Title: The Immediacy of Linguistic Computation
Abstract: Language unfolds over time. Thus, whether symbols in our minds or their physical realization in the world, if linguistic computations are not made over transient and shifting information as it occurs, they cannot be made at all. This talk explores some ramifications of this inherent temporal constraint—the Immediacy of Linguistic Computation—via two case studies in language processing.
First, I will present experimental evidence from a new perceptual learning paradigm which probes the content of speech representations. I demonstrate that real-time hypotheses comprise the probabilistic activation of discrete linguistic categories but include no meaningful retention of continuous acoustic-phonetic signal. Moving from perception to production, the second part of the talk will present an analysis of large-scale quantitative usage data on the English verb-particle alternation, providing a window into syntactic choices made during production. Results demonstrate how a mechanistic system can give rise to the appearance of highly efficient computational outcomes even without explicit optimization by individual language users.
Taken together, these studies highlight the utility of understanding the *intermediate representations* recruited during online processing and acquisition rather than strictly Input/Output mappings. We can place meaningful limits on the possible content of stable linguistic knowledge when we better understand what information is (or is not) maintained via ephemeral, on-the-fly, representations. I suggest that this may serve as a core explanatory bridge between areas of competence (linguistic representation) and performance (psycholinguistic behavior).
