sonora lectures 2013 1

18
Universidad de Sonora Walter Bisang Lecture Series 1 University of Mainz 04 – 08 de noviembre 2013 [email protected] ___________________________________________________________________________ Approaches to complexity I 0. Basic idea of this lecture Discussion of two very prominent approaches to complexity in linguistics: (i) hidden simplicity and recursion (Hauser, Chomsky and Fitch 2002; Givón & Shibatani 2009), (ii) overt complexity (McWhorter 2001, 2005; Dahl 2004; Kortmann & Szmrecsanyi 2012). Both of these approaches only look at language structure from the perspective of what is explicitly expressed. They neglect pragmatic inference and the economy side of complexity (cf. the competing motivations of explicitness vs. economy). The perspective of economy leads to “hidden complexity” (Bisang 2009) as a new type of complexity which will be introduced here. 1. Hidden simplicity and the production of complexity Formal linguistics: Minimalist Program (Chomsky 1995); Hauser, Chomsky & Fitch (2002), Fitch, Hauser & Chomsky (2005) Functional linguistics: Givón & Shibatani (2009) 1.1. The role of recursion in Universal Grammar Complexity is the result of the recursive application of a small number of simple universal principles. In that sense, what looks complex on the surface is due to hidden simplicity. Two non-linguistic examples for a start: Recursion: The Fibonacci sequence in mathematics (each of its new elements is created by summing up the two previous ones): (1) 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, 233, 377, 610, 987, 1597, ... . Complex behaviour as the result of underlying simplicity: Over and over again will we see the same kind of thing: that even though the underlying rules for a system are simple, and even though the system is started from simple initial conditions, the behavior that the system shows can nevertheless be highly complex. And I will argue that it is this basic phenomenon that is ultimately responsible for most of the complexity that we see in nature. (Wolfram 2002: 28) Recursion in linguistics: The beginnings: Recursion in language is often associated with Wilhelm von Humboldt’s famous statement that language makes infinite use of finite means. More recently: Recursion is the basic property of human language (Hauser, Chomsky and Fitch 2002), it is what actually makes human language special:

Upload: w

Post on 13-Sep-2015

217 views

Category:

Documents


1 download

DESCRIPTION

gyrtggijk

TRANSCRIPT

  • Universidad de Sonora Walter Bisang Lecture Series 1 University of Mainz 04 08 de noviembre 2013 [email protected] ___________________________________________________________________________

    Approaches to complexity I 0. Basic idea of this lecture Discussion of two very prominent approaches to complexity in linguistics:

    (i) hidden simplicity and recursion (Hauser, Chomsky and Fitch 2002; Givn & Shibatani 2009), (ii) overt complexity (McWhorter 2001, 2005; Dahl 2004; Kortmann & Szmrecsanyi 2012).

    Both of these approaches only look at language structure from the perspective of what is explicitly expressed. They neglect pragmatic inference and the economy side of complexity (cf. the competing motivations of explicitness vs. economy).

    The perspective of economy leads to hidden complexity (Bisang 2009) as a new type of complexity which will be introduced here.

    1. Hidden simplicity and the production of complexity Formal linguistics: Minimalist Program (Chomsky 1995); Hauser, Chomsky & Fitch (2002), Fitch, Hauser & Chomsky (2005) Functional linguistics: Givn & Shibatani (2009) 1.1. The role of recursion in Universal Grammar Complexity is the result of the recursive application of a small number of simple universal principles. In that sense, what looks complex on the surface is due to hidden simplicity. Two non-linguistic examples for a start: Recursion: The Fibonacci sequence in mathematics (each of its new elements is created by summing up the two previous ones): (1) 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, 233, 377, 610, 987, 1597, ... . Complex behaviour as the result of underlying simplicity:

    Over and over again will we see the same kind of thing: that even though the underlying rules for a system are simple, and even though the system is started from simple initial conditions, the behavior that the system shows can nevertheless be highly complex. And I will argue that it is this basic phenomenon that is ultimately responsible for most of the complexity that we see in nature. (Wolfram 2002: 28)

    Recursion in linguistics: The beginnings: Recursion in language is often associated with Wilhelm von Humboldts famous statement that language makes infinite use of finite means. More recently: Recursion is the basic property of human language (Hauser, Chomsky and Fitch 2002), it is what actually makes human language special:

  • 2 Faculty of LanguageBroad sense (FLB) Faculty of LanguageNarrow sense (FLN).

    Hypothesis 3: Only FLN is uniquely human. On the basis of data reviewed below, we hypothesize that most, if not all, of FLB is based on mechanisms shared with nonhuman animals ... In contrast, we suggest that FLNthe computational mechanism of re-cursionis recently evolved and unique to our species ... . According to this hypothesis, much of the complexity manifested in language derives from complexity in the peripheral components of FLB, especially those underlying the sensory-motor (speech or sign) and conceptual-intentional interfaces, combined with sociocultural and communicative con-tingencies. FLB as a whole thus has an ancient evolutionary history, long predating the emergence of language, and a comparative analysis is necessary to understand this com-plex system. By contrast, according to recent linguistic theory, the computations under-lying FLN may be quite limited. In fact, we propose in this hypothesis that FLN com-prises only the core computational mechanisms of recursion as they appear in narrrow syntax and the mappings of the interfaces. (Hauser, Chomsky and Fitch 2002: 1573)

    Merge and recursion in the Minimalist Program (Chomsky 1995): The computational system creates linguistic structures by merge, i.e., by the recursive binary combination of primitive elements of a language taken from the lexicon. Since merge always applies to two elements (binarity), it produces uniform binary branching structures.

    Recursion particularly allows the infinite embedding of constituents of the same type. Example 1: Sentences (CPs) embedded inside of sentences inside of sentences with no limitations: (2) John thinks that Mary said that Bill believes that Sue insisted [that X]. Example 2 (Kuno 1973: 7 on Japanese): Center embedding with relative clauses (RCs) in Japanese: [ ... [ ... ]RC2 ... ]RC1 A preliminary remark on relative-clause formation in Japanese: (3) Japanese: Simple declarative clause: ringo o tabe-ta. apple ACC eat-PST [X] ate an apple. Japanese relative clauses precede the head-noun. The verb must be in the non-polite form (as in example (3) for past -ta, the present ends in -(r)u): (4) [ringo o tabe-ta] hito apple ACC eat-PST man the man who ate an apple The syntactic function and the semantic role of the head noun within the relative clause must be inferred from context. In most cases, the head-noun is represented by zero in the relative clause. In a maximally simple relative clause that only consists of a verb (no overt subject, no overt object), the adequate coreference relation must be inferred from context and world knowledge:

  • 3 (5) a. Subject coreference: [tabe-ta] hito eat-PST man the man who eats [something] b. Object coreference [tabe-ta] ringo eat-PST apple the apple that [somebody] ate c. Instrument coreference: [tabe-ta] hashi eat-PST chopsticks the chopsticks with which [he] ate Recursion in Japanese relative clauses: (6) Japanese (Kuno 1973: 7): [[[John ga kat-te i-ru]RC1 neko ga korosi-ta]RC2 John NOM keep-CONV be-PRS cat NOM kill-PST nezumi ga tabe-ta]RC1 tiizu wa kusat-te i-ta. rat NOM eat-PST cheese TOP rot-CONV be-PST The cheese that the rat that the cat that John keeps killed ate was rotten. Step-by-step explanation of center embedding in (6): (6) [[[...]RC1 ...] RC2 ...] RC3 3. [[ ]RC2 nezumi ga tabe-ta]RC3 tiizu rat NOM eat-PST cheese the cheese that the cat ate 2. [[ ]RC1 neko ga korosi-ta]RC2 nezumi cat NOM kill-PST rat the rat that the cat killed 1. [John ga kat-te iru]RC1 neko John NOM keep-CONV AUX:PRS cat the cat that John keeps

  • 4 Compare center embedding in English: (7) *The cheese [that the rat [that the cat [that John keeps]RC3 killed] RC2 ate]RC1 was rotten. As can be seen from (7) as a literal translation into English, there are clear limitations to recursion in center embedding in English. And this limitation seems to be subject to a certain cross-linguistic variation: Japanese is more flexible than English. German is somewhat in between because it can have center embedding once: (8) German: Der Ksei, [deni die Rattej gegessen hat], ART cheese REL:ACC ART rat has.eaten [diej die Katzek, [diek Hans hlt], gettet hat] REL:ACC ART cat REL:ACC John keep:PRS:3.SG has.killed ist verfault. Is rotten While center embedding is hard in English, right branching is not: (9) Kuno (1973: 8): John owned a cat that killed a rat that ate cheese that was rotten. 1.2. Criticism Fitch, Hauser and Chomsky (2005: 203) mention the following observations that support the idea that recursion belongs to the Faculty of LanguageNarrow sense (FLN): (10) i. No animal communication system shows evidence of recursion ii. Monkeys cannot even process hierarchical phrase structure iii. There are no unambiguous demonstrations of recursion in other human cognitive

    domains, with the only clear exceptions (mathematical formulas, computer pro-gramming) being clearly dependent upon language.

    On (10i) and (10ii): Deacon (2003: 129) argues that recursion is a semiotic universal that can be expected in any symbolic system, no matter what device implements it. Based on Peirce (1897/1903), Deacon distinguishes three types of signs which developed during the evolution of species: (11) index > icon > symbol The use of symbols is specifically human. Symbolization is characterized by indirect reference and by the implicit combinability of one symbol with other symbols. If recursion is a property of the symbolic capacity of humans it does not come as a surprise that non-humans do not use it in communication and that monkeys cannot process hierar-chical phrase structure. Pinker & Jackendoff (2005: 230) argue that the occurrence of recursion in language is rather due to our ability to use recursion in thinking:

    Indeed, the only reason language needs to be recursive is because its function is to express recursive thoughts. If there were not any recursive thoughts, the means of ex-pression would not need recursion either.

  • 5 On (10iii): The dependence of recursion in mathematical formulas and computer programming on language is anything but uncontroversial. Varley et al. (2005): Three patients with diffuse brain damage to the left perisylvian language area who did not lose their mathematical capacities => mathematical abilities are not necessarily linguistically mediated. 1.3. Recursion as a product of economy and the symbolic capacity of humans Lets take (7) from English, repeated here, for a thought experiment on recursion as a symbolic capacity beyond language: (7) *The cheese [that the rat [that the cat [that John keeps]RC3 killed] RC2 ate]RC1 was rotten. How does a native speaker/reader of English analyse (7)? By a semiotic device: The reader adds the end of the entire relative-clause construction (ate) to the end of the first part of that construction (that the rat) and gets (i) that the rat ate. Then, he combines the elements of the next layer: that the cat plus killed, which yields (ii) that the cat killed. What remains is then the innermost, deepest embedded relative clause (iii) that John keeps. These three elements can then be combined as follows: (7) The cheese [that the rat ate]i [that the cat killed]ii [that John keeps]iii. This little thought-experiment shows that utterances like (7) that take recursion seriously can only be decoded by recurring to semiotic processes that are beyond linguistic rules. We have to note down the sentence and then start analysing it with paper and pencil. To try to account in purely linguistic terms for what goes on in a speaker/reader while she tries to understand this sentence is meaninglessher linguistic competence would simply leave her at a loss if she would hear that sentence and would have to process it incrementally. 1.4. Recursion and economy To understand why some recursive structures are found in the vast majority of the worlds languages, lets look at some recursive constructions which according to Everett (2005) do not exist in Pirah: (12) Coordination (Everett: 2007: 5): [NP John and Bill] came to town yesterday. (13) Disjunction (Everett: 2007: 5): Either [NP Bob or Bill] will come. (14) Relative Clause: The buffalo [CP I want to hunt] (15) Embedding: John thinks [CP that Bill will come]. In each of these constructions the solution offered by recursion is more economic than the solution without recursion illustrated below: (12) John came to town yesterday. Bill came to town yesterday. (13) Bob will come. Or. Bill will come. (14) the buffaloiI want to hunt that buffaloi (15) John thinks somethingi. [Bill will come]i.

  • 6 A closer look at recursion reveals that at least some of its instances such as the ones in (12) to (15) are due to economy. Their similarity to recursion can be seen as an epiphenomenon of economy. Thus, a considerable number of cases does not need to be ascribed to an arbitrary linguistic FLN. If there is recursion in these cases, it is there due to the symbolic capacity of humans ( 1.2 1.3) and/or because of economy ( 1.4). => If hidden simplicity is driven by recursion and if recursion does not seem to be a

    deep-rooted property of human language it cannot play a very important role in the discussion of complexity in language.

    2. The typological approach: Overt complexity Typological approaches to complexity are surface-oriented in the sense that they look at what is expressed by overt morphosyntactic marking/form. In the current discussion, overt complexity is mostly seen from a historical perspective: (i) Reconstruction of earlier stages of human language (Comrie 1992; not discussed here) (ii) Complexity as the result of the historical development of languages: McWhorter (2001, 2005, etc.) on creoles (in 2.1), Dahl (2004) on maturation. 2.1. McWhorter (2001, 2005): Creole grammars are the least complex grammars The structural properties of creoles are due to their origin from pidgins. Speakers of pidgins tend to eschew structures of secondary importance to their communication needs from their native languages. This simplicity of pidgin structure has its impact on creoles, at least at their initial stage: => Complexity is a result of later processes of grammaticalization. => Since creoles did not have enough time to accumulate more complex structures from

    their pidgin input they did not develop many properties we find in older natural languages.

    The general conclusion was that in older grammars, millennia of grammaticalization and reanalysis have given overt expression to often quite arbitrary slices of semantic space, the result being a great deal of baroque accretion which, while compatible with Universal Grammar, is incidental to it, as well as to even nuanced human expression. In not having existed for long enough a time for drift to encrust them in this manner to any great extent, creoles are unique in reflecting the innate component of the human language capacity more closely than other languages do. (McWhorter 2001: 126)

    Phenomena that are the result of later developments and are not attested in Creoles (McWhorter 2001: 163): - Ergativity - Grammaticalized evidential marking - Inalienable possessive marking - Switch-reference marking

  • 7 - Inverse marking - Obviative marking - Dummy verbs - Syntactic asymmetries between matrix and subordinate clauses - Grammaticalized subjunctive marking - Verb-second - Clitic movement - Any pragmatically neutral word order but SVO - Noun class or grammatical gender marking (analytic or affixal) - Lexically contrastive morphosyntactic tone beyond a few isolated cases (subtract Saramaccan and there is none at all).

    To compare complexity across languages, McWhorter proposes four diagnostics of gram-matical complexity that are based on the following intuition: (16) Complexity (McWhorter 2005: 45): [A]n area of grammar is more complex than the same area in another grammar to the extent that

    it encompasses more overt distinctions and/or rules than another grammar. The four diagnostics in some detail (McWhorter 2005: 45 46): (i) A phonemic inventory is more complex to the extent that it has more marked members (ii) A syntax is more complex than another to the extent that it requires the processing of more rules,

    such as asymmetries between matrix and subordinate clauses. (iii) A grammar is more complex than another to the extent that it gives overt and grammaticalized

    expression to more fine-grained semantic and/or pragmatic distinctions than another (iv) Inflectional morphology renders a grammar more complex than another one in most cases His more recent papers (collected in McWhorter 2011) focus on the elaboration of his prototype hypothesis (McWhorter 2005). With this hypothesis, it should be possible to see if a language is a creole even if there is no knowledge of its sociohistory: (17) McWhorter (2011: 60):

    A natural language is a creole (i.e. born recently from a pidgin and thus emerged from broken transmission) iff it has: 1) morphologically: little or no inflectional affixation, and among unbound inflectional

    markers, none of the contextual inflection, or of inherent inflection of the paradigmatically complex sort.

    2) phonologically: little or no distinction of monosyllabic lexical items or morphosyntactic distinctions via tone or register, and no typologically unusual proliferation of vowels, and

    3) semantically: little or no noncompositional combination of nonreduplicative derivational morphemes with roots.

    Even though complexity is not mentioned in this definition, it is still based on overt complexity. Some additional explanations of (17) may clarify this: On 1): The morphological part of the prototype hypothesis looks at how certain grammatical

    categories are overtly marked: McWhorter adopts Booijs (1993) distinction between contextual and inherent inflection: Contextual inflection is linked to syntactic context (structural case, agreement), while inherent inflection is (at least relatively) independent of syntactic context, and includes tense, aspect, gender and number markers (McWhorter 2011: 43).

    McWhorter (2011: 41)) additionally follows Kihm (2003), who also takes syntactic (morphologically unbound) markers of inflectional categories as instances of inflection.

  • 8 Taking these two approaches together, creoles only have inflection that is unbound

    and inherent. Finally, the inherent inflection of creoles is not paradigmatic, i.e., it does not involve

    the overt marking of subdivisions of semantic space such that the subdivision becomes denotationally arbitrary (declension classes in Latin) (McWhorter 2011: 44 on paradigmatic complexity).

    On 2): The phonological part of the prototype hypothesis is based on the overt presence of

    oppositions in terms of tone or register. The passage on unusual proliferation of vowels refers to the fact that loss of tonality or register may lead to an increasing number of vowel phonemes. This is illustrated by the loss of the register opposition between [breathy] in Khmer and the increasing number of phonological distinctions in vowel qualities. Of course, such a development presupposes a certain linguistic history of a language.

    On 3): This definition basically excludes compounds of the type understand, whose meaning

    cannot be derived from the two components of under and stand. 2.2. Dahl (2004): Complexity and maturation Dahl (2004) describes complexity in terms of maturation, i.e., as the result of accumulating grammatical material in the grammar G of a language that was not present at an earlier stage G of that language (cf. Ansaldo & Nordhoff 2009: 358, from which my simplification is inspired). An example from Dahl (2004: 105): Plural marking in Tok Pisin and English: Tok Pisin has a fully productive facultative plural marker ol (derived from Engl. all) (the example is not from Dahl): (18) a. Wanpela dok i kam. c. Ol dok i kam. One dog PM come PL dog PM come A/one dog came. (The) dogs came. b. Tupela dok i kam. d. Olgeta dok i kam. Two dog PM come All dog PM come Two dogs came. All the dogs came. In comparison to English, the relatively straightforward case of Tok Pisin lacks the following properties: Inflectional marking Numerous lexical idiosyncrasies (irregular plurals) Obligatory use of plurals even in contexts where number marking is informationally redundant, e.g. after quantifiers (cf. above (18a, b, c) vs. (18b)) involvement in syntactic agreement (verbs, demonstrative pronouns) existence of pluralia tantum (lexical items which are plural only: scissors, trousers).

    Systems of the English type must be the result of a historical development: It is reasonable to assume that a system like that of English and other Indo-European lan-guages can only come about after a historical development of significant length, in-volving a number of intermediate stages, where the earlier ones are more like that found in Tok Pisin (although the ultimate source of the plural morpheme may be different).

    (Dahl 2004: 105)

  • 9 Dahls more explicit definition of maturation:

    x is a mature phenomenon iff there is some identifiable and non-universal phenomenon or a restricted set of such phenomena y, such that for any language L, if x exists in L there is some ancestor L of L such that L has y but not x. (Dahl 2004: 105)

    Complex phenomena in terms of Dahl (2004: 114-115): complex word structure, including - inflectional morphology; - derivational morphology; - incorporating constructions; lexical idiosyncrasy, including

    - grammatical gender; - inflectional classes; - idiosyncratic case marking;

    syntactic phenomena that are dependent on inflectional morphology, including - agreement; - case marking (partly). word order rules over and above internal ordering of sister constituents; specific marking of subordinate clauses; morpheme and word level features in phonology. 2.3. An example on a cross-linguistic comparison of complexity creoles vs. other languages (Parkvall 2008) Parkvall (2008) intends to prove through an extensive typological database that creoles are indeed different from non-creoles, primarily in being less complex (Parkvall 2008: 265). For that purpose, he calculates the complexity score of individual languages on the basis of 53 features, mostly from WALS (Haspelmath et al. 2005). I list some of these features here: F01 Size of consonant inventories F02 Size of vowel quality inventories F05 Tone F06 Overt marking of direct object F11 Gender F13 Non-semantic gender assignment F15 Definite articles F20 Politeness in pronouns F25 Obligatory numeral classifiers F29 Grammaticalized perfective/imperfective F31 Remoteness distinctions of past F36 Grammaticalized evidentiality distinctions F41 Passive F46 Equative copula Locative copula F51 Total amount of verbal suppletion For each of these features, Parkvall (2008) calculated values between 0 and 1:

    Most features have binary values (occasionally with an intermediate value) which easily translate into the numbers 1 and 0 (or 0.5, as the case may be). Those which do not are F01, F02, F04, F05, F12, F18, F20, F26, F51 and F53. Some of these are numerically coded, and in those cases I compressed the values to fit the zero-to-one range.

    (Parkvall 2008: 273) The complexity score for individual languages corresponds to the average of the individual

  • 10 feature values. In his first experiment, Parkvall (2008) selected all the WALS languages with values for at least 30 out of 53 features. So, he analyzed 155 languages. I only mention the six languages with the highest and the lowest complexity scores, respectively: (19) Parkvall (2008: 274 275): 1. Burushaski: 0.62 150. Ndyuka: 0.22 2. Copainal Zoque: 0.57 151. Maybrat: 0.22 3. Khoekhoe: 0.57 152. Kobon: 0.20 4. Beja: 0.56 153. Hmong Njua: 0.20 5. Koasati: 0.55 154. Pirah: 0.18 6. Kannada: 0.56 155. Sango: 0.15 The 155 languages-sample consisted of only two creole languages. Both of them were among the six languages with the lowest complexity score. Ndyuka (French Guyana, Surinam) and Sango (Central African Republic). In a second experiment, Parkvall (2008) checked the same features for 30 pidgins and creoles (cf. the list in Parkvall 2008: 278). As it turned out, the values for all non-creole and non-pidgin languages was 0.41 (cf. some details in Table 2 below). Creoles and pidgins scored much lower, as can be seen from Table 1: (20) Language type Average complexity Creoles (excluding expanded pidgins) 0.24 Creoles (including expanded pidgins) 0.23 Sango & Ndyuka only 0.19 Expanded pidgins only 0.19 All pidgins 0.14 Table 1: Complexity values for creoles/pidgins (Parkvall 2008: 281) North-Caucasian produced a value slightly above 0.50. The complexity scores between 0.30 0.39 and 040 0.49 are found in the following geographic areas (GEO) and language families (GEN): (21) Complexity score 0.30 0.39 Complexity score 0.40 0.49 GEO Languages of Oceania, Languages of Africa, of Asia, of Europe, of South America, of North America, of the eastern hemisphere, of Southeast Asia, of the northern hemisphere, of the western of the southern hemisphere hemisphere. GEN Austro-Asiatic, Afro-Asiatic, Algic, Gunwingguan, Austronesian, Indo-European, Isolates, Niger-Congo, Pama-Nyungan, Nilo-Saharan, Penutian, Sino-Tibetan, Trans-New Guinea Uralic, Uto-Aztecan Table 2: Complexity scores between 0.49 0.30 and their areal/genetic distribution (Parkvall 2008: 279) The above data led Parkvall to the following conclusion:

    Typologically speaking, creoles stand out from languages in general, and the most salient difference is that they present a lower structural complexity. This does not necessarily have any bearing on issues regarding psycholinguistic complexity, however, and certainly not on their expressive potential. (Parkvall 2008: 283)

  • 11 2.4. On Creoles as a structurally distinguished group of languages (Bakker et al. 2011) Bakker et al. (2011) conducted 6 experiments for testing the following hypothesis (hypothesis 5 in their paper):

    Hypothesis: If one believes that universal patterns play a role in creole genesis, then one expects no clustering on the basis of lexifier, geography, substrate or historical connections (but it is likely that creoles themselves form a group). (Bakker et al. 2011: 15)

    Methodology: Language samples: Parkvall (2008): cf. 2.3 above Holm & Patrick (2007): 97 features that are assumed to be characteristic of Creoles, description of 18 Creole languages. Creoles in Holm & Patricks (2007) sample: Angolar (Portuguese, A), Berbice Dutch (Dutch, A), Cape Verdean (Portuguese, A). Dominican (French, A), Guinea Bissau (Portuguese, A), Haitian (French, A), Jamaican (English, A), Korlai (Portuguese, Indian Ocean), Krio (English, A), Nagamese (Assamese, Indian Ocean), Ndyuka (English, A), Negerhollands (Dutch, A), Nubi (Arabic, Africa), Palenquero (Spanish, A), Papiamentu (Portuguese, A), Seychellois (French, Indian Ocean), Tok Pisin (English, Pacific), Zamboangueo (Spanish, Pacific). [A stands for Atlantic]. Statistical method: Computational phylogenetics: Phylogenetic networks and trees Some results Study 4: Comparison of the 97 features analyzed by Holm & Patrick (2007) in 18 Creoles and 12 Non-Creoles The Creole languages correspond to the 18 languages in Holm & Patrick (2007) The other 12 languages are selected according to the following criteria:

    (i) a relatively isolating structure within the phylum (i.e. similar to the typological profile usually associated with creole languages) (ii) a complexity score around that established for creoles in Parkvall (2008), i.e., between 0.13 and 0.33.

    List of the Non-Creoles: Ainu (isolate, Japan), Akan (Niger-Kongo: Kwa, Ghana), Bambara (Niger-Kongo: Kwa, Mali), Brahui (Dravidian, India), English, Indonesian (Austronesian: Malayo- Polenesian), Kimbundu (Niger-Kongo: Bantu, Angola), Kolyma Yukaghir (isolate, Russia), Koyra Chiini (Nilo-Saharan, Mali), Mina (Afro-Asiatic: Chadic, Cam- eroon), Mandarin Chinese (Sino-Tibetan), Pirah (Mura, Brazil). Findings: Creoles and Non-Creoles cluster separately (cf. figure 1, next page).

  • 12

    Figue 1: (Bakker et al. 2011: 32)

    Study 5: Comparison of the 155 languages from Parkvall (2008) plus 32 Creoles analyzed for the same features (minimally 30 of them needed to be available for an individual Creole): Findings: (i) The Creoles are identified as a group. (ii) even a cursory look at the clustering of non-creoles shows that the non-creoles do not classify along genetic or areal lines. For example, Basque (iso- late, Western Europe), Hindi (Indo-European, India/South Asia), Burushaski (iso- late, North Pakistan), and Hunzib (East-Caucasian, Caucasus) cluster (Bakker et al. 2011: 33)

    (iii) the classifications do not follow the complexity scale for creoles (or non- creoles): the languages do not cluster from lowest to highest complexity, showing that it is not simply complexity that is measured (Bakker et al. 2011: 33-34).

    Conclusion: Creoles share structural similarities but these similarities are not based on complexity.

    Ansaldo & Matthews (2007:14) wrote: creole exceptionalism is a set of so- ciohistorically-rooted dogmas, with foundations in (neo)colonial power relations, not a scientific conclusion based on robust empirical evidence (our emphasis). We have shown that creole exceptionalism, or rather distinctiveness, is not a dogma, not an ideology, but a fact. (Bakker et al. 2011: 35)

  • 13 3. Explicitness vs. economy and complexity In addition to hidden simplicity ( 1) and overt complexity ( 2), there is a third way to look at complexity if one takes the perspective of what is not expressed morphosyntactically and thus needs to be pragmatically inferred. I will call this type of complexity hidden complexity.

    3.1. Introduction Morphosyntactic structures and their properties can never fully express the meaning they have in a concrete speech situation. They need pragmatic enrichment from context. An important reason for this is the articulatory bottleneck (Levinson 2000: 27-30). Human speech is characterized by its extremely slow transmission rateprocesses like prearticulation, parsing and comprehension produce and assess much more information in a much shorter time:

    inference is cheap, articulation expensive (Levinson 2000: 29). Thus, grammatical structures can be seen as the result of the competition between expensive articulation or explicitness and cheap inference or economy: = > Competing motivations: von der Gabelentz (1891: 251): Deutlichkeitstrieb vs. Bequemlichkeitstrieb Haiman (1983): iconic vs. economic motivations Optimality Theory: faithfulness vs. markedness constraints The competition between explicitness and economy creates two types of complexity: Explicitness: Economy: Articulation Pragmatic inference Overt complexity Hidden complexity

    Figure 2 3.2. The manifestations of overt vs. hidden complexity (Bisang 2009, forth) I. Explicitness: => Overt complexity: The structure of a language (a) forces the speaker to overtly express certain grammatical categories (obligatoriness) (b) provides a rich inventory of fine-grained grammatical categories II. Economy: => Hidden complexity: The structure of a language (a) does not force the speaker to overtly express grammatical categories that are part of its grammatical inventory (lack of obligatory categ- ories) (b) has multifunctional markers whose concrete meaning must be inferred from context.

  • 14 Examples: Ia. Explicitness: Obligatoriness Obligatory marking of subject (pronominal or by agreement), tense etc. as we know it from English main declarative clauses. Obligatory marking of number and (in)definiteness as we know it from English noun phrases Obligatory expression of evidentiality in various languages (cf. point Ib). Obligatory politeness marking in finite clauses (Japanese, Korean). Ib. Rich inventory of fine-grained grammatical categories Pronominal system in Fiji (Dixon 1988):

    Person Number 1st person 2nd person 3rd person Singular au iko Koya Dual (inclusive) kedaru kemudrau Rau Dual (exclusive) keiru Trial (inclusive) kedatou kemudou iratou (eratou) Trial (exclusive keitou Plural (inclusive) keda kemuni ira (era) Plural (exclusive) keimami

    Table 3 Obligatory expression of different degrees of remoteness in the past (recent past, distant past, remote past) combined with three types of evidentials (experiential, in- ferential, conjecture): Matses (Panoan, spoken in Amazonian Peru and Brazil; Fleck 2007). All statements with past reference must be classified as experiential, inferential, conjecture (or hearsay)1: Experiential: the speaker detects the occurrence of an event (or state), using any of the five

    senses, at the time that it transpires. It should be emphasized that the essential condition is that the speaker witnesses the event AS THE EVENT HAPPENS. (Fleck 2007: 595).

    Inferential: the speaker did not witness the event as it occurred, but is able to deduce its occurrence via some detected RESULTS (Fleck 2007: 597).

    Conjecture: the speaker wishes to report the occurrence of an event or state that s/he did not witness, did not hear aboutfrom somebody else, and for which s/he has not observed or otherwise detected any result. The actual source of information is the speaker, who may draw upon knowledge of the real or spiritual/magical world and power of logic to deduce the occurrence of the event, or the speaker may draw on his/her imagination to conjure up a possible past event (Fleck 2007: 601).

    The evidential suffixes are combined with three different categories of past remoteness:

    1 I will not discuss hearsay because it is expressed by a different expression format.

  • 15 Recent past (immediate past to about one month ago): Experiential: -o Inferential: -ak Conjecture: -ash Distant past (about one month to 50 years ago): Experiential: -onda Inferential: -ndak Conjecture: -ndash Remote past (about 50 years ago to max. human life span): Experiential: -denne (before speakers infancy) Inferential: -ampik (before speakers infancy) Conjecture: -ndampik

    If an event is reported away from the results (i.e. at a place from which evidence is no longer discernible), two tense forms must be used according to the following pattern:

    (22) I. II. III. Actual event state detection of the report (when the resulting evidence sentence is uttered) inferential suffix experiential suffix (23) Double Past (Fleck 2007: 599): a. shktenam kuen-ak-o-sh. white.lipped.peccary pass.by-RPST.INF-RECP.EXP-3 White lipped peccaries (evidently) passed by. [Fresh tracks were discovered a short time ago at a distant location.] b. shktenam kuen-ak-onda-sh. white.lipped.peccary pass.by-RPST.INF-DISTPST.EXP-3 White lipped peccaries (evidently) passed by. [Fresh tracks were discovered a long time ago at a distant location.] c. shktenam kuen-ndak-o-sh. white.lipped.peccary pass.by-DISTPST.INF-RPST.EXP-3 White lipped peccaries (evidently) passed by. [old tracks were discovered a short time ago at a distant location.] d. shktenam kuen-ndak-onda-sh. white.lipped.peccary pass.by-DISTPST.INF-DISTPST.EXP-3 White lipped peccaries (evidently) passed by. [old tracks were discovered a long time ago at a distant location.] IIa. Lack of obligatory categories in East and mainland Southeast Asian languages: No obligatory marking of arguments (radical pro-drop) Lack of obligatory tense-aspect marking in declarative matrix clauses Lack of number and definiteness marking in NPs Lack of morphological case marking Two examples from texts: (24) Modern Chinese: Lack of overt arguments (from Lu Xun, Fe@igo@ng Opposing aggression): ...

  • 16 ZI#xia~ de td Go@ngsu@n Ga@oi li zha#o M-zI#j, yI#jing ha#ojI# Zixia POSS disciple Gongsun Gao come seek Mo-Master already several hu le, j zo#ngshi b zi jia@, i jin bu zho j. Dyue@ sh times PF never NEG be.at home see NEG reach about be d-s huo~~zhe# d-wu# hu ba, zh ci qiqia#o zi mnko#u fourth or fifth time EXCL this finally by.chance at doorway i yjin j, ... meet Gongsun Gaoi, a disciple of Zixia, was looking for master Moj for several times and [hej] was never at home, so [hei] was unable to meet [himj]. It was at about the fourth or the fifth time that [hei] met [himj] in the doorway ... (25) Khmer (Bisang 1992: 7-8, 436): pdFyi k kraok laN i da tF$u i lF$k t$k mu$y khtE$h nu$h husband then get.up DIR:go.up | walk DIR:go | lift water one bucket DEM | i y$k tF$u sraoc lF$ sahayj nF$u knoN pI$N nu$h take DIR:go pour on lover LOC:be.at inside pitcher DEM | j slap tF$u. die DIR:go The husbandi got up, i went away, i raised the one bucket of [boiling] water and i poured it over the loverj [of his wife] in the pitcher and j died. Context: A wife hid her lover in a pitcher. Her husband comes home unexpectedly and finds him there.

    IIb. Multifunctional markers: Example: come to have-verbs The case of Khmer ba:n come to have (Bisang 2004, 2011, Enfield 2003): (26) Possible inferences of come to have in East and mainland SE Asian languages a. The event E is [+desired]: +> modal interpretation: can (potential meaning: abilitative or permissive) b. The event E is [-desired] +> modal interpretation: must, have to (obligation) c. In order for X to come to have E, E must have taken place:

    +> Past (E) (particularly if E is negated) d. In order for X to come to have E, E must be true: +> truth, factuality (27) KO~~t ban t~~u phsa. 3.SG TAM go market a. He was able to go to the market. (cf. 26a)) b. He DID go to the marked. (facticity of E against a wrong presupposition; (26c)) c. He went to the marke. (past, cf. (26d))

  • 17

    Outlook In the rest of my lectures, I will focus on hidden complexity and the relevance of pragmatic inference. As can be seen in 3.2, certain domains that are determined syntactically in languages like English, are determined by pragmatics in Chinese. We will look at Chinese and other East and mainland Southeast Asian languages (EMSEA) and we shall compare them with other languages, mainly West African and Creoles. As it will turn out, grammars of EMSEA languages allow more hidden complexity than West African and Creoles. In addition, we may look at other areas in more detail, depending on the preferences of the audience. For example: Classifier systems Serial-verb constructions Processes of grammaticalization and areality ... References Ansaldo, Umberto & Stephen Matthews. 2007. Deconstructing creole: the rationale. In Umberto

    Ansaldo, Stephen Matthews & Lisa Lim (eds.), Deconstructing Creole, 118. Amsterdam: John Benjamins Publishing Company.

    Ansaldo, Umberto and Nordhoff, Sebastian. 2009. Complexity and the age of languages. In: Enoch O. Aboh and Norval Smith (eds.), Complex Processes in New Languages, 345-363. Amsterdam: Benjamins.

    Bakker, Peter, Aymeric Daval-Markussen, Mikael Parkvall and Ingo Plag. 2011. Creoles are typologically distinct from non-creoles. Journal of Pidgin and Creole Languages 26, 5-42.

    Bisang, Walter. 2009. On the evolution of complexitysometimes less is more in East and mainland Southeast Asia, in: Sampson, Geoffrey, David Gil and Peter Trudgill (eds.), Language Complexity as an Evolving Variable, 34-49. Oxford: Oxford University Press.

    Bisang, Walter, forth. Bisang, W. Language contact between geographic and mental space, in: Auer, P., M. Hilpert, A. Stukenbrock and B. Szmrecsanyi, Linguistic Perspectives on Space: Geography, Interaction, and Cognition. Berlin: Mouton de Gruyter (Series Linguae et Litterae).

    Booij, Geert. 1993 Against split morphology. In: Geert Booij and Jaap van Marle (eds.), Yearbook of Morphology 1993, 2749. Dordrecht: Kluwer.

    Chomsky, Noam. 1995. The Minimalist Program. Cambridge, Mass.: MIT Press. Comrie, Bernard. 1992. Before complexity, in: Gell-Mann, Murray and John Howkins (eds.), The

    Evolution of Human Languages: Proceedings of the Workshop on the Evolution of Human Languages, Held August 1989 in Santa Fe, New Mexico, 193-211. Redwood, Calif.: Addison-Wesley.

    Dahl, sten. 2004. The Growth and Maintenance of Linguistic Complexity. Amsterdam/Philadelphia: Benjamins.

    Deacon, T. W. 1997. The Symbolic Species. The Co-evolution of Language and the Human Brain. London: Penguin.

    Deacon, T. W. 2003. Universal Grammar and semiotic constraints, in: M. H. Christiansen, and S. Kirby (eds.), Language Evolution, 111-139. Oxford: Oxford University Press.

    Dixon, R.M.W. 1988. A Grammar of Boumaa Fijian. Chicago & London: The University of Chicago Press.

    Everett, D. L. 2005. Cultural constraints on grammar and cognition in Pirah. Current Anthropology 46: 621646.

  • 18 Everett, D. L. 2007. Cultural constraints on grammar in Pirah: A reply to Nevins, Pesetsky, and

    Rodrigues (2007). http://ling.auf.net/lingBuzz/000427. Downloaded October 03, 2007 (W.B.). Fleck, David W. 2007. Evidentiality and double tense in Matses. Language 83, 589-614. Fitch, T. W., Hauser, M. D., and Chomsky, N. 2005. The evolution of the language faculty:

    Clarifications and implications. Cognition 9: 179210. Gabelentz, G. von der. 1891. Die Sprachwissenschaft, ihre Aufgaben, Methoden und bisherigen

    Ergebnisse. Leipzig: T.O. Weigel Nachfolger. Republished 1972 by Tbingen: Narr. Givn, Talmy and Shibatani, Masayposhi. 2009. Syntactic Complexity. Diachrony, Acquisition,

    Neuro-Cognition, Evolution. Amsterdam and Philadelphia: John Benjamins. Haiman, J. 1983. Iconic and economic motivation. Language 59: 781819. Hauser, M. D., Chomsky, N., and Fitch, T. W. 2002. The faculty of language: What is it, who has it,

    and how did it evolve? Science 298: 15691579. Kihm, Alain. 1989. Lexical conflation as a basis for relexification. Canadian Journal of Linguistics

    34: 35176. Kortmann, Bernd and Szmrecsany, Benedikt. 2012. Linguistic Complexity. Second Language

    Acquisition, Indigenization, Contact. Berlin: Mouton de Gruyter. Kuno, Susumu. 1973. The Structure of the Japanese Language. Cambridge, Mass., and London: MIT

    Press. Levinson, Stephen C. 2000. Presumptive Meanings. The Theory of Generalized Conversational Im-

    plicatures. Cambridge, Mass.: The MIT Press. McWhorter, J. H. 2001. The worlds simplest grammars are creole grammars. Linguistic Typology 5,

    125166. McWhorter, John H. 2005. Defining Creole. Oxford: Oxford University Press. McWhorter, John H. 2011. The creole prototype revisited and revised. In: McWhorter, John:

    Linguistic Simplicity and Complexity. Why Do Languages Undress? Berlin: Mouton de Gruyter. Parkvall, Mikael. 2008. The simplicity of creoles in a cross-linguistic perspective, in: Matti Miestamo,

    Kaius Sinnemki and Fred Karlsson (eds.), Language Complexity: Typology, Contact, Change, 265-285. Amsterdam: Benjamins.

    Pinker, Steven and Jackendoff, Ray. 2005. The faculty of language: Whats special about it? Cognition 95, 201-236.

    Peirce, C. S. 1897/1903. Logic as semiotic: The theory of signs, in: J. Buchler (ed.), The Philosophical Writings of Peirce (1955). New York: Dover Books, 98-119.

    Pinker, Steven and Jackendoff, Ray. 2005. The faculty of language: Whats special about it? Cognition 95, 201-236.

    Wolfram, S. 2002. A New Kind of Science. Champaign, IL: Wolfram Media Inc.