Towards a proper treatment of coercion phenomena

Abstract

The interpretation of coercion constructions (to begin a book) has been recently considered as resulting from the operation of type changing. For instance, a phrase of type o (object) is coerced to a phrase of type e (event) under the influence of the predicate. We show that this procedure encounters empirical difficulties. Focussing on the begin/commencer case, we show that the coercion interpretation results both from general semantic processes and properties of the predicate, and we argue that it is best represented at the lexical level. The solution is formulated in the HPSG formalism, where the lexical description of heads includes a specification of the argument and articulates syntax and semantics. We propose that the properties attached to the complement remain the same as they are oustside the construction, but that the semantics of the predicate is enriched to include an abstract predicate of which the complement is an argument.

Coercion: Definition and challenges, current approaches, and new trends

“…For the prototypical case of begin (cf. Example (1)), for instance, it has been argued (in particular by Godard and Jayez [1993], with evidence from gapping and anaphoric coreference) that no type change actually occurs on the complement when coercion applies. They propose including an abstract predicate (representing the event) of which the complement is an argument.…”

Aspectual Coercion and Logical Polysemy

“…Finally, there would appear to be constraints on coercion which suggest that agentivity may play an important role in licensing the operation. For example, as pointed out by Godard & Jayez (1993), sentences such as (7) and (8) are ill formed:…”

“…In this section, we review these apparent counterexamples and discuss each briefly to show why they are nonproblematic. We will concentrate, however, on the selectional properties of aspectual verbs such as commencer and begin, in order to show very clearly that sense enumerative approaches, such as that taken by Godard &Jayez (1993) are unable to capture linguistic and computational generalizations, with respect to how the lexicon contributes to the compositional semantics of natural language.…”

A computational model of logical metonymy

“…Our system first derives a set of possible metonymic interpretations from a large corpus. It then disambiguates them with respect to their word sense, using an existing sense inventory, and automatically organizes them into a new class-based conceptual model of logical metonymy that is inspired by linguistic theory [Vendler 1968;Pustejovsky 1991;Godard and Jayez 1993]. We then experimentally verify whether this representation is intuitive to humans, by asking human subjects to classify metonymic interpretations into groups of similar concepts.…”

“…(6) After three martinis John was feeling well. [Godard and Jayez 1993] The metonymic adage in (1) is a classical example. Here the pen stands for the press and the sword for military power.…”