Formal research on language has contributed greatly to our understanding of the brain. A priority for my work is asking how models of language developed in linguistic theory are implemented in the brain. Conversely, I also use behavioral and neuroimaging methods to inform linguistic theory.

The processes elicited by island violations

Some sentences are impossible for no good reason. Linguists call some of these sentences "island effects", so-called because they involve certain structures which prevent a grammatical operation called movement, in other words, these structures form islands that a phrase cannot escape from. Many different theories of island phenomena have been proposed, but it is difficult to distinguish these theories because they all predict that the sentences will be impossible. However, functional neuroimaging can potentially distinguish theories by revealing the processes that are elicited by these sentences. We have performed an fMRI study on subject islands, one class of islands, and compared the processes elicited by these islands with phrase structure violations (PSV) (Matchin, Almeida, Sprouse & Hickok, in preparation). Our results have shown that subject islands activate completely different neural systems from PSV. Subject islands elicit increased activation in areas involved in semantics, while PSV elicit increased activation in verbal working memory networks. These results suggest that subject islands involve primarily difficulty in semantic interpretation. These results are incompatible with prominent theories of island effects that hold that these violations result from an overload of working memory resources.

Good: Which candidate does the moderator of the panel think __ avoided the debate's questions about healthcare?


Island: *Which candidate does the moderator think [the speech by __] ruined the debate's questions about healthcare?


PSV: *Which candidate does the moderator of the panel think __ avoided the debate's about questions healthcare?


Both: *Which candidate does the moderator think [the speech by __] ruined the debate's about questions healthcare?

Island + Both > Good + PSV

PSV + Both > Good + Island


Argument structure & the brain

A fundamental linguistic distinction holds between selection and adjunction. In syntax, some elements are combined into head-complement configurations, while others are adjoined together. In semantics, a predicate takes arguments which together form events, while an argument can be combined with a modifier to enrich its meaning. In an fMRI study (Matchin, Liao, Gaston, & Lau, in preparation) we have shown that areas most strongly associated with combinatorial semantic processing, the anterior temporal lobe (ATL) and the angular gyrus (AG), do not discriminate between VPs (selection, argument structure) and NPs (adjunction, no argument structure) when lexical items are perfectly controlled between conditions. By contrast, areas more closely associated with syntax (IFG, pSTS) do discriminate these conditions.

© 2017 by William Matchin Personal

  • Twitter Clean Grey