Language Typology and Syntactic Description, Volume I: Clause Structure, Second edition
Reconsidering grammatical relations
Download 1.59 Mb. Pdf ko'rish
|
Lgg Typology, Synt Description v. I - Clause structure
4
Reconsidering grammatical relations In the conception of grammatical relations that has been assumed by our work so far, each np in a clause bears a single grammatical relation. One prob- lem for this view arises from syntactic ergativity: in some languages with 20 Dixon (1977a:393) observes that participants in a conference session devoted to whether various Australian languages were syntactically ergative or not were frequently doubtful of the correct treatment of their languages, often changing their minds in the course of preparing final versions of their papers. 198 Avery D. Andrews ergative features, it is unclear whether the subject grammatical relation should be regarded as expressing a/s or p/s functions, since the evidence is weak, or, as we shall see below, contradictory. But syntactic ergativity is not the only problem for grammatical relations. Another set of difficulties comes from the so-called ‘Philippine type’ of language structure, which seems in a sense to have two systems of grammatical relations functioning at the same time. In this section we will suggest a solution to both problems, which originated with some proposals by Keenan (1976c) about the nature of the subject concept, and has been developed by many other authors since then. 21 The basic idea of the solution is that the familar concept of subject should in fact be split into two concepts, one associated with the semantic role of agent, the other with the pragmatic role of topic. In English these two concepts pick out the same np in the sentence, but in certain others, such as syntactically ergative languages and the Philippine type, they don’t. Therefore in English we have the grammatical relation of subject, while in some other languages we must distinguish what we might call ‘a-subject’ (agent-oriented) from ‘p-subject’ (pragmatic, topic-oriented, pivot). We will first show how this idea helps with the problem of ‘mixed ergativity’, where a language shows a combination of ergative–absolutive and nominative– accusative organization. We will then show how it applies to the problematic features of the Philippine type. 4.1 Mixed syntactic ergativity Early work on syntactic ergativity assumed that languages would either show ergative–absolutive or nominative–accusative organization, depending on whether their sentence structures treated s and a alike (the majority), or s and and p. But this assumption proved initially to be dubious and, ultimately, false. An initial reason for doubt is that the evidence for setting up the grammatical relations one way or another in ergative languages is often rather weak, as we saw for example in Yidi . More serious is the fact that there are often contra- dictory indications about which way they should be set up. Sometimes one can make a case that the syntactic phenomena are showing ergative–absolutive orga- nization, and that the apparently nominative–accusative phenomena are being semantically conditioned, but there are also instances where it is clear that some syntactic phenomena are ergative–absolutive, while others are nominative– accusative. An example of the first type is Yidi (Dixon (1977a)), of the sec- ond, Inuit (Bittner (1994); Manning (1996)). Yidi also illustrates the rather common case where the available evidence about grammatical relations is rather scanty, so we will discuss it first. 21 Such as Schachter (1976, 1977), Foley and Van Valin (1984), Guilfoyle, Hung, and Travis (1992), Kroeger (1993), Wechsler and Arka (1998), and others writing on Austronesian languages, and Dixon (1979), Bittner (1994), and Manning (1996) on syntactic ergativity. The major functions of the noun phrase 199 We have already seen the evidence for an s/p grouping (putative p-subject) in Yidi (see (98)). But there are two constructions that treat a and s alike as opposed to p. The first is the imperative: imperatives require a second (or occasionally a first) person pronoun in s/a function (Dixon (1977a:370–1)). Imperatives with s and a addressees are illustrated below: (102) a. ( undu) guwa gali-n vou(sg) west go- imper ‘(You) go west!’ b. ( undu:ba) bu a wawa you(pl) woman watch(imper) ‘(All of you) watch the woman!’ The second involves a number of particles whose grammar treats s and a alike (Dixon (1977a:372–82, 387)). For example the particle gana: ŋ gar indicates that the referent of the np in s or a function was the first to perform a certain action: (103) a. ŋ ayu gana: ŋ gar gali:- I(nom) first go-past ‘I went first’ b. ŋ ayu gana: ŋ gar gunda:-l I(nom) first cut-past ‘I was the first person to cut [that tree]’ But these phenomena don’t constitute a truly compelling case for saying that the syntax is recognizing an s/a category, because what might be happening is that the phenomena have a semantic rather than a syntactic basis. In particular, the semantics of both the imperative and the particle constructions might be such that they involve an agentive argument in their interpretation, and impose constraints on it. Since ps are never agents, ps won’t be able to be involved in these constructions, for reasons that are quite independent of how the syntax is organized. In principle, one could investigate this issue by looking at intransitives with nonagentive s, and also antipassives, but Dixon doesn’t provide significant dis- cussion of this, although he does provide a suggestive example of the ‘cessation’ Download 1.59 Mb. Do'stlaringiz bilan baham: |
Ma'lumotlar bazasi mualliflik huquqi bilan himoyalangan ©fayllar.org 2024
ma'muriyatiga murojaat qiling
ma'muriyatiga murojaat qiling