Mandarin and English listeners' perception of Mandarin tones in isolation and 
in context

Tessa Bent, Northwestern University

A comprehensive model of cross-language speech perception must account for all 
aspects of phonological structure.  However, current models focus primarily on 
phonemic categories and exclude cross-language perception of prosodic 
structure. The cross-language interaction of prosodic categories in non-native 
listeners is difficult to model in current frameworks because multiple prosodic 
units must be considered as possible sources of transfer and interference 
including lexical tone, word stress, pitch accent, and phrasal intonation 
patterns. This study tested the hypothesis that perception of non-native 
prosodic contours is affected by both the relationship between the native and 
non-native languages' prosodic contour categories and the unit length over 
which the contours apply.  Native English and Mandarin listeners' 
discrimination of the four Mandarin lexical tones in isolated monosyllables and 
three-syllable sequences was tested. Reactions times to correctly discriminated 
"different" trials were entered into a multidimensional scaling analysis. In 
the English listener's similarity spaces, the configuration of the four lexical 
tones differed in the three-syllable context compared to their perception in 
isolation. In the Mandarin listener's similarity spaces, the isolated tones 
were arranged in triangular formation with the rising and dipping tones in 
close proximity but for the three-syllable stimuli, all tones were well 
separated.