Coarticulation Is Largely Planned *

Is coarticulation a reflection of planning an utterance or the automatic effect of producing speech? This question was examined by requiring speakers to begin reading nonsense aCV or abVCa strings (where C was fbi or IpI and V was Iii or lui) aloud before seeing the entire utterance. Those segments known before articulation began exerted normal anticipatory coarticulatory influences, while those seen after utterance onset usually did not. Perseverative coarticulation was, for the most part, present in both conditions, showing either that it is indeed a result of the mechanical nature of the speech organs or that the amount of planning necessary can be carried out in the time used for the first syllables. There was one anticipatory effect which began to emerge (even though unplanned) just before the stop closure, supporting an articulatory overlap explanation for some of the coarticulation. From the greater magnitude and extent of the coarticulatory influence of segments known in advance, we can infer that, for the most part, coarticulation is planned.