
Yidiny Coordination Reduction and Syntactic Ergativity Item Type Article; text Authors Frazier, Michael Publisher University of Arizona Linguistics Circle (Tucson, Arizona) Journal Coyote Papers Rights http://rightsstatements.org/vocab/InC/1.0/ Download date 28/09/2021 15:39:12 Item License Copyright © is held by the author(s). Link to Item http://hdl.handle.net/10150/253417 Yidiny Coordination Reduction and Syntactic Ergativity Michael Frazier Northwestern University 1. Introduction Syntactic ergativity refers to the phenomenon, found almost exclusively in a subset of languages with Absolutive/Ergative morphology, that some syntactic processes treat transitive objects and intransi- tive subjects as a natural class to the exclusion of transitive subjects. In this paper I present an analysis of a typologically unusual pattern of argument sharing or Coordination Reduction (CR)1 in Yidiny, a Pama-Nyungan language from North Queensland, Australia (Dixon (1977)), with an eye to what the theoretical status of syntactic ergativity is: is it a homogeneous phenomenon derivable from a parameter of the grammar (what we might call Hypothesis A), or is it a descriptive rubric for a distributional pattern than can emerge from multiple underlying mechanisms (Hypothesis B). Herein I propose that the assumption that (whatever the mechanism is) one mechanism accounts for these syntactically ergative patterns (Hypothesis A) is incorrect, and that syntactic ergativity can occur in a language for at least two reasons. It may occur either because of some operation yielding prominence of the internal argument of a transitive verb in the syntax, or because of the interaction of syntactic mechanisms with case-marking in a language with morphological ergativity. Yidiny shows two dissociated patterns of syntactic ergativity, one appearing to be dependent upon surface morphological features in CR and another indifferent to them in A0 structures, that are most naturally analyzed as arising due to separate mechanisms. Because within a single language syntactically ergative phenomena can dissociate, there must be at least two possible routes to syntactic ergativity. The evidence of Yidiny suggests that syntactic ergativity can result from sensitivity to a position available to internal arguments and the single arguments of intransitives, or from the sensitivity of a given syntactic operation to Case (or case)2 features. In Yidiny, both kinds of syntactic ergativity appear to be present.3 The outline for this paper is as follows. First, I review the points of Yidiny morphosyntax relevant to the argument at hand. Then, I discuss why it would be most natural to treat the pattern of syntactic ergativity exhibited in Coordination Reduction separately from that seen elsewhere in the language. I review some evidence that the absolutive case-form is a syncretic morphological default in Yidiny, and then discuss the similarity that the CR pattern shows, once the syncretic nature of the absolutive is known, to matching effects Grimshaw (1977) observed in other languages. After presenting an analysis of Yidiny CR in these terms, I close with suggestions for further research. 2. Yidiny Morphosyntax Yidiny has split-ergative case-marking of the typologically common NP-split pattern (Silverstein (1976), inter alia), which is also common within Pama-Nyungan Dixon (1976). Full nouns possess 1I use the term Coordination Reduction (CR) as a surface rubric for a family of constructions in which an argument in the first of a series of coordinated clauses licenses the omission—whether by ellipsis, a null pro-form, or some other mechanism—of a co-referent argument in subsequent clauses. 2Herein I capitalize the term ‘Case’, and Nominative, Accusative, etc., when reference is intended to abstract features involved in NP-licensing rather than to the morphological form of nominals (‘case’) which I take to reflect, among other things, this abstract licensing. I assume an assignment or valuation approach to Case-licensing broadly as in Chomsky (2000), Chomsky (2001). 3Recent examples of these classes of proposals include Aldridge (2008) and Otsuka (2010) respectively. 19 Table 1: Yidiny Case Marking Split man I/(1.sing) A wagudyanggu ngayu S wagudya ngayu O wagudya nganyany distinct ergative and absolutive forms, which is to say that the transitive subject is marked distinctly from the intransitive subject and transitive object, which are marked alike (in Dixon’s terms, A6=S=O)4. 1st and 2nd person pronouns, on the other hand, have distinct nominative and accusative forms, so that the transitive object is marked distinctly from the intransitive and transitive subject, which are marked alike, as shown in Table 1. (In Dixon’s terms, A=S6=O.) This kind of split, between speech act participants (i.e. 1st/2nd person pronouns) and non-participants (i.e. full nouns) is a pattern that is familiar from the work of Silverstein (1976), Keenan & Comrie (1977), and many others. In most of its clause-combining structures, Yidiny exhibits syntactic ergativity of a kind similar to the well-attested pattern where A0-processes like relativization are only grammatical, or are grammatical in more contexts, when the moved element originates in an absolutive (intransitive subject or transitive object) position (Dixon (1994)). The status(es) of these clause-types are not entirely clear: they often bear a plausible similarity to correlatives (Srivastav (1991)), though their exact analysis is not directly relevant to the point at hand. Dixon (1977) terms them “subordinatives,” a descriptive usage which I follow here. The generalization concerning these constructions is that an argument may be shared between the subordinative clause and the matrix clause if, in each clause, it is either an intransitive subject or transitive object (i.e. what would be marked absolutive if a full NP).5 When this is not the case, as in (1), the subordinative clause verb6 is obligatorily in an intransitivizing form, such as the antipassive in (1), which has the effect of placing the ‘logical’ transitive subject ngayu ‘I / (1psg-NOM)’ in the position of intransitive subject, allowing it to be omitted under coreference with the transitive object nganyany ‘me / (1psg-ACC)’ in the matrix clause. Note that morphological matching is not required here: although the pronouns take nominative/accusative case forms, the transitive object and intransitive subject behave alike in this construction. (1) gana nyundu:ba nganyany wamban, wurba:dyinyu:n dyambu:n-da TRY you.PL-NOM I-ACC wait-IMP, seek-ANTIP-DAT.SUBORD grub-LOC You wait for me, who am seeking grubs. (adapted from Dixon 1977:331) 2.1. Coordination Reduction It is in coordinated clauses that Yidiny displays a pattern of syntactic ergativity that is apparently unattested elsewhere.7 In most circumstances, Yidiny behaves like its more famous neighbor Dyirbal in having ergatively aligned argument sharing (or coordination reduction, hereafter CR) between apparently coordinated clauses. So in (2) below, it is the transitive object wagudya ‘man’, rather than the transitive subject bunya:ng ‘woman’, that is interpreted as the intransitive subject of mangga:ny ‘laughed’. Similarly, in (3), the intransitive subject of the first conjunct, dunda:lay ‘car’, licenses argument omission in the transitive 4A = transitive subject, O = transitive object, S = intransitive subject. 5In Apprehensional subordinatives, in contrast to the other classes Dixon (1977a) identifies, the shared element need only be absolutive in the embedded clause. 6Or, in some cases, the matrix clause verb. 7Though Yidiny has no overt coordinator, a construction similar to coordination, and which will be referred to for brevity’s sake as coordination throughout, can be identified. This construction is distinguished by, apart from the argument sharing restrictions that are at issue, the presence of obligatory tense matching between the verbs of the conjoined clauses, and potential occurrence of both clauses in a single intonational phrase, properties which do not obtain of adjacent independent clauses (Dixon (1977)). 20 object position of the second conjunct, rather than in the transitive subject position, as would be the case in any English analogue. (2) bunya:ng wagudya wurra:ny mangga:ny woman-ERG man-ABS slap-PAST laugh-PAST The woman slapped the man and (*she/he) laughed. (Dixon (1976)) (3) dunda:lay dyungga:ng gaba:ynydya, walba:ng bundya:dyinyu car-ABS run-PAST road-LOC, stone-ERG hit-:dyi-PAST A car was going along the road, and a stone smashed [into it]. (Dixon (1977):285) This is strikingly unlike the pattern observed in superficially similar constructions in English and other familiar languages, in which transitive and intransitive subjects are able to be shared between coordinated clauses to the exclusion of transitive objects.8 In Yidiny, there is an additional CR pattern that displays English-like argument sharing. In particular, when pronouns are arguments of coordinated clauses, they exhibit the English-like pattern of accusatively-aligned CR. Transitive and intransitive subjects behave alike in being able, unlike transitive objects, to be shared between conjuncts. The result of this is that, just as in English, a transitive subject will control the interpretation of a null intransitive subject coordinated with it, rather than the opposite pattern observed with full nouns, as in (2, 3). (4) ngayu nyuniny bandya:rr, wanda:ny I-NOM you-ACC followed, fall.down-PAST I followed you, and (I/*you) fell down. (from Comrie (1989)) The complexity of Yidiny’s CR pattern results from the possibility of full nouns and pronouns occurring together as arguments of a transitive conjunct. When a transitive conjunct contains a pronoun subject and a full noun object, both arguments of the transitive conjunct are able to control the interpretation of a null intransitive subject in the second conjunct, and the resulting, conjoined sentence is ambiguous, as in (5, 6). (5) ngayu bama bandya:rr, wanda:ny I-NOM person-ABS follow-PAST, fall.down-PAST I followed the person, and (I/he) fell down.
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages10 Page
-
File Size-