ODG – Salle 166
Robert Kluender (Université San Diego)
The Role of Working Memory in Syntactic Representation and Processing (3/4)
The session returns to the implications of results from both the verbal and visual working memory literatures for the nature of the human lexicon and syntactic representation. There are at least two main issues. First, if working memory truly reduces to the focus of attention operating over long-term memory representations, then what is the exact nature of the linguistic representations in long-term memory over which the attentional system operates in sentence processing? Various aspects of the working memory literature make reference to “chunks” and “chunking.” On certain theoretical accounts, there is no limit placed on the number of elements that can make up a chunk when it is compiled on the basis of information stored in long-term memory, while chunks that are not pre-existing but formed rapidly on the basis of new associations are limited to four (or, on some accounts, fewer) elements, due to constraints on how many items can simultaneously remain in the focus of attention. What does this imply about the nature of lexical entries in long-term memory and how they are assembled on line into syntactic representations? Second, there is ubiquitous reference in the working memory literature to featural representations (since much of it is based on evidence from visual working memory paradigms), and evidence from similarity-based interference paradigms in the verbal working memory literature likewise suggests that retrieval is primarily feature-driven. If this is true, what constraints does this fact impose on syntactic representations? To what extent do existing syntactic theories satisfy these constraints, and can we use such constraints as metrics to differentiate and possibly evaluate theories of syntactic representation against each other?