ALANA JOHNS | PROFESSOR

This paper examines the properties of two cases, ergative and absolutive, in Niuean, a Polynesian language of the Tongic subgroup. The particular question addressed is: Can clear equivalence relations be set up between ergative / absolutive and nominative / accusative cases? I argue that ergative case is not equivalent to nominative nor to accusative case in that it is direct, inherent and assigned internally to vP to the external agent argument. I also argue that absolutive case does not pattern purely with accusative nor with nominative case, but is hybrid in being internal, but primary and unmarked. The means that the property of being a primary unmarked case is not tied to INFL by features such as finiteness, EPP, or agreement, and this is shown to be empirically true in Niuean. This paper argues against the position of Bittner and Hale 1996a,b, and others, who argue that absolutive case is equivalent to nominative case in being associated with the most external case head. The paper suggests a direction for the study of parameterization of case systems: one in which case features can differ not only in their inherent properties, but also by virtue of appearing on different heads.

For some scholars, the "spurious antipassive" construction in the ergative language, Chukchi, has seemed to show that no restrictive, general theory of agreement is possible. By offering a principled analysis of this construction within a restrictive theory (Distributed Morphology), we show that this pessimism is unfounded. In the process, we present a novel theory of structural (as opposed to thematic) ergativity, based on the idea of multiple Case-checking by T, which is itself more principled than existing models. Analogies with French causatives are used to illustrate the wide applicability of this model.

While syntactic ergativity has long been regarded as a phenomenon independent of morphological ergativity, two facts suggest that the former should be understood as a consequence of the latter: a) morphologically accusative languages never show syntactic ergativity; and b) even when a language does show syntactic ergativity, an ergative pattern is often restricted to certain construction types. This study proposes that an ergative pattern arises when the relevant syntactic operation happens to be Case-sensitive and shows that syntactic ergativity involving relativization is an instance of such Case-sensitive operations. Based on the active Case approach (Levin and Massam 1984, Bobaljik 1993) and extending the proposals of Shlonsky (1992) and Suñer (1998), we argue that the distribution of gap and resumptive pronouns is governed by two features on C: i.e., [+pron], which can be strong or weak, and the associated Case feature. A strong [+pron] licenses a gap, while a weak [+pron] licenses a resumptive pronoun. The value of C’s Case feature may be either specified or unspecified in a given language. When its value is unspecified, either a gap (if [+pron] on the relevant C is strong) or a resumptive pronoun (if [+pron] on the relevant C is weak) can occur in any position regardless of the Case associated with that position. When specified, C’s Case feature is defined in terms of the value [±active]; a strong [+pron] feature is always associated with [+active] Case (i.e., NOM/ABS) whereas a weak [+pron] feature is always associated with [-active] Case (i.e., ACC/ERG). This approach accounts for the crosslinguistic generalization that if the distribution of gaps and resumptive pronouns is restricted in a language, a gap is permitted only in positions associated with the active Case and a resumptive pronoun only in positions associated with the non-active Case. Thus, at least one instance of syntactic ergativity is shown to be a manifestation of morphological ergativity.

The aim of this paper is to elucidate the syntactic mechanism of aspectually conditioned split-ergativity under the theory of feature checking recently developed by Chomsky (1995, 2000, 2001). It is argued that the distinction between accusative languages and ergative ones can be given a coherent explanation if Ura’s (2000) idea about the ergative parameter, which hypothesizes that SUBJ at the Spec of v can enter into a checking relation directly with v without movement, is appropriately revised and updated along the line of Chomsky’s (2000, 2001) theory of Agree and Case. Moreover, the theory of aspect checking, which has its root from the recent proposal by many researchers such as Borer (1994, 1998) and Ramchand (1997) is introduced with a Minimalist modification concerning the non-use of the new functional category, Asp or the like. With the proviso that the lexical aspect and the grammatical aspect are two determinants for the aspectual feature to be checked in syntax, this paper hypothesizes that the aspectual feature appears on v when the telicity due to the lexical aspect and/or the temporal boundedness due to the grammatical aspect is involved in the clause. By examining data mainly from Hindi and Georgian, both of which are languages with aspectually conditioned split-ergativity, it is demonstrated that, given both the theory of the ergative parameter and the theory of aspect checking, the case marking patterns found in intransitive and transitive clauses in those languages can be given a natural and consistent explanation.

This paper proposes a restrictive theory of ergativity, whereby ergative case is inherent (Woolford 1997), and absolutive case is eliminated in favour of nominative case and accusative case. Morphological case is dissociated from syntactic abstract case, such that morphology realizes the abstract case features in the most specific manner provided for by the morphological resources of the language. In Warlpiri, a language which lacks nominative and accusative case morphology, this results in both nominative and accusative abstract case being realized morphologically as unmarked, alias "absolutive". The proposed analysis of the Warlpiri case/ agreement system is shown to be superior to an analysis based on nonconfigurationality (Jelinek 1984).

This paper presents an account a case-shift phenomenon conditioned by Aspect: the ari construction, a certain type of progressive form in Basque, where ergative case is not assigned (1):

(1) a. emakume-a-k ogi-a ja-ten du
woman-det-E bread-det eat-impf has
"The woman eats bread"

b. emakume-a ogi-a ja-ten ari da
woman-det bread-det eat-impf prog is
"The woman is eating bread"

The "ari progressive" illustrated in (1b) has been claimed to be an instance of split ergativity or antipassive by Postal (1977). I follow Hualde & Ortiz de Urbina (1987) in assuming that the element ari is a verb that takes a nominalized clause as a complement. I also assume that progressive aspect is realized in syntax as a locative predication (Bybee et al (1994), Demirdache & Uribe-Etxebarria (1997)), yielding an unaccusative structure (Mateu & Amadas (1999)). Thus, I argue in detail that the contrast in (1) results from the fact that the ari construction involves a biclausal syntactic structure, where the main verb ari is unaccusative, and the embedded clause is a locative PP containing a nominalized clause (2):

(2) a. emakume-a [PRO ogi-a ja-te] n] ari da
woman-det [PRO bread-det eat-nom]loc] engaged is
'The woman is engaged in dance' (the woman is dancing)

This structure, in turn, necessarily yields the assignment of absolutive case to the main clauses’s subject, regardless of the nature of the verb embedded in the adjunct clause. Therefore, progressives do not cause a "split" in the Case system; rather, the contrast in (1) follows necessarily given the syntactic structure underlying it. Finally, I argue that there is a process of grammaticalization taking place in some eastern varieties of Basque, whereby ari has become a functional aspectual marker. In these varieties, the structure of the progressive is no longer biclausal, and thus ergative case surfaces in the case of transitive verbs.One last conclusion that can be drawn from this account is that ergative case in Basque appears to be indistinguishable from the agent theta-role, since the only case where this correspondence was thougth to fail were the subjects of examples like (1b), which, under this analysis, turn out to be instances of themes, not agents.

Halkomelem Salish displays ergative properties which are split along two dimensions: only 3rd person display an ergative pattern (and not 1st or 2nd) and only in the indicative mood (and not in the subjunctive). This paper develops an analysis without a special "ergativity" parameter. Rather, it is argued that the split ergative behaviour derives from the morpho-syntactic properties of agreement endings in the language: 1st, and 2nd person agreement as well as subjunctive agreement is analyzed as occupying a functional head high up in the clause (C) whereas 3rd person indicative agreement is analyzed as occupying a lower head (v). In addition, the paper argues that v only projects in the presence of a transitivizing suffix. Consequently 3rd person indicative agreement is only licensed in transitive clauses, whereas all other agreement affixes appear independent of the transitivity of the predicate.

In this paper, we address the following questions. Do ergative languages form some kind of homogeneous class? What are the (syntactic) ingredients of ergativity? To answer these questions, we begin by assessing the ergative analysis of Malagasy, a Western Austronesian language. Malagasy both fits and does not fit the definition of ergative. Further, languages such as Bahasa Indonesia have some constructions that appear to be ergative and some that appear to be nominative/accusative. We conclude that ergativity characteristics vary not only from language to language but even within languages and within particular constructions raising serious doubt as to whether there is a macroparameter of ergativity and even whether there is a necessary clustering of ergative characteristics in any construction.