Current models of spoken word recognition take into account factors such as
word frequency, word onset cohort size, and phonological neighborhood dens
ity. Using the word onset gating technique we tested word recognition when
bandpass filtering was used to allow subjects to hear the full prosodic pat
tern of a word (number of syllables and syllabic stress), deprived of segme
ntal information beyond that contained in the onset gate. Subjects also hea
rd either word onsets plus duration information or only word onsets. Result
s suggest that word prosody is represented in the mental lexicon and is eff
ectively used by listeners in spoken word recognition. (C) 1999 Academic Pr
ess.