about twenty years age”. (73) chi din snadhaakal padharik ghaar par laadhetik that day evening time tank fen wharf on red dhoti pess jana ghara lanala rahay crowd gather ptopl aur pet-agr. “the other way on the gathering of red-dhotied (people) (74) ahut din par paries son col abat rahii manu days after distant-;land from come-ptpl come-aux nat “After a long time, I was coming back form a distant land”.

7.9.2.2. Although the subject NP is overtly present in (73 ) , in all other sentences an optional subject pronoun deletion rule has been applied along with the movement rules. In (71) , (72) and (74) the understood subject is ham “I” This deletion somehow makes these sentences better as the beginning of a story. A placement of subject pronoun ham would either make these sentences odd or “matter of fact” statements, e.g..,

Page 422: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

419 (71) ham chidn jahanaa hath nE kakan keita chii ki ekTaa (72) sjjan pahunei gelahaa (73) ham aai son biis varkh ahilak baat kaha 74) ham bahut din par paries fon col akaita rahii 7.9.3.1. I shall give another example to show that adverb proposing and other order-changing rules are techniques that are very commonly found in the beginning of a story. I shall quite here form another collection of short stories of the same author (Jha.1960) published long after “RangaBealeaa” (Jha, 1949) appeared. The first paragraph of the first story of this collection rums like this:

(75) pandit kaakak sanan NE ait bii e. pass kaniaan pandit uncle –pess courtyard in today. B>A passed bride abi rahul chathinaha ii. Sunitahi samnse gaan nE come-ptopl prog aub-pres it hear-ptopl emph whole village in kutalak naath cbi gela, strikanek daya nE curiselty fleed come aux veonem-pess body in ptelp-yet-agr. Gunda gudii laagae lagalaun panditanikank karej tickling sem-petpl sux-pet-agr pandit wife pres heart bhaal tik ddat jakaa kaan pae laga lanhi saikaalki plantain pes leaves like shiver peepl aun-pet-Agr bii.e pathavalii kaniaank saannE juanb kathin B>A study ing bride-pess with vie difficult ….. hunakaa kuralii par bais kaa akhbar pathbak she to chair on sit-ptopl de-ptopl newspaper readtuf

Page 423: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

420

hissaka hastauni ehithaan hamaraa sadhak sankg habit may-have-Agr this place we-obj plural pess with einnbear niipal pear legntnhi ?20 goodese, alter anoit-inf able be-fut-agr “Today a B.A. passed (-graduate) bride is going to come to Pundit-uncle’s family. Seen after hearing this, there was a fleed of curiosity in the whole village, the women, as if, felt a tickling in their body. Pandi’s wife’s heart started shivering like a plantain-leaf. Nowadays it is difficult to vie with a daughter-in-law, studying in B.A. she must be accustomed to sitting on a chair and reading a newspaper (only) . Will she be able to anoint the goddess’ alter (in the courtyard) here with us? 7.9.3.2. There are altogether 10 instances of adverbial phrases, out of which 8 underwent the adverb-proposing rule, why it has been so can be understood by making the order of elements in the discourse normal. And then by comparing the resultants passage with the one in (75): (75’)bii. E. pass konicann pai dendit kaskack amran ND aab1 rahaln ahathick”kututhalak bathi ii sunitahi sanuune gess nE abi gela and gudii strugallak deha all langae kakaa rahi butte saikkathi bii e. dathavalit kaniank saan nikk junab kathin ….? Huma kaa akhaar oalbaar hissak kurniii par baiskaa haetaki hum kaa chiThaan humara cathak sang sinbear par lagataah

Page 424: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

421

7.9.3.3. Now, at least one of these sentences is completely weird, and another (i.e., hum-a-kaa-akhbaar Bakhhback…) has a different meaning altogether. Two other sentences which have a question-mark before them are odd. In the first case, kutuhalak Daalh-I being an animate subject should not come in the “normal” sentence-initial position, and in the second case, the introduction of an unspecified agent kar-a rak-u butt or kak-a rak-u lal-a “for somebody”, which is otherwise implicit, makes the construction strange. Some of the native speakers may not agree with me in judging these sentences in the way I have done, but there cannot be any doubt that given two passages—(75) and (75’) they would all dub the letter as stylistically inferiors to the former. 7.9.4.1. The reordering rules are preferred to a maintenance of normal word order in only the beginning of a story or an episode, but I believe the reverse is true in case of ending a paragraph or sunning up a chapter or a section (especially in a long and opus novel) by locking into an academy award winner novel “Naikaa ban-I jaara-I by Manipadma (1972) which was based on a folk-epic, I found tat of 43 chapters 24 and in narrative sentences-21, and baring one in all of these one would find one or more adverbs of different types. There are about 7 short sentences 4 of which show normal word order. The rest of them which do not have much things in common, have non-human NPs as subjects, e.g.,

Page 425: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

422

(76) bahar son naasatiii anssat sisakin rahal ehala outside form veual wind whitle ptelp preg aun pet-agr ;”the vermal winds was whitling form outside. (77) taken adhraatri biita rahala chaliak then half nighjt pes ptopl prg aur pet- ? “It was past midnight them” (78) dina mab har son biilanka chit kait calal diamond necklace from lightening flash ptopl preg-pres agr ‘the diamond necklace kept on giving flash of lightening

7.9.4.2. The same thing is true for the extra-long sentences. There are four of then which have 20 to 30 words in them, and except in one, all the others show normal word order. Again, the one which does not, has a non-human subject, a cow’s bell in this case, these four sentences have the following constituent structures:

(79) a. (Adj adj2 –n ) NPSU –AdvPl-Vptelpl Advpl Vaux

CONJUNCT –Cerel-Vaux (del – NP-CONJUNVT-NP Adv –Vaun a. (N-press N>NCONJUNCT –NP-pess NP-V-Ptopl –CONJUNCT N-press N>NCONJUNCT –NP-pess NP-V-Ptopl –CONJUNCT a. (N-press N>NCONJUNCT –NP-pess NP-V-Ptopl –CONJUNCT N-press N>NCONJUNCT –NP-pess NP-V-Ptopl –CONJUNCT . (Adj adj2 –n ) NPSU –AdvPl-Vptelpl Advpl Vaux CONJUNCT –Cerel-Vaux (del – NP-CONJUNVT-NP Adv –Vaun

Page 426: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

423

7.9.4.3. In (79)b, the subject NP occurs after two participative clauses because of identical subject erasing rule, one could delete the latter identical KP (which is NP in 79b) instead, by retaining the sentence-initial subject NP-1 which would have made not much difference meaning wise. However, is (79)b there has been no adverb before the subject NP, whereas in (79)d a reason adverb did proceed the same. The subject NP here was very much like the ones in (76)-(78). IT was ; tilanggak glaak samyaa naeriik ahautaa “The bell (hanging from) Tilanga’s (-a cow) neck –(weighing) six and one-fourth seet (-a measurement)

7.9.4.4.1. With the other sentences that end different sections in this novel, especially with the sentences which are 11 to 19 words long, the situation is not much different. There are 12 sentences of such a length, out of which 5 are of ecordinate structure. Only two ecordinate sentences show adverbials in the sentence-initial position, but at least in one of the instances the normal order would have been given a different reading, and hence adverb proposing was necessary in this case. Here are those tow sentences. (80) vriddhak aathkhi sin jhar jhar neta bahee laga lanhi old man-pess eyes from incessantly tears flow- sartr-pet-Ag. An dumna tarak aankhi sehe sajal bhae selani And both young men pes eyes also tearful be-ptopl come-pet Agr. Tears startle flowing inaccessibly from the old ? Man’s eyes and even the two younngmen eyes became tearful”.

Page 427: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

424 (81) raa jaak sanket son baiskii samaaytik nakkaara kig-pess gesture by sitting end-pess trumpet baji uThala an jayajayaakarak tumualnaada blow ptopl rise-pet-agr CONJ crulation pess tumuturns nar caataavana kna kanpt karas langlaa atmespher og tremble de-ptolp start-pet-agr ‘the trumpet (declaring) and of the sitting blow up by kings gesture and the tumptous uproar of exualtion started vibrating the atmosphere.

7.9.4.5. If we make the order of elements normal, both the sentences would show a different semantic interpretation, and this will be particularly evident in case of (80’) that follows: (80’) nara vriddhak aarkhi son jhar jhara behae laalanh aa dumn aanak aankhi sahe sajal bhal selalah (82) baisakii saaptik nakaara raajaak sanket son (83) baaj uThalaaa jaya jayaakaarak tumlaanada vaathavaraN kBa kampit kaaraa langala ? To me (80’) is odd because the second part of this sentence contradicts what the first part imp lies. The first part of (80’) after the normal” order was “restored” acquired a meaning such as this” “Tears started flowing incessantly from the old man’s eyes (and from no one else’s) the second part however informs us that “even the two young men’s eyes became tearfully. (80’) can be saved only if we assume that although the eyes of two young men became tearful, not a single drop of water fell from their eyes. But this is not implied by the original sentence in (80). The sentence in (81) has a similar interpretation so.

Page 428: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

425 Far as its first part is concerned: “The trumpet (declaring) an end of the sitting blow up(just because the ) king gestionlated (may be it wouldn’t have, had it been anyone else)’ Thus is surely not implied by (81) I did not mark (81) as odd because the second part here does not contradict the first. It is thus evident that the ending sentences of the medium length which have coordinate structure show a preference for retaining the normal word order unless such a retention gives an altogether different meaning. In two other cases where the recording tool place in manipadna (1972: 74 & 12 ) what happened was that in course of joining two actions which took place one after another or simultaneously. The event that occurred first or whichever is unimportant was placed in an adverbial phrase before the matrix sentence. They are thus not cases of simple adverbs being proposed for rhetorical purposes only in one instance a reordering was done with this purpose. This sentence is as follows: (82) srashana, prem an saufhadk vaaaavan nE respect loce and good wish oss environment in naikaa aa deapadna apna dal sahit ratsen sn Naika and devapadna self-pess group with earthsenea from Vidaa leahah Leavetake-pet-Agr. In an environment of (mutual) respect, love and good wishes, Naika and Devapadna with their on tool leave of Ratnasona”.

Page 429: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

426

7.9.4.6. If one looks back, there were 44 adverbs of different types in those 24 sentences, and only 10 of them were proposed. And, we also saw that in most of these cases, the reordering was unavoidable because of semantic reasons. All this shows is that the opening passages of stripe or novels are likely to have more and more violations of normal word order for a simple pragmatic reason that the authors have to make the readers accustomed to the setting of their tale, but the ending passages would normally show the reverse trend in his respect.

7.9.5.1. I shall discuss two more instances where adverb proposing and other order-changing rules are likely to fund place for different reasons. One would be a public lecture or a public announcement (preferably oral) of some kind, I could give a number of examples of this type to prove this point, but I shall deist form doing that, and give one long quote form a one-act play-33 in Maithili. In this satire, at one point a Brahmin is found to be “announcing” his intention of selling out all that he has for one rupee. In this discourse, all the adverbials are proposed within their sentence boundary except in one case:

(83) jati le jaati taka sea katti ek takaa de, caste take caste. A rupee a seer caste. One rupee give, Ham ekhanaha apan jaati beei daita chii I right-now self-pess castesell-ptopl give-ptopl be petr agr Raakaa lala braahama son dhobii bho jaai as Money for Brahim form Vasheuman be-ptpl ge-par and

Page 430: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

427 Dhobi kBn baabhan baman dii. Taakaak lala johan Washerman obj Brahmin make-ptopl give-pr many for whatever Kahil tehne vyavasthaa do dii. Taakaak nimitt

Ask-pr correl shastrie - give-ptopl give-pr money for in junctions. JhunTha kBn saanen ki dii; Taakaak lela Brahmin son Lie obj true make-ptopl give-pr money for Brahmin form Nuslamaan, Taakaak lela hindu son krishaan Taakaak lela phums gavaahii do dii. Taakaa hotn Money for false witness give-ptopl give-pr money for Paap kB2 punya saanii aa Takek vaaste nilea kBn Sin obj virtue assume-yr and money for mean obj Apan pitamaha banaa lii. Veda, dharma, kul-marjada Self-pess grandpa make-ptopl take-pr veda, religion, line age-honour. Yasa, bahappana san take sora Fume pride all a rupee a seer “Caste; take caste; one rupee a ser; give me a rupee. (and) I shall right away sell my caste. For many, I can become a Wassermann from Brahmin, and can make a Washerman a Brahmin. Only for money any shastra

Injunctions one wants can be given (without caring for the shastra) For money, I can make a lie a truth; For money (I can make) a Brahmin a Muslim or a Brahmin a Christian for money I can give a false evidence. For money I can regard sin as virtue and for money I can make a mean person my Grandpa. Veda, religion, lineage, honor fame and pride – all a rupee a seer”.

428 7.9.5.2. Another place where reordering rules are frequently found is in the play right’s instructions (to the actor-actresses and to the director) which intersperse

Page 431: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

dialogues and events. To my mind, the reason for this is the same as in case of beginning a short story. Since the characters make their presence felt very easily (because of the lines they have to utter). And since actually the actors portraying a character or the audience watching the palace become familiar with the character, all that a play right has to do in the directorial instructions is to describe either the actions that should go with a particular lime or to talk about the stage and other arrangements that he thinks is fit for a particular sense. Thus,. To make the place, time and manner of an action prominent, the another would chose to propose the adverb phrases is most of the cases. I selected a popular adoption of Neliere in Maithili to check my assumptions with it, and found that out of about 41 directorial isntstrctions-35 , adverb proposing and other movement rules go with it were applied in approximately 26 cases,

7.9.5.3. some of the striking examples are quoted here in support of this claims. (84) behut dor dhar1 duunn yaar dil khel1 kae long time for both friends heart open-ptopl de-ptopl hansi rahal dhath1 36 langh-ptopl preg-ptopl ax-press-Agr “Both the friends length their heads off for quite some time”.

Page 432: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

429 (85) taavat du5r nepathya son karkaln awar ND paaditn by them far aside from harah voice in pandit phuesal jhask garnjab sunaln jaaiahn 37 phuchai jha-poss howling hear-pass –ge-Agr. “My then, from far aside, in a haath voice. Pandit phychi jha (could) be heard howling”. (86) naayak isaaran sor rames caunkii tor nuka rahita maya posse gesture by ramesh eot below hide ptopl stazy-ptopl achi caunkii pa raakhala amda kEn dekh naayak aux0-pess oct on keep ptoplegg obj see-ptopl naya-pess vikal ta Helpessness. “By (lecking at) naya’s signal ramesh went to hide below the oct. nyas helplessness (is evident when she) sees an egg kept on the oct”

(87) darabejjan par ek caunkii tiin kurssii oact ektaa (88) deer jon one oct, thre chairs and one-class Turla raakhala achi caunki par naila san sata ranjii Steel kep pass-be-press cot on dirty like carpet. Achaaela chaika Spread pass be-press. “There is a dirty bed-spread lying on the cot” (89) …taah nE kaagaja ne lapatala ek aThannii chanhi (To him ) that in paper in wrap pass-oen 50p coin have-pres-agr-(SU) “In that (he) has a 50 paise coin wrapped in a paper”.

Page 433: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

430 (86) takhanahi paachaa so rameg cupacaap praves karait then emph back from tramesh silently enter-do-ptopl chathi aa…. Sun per agr and… (90) ….puram jaamanaak kapara pahiraa mail kiir kit old age-pess clothes wear ptopl dirty gleassy ekTa caedari garden par laperaE apanaa basil oen-class sear neafenck on cover ptopl self-obj cery nahatavapuran bunjahia gunahit gata son nahu nahua important think-ptopl proud speed in slowly slowly calaita kavitajii aabi rahala chathi going ayurvedha practitioner come ptopl pref amn press ag “wearing clothes of by-gene era, a dirty glossy scarf hanging form his neck, thinking too much of himself, slowly walking down with proud steps comes the Ayurvedic practitioner. 7.9.5.4. The sentence in (84) through (90) show how in directorial instructions of a play, adverbs are proposed, and normal order of elements disturbed. It is a time adverb (bahut der dhar-I) in (84), whereas in (85), time place and manner adverb phrases in that order (i. Teevat: ii.duur nenathva SO-n: iii. Karkas-a evar nR) are moved. (87) could be typically found wherever a play right describes a stage setting. (89) introduces an element of drama, whereas in (86) it is clear that had it not been for Maya’s gesture, Ramesh would not have gone below the act.

Page 434: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

431 To hide. It would be a matter-of-fact sentence if in (86) rames is brought back to its “normal”, i.e., sentence-initial position. (88), by its order of elements, almost describes the way a character would gradually unfold a mystery to everybody’s astonishment. finally in (90), by a placement of five adverbial constructions before the character of Kaviraai-a iii , the play right tells us more about this particular character than he could otherwise by a more description joy this ways. All these discussions point pit as top how reordering rules perform various functions which need a pragmatic consideration, and how creative writers make use of these different functions of movement rules successfully in their writing. 7.10.1.1. At the end of this chapter, I would discuss one more syntactic phonon omen that is not so well understood. Many of these rules have to be carefully studies at duns out and establish their exact relationship with presupposition and other pragmatic notions. The idea is to fund out as to what is it that forces or motivates a native speaker of Maithili to select (91) a. an optional rule as against non-selection of such a rule, and

one optional rule (or a set of such rules) as against another? 7.10.1..2. let us take the case of Equi as an example of (91)b. consider the following sentences which are options of one another:

(92)a. han caehaita chii (je han jiitii)

I want-ptopl aur-pres-cpl I win optative Agr “I want that I win”.

Page 435: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

432 b. (jiito) caahaita chii I win-Inf want-ptopl aur-pres-Agr. “I want to win” c. ham cashita chii (je) jiitii) d. I want –ptopl aur pres-CPL win-optative Agr. “I want (that) win”. e. ham (jiitii) so cashaita chii f. I win-optative PRO want-ptopl sun-pres-Agr. g. “I win it want”. e. ham eachira chi (jiitii) I wnt-ptopl aur-prs-win-Inf “I want to win”. f. *ham (ham jiitii) so eachaita chii I win-optative PRO want-ptopl aur-press-Agr. “I win it want’ (93) a. han eachaita chii (je raan codiar babun juutath) I want-ptopl aur-pres-OPL Ramachandra Mr.win-opt-Agr “I want that Mr.Ramachandra wins. b. ham (ramachandra naahum jiitathi) se eachiara chii I RamachandraMr. Win-opt-agr PRO want-ptopl Aur-pres-Agr “I Mr. Ramachandra wins it want”. 7.10.1.3. These sentences pose a very interesting problem. Let us assume for a moment that the “favored” word order for complement sentences is as in (94) a rather than as in (94)b :

Page 436: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

433

(94) a. 2n(x-(s)-n-y), where s2 is matrix and sc embedded sentence modes. b. Sn(X-N-(Sc)-Y) Sn

Now, having taken (94) a as the favored clause structure, if one looks at the sentences in (92) one has to start with an assumption that (92)f is the nearest to such a structure. But this structure as it stands is not an acceptable sentence in Maithili, unless one applies either an Equi true to delete the second occurrence of identical pre-form ham and thereby generating something like in (92)d, or a rule of Extra position which would move the embedded E-node to the eastern end of the matrix sentence as it happened. In (92)a. if one selects the first alternative, one has to delete the complementation from the structure. Similarly, in the second instance, although ie may be retained, the head pre-form se must be deleted. Thus, (92)d’ is very odd, whereas (92)a’ is not”

(92)d’ ham je jiitii se eachaita chii (92)a’ ham eachaita chii se eachaita chii But (92)a’ is unacceptable if the PRO is so: (92)a” ham se eachiata chii je ham jiitii Notice that if oen retains je as in (92)f, the result is worse; (92)f’ *ham je ham jiitii se eachiata chii (92)f” *ham ham je jiitii se eachita chii

7.10.1.4. It therefore seems that there are two possibilities before the native speaker to generate grammatical sentence from

Page 437: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

434

a sentence-structure as in (92)f, and they are as follows: (95 )a.i. Je-complementiner deletion, and ii. Equi-NP deletion. b. i. Extra position, and ii.Head-preform deletion.

But if we turn to other possibilities such as (92)b,c and e, we find out that the native speaker has other alternatives as well. Thus, apart from (95) a and b, he can also select one of the following derivations:

(95) c. i.Extra position, ii. Je-complementiser deletion, and

iii Equi. d. i.Extra position, and ii.Equi e. i.Extra position, ii.je-complementiner deletion.

iii. Equi, and iv. Infinitivisation. f. i. Je-complementier deletion, ii.Equi, iii. Infinitivisation, and iv. Head-pro-form deletion. 7.10.1.5. (92) a provides for examples of (95) c and d. in the same way, (92)e and (92)b are examples of (95)c and f, respectively. It would be interesting if one could find out as to why or when it is that the native speaker of a language selects one of the derivations in (95) rather than any other to make the “basic” sentence-structure acceptable (recall 91a). it would be possibly be worth mentioning at this point that there is nothing wrong with (92)f other than the fact that here the identical subjects of the two sentences are too close to each other. The proof of this (93)b which structurally identical to (92)f, and is grammatical. It is not clear at this stage as to why one must select one of the sets of rule as (95) for (92)f, whereas one need not use anything more than a je-

Page 438: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

435 complementiner deletion rule for (93)b (recall the problem of choice in 91b).

7.10.2. In conclusion, we have seen that Kata and Postal’s (1964) good old hypothesis that transformations do not change meaning is no more tenable. Transformations, especially once that reorder or line arise grammatical elements do chance so mantic and pragmatic interpretations of sentences. This was made clear by examples of a number of movement rules and rules of other types. Quotes from the novels, short stories and plays written by eminent literature of modern Maithili show that creative writers are very much aware of this meaning changing equality of syntactic rules, and they exploit this property of transformation wherever they fool it necessary. I have also tried to find out as to which are the typical or marked places in such creative writings were reordering of “favored” order of elements is done. I have shown that such reordering in in Maithili have three-fold functions mainly(i) emphatic functions (ii) discourse structural rhetorical functions, and (iii) presupposition motivation. Finally, I have briefly touched upon a rather unresolved problem of the selection of n optional rule as against its non-selection as well the problem of selecting one of the many possible derivations or liberalizations.

Page 439: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

436

NOTES TO CHAPTER VII

1. I shall not go into the origin and development of what is new known as “Pragnaties”. And its relationship with semantics. See Green and Morgan (1977) for a brief history of this field. The pioneering works in this field have been done by Austin (1962), Bearle (1969, 1975, 1976) Lakakoff 1962)Ross (1970b)stalnaker (1972), Mergan (1969ab), 1973, 1977, 1978), Robin Lakoff (1973), Badock (1969, 1972, 1975, 1978), Gordon and Lakoff (9171), Cries (1973, 1978), and Green (1978ab, 1973 , 1978).

2. I shall be using these names of transformation with which transformational grammarian in general are familiar this has been done simple to avoid confusion that may otherwise arise.

3. One can also say that the element in (7) e-e is a partials, whereas is in (7) a-b is a pre-form which is somehow related to “if’ in the some way as je ….. se “who …he “ (relative-correlative ) are related.

4. Heek (1974) and Aurora (1978) discuss the meaning changes brought by reversed compound verb sequences in Hindi.

5. heek (1978b) thinks the languages that show a clear OV/VO divide in a (syntactic ) dialectological map of Indo-Aryan, tend to have more of these inversed compound verbs. This he finds to be un confirmation with Greenberg’s (1963):85) universal 16 which says : “In languages with dominant order V90, am inflected auxiliary always follows the main verb”. If we read VO for VBO , and equity auxiliaries with aspectual or vector verbs ,then what Greenberg says would describe the Indo-Aryan situation perfectly well, Neek (1978b) thinks that Kashmiri is very near to such an OV/VO divide..

6. Heek (9175:4) has examples where he tries to show that some variety of Bengali have reversed compound verbs. While we do not deny that it could be a possibility, it is certain that the examples he gives are not very convincing. They look like what we have called Verbal Compounds in Chapter VI .,e., they are probably instance so from main verbs occurring in a sequence. His examples include the following.

PLACE NAMES DIALNOTS EXAMPLES GLOBS

Manipur Bishnupuriyas diyaa pithailee “sent”

Gylhet Bishnupurihan diyaa pithaile “sent’

Ceech behar Ranjbansii diyaa pethaail “sent”

Page 440: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

437

I think that these are either instances of two main verb sequences or are in stances of the verb “to send” in part passive where Udiyaa saves the purpose of an instrument marker

7. Note that I have eliminated certain details while writing these structures. For example, in case of (21)a, the well-formalness conditions of this language would demand that the complement marker je is shifted t the right of the sentence-initial constituent. Similar, I did not show that in (22) a je comlementiser could optionally be deleted as in (910 a.

8. For a somewhat detailed discussion n Raising rule, see chapter III of the this dessertation.

9. In some forms of speech , jamas cold be allowed t stay in such structures. This would give sentence such as the following instead of that in (33):

(33’) tarini baabul bangh jamaa garaj-I rahal-a ehath-I

meaning wise, it would be no different from (33)

10. A complete list of the works in Maithili referred to could be found in the ‘references’ separately. Wherever a sentence or a discourse has been taken from such a work as an example in his chapter. The details of the reference been taken from informants.

11. Cited from Vyathita’s (1968:63) play ‘Tripathgan’.

12. Cited from Lalit’s (1963:1) novel ‘Prithaciputra’.

13. Cited from Rajkamal Chaudary’s (1969:13) collection of short stories “Nirchii Baalam Kamar’

14. Cited from Yaatri’s (1963:48) novel ‘paare’

15. (69)ab are cited from Rajkamal chaudary (1969”13)

16. Cited from Mari Mohan Jha’s (1949”1) short story ‘Rang-a saalan’.

17. from Mari Mohan Jha’s (1949”1) short story ‘Kaalibaariik eer’

18. from Mari Mohan Jha’s (1949”1) short story ‘Saasurak eikna’

19. form Mari Mohan Jha’s (1949”1) short story ‘asbhut TeT-amaa

20. from Mari Mohan Jha’s (1949”1) short story ‘GrBjweT-a Patech-u

21. I have excluded these ‘closing ‘sentences that and in dialogues.

Page 441: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

438

22. From Nanipadna’s (1972:49) noval ‘maikaa ban-I jaraa’

23. From Nanipadna’s (1972:102).

24. From Nanipadna’s (1972:67)

25. See Naniapna (1972:3)

26. From Nanipadna’s (1972:12)

27. From Nanipadna’s (1972:91)

28. From Nanipadna’s (1972:1560. In (79) , pl indicates more than one occurrence of a particular element.

29. From Nanipadna’s (1972;9)

30. From Nanipadna’s (1972:65)

31. Note that in writing it would be spelled with a retro flexed rather than with the dental as it has been shown in the transcription.

32. Cited from Nanipadna (1972:138)

33. it is a Maithili adoption of Bharatendu Narish chandra’s Hindi play.

34. Cited form Prabodh Naryan Singh’s (1965”10) one-set paly ‘Amber” nagrii’. Here too, in writing one finds in words like brachnoU and ?

35. I have excluded from the total eleven occurrences of one-te-two-word instructions almost all of which have only adverse.

36. cited from P.N. Singh’s (1968:5) one-set play ‘premak reg’

37. from P.N. Singh (1968-180.

38. from P.N. Singh (1968:10

39. from P.N. Singh (1968:14-5)

40. from P.N. Singh (1968:19)

41. from P.N. Singh (1968;21)

42. from P.N. Singh (1968:29)

Page 442: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

439

CHAPTER VIII

SUMMARY AND CONCLUSIONS.

8.0. This chapter presents a summary of the topics discussed and the grammatical facts explained in the preview chapters it also brings out the conclusions reached in course of discoursing various aspects of Maithili syntax. In what follows each chapter is taken up separately to give a broad outline of the problem discussed the questions raised and the federalization proposed in the body of the work.

8.1. In this firs chapter the aims and perspective of the present study where discussed in brief . this discussions was followed by an account of organization of the present work. And on elaboration of the topics that where to be covered in different chapters of the dissertation.

8.2. Chapter II included a brief description of sound system transcription pattern) and morphological structure of the Maithili language. In this chapter relevant background information regarding Maithili “ (or Videha) present linguistic boundaries, dialect divisions of Maithili, historical stages of Maithili, Maithili and its genealogical relationship with other Indo-Aryan languages , Maithili and its three scripts (Tiruutan, Kaithii and Devanagari) were given in the beginning. It was pointed out that Maithili has 13 vowels and 31 consonants. The vowels included I,ii.i,u,un,u,o, and the consonants.

Page 443: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

440

Were; p, ph, b, bh, j, th, d, sh, T, Th , d dh k, kh g, gh, r u n c, ch, j, s r, R, rh, l, y and V.and Y.The peculiar development of the three vowels –lengths long short and very short was stressed here. The use of pluralizes sab, lakeni, and bahut. was explained and alternative ways of plural marking were also mentioned. The agreement in verbs and adjectives (the latter is new declining) the case-marking in modern Maithili and the presentable parading (including personal, relative, tense, mode. Aspect, personal terminations and order of elements within the verbal complex were briefly dealt. Towards the end chapter persential adjectives adverbs, and the question of “favored’ word order were also briefly discussed.

8.3.1. chapter III was divided into eleven sections all of which discussed the motions of grammatical relations and the role they played in theory of grammar in general., and in Maithili in particular. In the beginning the position of the classical transformational –generative grammarians in respect of destinies of usability f grammatical relations such as “subject of “ (SU), “Direct object of “ (DO), “Indirect Object of (10), and “Oblique object of “ (00) was discussed. It was pointed out that for the 24theorticians such as chenaky (1963) and Kats and Postal (1964) grammatical constituents such as NP, VP, PP, etc., were theoretical primitives for a synthetic analysis, and that it was widely believed at that point that the relational notions SU, DO, and IO could be defined in terms of the structural configuration of various grammatical constituents. This position w criticize in a long.

Page 444: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

441

section under four different heads; (a) Arguments to against the deep word order, (b) Arguments against the TE definition of surface relations. (c) Arguments based on language universals, and (d) Arguments based on characterization of rules. It was argued under (a) that Chenky’s (1965) definition of grammatical relations did not apply for the VBO and OBV languages. It was necessary to replace “Domineers” (his sole definitional criterion) by a more adequate notion to be able to characterize relational terms of object-initial languages such as Ayurina, Nadeb, Levant (all OBV) in the light of the room finding’s (of Deruyashire and Fullum, 1978) it was also shown that TE definition of BO and IO depended heavily on the assaying that the categories of PP and up were universal which they were not ( of. Johnson, 1974a). An attempt to save this classical position by using the notion of ‘Precedence’ to define grammatical are relations (which would be a follow-up of the proposals of No Cawley, 1970a: Fill mere, 1968 and Takeoff, 1970a) was also shown to be a miserable failure because of its arbitrary generalization concerning the “underlying word order” of all human languages, ever burdening grammars of many languages with many of here yet compulsory reordering rules, and because of its circularity of definition. Under (b) it was pointed out that TE grammarians never took surface grammatical relations seriously. The sub-section (e) showed that contrary to the belief of the classical transformation lists, a number of language universal based on both deep and surface grammatical relations have been proposed ever the years. This included the Accessibility Newsworthy (AK) of Keenan and Gowrie (1972, 1977),Parading case of stepwise counsel donation proposed by Carrie (1974), Advancement to subject.

Page 445: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

442

Chaining constraint (Johnson, 1974b), and a number of universals proposed by Perlmutter and postal (1974, 1978). Lastly, the sub-section (d) discussed traditional attempts to write the passive rule in terms of word order changes, case marking, and verb marking, and how all such attempts were specific to particular languages, and hence failed to give a universal characterization of passivisation (of. Perlmutter and Postal, 1977), whereas the same rule written in terms of grammatical relations could achieve this purpose.

8.3.2. In the next section, an outline of DG was presented where DG terms were introduced, organization of levels of grammar within the framework of DG was put forward, and the universal principles or laws such as cyclically law, relational annihilation law, verb agreement law, Reflexivisation law, cereferential Deletion law, breaking law, and advance laginess law were discussed briefly. In the section that followed, a few relation-changing rules fo Maithili were dealt with in detail which included raising X (sub-toSU), raising II (su-to-obj), and raising III (obj-to-SU), Passivisation, Dative Subject Formation, Conjunction Education (or 077), and The claims regarding “non-discreteness” of the notion “subject of “ (Koman, 1976:kashru of al.1976) were tested here by pleasing different kinds of subjects (such as SU of intransitive, transitive and di-transitive, Dative Subjects, Denoted oblique subjects, and presented or derived subjects) under different syntactic environments to find out whether they participated in raising and reflexivisation operations, and whether they both controlled and were accessible to equip. and OFF rules. It was shown that different

Page 446: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

443

Kinds of subjects in Maithili followed a subject heed Hierarchy which was different from that followed by Hindi-Urdu, Punjabi and Kashmiri (of. Kleeman, 1979 who claims that Hindi-Urdu does not follow any hierarchy). The subject heed hierarchy in Maithili was proposed as follows:.

(1) SU, Underlying <: S DER < S DAT < S OBI

8.3.3. Having pointed out that DS as proposed by permitted and postal (1974), 1978), and Johnson (1974) did not incorporate non-discreteness or fussiness to characterize relational nations, in the next section some other defects of DS were pointed out. Thereafter, an attempt within the framework of DS to “define” the notion of “subject of “ (Keeman, 1976) in terms of about thirty-odd properties was delineated along with its criticism by formalists (Johnson, 1977) who held that relational notions were “indefinable” primitives of syntactic theory. The following section pertrayed the belief of the generative sent cists that “….subject of and object of are not directly relevant for semantic interpretation” (Lakeoff and Ross, 1975:2: originally written in early 1967). It was argued that their failure to realize the importance of relational notions steamed from their seal for “reductivism”. It was demonstrated that even in some other works of generative scientists (waterbuck, 1976) this position was reflected. It was only latter (Lakeoff, 1975b) that they realized that instead of taking the notions such as “precedes”. “Dominates” and “Commands” as given primitives. One should take SU, DO, IO, etc as basic relations. This discussion was succeeded by a detailed consideration of the Lexicalists; treatment of traumatically relations (of. chemky, 1970, jachanderff, 1974).

Page 447: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

444

It was argued that Chemky’s (1970) account of differences between derived and gerundive nominal that led to the convention of I was at best language-specific because Maithili data went against many of his arguments. In this context, attempt of Dasgupta (1978) to show that gerundives were nominal category in Bengali was further celebrated and it was demonstrated that many points of similarity between Bengali and Maithili could be found in this respect. Jackendeff’s (1974) definition of “Subject” and “Object” was shown to be false on many counts, although it was admitted that I-Convention could “identify” these notions correctly even in VRO and object-initial languages.

8.3.4. The failures of classical T6 generative semanticists, and lexical sensitizes, and the shortcomings of the relational grammarians led to the following conclusions towards the end of this chapter” (2) a. While SU,DO,IO, and OO should be taken as primitives in a syntactic theory to capture the significant generalizations that could be stated in terms of them, it must be realized that a large number of syntactic rules and processes do not change or refer to these notions. For these reasons, a syntactic theory has to take both grammatical categories and grammatical relations as basic tools of language descriptions.

b. Although it is possible to characterize the notions of SU in terms of cross-linguistic prepositions, it is not definable universally. Therefore, the only alternative left would be to take relational notions as ‘undefined” primitives.

c. Lexical rules cannot take even all that transformations can do, but the former may be allowed to interposed the latter rules.

d. Relation-changing, Non-relation-changing as well as Linearisation rules all give rise to various semen tie/pragmatic interpretation in discourse and otherwise.

Page 448: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

445

e. “Basic word order” is more a matter of degree than an ideal level of grammatical description. Therefore, PS Rules could be done away with leaving the task of predicting surface word order to Linearisation and late adjustment rules.

f. The notion of “subject of “ is non-discrete, and different languages follow different hierarchy of subject heed statue. It is, therefore, advisable to talk about” subjective ness” of a nominal element.

g. In the light of the realizations such as in (2)a through f, “Transformational-generative” approach seems to be more like a research strategy than using a viable, coherent and feel proof model for grammatical description.

h. At this stage of linguistic research, the necessity of informative and extensive description of languages is more important than wild and fanciful theory building.

8.4. In the fourth chapter, three related and basic syntactic processes in Maithili were discussed. These were: Relative clause formation (RCP). Sentential complementation and Interrogation (or question formation). First, a detailed description of these processes were presented. An attempt was made to find out the nature of relative clause marking in different stages of Maithili language This was followed by a historical account of studies on Sentential complementation was outlined with reference to the notion of “activity” and how it worked in Maithili. In this respect, a comparison of Maithili was made with Bengal, especially with regard to the nature of complement construction in the latter. A classification of open or question types in Maithili succeeded this. This classification was shown to be applicable also to Assamese, Oriya and Bengali (of. Singh and Bandyopadhyay, 1978). The classification was supplemented by a brief discussion of open

Page 449: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

446 Question marking in Maithili, Assamese, Oriya, and Bengali, Then, the exact nature of K-Question rule was discussed with examples from Maithili and other Easter NIA Languages. The verb Interrogation Process was also discussed here. Having locked into complementation and interrogation, the attention was shifted to ROP. Various strategies of ROV (viz., particularization nominalization, equip, anaphora, verb heeding and gap) were closely examined to find out as to how many of them and which ones were found in Maithili.

8.5.1. Chapter V examined some generalization concerning Verb Agreement Rule and dealt with the nature of agreement rule in Maithili at different periods of this language. In the first part, different types of agreement patterns were looked into which showed the following things: (a) Verb Agreement was not a universal phenomenon because languages such as Igbe (or Ibe): Tankhul Naga and Thai did not show any sign of agreement. (b) the triggering element could optionally be deleted as in some Austria tongues. (c) a languages night show different patterns of agreement in different areas of grammar. Or could only show partial agreement patterns: Sanskrit, Kannada and Maithili were examples, (d) the some marker could be used to donate more than one category: for example, in Sanskrit and Chichewa the some suffices marked both plurality and politeness, (e) at times makers showing different categories took on inseparable or fused shape, as in Maithili, and (f) finally, there were languages where both derived and underlying terms cold control (as in Achness; of. Lawler, 1977)

Page 450: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

447

Agreement, and where non-subjects could also central it, as in Greenlandie Bakins, Peseman and Kala Lagan Langue (of . Weedbury, 1973: Buith-stark, 1976; and Bani and Klekeid, Uapub). In the light of the existence of was many varying patterns of Verb Agreement, it would seem that languages showed no commences so far as this rule was concerned. But a survey of about 94 languages of the world led one to purpose a tendency universal in the following terms. (of Singh, 1978b for further details):

(3) The Verb Agreement Kierarchv (VAK):

If a language had verb agreement, it would ideally show, at least, a Fereen-Agreement Pattern, and it might, in addition, show any, some, or all of the other agreement patterns that involved the categories of number, gender, and honorific.

8.5.2. In the latter half of this chapter, agreement rule in Maithili was discussed in detail to test the claim of Permute and Postal (1974) that only terms could trigger verbal agreement. It was argued that DE must be satisfied with a relativity weaker claim, namely, ‘the NPs which central verb agreement , if any, include b-subjects’ (Keeman, 19760, because it was demonstrated here (and also Singh, 1978b) that I Maithili even the ‘non-terms” or oblique objects (00s) of different types including genitive, goal, ablative, locative, etc central verb agreement marking in the same way as the subjects or other terms did, but that it could never be the case that in the selection of agreement suffixes, the (b) subjects did not have a role.

8.6.1. Chapter VI discussed with special reference to Maithili the problem of definition and identification of Compound Verbs.

Page 451: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

448

various tests for establishing a verb as a “vector’ , and classification of “polar” or main verbs on the basis of occurrence restrictions. Instead of advancing a feel-proof definition of what a compound verb was, the emphasis here was more on wht a compound verb did.

8.6.2. It was pointed out I the beginning that Maithili had a “polar vector” order normally, although the reversal was allowed in certain cases. It was noted that the first or polar member received an absolute suffix-I which underwent changes of different types especially when the verbs ended in vowels. Verbal Conjuncts (Variously known as main Verb sequences or Bi-polar Compounds) were also discussed to point out the differences between these and compound. verbs.

8.6.3. This part was followed by a critical evaluating of different tests for compound verbs in Maithili which had been suggested for other languages. These tests included Kaa-Tnsertion test (Heek, 1974), verb Interrogation test (Singh and Bandhophudhyay, , 1978), ,Negotion test (Geeffke, 1967: Necker, 1978: Neck,1974). And By the time test (Hook, 1974). This was supplemented with a discussion of Narbir Arora’s (1978) efforts to distinguish between compound and conjunct verbs. The types of combinations that compound verbs themselves could enter into and the possible structural arrangements within compound verbs were elaborated with Maithili examples. Here, arguments were given for not considering rah- and medals such sak – baa—daar ‘eam” and enk-“constellated main verbs” (or “Unit Ideations’) were also given After this, the nature of compound verbs in the Garyaa (9th e.-11th o A.D : the oldest text-in Boston NIA ) was discussed (of Singh, 1979b).

Page 452: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

449

8.6.4. This was followed by a rather long discussion of 19 “probable” vector verbs in modern Maithili that included jam “go’ as “come” de “give” , le “take”, uth “rise”, doP “fall”, rakh “keep”, hais “sit”, cal “move”, dhar “place”, Nathan ‘send”, nikal “exis”, nikaal bring out”, sam “bring”, nar “die”, baar ‘kill”, khan “fall” dekh “see’ , and phe-“threw”,. It was first shown that a few of these (jas, dder laR and UnarU) could be used as

Used as both polar and vector (e.g.de,-de “give away’) at the same times. One of the functions of these verbs was identified that the tests normally applied for judging perceptively could also be used for vector verb identification. With this aim, perceptively tests such as Phrasal Verb test (Kepeny, 1962; Periaka, 1967: Hock 1978b) Apprehension test (ferayth, 1970: verma, 1975), Vasudeva / contradiction test (Hock, 1974, 1978b), compound Verb Reduction test and completion test were applied on probales vector verbs in Maithili and the weakness of these tests were also pointed out.

8.6.5. Considering the fact that all the tests discussed earlier failed in identifying compound verbs in some cases, it was suggested that these tests could all be taken up together and applied on all probable vectors in a language and a “multiplex score sheet” could be prepared. If some of them posed in all or most of these tests while others failed or scored relatively low, the differences could be due to a slow rate of progression from verbal conjunct to compound verbs (or to the slow rate of lexical emptying) in the latter case. The results of this multiplex test could than be matched with the frequency of occurrences of these probable vectors. This was what was

Page 453: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

450

done in the later half of this chapter for Maithili. It was realized at this point that one would have to answer the following questions for successful evaluation of test-scores: (i) which probable vectors took how many verbs as polars? (ii) whether there were verbs which could not occur so polars of any of these vectors? (iii) whether polars showed some kind of select ional restrictions? (iv) what percentage of verbs could occur as polars with a particular probable vector? (v) whether one could classify main verbs of Maithili into different classes depending on which vectors they took, and on how many they could occur with?

8.6.6. To find out the answers to these questions, 219 main verbs of Maithili were taken up and various combinations and fermentations were tested. In a frequency

count, it was found out that of the 19 “probable” vectors, le, de, jea, nar, uTh, bais aa and paR occurred with 58%.99% to 10.60% of these main verbs, and that khas and

yhank occurred with none of them, and that therefore one could exclude these two from the list of “probable” vectors of 19 by reducing the number to 17 only. It also

emerged from these that there verbs 38 main verbs which could take only one vector each. These main verbs were termed ‘none-vectors verbs”. The name-vectors verbs,

interestingly, could occur with only frequent vectors noted above. Two main verbs were, however, found to be occurring with none of these vectors, and hence they were excluded from the total of 219 main verbs. Of the rest of the 217 works, 27.19% could

occur with any two vectors, 29.03% with three, 12.44% with four, 6.91% with five, 5.69% with six, 1.3% with seven, and 1.84% with eight of the nineteen vectors. The

17 vectors occurred about 321 times in

Page 454: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

451

None-vector to eight-vector combinations, and a frequency count that established as to whish vector took the maximum number of main verbs as polars matched exactly with the earlier frequency counts of these vectors. It was argued here that if the total number of probable vectors was 17, and if the vector-combinations could range from none- to seventeen-vector possibilities (i.e., if there were main verbs which could combine with all the vectors). If there were main verbs which could combine with all the vectors). Than the logically possible answer of combinations would be 131071 ( by addition: and by multiplication, the talk would be a 52-digit possibility). Against this, the “Linguistically possible combinations” were found out to be only 54. Thus, after showing the lacuna in the only serious attempt to classify main or polar verbs on the basic of the vectors they could occur with (namely, that of Hook, 1976a), it was claimed that in Maithili the eight-way (none-vector to eight-vector combinations0 could be further divided into these 84 sub-types. Finally, a multiplex score-sheet is prepared for the relatively less frequent vector verbs such as dekh, near, nika, nikaal, dhar and san And justification for accepting these in the final list of vectors is given.

8.7.1. In chapter VIX, idealization rules on well as some relation-changing (and other) rules were discussed in detail be demonstrate how pragmatic facts influenced sentinel interpretation. The liberalization rules discussed here included negative-Adverb proposing. Verb Proposing, Compound verb bordering, extra position-to-the-right, extra position-to-the-left, Heavy-NP-Shift, Sentence shift, slitting, Adverb-movement, Right dislocation, Topicalisation and others. Among the relation changing and other rules were raising Equip. Adverb-creating rule, unspecified Agent Deletion,

Page 455: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

452

Definitivisation, io-complemention deletion, and optimal subject presence deletion.

8.7.2 In the beginning, the markedly “emphatic” rules such as negative-Adverb proposing and Verb proposing were taken up to show how for emphasizing various elements in a sentence, or for strengthening the import of negation or on action, these rules could be applied. Compound verb reordering was pointed out to be a rule which served a similar purpose. The reversal of a “favored” “polar + Vector” order made a compound verb constructions somewhat “ relational” and “:mutative”, and an element of “guide ness” also came in. After leaking into different kinds of extra position and sentence-movement rules, it was established that these movements infused the festive interpretations in such a way that is Maithili (if not in other Indo-Aryan Languages), the following generalization held good:

(4) a. For a rule that moved a sentence, the “eastern site” (to the right of an S) of an embedding or matrix sentence was more prone to defactivisation than any other part, and

b. The “western site” (to the left of an S) of an embedding or matrix sentence increased the degree of press up position attacked with the content of the embedded S-mode.

8.7.3 Application of Raising, sentence-shift and Comparative Reduction were shown to allow intentional exaggeration. Sifting and unspecified agent deletion is Maithili also served the purpose of emphasizing but while the former brought the whole sentence in light, the latter de-emphasized the subject/Agent by deleting it, and thus brought the predicate in forms. The selection of unspecified agent deletion rule showed various intentions of the participants in a

Page 456: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

453

Speech act, and these were elaborated under the following heads” (a) Agent Incognito, (b) Unimportant Agent, (c) Speaker’s impartiality, (d) speaker/hearer’s Interdiction or Tabse, (e) Interprets customer. And (f) Non-Impretation of Responsibility.

8.7.4. Under the “discourse-structural rhetorical function”, rules such a Topicalisation, Right Dislocation and Adverb-Movement were discussed with a large number of examples from creative writing in Modern Maithili. In each case, the movement of adverb and the reordering the “favoured” word order of sentence, the resume behind such movements and the special effects created by these were explained. It was shown with statistical events that concrete examples that the novelists and short story writers tended to violate t he “favoured” order of elements in these sentences that spend a some. But it was also pointed out by examining one novel in detail that the “favoured” word order one more-or-less retained when the sentence concerned ended a chapter or summed up on event. Basing date on a famous play, it was demonstrated that in public lecture or announcement and into the directorial instructions of a play, adverbs were more often moved that not.

An unreserved problem was touched upon towards the end of this chapter. Efforts were made to find out as to what forced a native speaker of Maithili to select open optional rule as to against another, or to select an optional rule as against non-election of such a rule in this context equip. Extra position initiation complementer selection and optional subject pronoun deletion rules were discussed with special reference to Maithili,.

Page 457: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

454

REFFERENCES (A) In English

Adone, Dr. K.A Gamphell, V Gohan, J Levins, Mawell, G Eygrea J. deighard,

eds., 1971. Papers from the seventh regional meeting of the Chicago linguistic society. Chicago: University of Chicago.

Agineky, B.c 1955. A grammar of the Hindi language. Language dissertation ne. 20: baltimere: Linguistic Society of America. Anderson, S. 1977. “Comments on the paper by words” In P.W Gulicover et al, eds., 1977: 361-77. Anmanalai. B.1977. “Aspect of aspects in Tamil”. Paper presented at the International School of Dravidian Linguistic Seminar on Aspectual system, Bangalore. Aaki, Karea 1970. A grammar of Discusses. Berkeley: University of California Press. Aquilina, J 1959. The Structure of Haltine: A study in the Urdu grammar and

vocabulary. Fierina: The royal university of Holte. Aquique, Md 1974, ? History of Mithile. New Delhi: Abhinav publications. Areneea, N.L 1972. “Some notes on relative classes in co origin”. In F Peranteon et al, Eds., 1972a: 136-43. Arera, Harbir. 1978. Compound Verbs in Hindi . N. Litt. Dissertation, University Of Delhi, Delhi. Austin, J.E 1982. How to do things with words. New York : Oxford University Press. Bech, E 1968, Nouns and Noun Phrases:. In E. Each & R.T Karns, eds, 196s: 91-1 122. 1970. “Le Anharie am SOV language?” Journal of pthienian studies, 8: 3- 18. & R.T Karns, eds. 196s. Universals in Linguistic Theory. New York:

Melt, Rinchart & Winston.

Page 458: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

455 Bagahi, P.C 1938, “Materials for a critical edition of the old Bengali Garyapadas”. Journal of the Department of letters university of Calcutta. Calcutta: 30: 1-156. & Shanti Bhikshu Shastri, 1956. Caryagiiikoss. Shantiniketan: Vishuvahbarati. Bhani. B & T J Klokoid, n.d “Ergative switching in Kala lagan Langgon”. Me. Barannikev, A.P & P.A. Barannikev. 1956. Hindustani: Hindi: Urdu: Meskva: Indatel’stve literature Inestrannyz Jaaykar. Bahl, K.G 1964. A study in the transformational analysis of the Hindi Verb. Chicago; South Asia language and Area center, University of Chicago. 1967. A reference Grammar of Hindi. Chicago: South Asia Language and Area Center, University of Chicago Bell, S 1974. “Some notes on Cabuane and Relational Grammar”. No Berlin, B & Paul Kay. 1969. Basic color terms. The Hague: Mouton. Berman, He ward. 1972. “Subordinate clauses in York: A preliminary report”. In P. Prenatal at al, eds., 1972a: 256-61. Dhat, D.N.S. 1967. Descriptive Analysis of Tutu. Pune : Deccan College. 1977. “Vectors in Kannada”. Paper presented at the international school of Dravidian Linguistics seminar on aspectual system, Bangalore. Bhatia, K.G 1963. A Gypay Grammar. Ph.D Dissertation, University of Pennsylvania. Bierwisch, N. & K.E. Meidolph, eds. 1970. progress in Linguistics. The Hague : Mouton. Blech, Julea. 1970. The Formation of Marathi Language. Trans. By Devraj Chanana. Delhi: Motilal Banarsidass. Bleen field, L 1962. The Menomonee Language, New Haven: Yale University press. Botha, R.P 1976 “On the Rosetta Stone Strategy or how to canouflage non valid arguments”. Bloomington : Indiana University Linguistics Club.

Page 459: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

456 Brame, K.K 1976. Conjectures Refutations in Syntax and Semantics. Astoria: North-Holland.

Brass, P.R 1974. Language, Religion and Politics in North Trade New Delhi: Vikas Publishing House.

Bray, Demya De. S 1909. The Brahmi Language. Pt.I :Introduction and Grammar. Calcutta: Government printing.

Breanan, J.W 1970. “On Complementation: Towards a syntactic Theory of complement tigers”. Foundations of Language 6: 297-321.

1976a. “Toward a realistic model of transformational grammar”. Paper presented at the NIT-AT & T Convocation of communications, N.I.T.

1976b. “On the form and functioning of transformations” Linguistic Inquiry. 7.1:3-40.

1976a. “Konar gunmen’s for raising”. Linguistic Inquiry, 7.503-301.

Bright, William, 1957. The Karek language. Berkeley: University of California press.

Bules, A.A 1963. The Arabia Triliteral Verb. Beirut: Khayats.

Burgoss, E.1976 “Focus and topic in Kavante”. Ms. SIL, Bragil,

Campbell, M.A. J. Lindholn, A. Devison, W.Fisher, l Furbee, J.Levins, E Maxwell, J.Reighard & S Straight, eds.1970 Papers from the sixth regional meeting of the Chicago lingustic society. Chicago: University of Chicago.

Carion, S. 1968. English qualifiers. N&P Report 20, Competition Laboratory, Harvard University, Cambridge, mass.

1973. English qualifiers Tokyo: Teishukan publishing do.

Gardena, G. 1963. A qujarati reference grammar. Philadelphia University of Pennsylvania press.

Senses of India. 1961,Vel.V,Pt ia : Tables, New Delhi; Government of India.

Chae, Y.R 1968. Grammar of Spoken Chinese. Barjeket; University of California press.

Chatterji, S.K 1926. The origin and Development of the Bengali languages. Vals. Y.II Calcutta: Calcutta University Press

Page 460: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

457

A Babua Misra. 1940. Varnaratnaakara of Jyotitigwara kayisekhaanavya. Calcutta the royal Asiatic Scoiety of Bengal.

Chattepedhyay. Jayanti. 1976. Complement constructions in Bengali. M. Litt. Dissertation, university of Delhi. Delhi.

Chaudhary. R. 1976 A survey of Maithili Literature. Beeghar: Shanti Devi.

Chenky, n. 1957.Syntastic Structures. The have: Mouten,

1963. Aspects of the theory of syntax. Cambridge, name: NIT Press.

1966a.Gartesian Linguistics A Chapter in the historian nationalist doubt. New York: Harpor & Now.

1966b “ Topics in the theory of generative grammar” In T.A Rebook,ed., 1966:1-60.

1970. “Remarks on nominalization”. In R. Jacob & P. Encobann eds., 1970: 184-221. 1971. Problems of Knowledge and Freedom. New York: 1972.b. “Some empirical issues in the theory of transformational grammar”. In. S Sebers. Ed., 1972: 63-130 1973. “Conditions on Transformation”. In B Anderson & P. Kipareky, eds., 1973: 232-86 1977. “ On WE-Movement”. In P.W. Curlicues et al, eds., 1977: 91-132. 1978. “On binding”. MS. A.H Leenik. 1977 “Filters and Control”. Linguistic penury 8.3: 525-304. Gole, P., ed. 1976 Syntax and semantics 9 pragmatics. New York; Academic Press. & J.E Morgan., eds 1975. syntax and semantics 3: speech agts. New York: Academic Press. & J.N Shaddock., eds 1977 Syntax and semantics 3 grammatical relations. New York; Academic press.

Page 461: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

458

W.Newbert, S.Sridhar s. Nashimete, C.Nelson,& D. 1977. “Spme phrase assebillity an Island constraints”. In P.Cele & J.N. sadock ods. 1977: 17-46.

Celebreeke. N.T. 1801. “On the Sanskrit and Praerit languages”, Assistia Bescaches. 7: 199-231. Repr. In H.T,Celebreaks1977. Bysava History literature and religions of ancient India (Miscellaneous ) Vel. II New delhi Coome publications.

Cellinder, B.1957. Survey of the atlaiaLanguages. Strockhelms Almgvist & Vicsell.

Conrie, B,1974. positives and universal grammar” In the Transitions of the Philolegical Society) : 1974:1-32.

1978. “Linguistics is abot languages”, Lecture given at the Linguistic Institute, University of Illineis, Vurbhana Champaign.

Cernyn, W. 1944, Outline of Burmese grammar. Language dissertation no.38. Baltimere: Linguistic Society of America.

Corun, C., T.C. Snith-Stark & Ann Weiser. Eds. 1973. Papers from the Ninth Regional Meeting of the Chinese Linguistic Society. Chicago: University of Chicago.

Culicover, P.w., T. Wasow & A. Akniajan, ods., 1977. Formal syntax New York: Academic Press.

Cunnins, s. 1976, “Relation-changing rules in Nandaria” Mr. University of Toronto.

Cowell, N.W. 1964, Reference grammar of SyrianArabia (based on the dialist of Damascus). Washington, Georgetown University Press.

Darion: B.J. Bailey & alice Devising ods, 1968. papers from the Fouth Regional Meeting of the Chicago Linguistic Society Chicago university of Chicago.

Dasgupta, Proba. 1977, “The internal grammer of compound verbs in Bangla”, Indian Linguistics 30.2: 68-83.

Dasgupta, Probal, 1978. “The Bengla-wa/-ne form as participate and ground. Ms. New York University.

The preparation. Modern Bangala Yes/No Questi9ns Complement Clauses, and Relative Clauses: Their Synaptic and their Ammonic Interciation. Ph.B. Dissertation New York University.

Page 462: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

459

Davidson D& G.N. ods, 1972, The semention of Natural Languages Redirect : Reidel.

Dens: A. 1971. The Yarbesyntec / ?

Beryhire, B.&C.K. pullon, 1978, “Obejct –itial languages” paper preented at the lBA meeting: Linguistics Institute, University of Illineis, Urbana –Champaign.

DeRijk R>P>J> 1972. “Relative clauses in Bosque” A guided tour’. In P. erantdan et al, ods., 1972a; 115:33.

Dimon. R.N.V. 1972. The Bvirbal language of North England Cambridge, England & Cambridge University Press.

Cambridge, England: Cambridge University Press.

1977. Grammer of Iiddin Cambridge; Cambridge University Press.

Dinnon, F.P.S.j. od., 1966. Namagraph saries on Language and Linguistics. N0.19. Washington Collegeton university Press.

Dougherty R.C. 1975, The logic of linguistic research” Mines.

Be kman. P.R. 1976. “Empirical and non-empirical generalist ion in linguistics. In P.R. Firh ed 1976, 35-48

Brends, J.1970. Best and Structure preserving transformations.Pgh.B.Discortation N,I,T.

1972. “A reformulation of certain syntactic transformations In Peters, ed. 1972:21-62.

1976. A Transformational aooroach to English Syntex New.York /Academic Press.

Forguson,C.P.1963. “Assumptions about nasals: A sample study Phonological universals. In J.N. Greenberg , 1963b: 53-60.

Fillmere, C.J. 1968. “The case for case”’ In E.bash & R.T.Karma, ods,. 1968; 1-88.

1977. “The case for case respond “, In P.Cele & J.L.Mergon , ods, 1977; 59-81.

D.t.tLongsundon ods, 1971. studies in Linguistics semen ties. New York: Nelt, Rinchart & Winston.

Fereyth, J.1970. A Grammar of Aspect Cambridge: Cambridge University Press.

Page 463: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

460

Fortune, 6 1955. An Analytical grammar of shona London: languages green & Co.

Failure, O., ed. 1975. Three Dimensions of Linguistic Theory Tokyo; IEC corporation.

Caefike, F 1967. Unterenhnumn our Syntax of Hindi. The Margne: notes.

Cedr, J.W 1970. Gallesuial sinhalese elements structure Paris: Mouton

Gary, J.E & L. L. Keman. 1977 :On collapsing grammatical relations in universal grammer”. In P. cale & J.M Sedeck, eds. 1977: 83-13b.

Edbeen, J.A 1973. Syntax Grammatical structure, Ph.d Dissertation university of Hawali.Working factor’s n Linguistics. 5.5

Eill, K.B & M. A classes. 1969. A Reference grammar fo finished partial. Punjabi University.

Odven Talay 1975 “Premetion, accessibility and case markings towards understanding grammar”. Working papers on linguistics universals. 19:123.

Gerdon, D, & G Lokeff, 1971. “Conversational peculates”. In D Ada,e pf a;. ads. 1971”63-84.

Green, G 1974 Semen tie and syntactic regularity. Bleeding ton , Indiana University press.

1974B. “The function of form and the form of function”. In H.W La galy of al, eds., 1974: 196-97.

1975 “How to get people to do things with words: The hyperactive question. In P. Gale * J.L Mergan, eds., 1975: 107-41.

1976. “Pragmatic motivation and exploitation of syntactic rules”. Ma.

A J.L. Mergan. 1977. “Pragmatics and reading comprehension”. Ms

Green, M.M & G.E Igne. 1963. A Prescriptive grammar of Ighe darling verlag.

Greenberg, J>N 1963a, “Some universal of grammar with particular reference to the order of meaningful elements.” In J.K Greenberg, ed.., 1965: 75-113.

Page 464: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

461

Ed 1963. Universal of language Cambridge : NIT Frees.

1966. synchronic and diachronic universals in phonology language 42:508-17

od.1978. universald of human languages vold I-IV claifornia Stanford university frees.

Grice. N.P 1975 “Logic and conversation”. In.P.cals & J>L>Morgon ods., 1975: 41-98.

1978. “further notes on logic and conversation”,. In P.cels ed., 1978 113-27.

Grierson G.A, 1906 Linguistic survey of India Vol, IV Dravidian language Dep. In 1968 Delhi: National Donaidaas.

1906e. Introduction to the Maithili Dialect if the Bihari language as Spoken in North Bihar 2nd od. Calcutta the bayist hssion press.

Grindhar J.1978. super Equi NP Delection In N,A, campell ot al. ods., 1970 297-317.

Grees n., N. Halle & N.-P sehutamberger, ods., 1975 The formal analysis of Natural Lonaguage. The Hague : Hoston.

Greesson R.R. L.J. Sen & T.J, sehutanberger ods. 1975. The formal analysis of Natural Languages. The Nague: Houston.

Gressson R.R. L.J. & T. Venue ods. 1975. Papers form the Eleventh Regional Noting of the Chicago Linguistic Society.

Kaeker P. 1958. Our Punktion eition Nifiateran in Modern Hindi Nains, Academic and chaften and der literature.

Nale K.1976. “Linguistic academy and the linguistic of Carl Veegelin nthrepelogical inquisition.

Nall, R.A. 1966, Samariam Grammer language Monograph no 21. vol No.4, Baltimere lingustic society of America.

Hall, R.N.R. 1967. Yiddaiah synthetic press structure rules and optional singularly Transformation of the modern standard languages. Ph. B. Missortation New York University.

Page 465: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

462

Keep, Erie F 1972a “Introduction”. In P. Preteen at al, ods., 1972a; V-ix

1972b.Relative clauses in latin and Albanian” paper presented at the diathermies immense linguistics seminar university of Illinois, Urbana-Champaign.

Marris, E 1946. “Fre, ,er[jase tp itterames”. :Lontman, 22: 161-83.

Hire math, R.G 1961. The structure of kanada. Dharwar: Karnataka University.

Hook, peter E 1974. The compound verb in Hindi Ann Ayber: counter for south and southeast Asian studies, university of Michigan.

1975 “A conspiracy in the syntax of hindi”. Paper presented before the America Oriental Society, Meeting 185, 23 April.

1977, “The distribution of the compound verb in the language of North India and the question of its origin”. 6.2:336-51.

1976a. “The Hindi compound verb: what it is and what it does”. In K.S Singh, ed., 1978: 138-37.

1976b. “Perfecting a test for the perceptive: Aspectual parallels in Russian, Lithuanian, Modern Greek, Hindi and paste”. University of minimum papers in linguistics. 2.4.: 89-164.

Rudson-Williams, T 1936. A short grammar of old Persian Gardiff: University of wales press.

Jackendoff, R.S 1972. Sentie Interpretation in generative grammar Cambridge, mass: NIT Press.

1974. “Introduction to the X-conention”. Bleemington: Indiana University Linguistics Club.

Jacobs, R & P Beechum eds. 1970. Readings in English Transformational grammar Walthan, mass:

Jain, J P 1975. Some aspects of Hindi complement system. Ph.D Dissertation, University of California Berkelay.

Janna is, A.N 1968. An Historical Greek grammar, chiefly of the Attie Dialect. Coorg claims, Verlags buchhandling Nildochoin.

Jha, Sahara. 1958. The formation of the Maithili Languages London: Images & Co.

Page 466: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

463

Johnson, D.N 1974a. Edward & Theory of Relationaliy- Exceed Grammar. Ph.d Blessertation, University of Illinois. 1974b. “On the role of grammatical relations in Linguistic theory”. In N.W Locally et al, eds., 1974: 269-83. 1976a. “Ergativity in universal grammar”. Link Research New York: INK Thomas J. Watson Research counter. 1976b. “Om Keenan’s definition of ‘subject of;”IBM Research Report No 6033. New York: INK Thomas J. Watson Research counter. Also published as 1977a. Linguistic Inquiry. 8.4: 673-92. 1977b. “Om relational constraints on grammar”. In P gole & J.K Sedge, eds., 1977: 151-78. In preparation. “Passive typology”. No. &P.M Postal. Forth coming. Agro pair grammar Kasha, B.E 1969. A Reference Grammar of Kashmir, Urban Campaign University of Illinois. Kasha, B.E 1969 A Reference grammar of Kashibird. Urban Champaign: University of Illinois. Et al, eds. 1973. Usage in Linguistics papers in Honger of Hemry and Remee Kashma. Urbana-Champaign University of Illinois. asha, Yamaha. 1965. A transformational treatment of Hindi verbal syntax. Ph.d Dissertation, University of London. 1966. An Introduction to Hindi syntax. Urban-campaign: University of Ill9nais. 1972. “Causative sentences in Hindi revisited”./ In B.D Kasha et al, eds. 1973. 1976a. “Lexically complex verbs in south Asian and west African languages”. Ms. Paper presented at the 1976b. “Speaker intentions and attitudes and verb serialization”. Paper presented before the conference on ‘The signifying animal: The Grammar of Man’s language and of his Experiences’, University of Illinois, Urbana-Champaign. 1976e. “The quetative in South Asian language”, paper reed at the conference on South Asian languages and Linguistics, university of Illinois, Urbana-Champaign.

1979. “Pragmatics and compound verbs”. Lecture given at the Delhi University Linguistics Association (DULA).

Department pf logistics. University of Delhi

Page 467: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

464

BB. Keshru & Tej K. Khatia. 1976, “The notion ‘Subject’ A note on Hindi-Urdu, Kashmiri and Panjabd”. In N.K. Varma, od. 1976’ –

Kate, J.J. & P.M. Postal, 1964. An Integrated theory of Linguistic Description. Cambridge, Boss: NIT Press.

Cellegg, S.N. 1933. An Introduction to Hindi Syntax London; Routledge & Kegan Faul Ltd.,

Kannan, B.L. 1972. “Relative clause formation in Nalagary (and some related and some net so elated) languages”, In P.Peranteen et al, ods., 1972a; 169-89.

Koonan, B.L. 1975. “Some universals of positive in relational grammar”. In R.E. Cresnam, et,al, ods., 1975a; 340-32.

1976. “Towards a universal definition of subject”, In C.Lied., 1976; 303-33.

A.B.Courie. 1972. “Noun phrase accessibility and universal grammar”. Paper presented at kings college, London. Later published as 1977. Linguistic Tnsview 8.1:63-99.

Kont., R.G. 1933. Old Persian: grammar Texts, Lexian Connection American Oriental society.

Kaysor, S.J,ed., 1976 Recent Transformational Studies in European London. Linguistic Inquiry Homegraph 3; Cambridge. Boss. Press.

Kiparathy P.& G. Kipareky . 1970 “Fast’ In N.Bierwiseh & K.B.Hoidely ods., 1970’ 143-73.

Klaiman, Hindi. 1976. “Topicalisation and relativisation in Hindi”, Indian Ltnnistion 37:313-33.

1979. “On the statue of the subject hood hierarchy in Hindi”, IJBL 8.1; 17-31.

In preparation. Velitionality and subject in Bengali a study in Semantic Parameters in Grammatical Processes. Ph.B.Assertation ; University of Chicago.

Klokied, T. 1875 “On grammatical relations in Nitimat”. Ms. University of Calgary.

Kepeory, F. 1961. Glovesny Vid Gostine Prague. Respravy Os. Akadonia Ved 72.2.

Kesta, Jan. 1978. “Why subject sentences don’t exist”. In S.J. Keyser ed., 1978: 53-64.

Page 468: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

465 Krishnamurti, Ph.K.1975. “Gender and number in Prete-Dravidian” IJEL. 4.2.1. 320-50. Kruger, J.R. 1961. Chuvash Normal Introduction Grammar Reaer and Vocabulary. Bloomington: Indiana University Publications, Cralie and Atlais Beries, Vol.7. Kune, s.1975. The Structure of the Japanese Language. Cambridge Nase: NIT Press. Kureda, S-Y. 1968. “English relativisation and certain related problems”. Language. 44:244-66. Laffite, P. 1962. Gramaire Bosaue Bayone, France: Mition doesdu Mosee Bosque et Ikas’. Ln Galy, N.W., R.A. York & A.Bruck ods. 1974. Papers from the Tenth Regional Meeting of the Chicago Linguistic Society Chicago: University of Chicago. Takeoff, G. 1965. On the nature of Synthetic Irrerularing NEP Robert no. 16; Competition Labortroy Kaervard University Cambridge, Mass. 1967. “Deep and surface grammar’. Ms.Narvard University Cambridge, Mass. 1969, “Presupposition and relative grammatically In V.Tedd, ed. 1969: 103-16. 1970a. “Global rules”. Language. 46: 627-39. 1970b Irregularity in syntax New York: Nelt, Kinchart & Vinstron. 1971. “On generative semantics”’ In D.D. Stoinberg & L.A. Jakobe vits, eds. 1971; 232-96. 1972. “Linguist ions and natural logic”, In D.Davidson & G.Kharman, eds. 1972; 545-605. Sahkeeff, 6.1973a. “Some thoughts on Tran derivational constraints”. In B.Kshra et al, ods., 1973; 442-53. 1973b, “”Fussy grammar and the performance/competence terming game”. In C.German et al , ods, 1973:271-91. Lake off, G.& J.R. Ross, 1970 “Two kinds of and “ Linguistic inquiry, 1.2: 271-2. J.. R.Boss 1975. “Le deep structure necessary?’ Bloomington: Indiana University Linguistics Club. Also published in J. D Mecawley. Ed., 1976b: 159-66. ’

Page 469: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

466 Lakeff, Dehin, 1973, “The legle of politeness: or mind your D’s and q’s. In C germ of al, eds., 1973, ? Langunden, N. 1976. A grammar of ? Deckeley: University of California press. Lewlev. J. 1975 “On coming up terms in achieve ness the function of Verbal disagreement”. In B.E ? et al, eds., 1975b: 398-488. 1977. “A agrees with B in adhences: A problem for relational grammar. In P. cole & J.N Badeck, eds., 1977: 219-os. Loss, R.D 196a. The grammar of English nationalizations. The hague: Newton eds co., Ld, charles, ed. 1976. Defining a Linguistic : south code Chicago: The University of Chocago press. Mecawley, J.D 1969. :Lexical insertion in a transformational grammar”. Without deep structure”. In B.J.London et al, ods., 1960; 71-80.- 1969a. Review of current trends in linguistics. Yal :3 Theoritical publications” 44:996-95. 1970a. “English as a VAS Language”London 46: 289-99. 1976b. “Where do mean phrases come from T” In R. Jacobs & P. Rosenbun, eds., 1976: 166-88. 1972. “A program for logie”. In R Ravideen & C Hangman, eds., 1972: 418-344. 1976b. Syntax and Education. Vel.7 New York: Academic Press. 1976. “Language universal in linguistic ungrammantation”. Lecture given at the Linguistics Institution University of Illinois, urbana-campaign.

Page 470: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

467 Mihalie, Dev F. 1957. Grammar and Distionary of Ecc-Malanesian. Illinois: Mission

Press. Misra, Dipti, 1975. “Relativisation in Khasi”. M.A Term Paper, University of Delhi. Mishra, jayakant. 1976. History of Maithili Literature.. New Delhi sahitya academy. Misra, S.E 1968. A Comparative Grammar fo Sanskrit, Greek and Nittite Calcutta:

World Press. Mistry, P.d. 1978. “Some language-specific properties: Disagreement in agreement

in Hindi, Gujarat and Marathi”. Paper given at the conference on South Asian language and Linguistics, University of Illinois, Urbana-campaign.

Moravesik, Ddith A 1971. “Agreement”. Working papers on language universal 5. 1978. “Agreement”. In J.N Cremberg, ed 1978: 331-374 Mergan, J.L 1969a. “Om arguing about semen ion”. Papers in Linguistics. Y: 49-70. 1969b. “On the treatment of presupposition in transformational grammar”. In R.X

Binnick et al, eds., 1969: 167-78. 1972. “Some aspects of relative clauses in English and Albanian”. In P peranteem et

al, eds., 1972a: 63-72. 1973. Preposition and the presentation of Meaning recaution. Oh.D.Dissertion ,

University of Chicago. Ph.d blesertation, University of Chicago. 1977. “Conversational postulates revisited”. Language, 33: 277-84. 1978. “two types of conventions in indirect speech acts”. In p. gale, ed., 1978: 261-80. Mergen stierne, C. 1938. Indo-Iranian frontier language London 0550 Mukherji, Trapped. 1965. The old Bengali language and Text Calcutta: University of

Calcutta. Kilseem, U.E 1967. Intermediate Hindi. Madison Indian language and Are canter,

University of Wiseman. O’Leary, De lacy. 1969. Comparative grammar of the sanities language Austrian:

Philly Press

Page 471: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

468 Pahva, Munchi Tarakdss. 1936. The Pucca Mumchi peshawar contemned faquir chand harah. {amdjaro [ade. R 1978 “passive as on optical rule in Hindi, Marathi and Nepali”. Paper presented at the conference on south Asian language and Linguistics, University of Illinois., Urbana-Champaign. Paper, K.K 1955. phonology and marchelary of royal achievement ? Aun arber: University of richigun press. Park, B.-S 1975. “On the multiple subject constriction in Korean”. Linguistics, the 63-76. Paset, Kormen, 1974. Discussing Language The Hague & Paris: Houton. Partee, Barbara N. 1971. “On the requirement that transformations preserve meaning”. In G. Pillmore & B.T Langandeem, eds., 1971: Patnaik, B.N. 1976. Complementation in Oriya. Ph.D Delectation CIETZ, Hyderabad. Patnaik, R.E 1978. “On the bali-complement construction in Oriya”. Paper presented at the seldom jubilee celebrations and All India conference of Linguistics. University of Delhi Perantean, P., J.N Levi & G.C Phrase, eds. 1972a. The Chicago which kunti: papers from the relative classes festival Chicago: Chicago linguistic society. 1972a. Papers from the eighth Regional meeting of the Linguistic Society. Chicago; Chicago Linguistic Society. Perlmutter, D & p.N Pestle, 1974. “Lectures on relational grammar”. Lecturers given at the linguistic Institute, university of mass actuates, Amherst. 1977. “Towards a Universal characterization of passivisation”. In the Presentation of the Third Annual meeting of the Darkeley Linguistics Society. Berkley University of California. 1976 “Some proposed laws of basic clause structure”. Peters; Sed. 1972. Scale of linguistic Theory. Englaweed-Oliffo, New Jersey: Frontier-Wall. Pickering, W,N 1974. “Capping and constituent order in Apurina Ms University of Toranto.

Page 472: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

469 Piettman, R.s 1954. A Grammar of totalising (mereleg) kahnatle language Dissertation no 301 Ve. 30.1. pt. I. Maltimere: Linguistic society of America. Peppe, Nicholas, 1955. Introduction to Anglican comparative studies Nelainki: Budmalais ugrlainon Beuran Teinivuksia, Vel 110. Pepper, karl 1963. Conjectures and Refutations: The growth of scientific education. New York: Marper & Mu. Periska, Vinesne. 1967-9 “On the perfective verbal aspect in Hindi”. Archfy Oriental 33: 64-88, 208-31: 36: 233-31: 37: 19-47 345-84. 1977. “Perfective verbal expressions in Hindi (A new line of approach to the problem)”. Archfy oriental 45: 65-78. Pestal, P.M 1966. “On the so-called ‘pronouns’ in English”. In P.P.S.J Binean, ed., 1966: 117-206. 1970. “On the surface verb remind”. Linguistic Insuiny. 1: 37-120. 1974. On Baisings one rule of English Grammar and its Theoretical publications. Cambridge, meet NIT Press. Pray, Bruse B.a.d. On-cccurance destriations on Compound Verbs in Hindi-Urdu. Berkelay, California: University of California. Nimco Prideaux, .G.D. 1970. The syntax of Japanese enumeration. The mague: neuton. Prenetasana. P.N 1978. Relativisation in Ghai. N. Phill Disertatiln, university of Delhi, Delhi. In preparation. Some aspect of the Noun phrase in Thai Ph.d dissertation, University of Delhi . Delhi. Punte, N.C 1959. A Complete creation on the conjunction of English verbs. New York: Egivd Hokay. Pullon, G.K 1977. “Word order universals and grammatical relations”. In P.sole & J.N Sadeek, eds., 1977: 245-77. Rastergneva, V.S 1963. A short touch of paiik grammar. The Hague: Newton. Ray, Punya slake et al. 1966. Bengali language handbook. Washington, d.C: Conter of applied linguistics.

Page 473: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

470 Rischel, J. 1970. “Some characteristics of noun phrases in West Oremulandie”. Asta Linguistics hafnians. 13.2: 213-45. Rebins, R.N 1952. The yurek Language: Grammar, Texts and Design Berkeley University of California press. Resenbaun, P.E 1967. The Grammar of English Predicate general mark constructions. Cambridge, masss: NIT press. Ross, J.R. 1967. Constraints on variables in syntax. Ph.D. Dissertation, NIT. 1969. “Auxiliaries as main verbs”. Journal of Philosophical Linguistics. 1.1: 77-102; also published in W.Yeld, ed. 1969. 1970a. “Gapping and the order of constituents”. In M. Dierwisch & K.B Keidelph, eds., 1970: 249-39. 1970b. “On declarative sentences”. In R.A Jacobs & P.S Resembun, eds., 1970: 212-72. 1972. “The category aquish: End station homework”. In P. Peranteen et al, eds., 1972n. 316-25. 1973. “Sifting”. In N. cross et al, eds., 1975: 133-72. 1974. “Nominees”. In O Fujimara, ed., 1974: 137-237. Sachdeva, Vipla. 1977. “Relativisation in Tumkul mega”. M.A perm paper, University of Delhi. Sadeek, J.N 1969. “Hyper sentences”. Papers in Linguistics. It 205-370. 1972. “Speech not idons”. In. P Ferantean et al, eds. 1972b: 319-40. 1975. Towards a Linguistic Theory of Speech Art New york. Academic Press. 1978. “On testing for conversational implicative”. In P. Cole, ed., 1978: 281-97. Sahidullah, nd. 1940. “Buddhist Mystic Songs”. The pussa university Studies, 4.2. Semarin, W.J. 1966. Choral Language: Grammar, Texts and Vocabulary. Berkeley: University of California press.

Page 474: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

471 Sanches, N, & L. Slobin. 1973. “Numeral classifier and plural marking: An implicational universal”. Working papers on Language universals, II, I-22. Sapir. B & N Heijer. 1967. The Phonology and Morphology of Mache language. Berkeley: The university of California Press. Schorling, Susan. 1975. “Subject less sentences and the notion of surface structure”. In C. Cruun et al, ods., 1973; 577-86. Searle, J.R. 1969. Speech asts London; Cambridge University Press. 1975. “Indirect speech acts; In P.cals J.L. Morgon, ods., 1975; 1978. “Theirs of meaning”. Lecture given at the Linguistics Institute, University of Illinets, Urbana Campaign. Sebeek, T.A. ed., 1966. Current trends in Linguistics Val.J. Theoretical Foundations. The Mague; London. Sen. Nilration 1975. Barly Western IA Verification Simla; Indian Institute of Advanced Study. Sen, Sukumar. 1948. “Old Bengali Texts” Indian Linguistics , III. Repr. In 1965. Vol.91 61-133. 1936. Garyasgtindabeii Durdwan: Bardwasn Sahitya Babha. Sapire, N.1974. Aspects of Hindi Verbal Syntax Ph.D.Discertation, University of Chicago. Sharma, A.1958. A Basic Grammar of Modern Hindi. Delhi Ministry fo Education and Scientific Research. Singh, K.S.ed.,. 1978. Leadings in Hindi Hindu Linguistics. New Delhi; National Publishing House. Singh, Udaya Narayana. 1976a. “Diglessia in Bengali: A study of attitudes”. Papers in Linguistic Analysis. 1976. “Negation in Bengali and the order of constituents Indian Linguistics. 37.4: 295-385. 1977. “Reflexivisation in Bengali”. BINL 6.1: 8:23.

1978a. “Bele” Paper presented at the Halhod Seminar on Bengali Grammar. Jadavpur University , Calcutta.

Page 475: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

472 1978b. “Agreement rule, language universals and Maithili”. Paper presented at the conference on South Asian Languages and Linguistics, University of Illinois Urban-Campaign To eb published as 1979. South Asian Language Analysis. Vol. I, University fo Illinois. 1978c. “Verb agreement hierarchy and falsifiadbility”. Paper presented at the LSI Golden Jubilee Celebrities and All India Conference of Linguists, University of Delhi, Delhi. 1979a. “Grammatical categories and grammatical relations; why can’t we have the best of both the worlds?. Paper presented at the seminar on Grammar. Central Institute of Hindi. Hyderabad. S.K.Bandhopadhyay. 1978. “Some aspects of interrogation in Bengali Assamese and Maithili. Papers in Linguistics Analysis, II, 19-34. K.V.Subbarae & S.K.Bandhopadhyay.In preparation “polar Verbs in selected south Asian Language and their classification”. Sinha, Anil C. 1970. Predicate constrictions in Hindi and English. D.Phil Thesis. University for York. 1976. “A phrase structure rule for Hindi noun phrase and universal grammar”. Indian Linguisation. 33: 173-84. Sinha, anjani K. 1975. “On the stative passives an treatment of some idioms”. In C.Gerum et al, ods., 1973; 615-26. 1974. “How passive are passivesand treatment of idioms”. In N.W. Lagaly et al, ods., 1974: 631-42. Smith-stark, T. 1976. “Dragativity, grammatical relations accessibility hierarchies, Feeman, and consequences”, Ms. University of Chicago. Steinberg, D.D. & L.A. Jakebevits ods. 1971. Semantion in interdiaplibnary Render in Philosophy Inquisition and physiology, London, Cambridge University Press. Steckwell, R.P., Schachter and B.N.Partee.. 1968. Integration of Transformational Theories in English syntax. B.B.Air Force, 68-419. Oxford . Subbarae, K.V.1972. “Is Hindi an SVE language ?” Ph.B. University of Illinois.

1974a. New phrase complementation in Hindi Ph.B.Dissertation, University of Illinois.

Page 476: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

473 1974b, “Phrase structure rule for Hindi noun phrase and universal grammar. Indian Linguistics . 35: 173-84. 1976. “Complementisers in Hindi” .Indian Linguistics 37. 115-32. 1977. ‘Aspectual verbs in Telugu”. Paper present at the International school of Dravidian Linguistics Seminar on Dravidian Aspectual system, Bangalore. Dipti Misra. 1976. “Pronouninalisation or deletion in relative clauses. Papers in Linguistic Anaysis. It 37:50. Stalnaker, R.C. 1972. “Pramation ‘, In D Davidson & G.Harman ods, 1972; 380-97. Subramanyan, P.S. 1968. A Descriptive Grammar of Gondi. Annamalianagar: Annamalai university. Tedd, W., ed, 1969. Studies in Philosophical Linguistics series one. Cambridge, Illnois; Great Expectation Press. Trithart, R.1975 “Relational grammar and Chichewa subjectivisation rule”. In R.N.Gresman et al, ods., 1975a; 615-24. Tucker, A.N. 1967. Eastern Premiere Language London: International African Institute. N.A. Bryan 1966. Lingustic Analysis; The Non-Bantu Language of North-Eastern African, with a suppliment on Ethiopia language (by Welf leslan) London: Oxford University Press. Thunb, Albert 1964. uA Handback of the Modern Greek Languages. Trans. By S.Angus. Chicago : Argument Publishers. Thurneyson, R. 1970 A Grammar of old Irish Dublin: The Dublin Institute of Advanced Studies. Vale,R.N. 1948. Verbal Composition in Indo-Aryan Pune: Deccan College,. Van Olphen H.K.1970. The Structure of the Hindi Verb Phrase. Ph.D.Dissertation, University of Texas. Austin. Verma, H.K. 1967. “Structural notes” In U.S. Wilson, 1967: I-xx. 1971. The structure of Noun Phrase in English and Hindi Delhi: Motilal Banarsidass.

Page 477: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

474 Ed, 1976. The notion of subject in South Asian Language. Madison’s: South Asian of publications no 3, University of Wisconsin. Verma, Usha. 1979. “Some basic transformations in Angani”. N.A term paper, University of Delhi, Delhi. Vascow, T. 1977. “Transformation and Lexicon”. In P.W Culicover et al, eds., 1977: 327-60. Watkins, W.E 1957. A grammer of Chewas. A Bostn Language of British Central Aftica. Language Dissertation No.24 Baltimore: Linguistic Society of America. Wells. R. 1947. “Immediate constituents”. Language, 23: 81-117. Welmore, W.B. 1946. A Prescriptive grammar of punti. Language Dissertation No. 39. Dultimore: Linguistic Society of America. Wierabieka, Anna. 1976. “Mind and Body”. In J.D Macawlay, ed., 1976: 129-57. Williamson, Kay. 1963. Grammar of the Kalshuma Dialest of Iie Cambridge: Cambridge university press. Wirth, J.R., ed. 1976. Assessing Linguistic Arrments. New York: John Wiley & Bone. Weedbury. A. 1975. “Negativity of grammatical processes: A study of screenland bakine”. N.A thesis, University of Chicago. Zeyer, Louis. 1074. “Is English a VSO Language?” Flemington: Indiana University Linguistics Club. Ziv, Y. 1975. “On the relevance of content to the form-function correlation”. In. R.Greenman et al, eds., 1975b: 1976. Function of Relative Clauses in English and Hebrew. Ph.D Dissertation, university of Illinois. Gull, G.G 1966. A penal system for generating French Verb Paradians: A study in linguistic features. Ph.D. Dissertation, University of Wisconsin, Madison.

Page 478: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

475 In Bengali

Shastri. Hariprasad, 1916, Easier Bscharar purame Brannalaa Yhaagaae Bandhagan e Dehna. Calcutta Bangiya Bahitya Parishat. Singh, Udaya narayana. 1979b. “Garya-beakya-byabashed-bisayak prastaab/ii Eriyamped”. Gengeo patre.T: 54-64. ( c ) In Maithili Choudhury, Rajkamal. N.d Nirmathi Beelan Kumar. Laheriasarai: maithili pocket books. Jha, Bhelamath. 1970. Maithili Theories: Aitihasaik Tatham Bhassavaimenik Viveem. Beeghar. Jha, dinabondhu. 1950. Mithilas Dhapan Vidvetnna. Darthanga. Jha, Govinda. 1968. Maithilik udane &vikaasa. Calcutta: Maithili Frakashan Samiti. Jha Harincham. 1949. Ranghesheelas Patna: Pustak Bhandar. 1960, Gorearii, Calcutta: maithili pralashan.

Jha, Bangnath. 1955. Mithilian bhagan prakashan darthaga: oranthalaya pralashan. Lalit. 1963. Prithrivutra. Darthanga: Vidyapati Prakashan. Manipadma, Dvajkisher Varma. 1972. Kafkaa buriearea. Mann, prayhulla kumar singh. 1972. Wepaslak Maithili gashitwak Itikaads. Diratnagar, Nepal: maithli Sahitya parishad. Pukhrol, Balkrishna. 1972. “Ekuumikaa” In Nam, 1972: 3-4. Singh, jayadhari. 1969. Randhagaam nB tnentrik siddhant- madhubani, darthanga. Singh, prabodh narayan. 1965. Amber Wesnrii Calcutta: Mithila Darshan. 1968. Fremak Reg Calcutta: Leka sahitya parishad. Vyathita, Balgavinda jha. 1968. Trinathagna. Jhan Jharpur, Barthanga: Van Dana Prakashan. Yastrii. 1965. Paare. Darthanga; Granthalaya Prakashan.

Page 479: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

476

VITA Udaya Narayana Singh was born in bilingual family at Calcutta, west Bengal on 23rd November, 1951. He got Maithili from his father who belongs to Bihar and Bengali from his mother, a Bengali. He posed Higher Secondary Examination in 1969 from Kare school, Calcutta with a first division and decided to join the government Sanskrit college to study B.A. Honors in Linguistics. He passed the Bachelor of Arts (Exam) Examination in 2972 from Calcutta University standing first in the first class. He had also the honor of being the Bushan scholar of Calcutta university in 1972 for getting the highest marks in the whole university in the B.A examinations. He also received Jubilee Award on this occasion. He came to Delhi with a national Scholarship to join M.A in Linguistics course at the Department of Linguistics, University of Delhi and passed out in 1974 standing first in the first class and thereby becoming the first recipient of the aggressing Memorial Gold Modal. He was awarded a University grants commission junior Research Fellowship in December, 1974 and since then he has been working on the present dissertation and also assenting in the teaching programmers in the Department of linguistics, University of Delhi, Delhi. Since August 1, 1979, he has been teaching in the department of linguistics as visiting lecturer. He has, between June, 1979 and July, 1979 worked on a masses project on ‘Language Teaching in a Bi-or plurillingual and Multicultural contexts’. He has published a number of books and articles in the field of Literature and participated in different conferences and symposia in India and abroad to present papers on different aspects of

Page 480: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

477 Linguistics many of which have subsequently appeared in the professional journals. The conferences he attended included the following: Ivth All India Conference of Linguistic, Jawaharlal Nehru university, New Delhi, 30-31 Dec. 1974: Seminar on Linguistics and Language Teaching. Northern Regional Language Centre (CIIL), Patiala, 18-19 Dec. 1975: VIth All India Conference of Linguistic, Sri Venkateswara University, Tirupati, 13-14 Jan. 1976: Vith Dravidian Linguistics conference, Andhra University, Waltair, 3-5 June, 1976: Workshop on Translation of Poetry in Indian Languages, Sahitya Academy, Bhopal, Nov, 1976: VIIth all India Conference of Linguists, University of Bihar, Nusaffarpur, 28-30 Dec. 1977: Workshop on the preparation and Evaluation of grammar books, NGEET, New Delhi, 13-18 March, 1978: Workship on One-ast plays, Sahitya Academi, Banaras, 20-23 March, 1978: Conference on South Asian Language and Linguistics, University of Illinois, Urbana-champaign, U.S.A., 14-16 July. 1978: LIS Golden Jubilee celebrations and All India conference of Linguistics, University of Delhi, Delhi, 14-15 ect., 1978: Sahitya Academy ‘Seminar on the contribution of the underprivileged to the Indian Postie Tradition’, Kala parishad, Bhopal, 10-11 January 1979: and ‘seminar on grammar’, kendriya Hindi same then. Hyderabad, 22-24 February 197. Kis publication include the following:

LINGUISTICS 1 Fihlrddis in Nrnhsl A Study of attitudes. In apaers in Linguistics Analysis I 94-99(1976) 2 Ptsnoyh N Pandit life and works. In pakha benjman, 9.1: (1967) 3 Negation in Bengali and the order of constituents. In India Linguist ion, 37.4.293 (1976)

Page 481: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

478 4. Reflexivisation in Bengali. In International Journal of Dravidian Linguistics, 6.11 8.83 (1977) 5. Bektigate Vidyasagar. In gaangee Patra, 6: 36-49 (1978). 6. Some Aspects of Interrogation in Bengali, Assamese and Maithili. In papers in Linguistic Analysis, 19-34 (1978) 7. Charyaa-Vaskya-Vjavachade-Vishayak prasteau/ii driyaspeda in ganges patra, T. 54-64 (1979) 8. Agreement rule, language Universals, and maithili. In south Asian Language Analysis, University of Illinois at Urbana-Champaign (1979).

LITERATURE 9. Kayaye Yadanti. Calcutta: Maithila darshan (1966) 10. Amritasya Putrash; Calcutta: laka Sahitya Parishad (1971) 11. Kasyakak Naan Jlivan. Calcutta: Akhil Bharatiya mithila gangha (1971). 12. JK Chal Besics. Calcutta: maithili Bangamanaha (1974). 13. Kastakak Lal. Calcutta: Mithila Darshan (1975). 14. Maharshi, Aacharya ser Dhashan. In Mithilaa Dhuni; (1975). 15. Asanmanya Lekak Lekhak: Asammesaya Lekak metu sashitya In Agnipatra: (1975). 16. Naatakak Swarmp Nirnaya as maithili Neatak. In Rangamensha. I (1975). 17. Pratvasvartas. Calcutta: mithila darshan (1976). 18. Baangla a Upanyaser Shariir: Sunil Gangopadhyay. In Ganga patra, 2: 31-45 (1976). 19. Somaslechanaar Naamtaa O Oupanyaseik Satinath Bhaduri. In gangee patra, 3: 12-29 (1976). 20. Benkinchandra: Garalgaachaa O Saagarmanthan. In gangee patra. 4: 35-51 (1977). 21. Asndelan. Calcutta: Mithila Dare Shan (1977). 22. Rasmliles. Calcutta: Mithila Darshan (1977). 23. sashityatattva O Aalgaa Cinteabhasbanaa. In paharteli, 5: 1-12 (1978)

Page 482: some aspects of maithili syntax: a transformational-generative approach

Central Institute of Indian Laguages

479

SOME ASPECTS OF MAITHILI SYNTAX:

A TRANSFORMATIONAL-GENERATIVE APPROACH

THESIS SUBMITTED TO THE UNIVERSITY OF DELHI

FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

R.94/P 7.7.8 Some Aspects of Maithili Syntax: A Transformational generative approach By Udaya Narayana Singh. University of Delhi. Delhi Aug 79. Mag 26.9(168)101 panel 140 (628( 588

End 8 7 80

By UDAYA NARAYANA SINGH

DEPARTMENT OF LINGUISTICS UNIVERSITY OF DELHI

DELHI

AUGUST, 1979