|Full Title:||Identity in Ellipsis|
|Start Date:||20-Sep-2013 - 21-Sep-2013|
|Meeting Email:||click here to access email|
Identity in Ellipsis
LUCL, Leiden University, 20-21 September 2013
Identity in ellipsis is a long-standing issue in linguistics. Virtually all accounts of ellipsis phenomena assume that the elided material must be identical in some way or other to material in a suitable antecedent available in the discourse. The nature of the identity condition and its precise formulation is under heated debate in current research. The main question is whether identity holds at some syntactic level of representation, or some semantic one, or both.
The syntactic account of identity (Sag 1976, Williams 1977, Fiengo & May 1994, Chung et al. 1995, Baltin 2012, Johnson 2012, Chung 2013) holds that ellipsis is recovered under a form of structural identity with the antecedent, defined over phrase markers of some sort, in most but not all cases, identical LF representations. Clearly, surface identity is not required, as shown by possible inflectional differences on verbs (see Warner 1985 for exceptions, and Lasnik 1995 for an account), cf. (1) (elided material appears in square brackets).
(1) a. They eat nattoo and John [eats nattoo], too.
b. John has eaten nattoo and Bill may [eat nattoo], too.
Semantic accounts, on the other hand, operate with an identity relation stated purely over semantic representations (Dalrymple et al 1991, Sag & Hankamer 1984, Hardt 1993, Ginzburg and Sag 2000, Merchant 2001, Anderbois 2011). Support for the semantic approach comes from various phenomena. One is ‘vehicle change’ (Fiengo and May 1994), others are syntactic mismatches of the kind in (2) (Merchant 2005):
(2) I remember meeting him, but I don’t remember when [I met him].
Semantic accounts are also supported by the availability of non-isomorphic antecedents in clausal ellipsis like sluicing (Postdam 2007, Vicente 2008, van Craenenbroeck 2010).
Syntactic identity is heavily supported by the lack of argument structure alterations (Chung 2006) and the lack of active-passive mismatches in clausal ellipsis (Merchant 2008), such as (3), since active and passive sentences differ from each other only in their syntax but not in their semantics.
(3) *Joe was murdered, but we don’t know who [murdered Joe].
In addition to purely syntactic and semantic approaches, there also exist ‘hybrid’ approaches, according to which both semantic and syntactic identity can guide the recovery of the elided material. In the most prominent type of hybrid approach (Kehler 2002), it is the discourse relation between the elliptical clause and its antecedent that determines the choice between syntactic and semantic identity and impacts the elidability of constituents.
Psycholinguistic research on ellipsis has also begun to address the issue of ellipsis identity, in order to understand what guides the parser in the recovery of the antecedent. It has been found that various forms of antecedent-ellipsis mismatches receive a systematic cline of acceptability (Kim et al 2011), which might suggest that identity is syntactic but non-identical ellipsis can be ‘repaired’ and become acceptable if the repair is not too difficult (Arregui et al 2006).
We are pleased to announce that the following invited speakers have agreed to give a talk:
Jeroen van Craenenbroeck
|Linguistic Subfield:||Psycholinguistics; Semantics; Syntax|
|Calls and Conferences main page|