Nonsegmental Factors in Foreign Accent

Untrained native English listeners assigned foreign accent scores to sentence and narrative utterances that had been rendered unintelligible through low-pass filtering. Utterances produced by native English talkers were assigned consistently higher ratings than those produced by Mandarin-speaking learners of English, even when the listeners were unfamiliar with the content. Because these filtered speech stimuli contained very little of what could be considered segmental information, the results suggest that untrained listeners can identify foreign-accented speech on the basis of nonsegmental information alone, whether they are presented with material of known or of unknown content. Acoustical analyses of the stimuli suggested that differences in speaking rates, intonation patterns, and timing may have played a role in the listeners' assessments, although the cues to foreign-accentedness may have varied from talker to talker and from utterance to utterance. Surprisingly, no relationship was observed between the ratings of the filtered and unfiltered versions of the nonnative stimuli.