Listeners use speaker gender information in non-native phoneme categorization

Speaker gender is known to affect phoneme categorization, and this effect is especially well-established in fricative perception: For example, in /s/-/θ/ categorization tasks, more /θ/ responses are given when the sound is paired with a male than female voice, henceforth referred to as the speaker gender effect. However, so far this phenomenon has only been tested in native (L1) listeners. The present study examines how speaker gender is used in phoneme categorization in a second language (L2), specifically when listeners are presented with a non-native phoneme contrast that does not occur in their L1 (/s/-/θ/ for German and Polish learners of English). Learners’ categorization results of a continuum from /s/ to /θ/ spliced onto words spoken by a male voice and a female one were compared to the results of a native British English listener group. No significant difference was found between the L1 and L2 groups in their use of speaker gender information in the categorization task, despite the lack of the phoneme contrast in the learners’ native languages. This study hence demonstrates the robustness of the speaker gender effect in phoneme categorization. Moreover, the results suggest that this effect is not language specific, as it also appears accessible to non-native listeners.