Abstract:
Speech naturally occurs over time---words unfold from ``left-to-right.'' This structure inherently confounds two logically distinct factors: the amount of phonetic information presented, and the amount of time the perceptual process has been working. For example, in the word ``acoustical,'' both the elapsed processing time and the number of processed phonemes (or syllables) would be greater for the second /k/ (``c'') than for the first one. The current study attempts to unconfound these two factors by using speech compression and expansion techniques, coupled with phonemic restoration methodology. The strength of phonemic restoration can be used as an index of the strength of lexical activation. For example, restoration is stronger for word-final phonemes than for phonemes in earlier word positions, due to the higher lexical activation later in words. The current study tests phonemic restoration when available processing time is manipulated via speech compression/expansion. An interesting additional variable that appears to interact with time is the active lexical cohort size: Additional processing time seems to be more useful when the cohort has been narrowed to a small set. The results support and constrain activation models of lexical access. [Work supported by NIMH.]