Perceptual consequences of clear speech

Abstract

When faced with communication challenges talker often modify their way of speaking producing clear speech. In a production study we investigated how American English talkers adapt medial flaps (e.g., petal and pedal are pronounced as [ˈpɛɾəl]) following miscommunications with a computer partner. When subjects said a word (e.g., petal) the computer either interpreted the word correctly or incorrectly (it guesses pedal, kettle or ???). Talkers produced longer words, vowels, and stop closure in response to an incorrect versus correct guess. Talkers also changed their flapped productions to stops and lengthened stop closures more following voicing errors compared to other error types. Thus, talkers’ adjustments appear to depend on the miscommunication type. Next, we investigated how listener perceptions were influenced by the talkers’ adjustments. Listeners heard talkers’ productions and indicated what they heard. Listener perception was more accurate for lengthened compared to shortened productions and better for stops than flaps. However, listeners misperceived /d/ for /t/ when talkers lengthened their /d/ closures. Moreover, production changes talkers made following a misrecognition improved listener perception for medial /t/, but not medial /d/, which was at-chance. These findings indicate that talker adjustments made to remedy a misrecognition do not universally aid listeners’ perception.