Shadowing latency and imitation: the effect of familiarity with the phonetic patterning of English

Speakers imitate the speech they shadow. However, speech is not wholly imitative; speakers use their own speech habits or language knowledge in shadowing as well. We examined the interplay between the effects of input variables and of knowledge of the language on shadowing. We asked speakers to shadow utterances composed of phonetic sequences that varied in their order of approximation to English. Shadowing latency and errors reduced as order of approximation increased. This is consistent with the inference that knowledge of the language (e.g., speech habits, lexical or phonotactic knowledge) guides shadowing. To assess the interplay between this knowledge and the effect of imitation of the input on shadowing, we asked whether imitative fidelity varied with order of approximation. We used an AXB test in which X was a shadowed utterance, A (or B) was a shadowed response to X and B (or A) was a read version of the same utterance produced by the speaker of A (or B). Listeners were asked which of A or B was a better imitation of X. Generally, they chose the shadowed utterance; however, they did so significantly more frequently when the utterance was a 1st than a 12th order of approximation to English. (C) 2003 Elsevier Science Ltd. All rights reserved.