LSA 13 Workshop: How the brain accommodates variability in linguistic representations

Basic Info

  • Date: 7/14
  • Time: 08:30-18:30
  • Location: LSA Institute, Auditorium C, Angell Hall (435 S. State), University of Michigan, Ann Arbor

Invited Speakers

Sheila Blumstein Gary Dell Joan Bresnan Molly Babel Ann Bradlow Jerome Bellegarda
Professor of Cognitive, Linguistic & Psychological Sciences Professor of Linguistics and Psychology, The Beckman Institute Senior Researcher at Stanford's Center for the Study of Language and Information Assistant Professor of Linguistics Professor of Linguistics, Chair of the Department of Linguistics Apple Distinguished Scientist
Brown University University of Illinois Stanford University University of British Columbia Northwestern University Apple Inc.


Each time that we communicate, we encode an intended message into a series of motor commands. The resulting sound wave travels to our interlocutors, who decode the auditory signal to infer our intentions. This process is necessarily noisy: no two instances of a /p/ are exactly alike in terms of the physical signal that corresponds to them. Combinations of acoustic features seem to define probability distributions over phonemes, words, and sentences rather than deterministically mapping onto them. Decades of research in phonetics and psycholinguistics have not been able to identify an invariant base, in which sounds are deterministically distinguishable based solely on the properties of the physical signal. Yet, we usually understand each other – a fact even more remarkable once we consider that speakers differ in their linguistic realization preferences. For example, the acoustic distributions corresponding to one speaker’s /p/, can be physically more similar to the acoustic distributions of another speaker’s /b/ than the distribution of that speaker’s /p/. But variability does not stop here: even productions by the same speaker can differ over time. This variability within and across speakers is not limited to accents – speakers also seem to differ in their lexical and structural preferences. How then do we understand each other and to so usually without much effort?

To this date, it has been impossible to find a mapping from variable input onto an invariant base (cf. the infamous ‘lack of invariance’ problem). To the contrary, recent work from phonetics to syntactic processing suggests that we continuously adjust our linguistics expectations and, perhaps, even representations, thereby raising questions about the nature of linguistic representations.

This workshop aims to bring together researchers in cognitive psychology, psycholinguistics, computational linguistics, sociolinguistics, and generative as well as usage-based traditions in linguistics, who work on variability in production and comprehension. How do we learn to understand a new speaker? To what extent are we able to maintain what we have learned about a new speaker? To what extent can we extrapolate and generalize what we have learned in the past to new speakers we encounter? And, what is the nature of linguistic representations that seem to afford such remarkable flexibility? Are all levels of linguistic representation and processing affected by variability to the same extent? Are the brain’s solutions to variability at different levels of linguistic representations the same or different? What is the relation between productive and receptive competence and how can similarities and differences between these two types of abilities be reconciled with existing linguistic theories of ‘grammar’?

Understanding how the brain overcomes variability is one of the central questions in the language sciences and beyond. Answers to this question affect linguistic theories as much as theories of sentence production and language understanding.

The workshop will consist of a series of invited plenary talks, for which we invite audience discussion. During the lunch break, we will have an extended poster session to facilitate discussion between researchers from the various disciplines we hope to bring together.

Please direct inquiries about this workshop at the organizers: