Ellipsis: computation of

Abstract

A computational account of ellipsis should specify not only how the meaning of an elliptical sentence is computed in context but also a description of what is being computed. Many proposals can be divided into two groups, as per whether they compute the meaning of an elliptical sentence based on the semantic or the syntactic parts of its context. A unifying theme of these proposals is that they are all based on the idea that the meaning of an elliptical sentence is determinable based on a structured representation which is transformationally related to its surface syntactic structure. WIREs Cogn Sci 2012, 3:411-418. doi: 10.1002/wcs.1168 For further resources related to this article, please visit the WIREs website.

Action-Based Grammar

“…Categories, or so-called nowadays syntactic types, still remain sets of strings expressing hierarchical constituency assumptions only now paired with semantic interpretation. We do find accounts based on Categorial Grammar and the general framework of Minimalist Grammars (MG: Stabler 1997;Kobele 2012aKobele , 2012bKobele , 2015) much closer to ours than standard syntactic accounts like the traditional minimalist/HPSG ones we mentioned in the paper. Additionally, we find Kobele's (2015) account of ellipsis and Kobele and Merchant's comments more convincing than their previous accounts (Kempson et al forthcoming).…”

“…Secondly, such accounts adopt a standard competence-performance distinction, interpreting Marr (1982), in seeing the grammar as providing "a description of what is being computed" (Kobele, 2012b), i.e., propositional knowledge-that, with the parser/generator implementing algorithms based on the grammar. DS explicitly rejects such a view of the distinction between computation vs. algorithm which is based on a view of language as a code, to be analysed on the model of formal languages (Gregoromichelaki 2013a).…”

Language as Mechanisms for Interaction

“…Given that current formal linguistic theories take competence-performance (Chomsky 1995), modularity (Fodor 1983), and Marrian computational vs algorithmic level distinctions (Marr 1982;Steedman 2000;Kobele 2012b) as fundamental, 3 it is worth examining whether standard models can be enhanced with tools that handle the dialogue data while maintaining standard foundational assumptions. In such accounts, the grammar deals only with the licensing of "expressions" (structured stringmeaning pairs) in order to satisfy the required bottom-up compositionality of content for sentence-strings, with this level being unrelated to the order of words.…”

Eliding the derivation: A minimalist formalization of ellipsis

“…One of the biggest stumbling blocks to a unified theory of ellipsis (one which treats all elliptical phenomena as being the product of a single 'ellipsis' mechanism) is the fact that different 'sorts' of ellipsis have different properties (for more information see Kobele (2012a) and references therein). Most interesting to us here, as I will in fact be advocating for a theory involving exact syntactic identity, are the differences between sluicing (see Merchant, 2001, and references therein) and verb phrase ellipsis (vpe) (see Hardt, 1993, and references therein) with respect to the nature of the formal relation between antecedent and (supposed) ellipsis site.…”

“…The above account of ellipsis is stated at Marr's computational level, which describes what is being computed but not how. The most natural way of implementing the recognition of ellipsis in this context is to separate the detection of ellipsis sites from their resolution, at least logically (Kobele, 2012a) (although this separation can and should be 'parallelized' on-line). This allows perfectly standard minimalist grammar parsing algorithms (Harkema, 2001) to be used to construct parse trees with unresolved ellipsis sites, which are written in fraktur as E (upper case 'E').…”