Speech sounds are never perceived in isolation, but always in the context of adjacent sounds and general characteristics of the speech signal. It has been shown repeatedly that the perceived gender of the speaker is taken into account in the interpretation of sounds. For example, in English, acoustically ambiguous sounds between /s/ and /θ/ (also known as “th”) are more often perceived as /θ/ when the voice is perceived as male than when it is perceived as female.
The recently published study investigated whether people who have learned English as a foreign language have a similar preference. To this end, 55 people with English as their native language, 55 people with German, and 50 people with Polish as their native language were invited to participate in a listening study. The language material presented was in English. Sounds along an artificially generated continuum between the speech sounds /s/ and /θ/ were presented in English word pairs, for example sink-think or face-faith. The listeners' task was to decide which of the two words they heard. While the sound continuum was identical in all audio examples, the remaining parts of the words were spoken by a man or a woman. It was shown that all three groups of study participants, regardless of their native language, categorized the sound continua differently depending on the voice, and all groups showed similar differences, even though neither German nor Polish have the sound /θ/.
The results show that social information from the voice of speakers is taken into account in speech perception and that this process is so automatic and efficient that it is also applied in foreign languages and to newly learned speech sounds (in this case /θ/).
Read the full journal article in the Journal of the Association of Laboratory Phonology.