Maryellen MacDonald

Department of Psychology

University of Wisconsin – Madison

http://lcnl.wisc.edu/people/mcm/

Friday, October 2

12:00-1:30 pm

The PDC -- Where Language Production, Comprehension, and Acquisition Meet

Inferences from behavioral data to perceptual or cognitive processes are

at the heart of theorizing in many areas of cognitive science. In

language research, these inferences shape hypotheses about the language

comprehension system: If a sentence with syntactic structure A is

easier to understand than one with structure B, then the architecture of

the comprehension system might be hardwired to prefer A-type structures,

and/or B-type structures might exceed humans' verbal working memory

capacities while A-types do not. The Production-Distribution-Comprehension

(PDC) account provides an alternative view that rejects inferences of this

type. Instead, it places substantial explanatory burden on people's statistical

learning from linguistic experience, and it traces many classic results not to

the comprehension architecture but to the way that language production

works. Using language production studies, corpus analyses, natural and

artificial language learning studies, and language comprehension

studies, I will argue that language production processes promote certain

kinds of utterances over others, leading to robust distributional

patterns in language use; that implicit learning processes track these

distributional patterns in fine detail; and that comprehension processes

bring this statistical learning to bear on new input, so that behavior

previously taken to reflect the nature of the comprehension architecture

instead reflects learning about other people's productions.