Sheri Wells-Jensen and Jason Wells-Jensen
Bowling Green State University
LASSO (Linguistic Association of the Southwest) 2001 conference
In studies of speech errors, or "slips of the tongue", phonological anticipations are usually found to outnumber perseverations. For example, given the target "cats and dogs", a speaker would be more likely to produce "dats and dogs" than "cats and cogs". However, the majority of published speech error research has involved English and other Germanic languages, and may not be representative of speech production cross-linguistically.
Several recent studies of non-Germanic languages have, in fact, failed to find an anticipation bias, and some researchers have suggested that anticipations are favored in English because of its prosodic structure: the most prominent or "tonic" syllable in an utterance, where semantic or pragmatic focus is realized with an intonation peak, is usually the source of anticipations in English. Thus, languages which have other strategies for marking new information, and which lack a tonic syllable, would not be expected to show the anticipation bias.
The present paper concerns a laboratory study of errors in a typologically-diverse sample of five languages, in which speaking conditions were held constant across speakers and across languages. Because these speakers were under pressure to narrate a fast-paced silent film, they had relatively little opportunity to plan ahead in formulating their utterances and tended to produce many hesitations, restarts, and sentence fragments. Under these conditions, any bias related to tonic-syllable prominence would have been partially suppressed due to the relative infrequency of completed utterances.
The ratio of anticipations to perseverations, in fact, was approximately 1-to-1 in Hindi, Japanese, and Spanish, while a very weak anticipation bias was found in English; these results are compatible with the tonic-syllable hypothesis.
The 2-to-1 ratio of anticipations to perseverations found in Turkish, however, was much larger than in any of the other four languages. We propose that the stress system of Turkish may have contributed to this rather surprising result. The consistent final-syllable stress in Turkish words may exert a prominence effect similar to that of the tonic syllable in English, acting within the domain of individual words and short phrases rather than entire utterances.