| |
Abstract:
Models of word recognition differ in their placement of
inhibition: some treat it as suppression of lexical activation
(e.g. Marslen-Wilson 1987) and others as a post-lexical conscious
strategy (e.g. Neely 1977, Forster 1981). We investigated the time
course of inhibitory effects in lexical decision using
magnetoencephalography (MEG). Based on previous studies (Gonnerman
1999, Rastle et al. 2000), we predicted reaction time (RT)
inhibition for pairs such as SPINACH-SPIN. In the MEG data, we
focused on the M350, a 300-400ms MEG response component, which can
be argued to reflect lexical activation. First, the M350 is the
first component whose latency predicts the frequency and repetition
priming effects on reaction times (RT) (Embick et al., to appear,
Pylkkanen et al., to appear). Second, M350 latencies vary
independently of RTs in tasks where activation contrasts with the
post-access decision process: stimuli with a high phonotactic
probability elicit fast M350s due to a sublexical frequency effect
but longer RTs due to a neighborhood competition effect which slows
down the word/non-word decision (Pylkkanen et al, to appear). The
present study used Gonnerman's crossmodal priming materials to test
whether the M350s of the targets in SPINACH-SPIN type pairs show
shorter latencies, due to phonological priming, or longer
latencies, due to inhibition. Post-lexical theories inhibition
predict the shorter latencies while lexical theories predict the
longer.
|
