ASA 130th Meeting - St. Louis, MO - 1995 Nov 27 .. Dec 01

5aSC24. Detection of speech targets under the divided attention manipulations.

Chin-Hsing Tseng

Feng-Ming Tsao

Commun. Disord., Dept. of Special Education, National Kaohsiung Normal Univ., Kaohsiung, Taiwan

Divided attention tasks were designed to test the following hypothesis: When two dimensions of a single speech stimulus are simultaneously attended to for target detection, one dimension can be processed faster than the other if the former has a higher information value. Targets were (1) a given vowel or any vowels produced by a given speaker, or (2) a given plosive or a specified lexical tone embedded in a syllable, or (3) a specified lexical tone embedded in disyllabic noun or a name for man-made objects. The target occurrence probabilities for the two dimensions were set at 0.10 vs 0.40, 0.25 vs 0.25, or 0.40 vs 0.10 in a given block of trials. Chinese-speaking college students participated in these experiments. It was found that the target detection latency varied inversely with target occurrence probability for either dimension under all these manipulations. A clear performance trade-off pattern was observed for each manipulation; that is, as one dimension received more attention, the processing of the other became less efficient. These results demonstrated that attention can be divided between dimensions of ``speech processing space'' as cued by the information value. [Work supported by NSC, ROC.]