James S. Magnuson,
Reiko A. Yamada
The Effects of Talker Variability on the Perception of American English /r/ and /l/ by Japanese Listeners, II: Subject differences, acoustic and temporal correlates of talker effects, and some technical considerations
Abstract:In an earlier technical report [Magnuson and Yamada, TR-H-110, 1994.12.6], we presented the results of three experiments which investigated the effects of talker variability on the perception of American English (AE) /r/ and /l/ by Japanese adults. In that first report, we found that Japanese subjects set criteria for /r/-/l/ decisions based on the range of cues they experienced in any given block of trials, independently of talker-specific differences in cues to /r/ and /l/. This led to significant differences in how often subjects responded "R" (R-rate) to particular talkers in blocked (single) and mixed (multiple) talker conditions. At that time, we reported that we were unable to find acoustic explanations for the talker effects we observed. We extend and comment on our previous report in three ways. First, we describe a new analysis of subject differences in those experiments. Second, we describe a series of analyses that have revealed some weak correlations between acoustic and temporal measurements and talker effects observed in the first two experiments in the previous technical report. Finally, we describe a minor flaw in the design of the second experiment and recommend a scheme for avoiding similar flaws in future experiments.