The effect of removing linguistic information upon identifying speakers of a foreign language

Abstract

The native-language background of a listener has been shown to have an effect upon identifying speakers of a foreign language. Previous experimental research showed that a German target speaker was identified significantly better when listeners were native speakers of German, or native speakers of English who had some knowledge of German, than when they did not know the language of the target speaker (Schiller and Koster 1996). This result was taken as support for the hypothesis that familiarity with the language of the target speaker has a positive effect upon identifying that speaker. This paper reports the results of a follow-up experiment that investigated the identification of a speaker in a voice line-up using nonsense speech. The results show that German natives, monolingual English natives, and English natives with some knowledge of German do not significantly differ in identifying a German speaker when most of the linguistic information of the language of the target speaker is removed from the stimulus materials.

Infant ability to tell voices apart rests on language experience

“…Although the Dutch stimuli in our Experiment 2 were rhythmically and prosodically more like English than the stimuli in the other two languages, the English listeners performed worst with the Dutch stimulus set. In this respect our Experiment 2 results confirmed the earlier findings of Schiller and colleagues (Schiller et al ., 1997; Köster & Schiller, 1997) with other language pairings and different methodologies.…”

“…Support for sensitivity to the sound structure of language as a determinant of voice recognition ability is, however, not strong. The prosodic structure of a language alone, as retained in reiterant speech, does not support an effect of language familiarity on voice recognition (Schiller, Köster & Duckworth, 1997). Moreover, linguistic similarity between familiar and unfamiliar languages does not seem to modulate the language familiarity effect on voice recognition; English, Spanish, and Chinese speakers all find it equally difficult to identify German speakers (Köster & Schiller, 1997).…”

Multiple levels of linguistic and paralinguistic features contribute to voice recognition

“…These observations of robust, similar performances across the German and English conditions contradict earlier reports of reduced voice recognition in an unfamiliar language such as German 23 24 . It should be noted that those studies utilized excerpts of text written in or translated to German; when linguistic information in this text was reduced, such that all syllables are replaced with /ma/, language familiarity no longer influences voice recognition 25 . The lack of differences in voice recognition observed in both our study — where bisyllabic words/pseudowords were presented — and Schiller’s study suggests that lexical and semantic information are not necessary for successful voice recognition, at least in these particular experimental procedures.…”

Bilingual speaker identification: Chinese and English

“…Goggin et al (1991) showed that, when the passage being spoken is made increasingly less similar to English by rearranging words, rearranging syllables and reversing text, English-dominant listeners' voice recognition deteriorated systematically across these conditions. Schiller, Köster and Duckworth (1997) removed linguistic information by replacing all syllables in a natural German passage with 'ma' in order to minimize the linguistic cues to the target language, while keeping some phonetic and phonological features. They found that German and English monolingual listeners as well as English listeners with some knowledge of German did not differ significantly in their recognition ability.…”