Temporal extent of anticipatory lip rounding in child and adult speech

Abstract

Look-ahead models predict that anticipatory lip rounding is defined segmentally: for example, extending backwards from the target to the first segment specified for rounding. Gestural models predict that it is limited by the duration of the activation interval, which is why coarticulatory effects are weaker in slow speech compared to fast speech. This study investigated the temporal extent of anticipatory rounding in child versus adult speech with these model predictions in mind. Child speech is slower than adult speech. It is also purportedly equally coarticulated as adult speech, but long-distance effects are underexamined. Two groups of school-aged children, 5- and 8-year-olds, and one group of adults produced simple subject-verb-object sentences. The vowel of the object noun contained either the same unrounded vowel as the verb or a different rounded vowel. An unstressed determiner intervened between the verb and noun. Coarticulatory extent was measured perceptually using gated AV speech. Rounding was found to extend backwards to the preceding stressed vowel in both child and adult speech, and so up to (roughly) 400 ms before the target in 5-year-olds’ speech compared to 340 ms in 8-year-olds’ speech and 235 ms in adults’ speech. These results are more easily reconciled with look-ahead model predictions than with gestural model predictions. [Work supported by NICHD grant #R01HD087452.]Look-ahead models predict that anticipatory lip rounding is defined segmentally: for example, extending backwards from the target to the first segment specified for rounding. Gestural models predict that it is limited by the duration of the activation interval, which is why coarticulatory effects are weaker in slow speech compared to fast speech. This study investigated the temporal extent of anticipatory rounding in child versus adult speech with these model predictions in mind. Child speech is slower than adult speech. It is also purportedly equally coarticulated as adult speech, but long-distance effects are underexamined. Two groups of school-aged children, 5- and 8-year-olds, and one group of adults produced simple subject-verb-object sentences. The vowel of the object noun contained either the same unrounded vowel as the verb or a different rounded vowel. An unstressed determiner intervened between the verb and noun. Coarticulatory extent was measured perceptually using gated AV speech. Rounding was...