agentless transitives 51.3 4.tuite

7/31/2019 Agentless Transitives 51.3 4.Tuite

1/28

Agentless Transitive Verbs in GeorgianKevin Tuite

Anthropological Linguistics, Volume 51, Numbers 3-4, Fall and

Winter 2009, pp. 269-295 (Article)

Published by University of Nebraska PressDOI: 10.1353/anl.2009.0010

For additional information about this article

Access Provided by University of Montreal at 10/24/10 6:37PM GMT

http://muse.jhu.edu/journals/anl/summary/v051/51.3-4.tuite.html
http://muse.jhu.edu/journals/anl/summary/v051/51.3-4.tuite.htmlhttp://muse.jhu.edu/journals/anl/summary/v051/51.3-4.tuite.html


2/28

269

Agentless Transitive Verbs in Georgian

KEVIN TUITE

University of Montreal

Abstract. The Georgian language has an unusual abundance of indirect(dative-subject) verbs. Most of these are intransitive, but several dozen areformally transitive. The focus of this article is on the subset of Georgian indirecttransitives that lack overt grammatical subjects (e.g., I shiver, lit., it makesme shiver). The semantic, morphological, and syntactic features of Georgianagentless transitives are presented and compared to those of similar verb typesfrom other languages. Of particular interest is a small group of bodily emana-tion verbs, such as yawn and belch, that are paired with syntactically inversedirect-transitive verb forms. A scenario is reconstructed for the origin of suchdirect-indirect pairings, which are otherwise unknown in Georgian.

1. Introduction. Grammars and dictionaries of Georgian customarily classifyverbs by transitivity, voice, or both, in order to accommodate, at least partially,the complexities of verbal morphology and case assignment. As is well known,the Georgian case inventory includes an ergative case (also known as narra-

tive case), assigned to the agents of certain verbs, but only when the latter arein the aorist or optative paradigms. As a consequence, descriptive grammars of Georgian characteristically feature tables such as the one below, in which theassignment of case to the principal clausal arguments is correlated with twoparameters: verb class, which Georgian linguists usually refer to as voice( gvari), and many non-Georgian linguists as conjugation; and tense-aspect-mode paradigms, grouped into three series according to stem form and case-assigment properties. The verb classification scheme formulated by A. Shanidzein his highly influential Fundamentals of Georgian Grammar (1953) separates

those verbs that can assign the ergative from those that cannot, as well asapplying a cross-cutting distinction between medial (medioactive and medio-passive) and non-medial verb types.

Whereas Shanidzes voiced-based verb-classification scheme and others likeit are used in grammatical works destined for native Georgian speakers, a con-siderable number of linguists writing for general (and therefore mostly non-Georgian) readerships reconfigure the classification to include a separate classof indirect or inverse verbs (Tschenkli 1958:44690; Aronson 1982a:33244), as in table 2. 1 As indicated by the choice of label, the most noteworthy fea-ture of indirect verbs, at least for native speakers of West-European languages,is the apparent inversion of the normal relation between case (and also agree-ment) marking and grammatical roles. The noun phrase assigned dative case by


3/28

270 ANTHROPOLOGICALLINGUISTICS 51 NOS. 34

indirect verbs has subjectlike attributes, such as the capacity to govern reflexiveand reciprocal pronominals, as in (1).

Table 1. Georgian Verb Classes and Case Assignment

VOICE ( gvari) ACTIVE P ASSIVE MEDIOACTIVE MEDIOPASSIVEagent patient IO theme IO agent IO theme IO

SERIESI (present,future,imperfect)

NOM DAT DAT NOM DAT NOM DAT NOM DAT

II (aorist,optative)

ERG NOM DAT NOM DAT ERG DAT NOM DAT

III (perfect,pluperfect)

DAT NOM () NOM DAT DAT () NOM DAT

NOTE: After Shanidze (1953). Columns in bold type correspond to arguments whose casemarking shifts according to series. IO = indirect object.

(1) bavvebs k ertmanetik uqvart.children-PL-DAT each.other-NOM O3.VM-love-PLThe children love each other.

Table 2. Aronsons Georgian Verb Classes (Conjugations)

CONJUGATION 1. ACTIVE 2. P ASSIVE 3. MIDDLE 4. INDIRECTSERIES Subj. DO IO Subj. Obj. Subj. IO Subj. Obj.I. Present, Future NOM DAT DAT NOM DAT NOM DAT DAT NOMII. Aorist ERG NOM DAT NOM DAT ERG DAT DAT NOMIII. Perfect DAT NOM () NOM DAT DAT () DAT NOM

NOTE: From Aronson (1982:344). Subj. = subject; Obj. = object; DO = direct object; IO =indirect object.

The recognition of indirect verbs as a distinct class or conjugation strikesme as problematic for several reasons. First of all, it introduces a syntactic fea-turegrammatical subjecthoodinto what is otherwise a purely formal classi-fication of verbs according to stem morphology and case assignment. Second, thedative-subject intransitive verbs classified by Tschenkli, Aronson, and othersas indirect are formally quite heterogeneous (see the detailed study of Geor-gian indirect verbs by Cherchi [1997]), and the morphological properties they doshare, such as the shape of their future and aorist stems, do not exclude a small

number of nominative-subject verbs. Finally, indirect syntax, as indicated bysubject properties accruing to an argument marked as a morphological object(according to case assignment and agreement), is by no means limited to indirect


4/28

2009 KEVIN TUITE 271

verbs. A significant number of bivalent intransitive verbs (which Tschenkli andAronson group into class 2) assign syntactic subject privileges to their formalindirect objects. Furthermore, there are even a handful of transitive verbs (i.e.,class 1) characterized by indirect syntax, as in (2). As a consequence, I prefer toretain a classificational scheme such as Shanidzes, based on the cross-cuttingcriteria of ergative case assignment (classes 1 and 3) and the form of the future-aorist stem (classes 3+4 vs. 1+2), without consideration of syntactic criteria. 2

(2) bavvebs k ertmanetik ainteresebt.children-PL-DAT each.other-NOM VM-interest- SM-PL The children are interested in each other. (lit., each other interests the children)

The topic of this article is a subtype of Georgian indirect transitive verb thathas received little attention from linguists, the agentless transitive verb. Theagentless transitive verb is characterized by transitive morphology (most oftenthat characteristic of causative verbs: version vowel a3 and present-seriesmarker eb); but only one argument, formally marked as an object, is subcate-gorized. In other words, agentless transitive verbs look like ordinary Georgiantransitive verbs, but are not accompanied by overt grammatical subjects (B. Jorbenadze 1985:16465; N. Jorbenadze 2006:26; Melikishvili 2001:23940).Their one surface argument is assigned dative case and controls object agree-ment marking in the verb. Three examples of agentless transitive verbs aregiven in (3)(5)two of them from Georgian-language Internet chat groups, andone from a short story by a popular contemporary writer. The single argument of each of these verbs is marked by the first person singular object marker m.Each agentless transitive verb also ends with a third person singular subjectsuffix (present and subjunctive s, past indicative a), which is not cross-indexedto a surface noun phrase.

(3) ertxel kimiis lekciaze da=mamtknara .once chemistry-GEN lecture-at PV=O1SG-VM-yawn-AOR .S3SGOnceI yawned at a chemistry lecture. (chat group Tbilisi forum)

(4) dzalian a=civda, gairvebulma maisuric kivery got.cold-S3SG distressed- ERG T-shirt- NOM-also PRT

amo=viime, makankalebda mainc. PV-S1-VM-stretch-AOR O1SG-VM-shiver-SM-IMP -S3SG nonethelessIt got very cold. Suffering (from the cold) I stretched my T-shirt downward, but still

I was shivering . (Guram Doanivili Erti ramis siqvaruli )


5/28


(5) vkvdebi liv tailerze mabodebs masze!! S1-die-SM-PRS Liv Tyler-on O1SG-VM-craze-SM-S3SG her-onIm dying over Liv Tyler, I am crazy about her. (chat group Netgamer )

Compare the syntactic frame of the agentless transitive verb axvelebsXcoughs in (6) to that of an ordinary causative verb, such as amg erebsXmakes Y sing in (7).

(6) bis axvelebs. boy-DAT VM-cough-SM-S3SGThe boy is coughing.

(7) megobari bis amg erebs.friend-NOM boy-DAT VM-sing-SM-S3SGA friend makes the boy sing.

The verbs axvelebsand amg erebshave identical morphology, but whereasthe latter subcategorizes for both a dative-case noun phrase denoting the singerand a nominative-case argument denoting the person or situation that causesthe singer to sing, axvelebsis only accompanied by a single noun phrase,designating the one who coughs. At first glance, sentence (6) looks like it oughtto mean X makes the boy cough, but no X ever appears with the habitual mark-ers of a causative agent (nominative case in the present series of paradigms,ergative case in the aorist and optative). The author of (5) claims to be drivencrazy, and specifies the actress Liv Tyler as the proximal cause, but the nounphrase referring to her is marked by a postposition ( zeon, at), and does notoccupy the role of grammatical subject. For all intents and purposes, axvelebs,abodebs, and other agentless transitive verbs are monovalent. 4

I have so far identified around three dozen agentless transitive verbs inModern Georgian, which are shown in table 3. In semantic terms, Georgianagentless transitive verbs appear to form a coherent group. All the examplesthat I have found denote observablebut usually involuntaryresponses tointernal physiological conditions. One subset clusters around the physiologicalsymptoms of fever or chills: shivering, trembling, delirium, and the like. Anothergroup, semantically less tightly centered, refer to the experience of sharp pains,stomach distress, or muscular discomfort. A third subsetto which I devote par-ticular attention further on in this articlecomprises a small, semanticallyfocused group of verbs denoting different types of audible reaction to internalstimuli: coughing, sneezing, belching, yawning, hiccupping, and vomiting.


6/28


Table 3. Georgian Agentless Transitives and Causatives

FACULTATIVE AGENTLESS TRANSITIVE VERBS

CAUSATIVE

=/ga=macaxcaxebsI tremble=mataxtaxebsI am overcome by quivering, trembling=madzagdzagebsI tremble=madzigdzigebsI tremble ga=/da=mazrialebs (tani)I am overcome by shaking, trembling (in my body) ga=/e=ma rialebs (tani)I am overcome by shuddering, my whole body

tremblesmatrtolebsI am overcome by shaking, tremblinga=/ga=/e=ma r olebs (tani)I am overcome by shuddering, my whole bodytrembles

a=/ga=/e=makankalebsI begin to shake, am overcome by shaking=/a=mababanebsI shiver=/ga=/e=maci(v)ebsI have/get hot and cold spellse=mamcivnebsI shiver (especially from fever)macxelebsI have a fever (from malaria)macxroebs(


7/28


alignment, such as Latin, Icelandic, Russian, or German, agentless transitiveverbs, such as those in (8a)(8c)usually referred to in the literature as imper-sonal transitives, transimpersonals, or accusative-subject verbsassignaccusative case, as do ordinary transitive verbs, but are accompanied by non-referential nominative-case arguments (typically third person singular neuterpronouns) or no nominative-case noun phrase at all (Fay 1917; Lehmann 1974:40; Gonzalz 1984; Babby 1994, 1998; Bowers 2002; Bardal and Eythrsson2003; Creissels 2007).5

(8a) LATIN IMPERSONAL TRANSITIVESme paenitet I regret, repentme pudet I am ashamedme taedet I am weary

(8b) R USSIAN IMPERSONAL TRANSITIVESmenja znobit I feel chilly, feverishmenja rvt I vomit (X rends me)menja tonit I feel sick

(8c) GERMAN IMPERSONAL TRANSITIVES Mich friert (es)I am freezing Mich hungert I am hungry Mich gelstet (es) nach X I have a craving for X

In Amharic, which is double-marking, the single surface argument of agentless transitive verbs is assigned accusative case and governs objectagreement in the verb.

(9) rab (Amharic)hunger. PERF.3MASC-1SG.OBJI am hungry (lit., it hungers me) (Amberber 2000:325, 2005:309)

The phenomenon of accusative-subject predicates has been extensively stud-ied with respect to Icelandic, which, for a West-European language, has anunusually high number of verbs specifying nonnominative subjects (Bardal2001; Eythrsson 2000; Svenonius 2001). Besides monovalent agentless transi-tive verbs, such as (10a), the Icelandic lexicon includes twenty or so bivalentverbs which assign accusative case to both core arguments, the grammaticalsubject as well as the object, such as (10b).

(10a) na lagi. (Icelandic)river. ACC frozeThe river froze. (Bardal 2001:203)


8/28


(10b) Mig dreymdi mmu. (Icelandic)I.ACC dreamed grandma. ACCI dreamed of grandma. (Bardal 2001:203)

The identification of agentless transitive verbs in dependent-marking lan-guages of consistently ergative alignment cannot in principle be based on caseassignment alone, since the single argument of an agentless transitive verbwould receive the same absolutive marking as the subject of an intransitiveverb. Johanna Nichols (p.c. 2005) has nonetheless detected a small number of agentless transitive verbs in the northeast Caucasian language Ingush, shownin table 4. These expressions either employ a transitive auxiliary verb ( loac catch, capture, as in the first example in the table, or C.u make [where C is aclass-agreement marker]), or are transitive verbs that are facultativelyagentless.

Table 4. Ingush Agentless Transitives

maalx/but loac solar/lunar eclipse occurs (lit., X catches sun/moon)dosh d.uuc go on trial, be on trial (lit., X narrates word/matter)

AGENTLESS CONSTRUCTIONS WITH AUXILIARY VERB C.u make [C = class prefix]muq=d.u rust, get rusty, rust through, be rust-eaten, corrodesha=b.u freeze, turn to ice (lit., X makes ice (sha))qeika=d.u cough; be sick, have a cough (qeik cry, shout)sotta=d.u stretch (on waking) (sott bend, curve)loarha=d.u make up ones mind (to); decide, dare (loarh count, respect, consider,decide)

sa=got(ta)=d.ube worried, be upset (sa soul, gotta narrow, cramped)qoa=d.u manage, find time, manage to find time, manage to finish

SOURCE: Johanna Nichols (p.c. 2005).

The number of agentless transitive verbs in languages for which I haveinformation ranges from a handful (three to five) to the two or three dozenascribed to Icelandic (see table 5). More importantly, the semantic characteris-tics of these verbs, in Indo-European and non-Indo-European languages alike,are strongly similar. On the whole, agentless transitive verbs denote sensations,symptoms and changes of state that normally occur spontaneously or withoutthe experiencers volition:

(i) psychophysiological sensations (be cold, hungry, tired )(ii) symptoms of illness or other internal states ( shiver, sneeze, have cramps)(iii) emotional reactions, almost always negative ( feel fear, disgust, shame)(iv) changes of weather, state, bodily health or life-cycle phase ( freeze, rust,

age, die)(v) passive movement (only described for Icelandic, as far as I know: drift, becarried )


9/28


Table 5. Semantic Range of Agentless Transitive Verbs in Various Languages

GEO LAT R US ICE GER AMH SHI NGA INGP SYCHOPHYSICALSENSATIONShunger, thirst + + + +heat, cold + + + + + + +dizziness, delirium + +pain + + + +sleepiness, exhaustion +

OBSERVABLE SYMPTOMSshivering, fever, chills + +audible (cough, belch, etc.) + + +nausea, diarrhea, cramps + + + +paralysis, stiffness + + + +swelling, fatness +stumble, fall + +

EMOTIONS(NEGATIVE)fear, worry, sorrow + + + + +shock, surprise + + +moral reaction + + +need, lack, longing + + boredom +positive emotion (happiness) + +

COGNITIVEmemory +dream + +think, imagine +

SPONTANEOUS CHANGEfreeze, thaw + +rust, rot, curdle + + +life cycle: be born, die, age + +weather, celestial event + +

P ASSIVE MOVEMENT +NOTE: GEO =Georgian, LAT = Latin, R US = Russian, ICE = Icelandic, GER = German,AMH = Amharic, SHI = Shina, NGA= Ngangityemerri, ING = Ingush.

Whereas the inventory of agentless transitive verbs for some languagesspreads over most or all of the semantic fields enumerated above, those of otherlanguages cluster in one or two fields. The five Latin impersonals that assignaccusative case describe negative reactions of an emotional or moral nature. Thehalf-dozen lexical agentless transitive verbs of Russian, like the larger group in

the Georgian lexicon, denote physical symptoms.6

I note in passing that the classification of some of the verbs described aboveas agentless transitive verbs has been contested by certain linguists. Moravcsik


10/28


(1978:24142; see also Plank 1984:352; Wunderlich 2006; Creissels 2007) char-acterizes Latin, German, and Amharic agentless transitive verbs as instances of the extension of accusative markers to intransitive subjects, and as such, com-parable to the split-ergative or split-S alignments described for numerouslanguages (Dixon 1994:73). According to this analysis, a phrase such as me pudet or mich hungert is generated by an intransitive verb that assigns accusa-tive case to its subject. A similar interpretation has been proposed for the classof superficially monovalent verbs in head-marking languages such as Wichita(Rood 1971), Lakhota (Mithun 1991:51418), and Caddo (Mithun 1991:52528),which cross-reference their single argument with object-agreement affixes. InCaddo, for example, the prefix ku cross-references the first person singularpatient of transitive verbs such as (11a), and also the single surface argument of

certain verbs denoting states and involuntary events, such as (11b).(11a) ku :wida:kuhnah (Caddo)

he grabbed me (Mithun 1991:525)

(11b) ku kah w nah (Caddo)I burped (Mithun 1991:527)

Since third person singular subject-agent agreement in these languages is zero,such verbs could equally well be regarded as agentless transitive verbs. I do not

know if there are independent, language-internal grounds in Caddo and theother languages mentioned that would compel analysis of (11b) and similarverbs as intransitive rather than impersonal transitive; they may well, in fact, be indistinguishable for languages of this type. 7

3. Facultative agentless transitive verbs. Georgian agentless transitiveverbs divide into two groups on the basis of morphological and syntacticproperties. Most verbs of the first group are formally causative, as marked bythe transitivizing version vowel a and the series marker eb. Many have ex-pressive roots, as indicated by full or partial reduplication and phonetic sym- bolism (Holisky 1981). Shanidze (1953:19596) qualifies the agentless transitiveverbs of the first group as polysemic, since they allow both bivalent and mono-valent syntactic frames, as in (12a) and (12b). In the bivalent frame in (12a), thenoun phrase denoting the cause of the trembling is marked as a transitive agent,which is assigned ergative case by an active verb in the aorist tense. In (12b), theverb has only one core argument, denoting the one who trembles, whereas thecause is expressed by a postpositional phrase.

(12a) INDIRECT TRANSITIVE, BIVALENT FRAME

ima isev a=makankala.fear-ERG again PV=O1SG-VM-shiver-AOR .S3SGFear made me tremble again. (Terenti Graneli, g ame otaxi)


11/28


(12b) AGENTLESS TRANSITIVE, MONOVALENT FRAMEsicivisagan a=makankala.cold-GEN -from PV=O1SG-VM-shiver-AOR .S3SG

I trembled from the cold.

Most of the bivalent homologues of agentless transitive verbs are associatedwith indirect syntaxthat is, the formal direct object, which tends to refer tohuman experiencers, takes on certain of the syntactic attributes of a gram-matical subject at the expense of the agent, which almost invariably has inani-mate or abstract reference (Tuite 1987). In accordance with the split-ergativepatterning characteristic of Georgian and some of the other Kartvelianlanguages, the case assignment properties of transitive verbs (and some intran-sitives) shift from an accusative alignment in the present series of conjugationalparadigmswhere the dative case doubles as an accusative markerto anergative alignment in the aorist and optative. Bivalent indirect intransitivesconform to this pattern; compare the example with a present-tense verb in (13a),and the same verb in the aorist in (13b).

(13a) kals usiamovno mogoneba a r olebs.woman-DAT unpleasant recollection- NOM VM-shudder-SM-S3SGAn unpleasant memory makes the woman shudder.

(13b) kali usiamovno mogonebam e=a r ola.woman-NOM unpleasant recollection- ERG PV=VM-shudder-AOR .S3SGAn unpleasant memory made the woman shudder.

When the same verbs are employed with monovalent syntactic frames, how-ever, their case-assignment properties change. The single argument is assigneddative case in both the present and aorist series as though it were an indirectobject, as shown in (14a) and (14b).

(14a) xazarulas sicivisagan a r olebs.

xazarula -DAT cold-GEN-from VM-shudder-SM-S3SGThexazarula (name of an apple tree) shudders from the cold.

(14b) xazarulas sicivisagan e=a r ola.xazarula -DAT cold-GEN-from PV=VM-shudder-AOR .S3SGThexazarula shuddered from the cold. (Nodar Dumbadze, Xazarula )

The shift from direct-object to indirect-object marking brings the agentlesstransitive verbs into conformity with the vast majority of Georgian verbs thatdisplay indirect syntax. Most such verbs belong to the passive (class 2) or medio-

passive (class 4) types, which do not undergo case shift. The morphological subject is assigned nominative case and the morphological indirect objectwhich,in the case of indirect verbs, receives the syntactic attributes of subjecthoodis


12/28


assigned dative case. The case-assigning properties of the two types of indirecttransitives are juxtaposed to those of indirect passive and mediopassive verbs intable 6.8 (A typical construction with bivalent indirect passive would be gogosdeda enatreba[girl-DAT mother. NOM VM-miss-SM-S3SG] the girl-DAT missesher mother; a typical monovalent indirect passive would be gogos emg ereba[girl-DAT VM-sing-SM-S3SG] the girl-DAT feels like singing.)

Table 6. Case Assignment by Georgian Indirect Verbs

INDIR . TRANS.(BIVALENT)*

AGENTLESS(TRANS.)**

BIVALENT (MEDIO)-PASS.

MONOVALENT INDIR .(MEDIO)-PASS.

SERIES agent patient agent patient theme experiencer theme experiencer

present NOM DAT DAT NOM DAT DATaorist ERG NOM DAT NOM DAT DAT

perfect (DAT NOM) ( DAT) NOM DAT DAT

NOTE: The syntactic subject is marked by underlining. Case marking for indirecttransitives that contrasts with that of other verbs (aorist series) is in bold type. Indir. =indirect; pass. = passive; trans. = transitive.*The indirect transitive frame is seen in (13a) and (13b).** The agentless frame is seen in (14a) and (14b).

The phenomenon described above seems not to have spread to agentlesstransitives in wish and curse formulas, which are generated by the impersonaloptative construction described by Suxishvili (1979; 1986:9093) and Amiridze(2005). In such formulasparticularly common in the highland dialects of northern Georgiathe optative particle netavi or its variants is combined witha transitive verb in the aorist tense, as in (15a). The agent of the verb wasformerly expressed as an indefinite pronoun ( vin, traces of which subsist in theparticle netavi, netain < netar blessed +vin someone [Shanidze 1953:636]), but its presence is no longer apparent to speakers, and, in fact, the particle can be dispensed with entirely, as in (15b). Impersonal optatives, which could beclassified as quasi-agentless transitivesin order to distinguish them fromthose which undergo direct-object-to-indirect-object shift as in (14b)can becreated in principle from any verb stem.

(15a) netain ma=mkla mtaia, da=mmarxaOPT PV=O1SG-kill-AOR .S3SG mountain-in PV=O1SG-bury-AOR .S3SG

bunebaia.nature-in

May I die in the mountains, may I be buried in nature. (lit., may [someone] killme . . . bury me ) (Xornauli 1949:216; translation from Tuite 1994:121)


13/28


(15b) daokili mexveceboda erti d g e entan mamqopa .kneeling he.begged.me one day you-with O1SG-VM-be-AOR .S3SGOn his knees he begged me, May I spend one day with you! (Giorgi Leonidze,

Pupala )

As would be expected on pragmatic grounds, the beneficiary of wish formu-las is generally in the first, or less often, the second person. Impersonal optativeconstructions with a single third-person noun phrase are very rare in spon-taneous speech or literature. In the few examples I have come across, the singlenoun phrase receives direct object coding (i.e., nominative case).

(16) netamc ki okros tavtavad gada=akcia qvela o! OPT indeed golden ear-ADV PV=VM-turn- AOR .S3SG everyone: NOM -QUOT

Mayeveryone turn into golden ears (of grain)! (io Mg vimeli, Okros tavtavi)

4. Lexical agentless causatives and syntactic flip-flop. The ten or soverbs composing the second group of agentless transitive verbs in table 3 arelexically agentless, that is, they are always monovalent. Although membershipin this group is determined by morphosyntactic criteria, the verbs clusteraround a semantic prototype as well: audible actions of the mouth or nose,sometimes accompanied by the expulsion of a gaseous or liquid substance, andusually involuntarily provoked by internal states of the body, although most of

the denoted actions can be controlled by the subject. All lexical agentlesstransitives known to me are morphologically causative. Unlike the other types of agentless transitive verbs, and indeed unlike any other verb type described forGeorgian, the single argument of lexical agentless transitive verbs can flip frommorphological subject to morphological object status (Melikishvili 2001:117, 240; Jorbenadze 2006:2639). The verb meaning yawn, for instance, has the twoparadigms in the present tense shown in table 7.

Table 7. Alternative Present-Tense Paradigms of Yawn

INDIRECT CONJUGATION DIRECT CONJUGATIONme m amtknarebs me vamtknareb I yawnI/me O 1 SG VMyawnSMS3SG I/me S 1 -VM-yawn-SMen gamtknarebs en amtknareb you (sg.) yawnmas amtknarebs is amtknarebs he/she yawns

What is so unusual about the two paradigms in table 7 is not their form, butrather their near equivalence in meaning. The single argument in each instancedenotes the one who yawns. Formally comparable paradigms can also be formed

from ordinary causative verbs, but the role of the argument varies according tothe agreement marker it governs, as in (17a) and (17b).


14/28


(17a) me m amg erebs (is). I/me O 1 SG -VM-sing-SM-S3SG (she/he. NOM)She/he makes me sing.

(17b) me vamg ereb (mas).I/me S 1 -VM-sing-SM (she/he. DAT)I make him/her sing.

All eleven lexical agentless transitive verbs known to me can undergosyntactic flip-flop as in table 7. The alternation between the indirect and directconjugations has certain semantic implications. According to Nino Amiridze,whereas direct-syntax forms can be used for an unintentional ( uneburi) actionas well as for sneezing or coughing on purpose, the indirect-syntax forms areonly for unintentional [actions] (p.c. 2005). Note the contrast between theindirect and direct uses of axvelebscoughs in the two examples in (18) and (19), both from the play ebindebidan gatenebamde by Bao Kvirtia. On the otherhand, the direct use of the verb in (20), from an antismoking tract posted on theweb page of the Georgian Orthodox patriarchate, does not appear to becorrelated with intentional coughing.

(18) INDIRECTcamosvlis d g es gaciebuli vqopilvar; sainlad

leaving-GEN day-DAT chilled I.was.PERF terribly m axvelebda. O 1 SG -VM-cough-SM-IMP -S3SGThe day I was to leave I had a cold; I was coughing terribly.

(19) DIRECTdidi kacivit mui vaxveleb. big man-like palm-in S 1 -VM-cough-SMLike an adult, I cough into my hand.

(20) DIRECTd g es gacilebit uket vgrdznob tavs, ag arc vaxveleb datoday considerably better I.feel myself no.longer S 1-VM-cough-SM and

ag arc naxveli maxrobs.no.longer coughed it.chokes.me

Today I feel much better, I am no longer coughing and no longer gagging onsputum. (http://www.patriarchate.ge/su/312/7text.htm)

According to the small sample of Georgian speakers whom I consulted, the

indirect conjugations of lexical agentless transitive verbs tend to be morefrequent in the imperfective paradigms, such as the present and imperfect, thanin the aorist-series paradigms (see table 8). 9 Perfect-series paradigms formed


15/28


from indirect lexical agentless transitive verbs are uniformly judged unaccept-able. According to Jorbenadze (2006:29), some speakers consistently employ theindirect conjugation in the present and imperfect tenses, and the direct conjuga-tion in the aorist. A sampling of Georgian texts found on the internet (mostlychat groups) implied that actual usage frequencies might vary significantly fromone verb to another, but the greater likelihood of the aorist to be used in thedirect conjugation was confirmed (see table 9).

Table 8. Frequency and Acceptability of Perfect-Series Paradigms for LexicalAgentless Transitive Verbs

TENSE INDIRECT CONJUGATION DIRECT CONJUGATIONpresent m axvelebs > vaxveleb I coughaorist da=m axvela < da=vaxvele I coughedpresent perfect (*da=vuxvelebivar) da=m ixvelebiaI have coughed

NOTE: > = more frequent than; < = less frequent than.

Table 9. Internet Survey of Georgian Lexical Agentless Transitive Verbs (FirstPerson Singular Only), February 2010

VERB TENSE INDIRECT DIRECTI yawn(ed) [mtknar] present 56 87

imperfect 11 21aorist 29 57

I cough(ed) [xvel] present 155 76imperfect 47 35aorist 25 29

I hiccup(ed) [slokin] present 54 19imperfect 32 1aorist 8 4

Superficially similar phenomena have been described in other languages.One might juxtapose the syntactic flip-flop of Georgian lexical agentless transi-

tive verbs to the shift from indirect to direct syntax for so-called psych-verbs inMiddle English, such as the oft-discussed transition from am cynge licodon peran [the.DAT king.DAT liked.3PL pears. NOM] to its modern equivalent the king liked pears (Trask 1996:13839). In reality, the two cases have different causes.The shift from indirect to direct syntax in English and some other Germaniclanguages was associated with the erosion and eventual loss of case suffixes andthe increasingly rigid preference for subject-verb-object word order. Nothing of the sort is happening in Georgian. Its double-marking morphosyntax has beenremarkably stable since the earliest attestation of the language fifteen centuries

ago, and not even the slightest indication of a generalized syntactic drift awayfrom indirect constructions can be detected in any Georgian dialect. Dative-subject verbs number in the hundreds, and new ones are easily (and frequently)


16/28


added to the lexicon. They are also among the very first verbs acquired bychildren (Imedadze and Tuite 1992:6364).

Typical causatives, like amg erebsX makes Y sing in (7), as well as facul-tative agentless transitive verbs such as akankalebsX makes Y shake; Yshakes in (12), are derived from intransitive medial (medioactive) verbs, orsometimes nouns or adjectives. Their stems also appear in inchoative-intransitive verbs such as amg erdebabegins to sing, akankaldebabegins to shake (see table 10). Compared to other types of causatives, the lexi-cal agentless transitive verbs are derivationally rather isolated. Except forslokinebshiccups,boqinebsburps, and the obsolete iqwels (ixvels)coughs,attested in medieval medical texts (Panaskerteli-Cicivili 1978), Georgianlexical agentless transitive verbs lack medioactive or inchoative counterparts

(see table 10).10

Table 10. Georgian Agentless Transitive and Associated Verb Forms

MEDIOACTIVE INCHOATIVE INTRANSITIVE CAUSATIVEI am X-ing I will begin X-ing (Y) makes me X

ORDINARY MEDIOACTIVEsing vmg eri(var) a=vmg erdebi mamg erebs

FACULTATIVE AGENTLESS TRANSITIVE

tremble vcaxcaxeb a=vcaxcaxdebi macaxcaxebsquake vdzagdzageb a=vdzagdzagdebi madzagdzagebsshake vkankaleb a=vkankaldebi makankalebsquiver vtrti a=vtrtoldebi matrtolebs

LEXICAL AGENTLESS TRANSITIVEhiccup vslokineb a=vslokindebi maslokinebscough (vixvel) a=vxveldebi maxvelebsbelch vboqineb a=vboqindebi maboqinebsbelch maloqinebsyawn mamtknarebssneeze maceminebssneeze macxikvebsvomit marcqevsvomit mag ebinebs

How did syntactic flip-flop arise? The near-synonymy of maxvelebs andvaxvelebis unlike anything else in Georgian morphosyntax. 11 This is not to saythat the alternation between direct and indirect syntax is limited to the smallgroup of lexical agentless transitive verbs. Quite a few Modern Georgian verb

types, mostly bivalent intransitives but also a few dozen transitives, allow eitherof their principal arguments to accrue some or all of the attributes of syntacticsubjecthood, such as the capacity to bind reflexive and reciprocal pronouns and


17/28


number agreement with the verb in all three persons (Tuite 1998:3639). Con-sider the following syntactic minimal pair, in which the argument having mostsubject properties is indicated by underlining. In (21a), the nominative-caseargument (intransitive subject) has the properties of syntactic subjecthood, andcontrols plural number agreement in the verb. In (21b), the roles are reversed:dedebs mothers, formally an indirect object, has most of the privileges of syntactic subjecthood, including the control of number agreement in the thirdperson as indicted by the choice of suffix. In the following examples, ian agreeswith the third person nominative noun phrase, whereas tagrees with the thirdperson dative noun phrase.

(21a) DIRECTbavvebi emalebian dedebs.child-PL-NOM O3.VM-hide-SM-S 3PL mother-PL-DAT

The children are hiding from (their) mothers (elicited)

(21b) INDIRECTdedebs emalebat bavvebi.mother-PL-DAT O 3 .VM-hide-SM-S3SG-PL child-PL-NOM

The children are hiding from (their) mothers (elicited)

The change of meaning between the direct and indirect variants of emaleba

X hides from Y is difficult to characterize precisely. According to native speak-ers consulted in Tbilisi, the direct variant implies intentional activity on thepart of the hiders, whereas the indirect variant does not. A similar semanticentailment was noted earlier for the direct variants of lexical agentless transi-tive verbs. The crucial difference between instances of direct-indirect syntacticalternation such as (21a)(21b) and the flip-flop associated with lexical agent-less transitive verbs is at the level of argument structure. In both (21a) and(21b), the nominative noun phrase denotes the ones who hide, and the dativenoun phrase denotes those from whom they hide. Though the shift in subject-hood between the two arguments is linked to the meaning difference mentionedabove, and sometimes a shift in discursive focus or empathy as well, thethematic roles remain unchanged. Lexical agentless transitive verbs, beingmonovalent, specify a single argumentdesignating the one who coughs, yawns,etc.but both the thematic and syntactic roles linked to that argument changefrom one variant to the other.

According to information supplied by Lela Samushia (University of Frank-furt), the closely related Mingrelian language does not have syntactic flip-flopas such. Mingrelian lexical agentless transitive verbs, some of them based onroots cognate with those of Georgian lexical agentless transitive verbs (e.g.,moxvalapuansI cough [cf. Geo.maxvelebs]; moikinapuansIhiccup [cf. Geo.maslokinebs]), are indeed in contrast with direct-syntaxconstructions, but these latter are medioactive rather than causative in form


18/28


(xvaluns coughs, ikinunshiccups; see also Kajaia [2001, 1:255; 2001,3:260, 537]). Comparative evidence from Mingrelian, therefore, would lead oneto expect that the original alternation was between a medioactive (such as theobsolete vixvelI cough) and an indirect causative ( maxvelebsit makes mecough > I cough). If this was the case, then it is the origin of the direct-syntaxvariant ( vaxveleb) that requires explanation.

Unfortunately, but understandably, verbs meaning belch, hiccup, andvomit are not especially frequent in the textual genres composing the bulkof the medieval Georgian corpus. A preliminary search in Georgian medicalmanuals, such as the fifteenth-century Samkurnalo cigni Book of healing(Panaskerteli-Cicivili 1978) has yielded a handful of examples, but so far notnearly enough to begin sketching out the history of how vaxveleband similar

causatives came to be interpreted as synonymous with the simple medioactivesthey later supplanted.

Table 11. Georgian and Mingrelian Lexical Agentless Transitives andAssociated Medioactives

AGENTLESS TRANSITIVE MEDIOACTIVE belch Georgian aboqinebs boqinobs

Mingrelian obo inapuans bo inunshiccup Georgian aslokinebs slokinebs

Mingrelian oikinapuans ikinunscough Georgian axvelebs (ixvels)

Mingrelian oxvalapuans xvalunssneeze Georgian acxikvebs

Mingrelian oionapuans ionuns

SOURCE: Lela Samushia (p.c. 2005); Kajaia (2001).

Semantics may provide the crucial clue. The Georgian lexical agentlesstransitive verbs comprise a small set of verbs with strongly similar meanings:

all denote some sort of expulsion of sound or fluid (or both) from the mouth. Assuch, they resemble what Blake (1994:69) called bodily-emanation predicates.In many languages, verbs of this type have unexpressed or optional patientarguments denoting the emanating substance or phenomenon, and the case-assigning properties of transitives. Several lexical agentless transitive verbs arein fact commonly used with such object noun phrases, e.g., amo=vaxvelebIcough something up and amo=varcqevI vomit up (blood, food), as in (22a)and (22b). By contrast, the facultative agentless transitive verbs, which denotea condition manifested by motion (trembling or shivering) or an internal sensa-tion of cold, heat, or pain, do not subcategorize for a patient argument of thiskind.


19/28


(22a) zogi haers iolad amo=aboqinebs.some-NOM air-DAT easily PV=VM-burp-SM-S3SGSome (babies) burp up air easily. (chat-group)

(22b) sisxli amo=vaxvele. blood-NOM PV=S1-VM-cough-AOR .S1SGI coughed up blood.

Georgian bodily-emanation verbs can also be used without an expressedobject; this is almost always the case for the two Georgian verbs denoting theexpulsion of intestinal gas: ga=akuebsfart audibly and ga=acuebsfartsilently, both of which have the form of causatives but rarely if ever appear withan overt direct object.

If my hypothesis comes close to the truth, then maxvelebsand vaxveleb had their origins in distinct constructions with different argument structures.The former variant would have had the same structure as makankalebsI am shivering and the other facultative agentless transitive verbs, and wouldhave undergone the shift of direct object to indirect object, as in (23a). Thedirect-conjugation variant ( vaxveleb) evolved from a semitransitive bodily-emanation predicate of the kind described by Blake, with an underlying agentargument and an optional (and almost always unexpressed) patient. The directcausative vaxveleb,due to its frequent use without an explicit direct object,

expanded into the semantic range of the medioactive vixvelI cough, andeventually supplanted it, as in (23b).

(23a) maxvelebsit makes me cough > I cough

(23b) (amo=)vaxvelebI cough up something > I cough(displaces vixvelI cough)

As a consequence, a small, semantically coherent verb class emerged, with anunusual type of formal opposition between direct- and indirect-conjugationparadigms.

5. Conclusion: syntactically marginal verb classes in Georgian. In theshadows of the major formal categories of Georgian verbs lurks a bestiary of minor verb classes that share the same stem morphology as the major classes, but differ in their syntactic properties. Alongside the large and productivecategory of prefixal class 2 (passive) verbs, one encounters such curiosities asthe so-called deponents (which are passive in form, but have the syntactic andsemantic properties of antipassives [Tuite 2002, 2007]), and a small but poten-tially open class of monovalent indirect verbs that signify that the referent of thedative-case noun phrase feels like doing the activity or enjoying the substance


20/28


denoted by the verb root, e.g., meokoladebaI have a craving for chocolate(Shanidze 1953:299301). What sets these minor classes apart from thedominant ones is the mapping of actant roles (or thematic relations) ontogrammatical roles. The marginal verb types assign subjecthood to an argumenttype that the majority of verbs of their morphological class would assign to adifferent syntactic role, or even leave unexpressed. For example, the subject of atypical class 2 verb in i is the patient of the corresponding active verb, whereasthe agent is either omitted or relegated to a postpositional phrase. It is preciselythe agent, by contrast, that serves as the subject of a class 2 deponent. Of thetwo types of sneezing-coughing verbs, the direct-conjugation variant representsa case of syntactic marginality of the same kind as that just described. Theindirect-conjugation variant, on the other hand, retains the same mapping of

thematic relations to grammatical roles as the Georgian transitive causative,albeit with omission of the agent. The syntactically marginal class 2 and class 1verb types are compared in tables 12 and 13.

Table 12. Two Georgian Verb Classes and Their Syntactic Variants

CLASS2 PREFIXAL INTRANSITIVES CLASS1 CAUSATIVE TRANSITIVESAGENT P ATIENT CAUSER EMITTER /EFFECTOR *

P ASSIVE (bkgrd) S/subj. class 1 transitive S/subj. ODEPONENT S/subj. (bkgrd) ATV (maxvelebs) O/subj.

vaxveleb S/subj.

NOTE: S, O = person markers; subj. = syntactic subjecthood; (bkgrd) = backgrounded, i.e.,demoted to oblique object status or unexpressed; ATV = agentless transitive verb.*These terms for actant roles have been adopted from Van Valin (2002).

Table 13. Semantic Properties of Syntactic Variants (Deponents and Class 1Monovalents

SYNTACTIC VARIANT BASIC VERB CLASS SEMANTICS CONTRASTING FORMdeponent class 2 passive antipassive class 1 transitive or(vig eebiI (vibadebiI (focus on contours class 3 medial

[AGENT] chew, [PATIENT ] am born) of activity as (vg eavI [AGENT]masticate) characteristic chew, masticate sthg)of agent)

class 1 monovalent class 1 causative nearly synon- class 1 agentless(vaxvelebI (vatirebI ymous with transitive[EMITTER ] cough) [CAUSER ] make sb agentless trans- ( maxvelebsI

[EMITTER / EFFECTOR ] tive, but can also [EMITTER ] coughcry) denote inten- unintentionally)

tional coughing

In the contours of emerging paradigms one can detect evidence of diachronicshifts observable elsewhere in Georgian morphosyntax over the fifteen centuriesof documented history of the language. It was mentioned above (see section 4)


21/28


that the single argument of Georgian agentless transitives assumes the pro-perties of an indirect object. Readers familiar with Germanic historical syntaxmight wonder if such a shift from direct object to indirect object marking wouldindicate that Georgian is suffering from a case-assignment malady similar tothe dative sickness said to afflict Icelandic and related languages (Eythrsson2000; Bardal and Eythrsson 2003). This term, invented by prescriptive gram-marians, labels the tendency for impersonal transitives that once marked theirprimary argument with accusative case (e.g., German mich hungert, michdrstet) to assign dative instead (* mir hungert, *mir drstet). Any such cross-linguistic comparison must, however, take into account certain changes under-gone by the Georgian dialects over the past several hundred years, many of them manifestations of a drift away from the ergative-absolutive alignment

apparent in Old Georgian morphology and syntax (Harris 1985). One suchchange is the increasing morphosyntactic prominence of core arguments withhuman reference, accompanied by a decline in the prominence of the nounphrase assigned nominative case, especially when the latter has the status of direct object (Tuite 1998). A possible correlate of the greater prominence of animacy might be a marked dispreference in Modern Georgian for assigning therole of patient to noun phrases with human reference when the agent is demotedor absent. Whereas the patients of ordinary transitive verbs very commonlyhave human reference in Georgian, this is not the case when there is no agentamong the core arguments of the clause. Shanidze (1953:29091) observed agrowing tendency for Georgian class 2 verbs (the verb class traditionally labeledpassive) to allow a genuinely passive interpretationin the strict sense of apatient promoted to subjecthood and a demoted agentonly when the patientnoun phrase has nonhuman reference. The Old Georgian corpus containsnumerous instances of class 2 verbs with human subjects that have passivemeaning, such as mo=iklahe was killed (2 Kings 11:26) ormi=iqvanahewas taken away (Luke 16:22). Such constructions are increasingly rare, if notimpossible, in the modern literary language. 12 The assignment of indirect objectstatus to the single argument of agentless transitive verbs (which almost alwayshas human or at least animate reference) manifests a similar avoidance by ani-mate noun phrases of patienthoodor at least, avoidance of the direct objectsyntactic role assigned to patients by transitive verbswhen there is no agent,whether due to syntactic demotion (as in passives) or absence from the initialcase frame (as in agentless transitive verbs).

This phenomenon can also be interpreted as further evidence of the morpho-syntactic marginality of direct objects in comparison to indirect objects in Geor-gian. Third-person direct objects no longer govern agreement in most varieties of Modern Georgian. When both are present in the clause, indirect objects are

favored over direct objects in competition for the preverbal object-agreement slotin the verb (Boeder 2002:9698).13 In general, the Georgian indirect object playsa particularly prominent role in both clause-internal and clause-external syntax.


22/28


1. In classifications such as Aronsons, the passive class includes some of Shanidzes mediopassives, whereas the remainder are assigned to the indirect class.Aronson's middle verbs correspond to Shanidzes medioactives.

Many dative-case noun phrases function as syntactic subjects outright, andmany others share at least some features with subjects: their tendency to referto humans, to appear toward the front of the clause, and to govern agreement inthe verb. Especially in contrast to the particularly close relation between thedirect object (more precisely, the absolutive argument) and the verb, indirectobjects come across as tantamount to secondary subjects. When no initial ex-ternal argument is assigned by the verbas can be assumed for the large number of passive and mediopassive verbs that have formal indirect objectstheindirect object takes on all or most privileges of subjecthood, while at the sametime signaling, through its dative marking, various shades of contrast to canon-ical agenthood (experiencer or beneficiary role, decreased volitionality, indirectevidentiality, etc.). 14

As phenomena emergent from and renewed by the communicative practiceof a speech community, the association of verb morphology to argument struc-ture, and the grouping of verb forms into paradigms (i.e., the perception thata suite of forms pertain to the same verb), are susceptible to change, recon-figuration, or the spawning of new form-meaning links. In earlier writings, Ilikened the emergence of new verb paradigms in Kartvelian to the coalescence of planets and moons from swirling clouds of dust (Tuite 1996). I still find thesimile useful, although now I would modify it to allow the seemingly solid bodiesrepresenting well-established verb classes to occasionally lose matter, which inturn can condense into smaller objects or be drawn into new formations bymaterial from other planets.

Notes

Acknowledgments. The initial version of this article was read at the 2005 CentralEurasian Studies Society conference. Revised versions were presented at the TypologySeminar of the Oriental Institute of the Georgian Academy of Sciences (March 2006), andthe Linguistics Department of the Max-Planck Institute for Evolutionary Anthropology(April 2009). I wish to express my deepest thanks for the help and data offered by NinoAmiridze, John Colarusso, Johanna Nichols, Lela Samushia, Medea Saghliani, RusudanIoseliani, Winfried Boeder, Bernard Comrie and an anonymous reader for Anthro- pological Linguistics.

Transcription . The transcription used here is for the most part consistent with theconventions favored by specialists in Caucasian linguistics. Ejective occlusives aremarked by an apostrophe. The symbol j represents the voiced palato-alveolar affricate, asin English judge.

Abbreviations. The following grammatical abbreviations are used: ACC= accusative;ADV= adverbial case; AOR = aorist; DAT = dative; ERG = ergative; GEN = genitive; IMP =imperfect; NOM = nominative; O1SG object marker (first person singular); OBJ = object;OPT = optative particle; PL = plural; PRT = particle; PV = preverb; QUOT = quotative; S3SG =subject marker (third person singular); SM = series marker; VM= version marker.


23/28


2. Harris (1981:131) makes a similar argument, although on the basis of differentpresuppositions about the relationship between deep and surface structure.

3. On the Kartvelian grammatical category traditionally called version, which inmany respects corresponds to middle-voice and applicative categories in other languages,see, among others, Boeder (1968), Aronson (1982b), and Lacroix (2009).

4. Not everyone would agree that Georgian agentless transitive verbs are mono-valent. Hewitt (2008:9798) postulates the existence of phantom agentsdenoting themedical condition, Providence, the circumstances, or whatever a folk-semantictheory would hold responsible for the condition describedsomewhere in the deepsemantic structure of these verbs. Such an account, however, fails to provide an explana-tion for the shift in case-assignment behavior discussed below and illustrated in (14b).

5. I note in passing that impersonal constructions are discussed in a genre of linguistic literature distinct from that employed in this article, in connection with theiralleged link to certain facets of the mentality or worldview of speakers of languages inwhich such constructions occur frequently (Sriot 2000).

6. Here are glosses of agentless transitive verbs reported for three more languages:(i) Amharic (Bender and Fulass 1978): worry, be tired, be sick, feel gloomy, bleed(nose), be bored, be comfortable, be thirsty, yawn, have a cramp, stumble, bedisturbed; (ii) Ngangityemerri (Australia; Reid 2000): feel sad, feel shamed, be cold,be happy, need to get ones breath back, have a toothache, feel ill at ease, feeluncomfortable talking together; (iii) Shina (Indic; Hook and Zia 2005): feel hungry,thirsty, bored, ashamed, afraid, hot, cold, weak, cloyed, dizzy, exhausted;become old, fat, blind, paralyzed, startled, restless, tormented, rusty, gassy,fed up [with X]; stumble, fall, swell up, curdle, shine, die.

7. On the distinction between impersonal transitives and intransitives with obliquesubjects, and instances where the former has given rise to the latter, see Malchukov(2008).

8. Transitive verbs in the present perfect and pluperfect (series III) undergoinversion of the case and agreement marking assigned to their agents and patients.The agent noun phrase receives dative case and controls object agreement in the verb.Series III forms of agentless transitive verbs are accepted by at least some speakers(albeit rejected by Melikishvili [2001:240]), but without genuine inversion.

9. As is the case with facultative agentless transitive verbs, lexical agentless tran-sitive verbs assign dative case to their single noun phrase in the aorist series, that is, it iscoded as an indirect object rather than a direct object, as in (i).

(i) mas da=axvela.she/he. DAT PV=VM-cough-AOR .S3SG She/he coughed. (Lela Samushia p.c. 2005)

10. Whereas many medioactive verbs are also accompanied by an indirectintransitive derived from the same root meaning feel like X-ing (e.g., memg erebaIfeel like singing), almost no Georgian agentless transitive verbs have such counterparts.This might well reflect a perceived incompatibility between the blind psychophysicalcompulsion to shake, tremble, vomit, etc., signaled by agentless transitive verbs, and thetypical entailments of indirect passives in e, which are used to indicate the perception of a quality (mebevrebait seems a lot to me), or the possibility or desire to act in acertain way (Shanidze 1953:299301).

11. Constructions that resemble Georgian flip-flop, at least at first glance, do occurin other languages (John Colarusso p.c. 2010). Some languages of the North Caucasushave a sort of antipassive construction that results in an inversion of the cases assigned


24/28


to the principal clausal arguments. Ordinary transitive verbs in these languages assignergative case to their subjects and absolutive case to their direct objects, whereas theirantipassive counterparts assign absolutive case to their subjects, and an oblique caseidentical to the ergative case to their direct objects. An especially striking instance is theDargi antipassive verb form, which has no special morpheme distinguishing it from thecorresponding transitive (Colarusso 1992:17778; Berg 2005:178; Hewitt 2005:12325).Much of the similarity of these constructions to Georgian maxvelebs/ vaxveleb pairs,however, results from language-specific morphophonemic rules relating to case para-digms and person agreement.

12. Modern Georgian class 2 verbs certainly can be used with human subjects, butthis almost invariably imposes a middle or even active interpretation. This trend maywell have contributed to the growth of the minor class of deponents mentioned above.

13. Plank characterized the canonical direct object as the polar opposite of themost active participant (agent) specified by bivalent or trivalent predicates, in the sensethat it denotes the participant which is least active, completely under the control/influence of the agent (1984:34345). The Modern Georgian direct object, it would ap-pear, is very much a polar opposite in Planks sense, in that it requires the copresenceof an agent in argument structure; otherwise it undergoes movement to subject orindirect object position.

14. It is noteworthy in this respect that some linguists working within the Minimal-ist tradition have sought to capture the syntactic prominence of indirect objects anddatives in many languages by assigning them configurational positions outside of theverb phrase such that indirect objects are treated as external arguments of a sort, albeitnot as external as syntactic subjects (see, among others, Pylkknen [2000] on highapplicative constructions, McGinnis [1998], and Woolford [2006]).

References Amberber, Mengistu

2000 Valency-Changing and Valency-Encoding Devices in Amharic. In ChangingValency: Case Studies in Transitivity, edited by R. M. W. Dixon andAlexandra Aikhenvald, 31232. Cambridge: Cambridge University Press.

2002 Verb Classes and Transitivity in Amharic. Munich: Lincom Europa.2005 Differential Subject Marking in Amharic. In Competition and Variation in

Natural Languages: The Case for Case, edited by Mengistu Amberber andHelen de Hoop, 295320. Amsterdam: Elsevier.

Amiridze, Nino2005 Mismatches between the Form and the Meaning: Reanalyzing Wish and

Blessing Formulae in Georgian. MS, Utrecht Institute of Linguistics.Aronson, Howard I.

1982a Georgian: A Reading Grammar. Columbus, Ohio: Slavica.1982b On the Status of Version as a Grammatical Category in Georgian. Folia

Slavica 5:6680.Babby, Leonard H.

1994 A Theta-Theoretic Analysis of Adversity Impersonal Sentences in Russian. In Annual Workshop on Formal Approaches to Slavic Linguistics: The MITMeeting, 1993, edited by Sergey Avrutin, Steven Franks, and LjiljanaProgovac, 2567. Ann Arbor, Mich.: Michigan Slavic Publications.

1998 Voice and Diathesis in Slavic. Paper presented at the Workshop on Com-parative Slavic Morphosyntax: The State of the Art, Indiana University,Bloomington, 57 June 1998. (www.indiana.edu/~slavconf/linguistics/ babby.pdf).


25/28


Bardal, Jhanna2001 Case in Icelandic: A Synchronic, Diachronic, and Comparative Approach.

Ph.D. diss., Lundastudier i Nordisk sprkvetenskap A57. Department of Scandinavian Languages, Lund.

Bardal, Jhanna, and Thrhallur Eythrsson2003 The Change that Never Happened: The Story of Oblique Subjects. Journal

of Linguistics 39:43972.Bender, M. Lionel, and Hailu Fulass

1978 Amharic Verb Morphology: A Generative Approach. East Lansing: AfricanStudies Center, Michigan State University.

Berg, Helma van den2005 The East Caucasian Language Family. Lingua 115:14790.

Blake, Barry J.1994 Case. Cambridge: Cambridge University Press.

Boeder, Winfried1968 ber die Versionen des georgischen Verbs. Folia Linguistica 2:82151.2002 Syntax and Morphology of Polysynthesis in the Georgian Verb. In Problems

of Polysynthesis, edited by Nicholas Evans and Hans-Jrgen Sasse, 87111. Berlin: Akademie Verlag.

Bowers, John2002 Transitivity. Linguistic Inquiry 33(2):183224.

Cherchi, Marcello1997 Modern Georgian Morphosyntax. A Grammatico-Categorial Hierarchy-

Based Analysis with Special Reference to Indirect Verbs and Passives of State. Wiesbaden: Harrassowitz.

Colarusso, John1992 A Grammar of the Kabardian Language. Calgary: University of Calgary

Press.Creissels, Denis

2007 Impersonal and Anti-Impersonal Constructions: A Typological Approach.MS, Universit Lumire (Lyon 2).

Dixon, R. M. W.1994 Ergativity. Cambridge: Cambridge University Press.

Eythrsson, Thrhallur2000 Dative vs. Nominative: Changes in Quirky Subjects in Icelandic. Leeds

Working Papers in Linguistics and Phonetics 8, edited by Diane Nelson andPaul Foulkes, 2744. Leeds: Department of Linguistics and Phonetics, Uni-versity of Leeds.

Fay, Edwin W.1917 Syntax and Etymology: The Impersonals of Emotion. The Classical

Quarterly 11(2): 8893.Gonzlez, Jos Antonio Enrquez

1984 Verbos impersonales del sentimiento. Faventia: Revista de filologia clssica6(2):2532.

Harris, Alice C.1981 Georgian Syntax: A Study in Relational Grammar. Cambridge: Cambridge

University Press.1985 Diachronic Syntax: The Kartvelian Case. Orlando, Fla.: Academic Press.

Hewitt, George2005 North West Caucasian. Lingua 115:91145.2008 Cases, Arguments, Verbs in Abkhaz, Georgian, and Mingrelian. In Case

and Grammatical Relations: Studies in Honor of Bernard Comrie, edited


26/28


by Greville Corbett and Michael Noonan, 75104. Amsterdam: JohnBenjamins.

Holisky, Dee Ann1981 Aspect and Georgian Medial Verbs. Delmar, N.Y.: Caravan Books.

Hook, Peter Edwin, and Muhammad Amin Zia2005 Searching for the Goddess: A Study of Sensory and Other Impersonal

Causative Expressions in the Shina of Gilgit. In The Yearbook of SouthAsian Languages and Linguistics: Special Issue on South Asian Syntax,2005, edited by Rajendra Singh and Tanmoy Bhattacharya, 16588. Berlin:Mouton de Gruyter.

Imedadze, Natela, and Kevin Tuite1992 The Acquisition of Georgian. In The Crosslinguistic Study of Language

Acquisition: Volume 3, edited by Dan Isaac Slobin, 39109. Hillside, N.J.: L.Erlbaum Associates.

Jorbenadze, Besarion1985 kartuli enis pakultatiuri gramatikis sakitxebi [Issues in the Facul-

tative Grammar of Georgian]. Tbilisi: Tbilisis Saxelmcipo UniversitetisGamomcemloba.

Jorbenadze, Nino2006 saualo gvaris zmnata u g lebis sistema tanamedrove kartuli [The

Conjugational System of Middle-Voice Verbs in Modern Georgian]. Ph.D.diss., I. Javaxivili Tbilisi State University.

Kajaia, Otar2001 megrul-kartuli leksikoni [Mingrelian-Georgian Dictionary]. Tbilisi:

Nekeri.Lacroix, Ren

2009 Description du dialecte laze dArhavi: Grammaire et textes. Ph.D. diss.,Universit Lumire (Lyon 2).

Lehmann, Winfred P.1974 Proto-Indo-European Syntax. Austin: University of Texas Press.

Malchukov, Andrej2008 Split Intransitives, Experiencer Objects and Transimpersonal Construc-

tions: (Re-)Establishing the Connection. In Typology of SemanticAlignment, edited by Mark Donohue and Sren Wichmann, 76101. Oxford:Oxford University Press.

McGinnis, Martha1998 Locality and Inert Case. In Proceedings of North East Linguistic Society 28:

University of Toronto, edited by Pius Tamanji and Kiyomi Kusumoto, 26781. Amherst, Mass.: Graduate Linguistic Student Association, Departmentof Linguistics, University of Massachusetts.

Melikivili, Damana2001 kartuli zmnis u g lebis sistema [The Conjugational System of the Georgian

Verb]. Tbilisi: Logos Press.Mithun, Marianne

1991 Active/Agentive Case Marking and Its Motivations. Language 67:51046.Moravcsik, Edith

1978 On the Distribution of Ergative and Accusative Patterns. Lingua 45:23379.

Panaskerteli-Cicivili, Zaza1978 samkurnalo cignikarabadini [The Book of Healing]. Edited by Mixeil

Shengelia. Tbilisi: Sabota Sakartvelo.


27/28


Plank, Frans1984 Verbs and Objects in Semantic Agreement. Journal of Semantics 3:30560.

Pylkknen, Liina2000 What Applicative Heads Apply To. In Proceedings of the 24th Annual Penn

Linguistics Colloquium, edited by Michelle Minnick Fox, 197210. Univer-sity of Pennsylvania Working Papers in Linguistics 7(1). Philadelphia:University of Pennsylvania.

Reid, Nicholas2000 Complex Verb Collocations in Ngangityemerri. In Changing Valency:

Case Studies in Transitivity, edited by R. M. W. Dixon and AlexandraAikhenvald, 33359. Cambridge: Cambridge University Press.

Rood, David1971 Agent and Object in Wichita. Lingua 28:100107.

Sriot, Patrick2000 Le combat des termes et des relations ( propos des discussions sur les

constructions impersonnelles dans la linguistique en Russie). In Le para-doxe du sujet: les propositions impersonnelles dans les langues slaves etromanes, edited by Patrick Sriot and Alain Berrendonner, 23555.Lausanne: University of Lausanne.

Shanidze, Akaki1953 kartuli gramatikis sapudzvlebi, 1: morpologia [Fundamentals of Georgian

Grammar, 1: Morphology]. Tbilisi: Tbilisis Universitetis Gamomcemloba.

Suxivili, Murman1979 uaxloesi obiektisa da subiektis gauinareba gardamavali zmnis emcvel

esitqvebai [Omission of the Direct Object and Subject in ConstructionsContaining Transitive Verbs]. iberiul-kavkasiuri enatmecniereba 21: 5767.

1986 gardamavali zmnebi kartuli. sistemisa da istoriis zogi sakitxi.[Transitive Verbs in Georgian: Some Questions on System and History].Tbilisi: Mecniereba.

Svenonius, Peter2001 Case and Event Structure. In Syntax of Predication: Proceedings of the

Workshop on Syntax of Predication, Nov. 23, 2001, edited by Niina NingZhang, 121. ZAS Papers in Linguistics 26. Berlin: Centre for GeneralLinguistics, Typology, and Universals Research (ZAS).

Trask, R. L.1996 Historical Linguistics. London: Arnold.

Tschenkli, Kita1958 Einfhrung in die georgische Sprache. Zrich: Amirani Verlag.

Tuite, Kevin1987 Indirect Transitives in Georgian. In Proceedings of the Thirteenth Annual

Meeting, 1416 February: General Session and Parasession on Grammarand Cognition, edited by Jon Aske, 296309. Berkeley: Berkeley LinguisticsSociety.

1994 An Anthology of Georgian Folk Poetry. Rutherford, N.J.: Fairleigh Dickin-son University Press.

1996 Paradigm Recruitment in Georgian. In Linguistic Studies in the Non-SlavicLanguages of the Commonwealth of Independent States and the Baltic Re-publics, edited by Howard I. Aronson, 37587. Chicago: Chicago LinguisticsSociety, University of Chicago.


28/28


1998 Kartvelian Morphosyntax: Number Agreement and MorphosyntacticOrientation in the South Caucasian Languages. Munich: Lincom Europa.

2002 Deponent Verbs in Georgian. In Philologie, Typologie und Sprachstruktur:Festschrift fr Winfried Boeder zum 65. Geburtstag, edited by WolframBublitz, Manfred von Roncador, and Heinz Vater, 37589. Frankfurt amMain: Peter Lang.

2007 Liminal Morphosyntax: Georgian Deponents and Their Kin. Proceedingsfrom the Annual Meeting of the Chicago Linguistic Society 39:77488.

Van Valin, Robert D. Jr.2002 Semantic Macroroles in Role and Reference Grammar. (wings.buffalo.edu/

linguistics/people/faculty/vanvalin/rrg.html).Woolford, Ellen

2006 Lexical Case, Inherent Case, and Argument Structure. Linguistic Inquiry37(1):11130.

Wunderlich, Dieter2006 Towards a Structural Typology of Verb Classes. In Advances in the Theory

of the Lexicon, edited by Dieter Wunderlich, 57166. Berlin: Mouton deGruyter.

Xornauli, Ivane, ed.1949 pav-xevsuruli poezia [Pshav-Khevsur Poetry]. Tbilisi: Saxelgami.

agentless transitives 51.3 4.tuite

Documents