Identifying the neural structures and dynamics that regulate phonological structure

点击翻译按钮获取中文摘要

The systematic patterning of language is a fundamental property of cognition. One aspect of this patterning, constraints on the combination of speech sounds to form words (phonotactic structure), has been implicated in constraining diverse processes related to language acquisition, perception, and production, bilingual language use, memory and even influences non-linguistic processes including the memorability of novel words and consumer reaction to novel brand names. This patterning changes in a variety of common acquired and developmental communication disorders. We will explore two explanations for these effects. One, developed in linguistic theory, argues that language users discover a set of abstract, language-specific rules or constraints that shape language use. The other view, developed in connectionist and dynamic systems theory, argues that phonotactic constraints emerge from top-down lexical influences on speech perception. Discriminating between these approaches is difficult because both explain behavioral data well. It is essential to discriminate between these accounts for two reasons. This question offers an excellent opportunity to resolve the debate over whether abstract linguistic rules/constraints create or simply describe the patterning of language. The resolution of this question has fundamental implications for the way linguistic formalism and connectionist simulations relate to human processing. At a more immediate level, this research offers the opportunity to identify a common core mechanism (either the leveraged use of abstract linguistic rules or top-down lexical influences) that explains and unites diverse linguistic and cognitive phenomena. Past efforts to resolve these issues have failed because of fundamental inferential limitations of behavioral and BOLD imaging paradigms. Accordingly, we have developed new tools and research strategies that allow us to identify patterns of directed interaction between brain regions (effective connectivity), and use these analyses to draw much stronger inferences about the dynamic processes that shape cognition. Observers in the field have argued that our methods have already provided “definitive” evidence to resolve the decades old debate over the role of top- down processes in speech processing. This proposal would extend those methods, and introduce innovative neural decoding analyses that we will use to characterize the categories (e.g. rules, words, abstract phonological representations needed to support rule application) that are encoded in localized brain activity. Using these methods, we will determine whether top-down lexical processes that we have shown produce phonotactic phenomena related to the processing of patterns that occur in speaker's language generalize to unfamiliar patterns. We will also use them to identify the substrates of rule- versus word-mediated processing, to provide a baseline for interpreted the representations and dynamic processes that support phonotactic effects in natural language processing.