When arguments don't fit with verbs: Interdisciplinary approaches to metonymy and coercion


In Stuttgart, Germany

Sponsored by Stuttgart University, SFB 378.

Metonymy (John began the book, Mary enjoyed the beer, Ben read Dickens yesterday) is a classical case of verb-argument mismatch and therefore poses a challenge to strong compositional approaches. The topic has always been of great interest to researchers from different communities concerned with linguistic research, who have provided very different accounts of the phenomenon.

Nevertheless, there seems to be little agreement on what type of resources are involved in the processing of these phenomena, and in the end on their very nature. The Generative Lexicon approach (Pustejovsky 1995, Jackendoff 1997) accounted for metonymy as a case of type-clash (verbs like begin sub-categorize for events) and type-coercion (the book is coerced into an event-denoting argument, determined by its telic quale to read). Criticism of this approach has been voiced from various sides (Fodor and Lepore, 1998; Asher, 2007; De Almeida and Dwivedi, 2008), who argue that metonymy resolution is driven by pragmatic, context-driven inferences rather than the internal structure of lexicon entries in isolation. Psycholinguistic research (see Pylkkänen and McElree 2006 for a review), corpus linguistics (Briscoe et al. 1990, Vespoor 1997), and more recently computational models (Lapata et al. 2003) and neurolinguistic research (Pylkkänen and McElree 2007, de Almeida et al. 2009) have seeked to support theories of metonymy and to provide empirically-based accounts of the phenomenon. Furthermore, work by McRae and colleagues (Matsuki, et al. 2009) challenged the idea that selectional restrictions as lexically-based, suggesting that they might have the same nature of event-based knowledge.

Also the nature of other phenomena of metonymy, such as Ben read Dickens yesterday, pose the question of what cognitive resources are involved in their processing (Hahn & Markert 1997, McElree & Frisson 2006, Frisson & Pickering 2007): to what extent are metonymies conventionalized, and how can we account for creativity in language?

The recent advancements in psycholinguistic and neurolinguistic research, together with the results achieved by corpus-based models, have led to a recent stronger interest in metonymy and selectional preference/restrictions, with a new awareness that the common effort from different communities can help shed new light into these phenomena. The aim of this interdisciplinary workshop is to bring together researchers from a wide range of fields such as linguistics, computational linguistics, computational lexicography, psycholinguistics, cognitive science and neuroscience, to foster the development of interdisciplinary projects and to gain a better understanding of the (computational) linguistic and cognitive properties of metonymy.