| |
Abstract:
Abstract: The reaction-time in a lexical decision task
increases as the frequency of words decreases. Some indication of
cortical responses reflecting this effect has come from
sentence-based ERP studies (e.g.King&Kutas(1995), but the
precise nature of frequency-sensitive neural responses in lexical
decision has remained unclear. The present study employed MEG in a
word/ non-word lexical decision task. Stimuli were divided into
eight bins: two non-word bins (pronounceable and
non-pronounceable); and six bins of open-class words. The words
were selected such that the average log- frequency (lemma frequency
from the COBUILD corpus) of the bins decreased linearly (2.8/mill
to -0.7/mill). Stimuli were presented visually; data were collected
in a 64 channel axial gradiometer MEG system. Preliminary results
(5 subjects) showed three primary components, consistent with
previous MEG studies of visually presented words (cf. Kuriki et
al.1996, Salmelin et.al 1996): at ~150ms (occip), ~250ms (temp) and
300+ms (temp) post stimulus onset (precise localization will be
reported in SF). The latencies of the first two components remained
constant across stimulus classes whereas the latency of the third
component (p <.05) and the RT latency (p <.001) increased
as the frequency of the stimuli decreased. Because lexical look-up
is estimated to occur earlier than this component (e.g.
Marslen-Wilson(1993)) the results suggest that word frequency
effects post-lookup processes in lexical decision rather than
pre-lexical processes or lexical look-up itself.
|
