Language-independent talker specificity in bilingual speech:
Using a multi-lingual speech corpus to identify individual “traits” that persist across L1 and L2 “states”

Ann Bradlow

Northwestern University, Department of Linguistics

Current phonetic theories emphasize the acoustic and perceptual integration of linguistic information (the “what” of the utterance) with indexical information (the “who,” “where,” and “when” of the utterance).  For monolingual speakers, linguistic and indexical information are simultaneously conveyed by phonetic variations of a single language.  For bilingual speakers, however, linguistic and indexical features are potentially decoupled, thus offering a unique window into the nature of speaker-specific features as a source of systematic, language-independent signal variation.  In this talk, I will present data that establishes a strong link between variation in second-language and first-language speech production in a group of bilinguals (n=86) from a various language background (n=10 languages).  Specifically, I will show that L1 speaking rate significantly predicts L2 speaking rate within individuals and similarly that a significant portion of variation in overall L2 intelligibility can be predicted by variation in overall L1 intelligibility.  That is, relatively fast or slow talkers in L1 were also relatively fast or slow in L2, respectively, and relatively high or low intelligibility talkers in L1 were also relatively high or low intelligibility talkers in L2, respectively.  Together, these results indicate a persistent influence of talker-specific trait characteristics that combine with, rather than are overwhelmed by, language-specific and dominance-dependent influences in the speech of bilingual individuals.

If you would like to meet the speaker, please contact Bernd Möbius.