| Abstract | Coarticulation, the influence of one phoneme on another in the same utterance, has been the crux of studies of the phonetic shape of language. But are the influences specifically planned or are they the automatic result of successfully producing a string of phonemes? This question was examined by forcing speakers (N = 3 native English speakers) to be talking before they knew the entire utterance. Simple nonsense utterances of two or three syllables provided various measures of coarticulation due to vowels (/i/ & /u/) or consonants (/b/ & /p/), either of which was unknown when the utterance began. Segments known at initiation time exerted their normal influence, even across unknown segments. Segments not known mostly did not: anticipatory effects were virtually absent though perseverative (carry-over) effects were normal. Unplanned-for vowels exerted some influence on preceding vowels when the intervening consonant closure was short enough, indicating that some of the coarticulation was necessary to production. From the greater magnitude & temporal extent of the influence of the known segments, however, it can be inferred that, for the most part, coarticulation is planned. |