Generalized Event Knowledge in Logical Metonymy Resolution
Alessandra Zarcone, Sebastian Padó · eScholarship (California Digital Library) · 2011
(1) John began the book ➮ John began reading the book (2) John is a famous wrestler. He really enjoys a good fight ➮ He enjoys fighting a good fight (3) John is a wrestling fan. He really enjoys a good fight ➮ He enjoys watching a good fight the interpretation of LMs involves the integration of a covert event how is this implicit knowledge retrieved? 2. Lexicon vs. world knowledge The lexical hypothesis (Pustejovsky 1995, McElree et al. 2001): type-mismatch (event-subcategorizing verb + entity-denoting object) qualia structure in the lexicon (book: reading OR writing) plausible, but not dynamic enough The pragmatic hypothesis (Fodor and Lepore 1998, De Almeida and Dwivedi 2008): dynamic inferences based on context and world knowledge not specified enough: what constraints apply? 3. An alternative hypothesis Generalized event knowledge (GEK) (McRae and Matsuki 2009): knowledge of typical events, first and secondhand experience, available in our memory wash hair ➮ shampoo, sink, bathroom, indoor wash car ➮ hose, outdoor words in isolation immediately activate GEK words can rapidly combine in sentences to cue specific concepts that are relevant to GEK scenarios syntactic cues modulate expectations for certain aspects of GEK GEK can immediately modulate expectations for syntactic structure Matsuki et al. (in press): (1) Dana used the hose to wash her filthy car. typical (2) Dana used the shampoo to wash her filthy car. atypical (3) Dana used the shampoo to wash her filthy hair. typical (4) Dana used the hose to wash her filthy hair. atypical (1) and (3) show shorter reading times and eye fixations at the patient position than in (2) and (4): facilitation effect when expectations about “typical ” washing scenarios are met different linguistic contexts tap into different GEK scenarios dynamic model Research question: can GEK account for covert event retrieval in