The perception of rate induced resyllabification in English

Stetson (1951) noted that, when repeated, singleton coda consonants (VC) appear to modulate into onset consonants (CV) as the rate of repetition increases. The current study examines whether nave listeners perceive such resyllabifications, and whether such perceptions are affected by the voicing of the resyllabified consonants. Stimuli were extracted from production experiments in which talkers repeatedly produced singleton voiced and voiceless stops in the CV or VC position. Speakers entrained to a metronome which increased tempo from 450 to 150 ms/syllable. Open-set identification revealed that (1) while slow VCs are identified as such, fast VCs are identified as CVs a majority of the time and (2) CVs are rarely identified as VCs; however (3) both CVs and VCs at fast rates are often identified as CVCs, especially when the consonant is voiceless. A forced-choice identification task indicates that fast VCs and CVs, while clearly differentiated at slow rates, are both identified as CVs 80% of the time at fast rates. These results support Stetson’s observations, but like previously reported production results, indicate that fast rate tokens are partially ambiguous between CV and VC forms, an ambiguity which can get resolved by splitting the consonant in question. [Work supported by NIDCD and NSF.]