The perception of speech in an unfamiliar language

Abstract

Experiments in speech perception almost invariably use speech tokens and response categories drawn from the language native to the listeners participating in the experiments. This has served to confound the listeners' extensive knowledge of their own language with their ability to perceive speech stimuli as such. The present study attempts to separate these two components of speech perception. Short (4–5 s) excerpts of English and non-English speech were altered in one of two ways: by the deletion of a closure silence or by the addition of a 100 ms silence to an existing closure silence. Native English listeners were highly adept at detecting and identifying alterations in English. However, their performance was at a chance level on non-English excerpts. The original foreign talkers also listened to the excerpts: they performed well with excerpts of their own language and poorly in all other languages. These experiments underline the importance of experience in speech perception and suggest that the perception of even elementary speech events may depend on the listener's familiarity with the context in which they occur.