|
Agreement Production Through Statistical Learning| old_uid | 3486 |
|---|
| title | Agreement Production Through Statistical Learning |
|---|
| start_date | 2007/11/20 |
|---|
| schedule | 11h-12h |
|---|
| online | no |
|---|
| summary | The production of grammatical agreement in sentences (par example, Le chat mange vs. Les chats mangent) is usually thought to reflect passing of abstract features (such as +plural) through a syntactic tree. I will describe an alternative constraint satisfaction account of agreement production in which agreeing forms (mange, manges, mangent, etc.) are activated as a function of the context?the message to be conveyed, the syntax of the utterance, and the lexical items activated for the utterance. I will argue that these contextual effects on agreement production emerge via statistical learning over a lifetime of language use. Data from several English agreement production studies will be presented in the context of manipulations of various kinds of language statistics, including noun-verb co-occurrences and linear order in sentences. I will also present data from agreement priming studies, in which we manipulate the participant's experience with agreement patterns in English, in order to observe the effects of statistical learning on agreement production in the laboratory. |
|---|
| oncancel | Exceptionnellement un mardi |
|---|
| responsibles | Pélissier |
|---|
| |
|