You are here

N400 Evidence for Parallel Lexical Predictions

Date and Time: 
Thursday, April 18, 2019. 03:00 PM - 04:30 PM
Meeting Location: 
Building 460, Room 126
Workshop: 
Cognition and Language Workshop 2018
Meeting Description: 

There is a growing consensus, stemming from both behavioral and neurolinguistic data, that comprehenders generate probabilistic predictions about upcoming words. For example, N400 amplitudes vary gradiently with the cloze probability of a target noun. This is taken as evidence that word representations are pre-activated in proportion to their probability of being seen next (given the context). Computational theories of prediction have often assumed that this entails parallel lexical pre-activation, with multiple lexical items being predicted in parallel, each with different probabilities. We used ERPs to explicitly ask whether multiple probable candidate words are predicted in parallel in medium-constraint contexts. This account is widely assumed, but has not been explicitly tested in previous ERP paradigms of gradient lexical prediction. This is because most previous ERP studies have effectively confounded the probability of a given critical word (its cloze probability) with the probability of the most expected word, given its preceding context (its contextual constraint). We carried out an ERP study in which participants read short story contexts that generated predictions for at least two probable words. We found facilitation for all continuation words in proportion to their probability, regardless of whether a word was the single best completion for its context. Our findings supports a parallel gradient account of prediction in which multiple words can be pre-activated simultaneously in proportion to their probably of occurrence in context.

Emily Morgan is professor of linguistics at University of California, Davis.

Workshops Calendar

S M T W T F S
 
 
 
 
 
1
 
2
 
3
 
4
 
5
 
6
 
7
 
8
 
9
 
10
 
11
 
12
 
13
 
14
 
15
 
16
 
17
 
18
 
19
 
20
 
21
 
22
 
23
 
24
 
25
 
26
 
27
 
28
 
29
 
30