The influence of speaker origin and individuality on rhythmic features of non-native speech

Abstract

How do rhythmic and temporal features of non-native speech allow us to take guesses about a speaker’s origin or identity? Research showed temporal phenomena to correlate with non-native speakers’ intelligibility and strength of foreign accent; however, the contribution of rhythmic cues to the identification of non-native speaker origin is not yet completely understood. Likewise, little research has investigated speaker-individuality in non-native speech.
We used a perceptual approach to study the influence of speaker origin on non-native speech, and applied a range of signal manipulation methods to reduce French- and English- accented German so as to convey different types of primarily or exclusively temporal information. Findings revealed that listeners can identify speaker origin above chance in a non-native speech containing temporal information alone. Our results suggest a weighting of cues in foreign accent identification, an additive trend of time and frequency domain information, perceptual salience of French-accented German and variability due to speakers.
We further investigated the influence of speaker-individuality on non-native temporal features from a speech production point of view, finding high between-speaker variability in speakers’ native Zurich German and in their non-native French and English. Speaker- individual behavior was also evident within speakers, with most speakers exhibiting proportionally constant behavior in Zurich German, French, and English.