Acoustic measures of planned and unplanned coarticulation

We present a replication of an experiment [23] where vowel-to-vowel coarticulation was found when an utterance was initiated before an intervening consonant’s identity was known. English nonsense strings [əәˈbVCɑ] were used. In one condition, the V was known but the C was not, with the reverse in the other condition; missing information was presented once phonation began. Results show anticipatory vowel effects on the final portion of the schwa (transitions into the stop) occurred only when the vowel was known ahead of time. Anticipatory effects of the V on the schwa’s F2 were found throughout its duration when the V was known, but only for the speaker with the shortest V duration. Perseverative effects on the final vowel were similar in both conditions, as was consonant coarticulation. The implications of these results for planning are discussed, and the value of replication in the social sciences is emphasized.