| |
Abstract:
Abstract: This study investigates temporal integration times
in the context of speech stimuli. Vowels in which the first formant
(F1) sets on asynchronously are associated with a change in percept
(e.g. from /a/ to u/). Vowels /u/ and /a/ have different
neuromagnetic evoked response latencies for the M100 component,
detectable by MEG; we evaluated a continuum of F1 onset asynchrony,
in terms of perception of 1 vs. 2 tokens (/u/ vs. /a-u/) and in
terms of M100 latency. 9 subjects underwent MEG recording; 18
stimuli were presented (from /u/ to /a-u/, with F1 onset asynchrony
varying up to 200ms). In six of nine subjects, the following was
observed: for F1 onset asynchrony up to 20-35ms, evoked response
latency was constant. At greater asynchrony, increasing M100
latency was observed, appearing commensurate with F1 onset
asynchrony (the M100 response component appeared elicited by onset
of the second token, /u/). The threshold at which this occurred was
the same as that separating a percept of one vs. two tokens. We
conclude that M100 latency for stimuli formed with F1 onset
asynchrony is indicative of the percept of 1 vs. 2 tokens within
the stimulus and the magnitude of onset asynchrony. As such M100
latency reflects temporal processes of perception formation.
Supported by NSF-LIS grant SBR9720398.
|
