Testing the psychological reality of phonemes

One important aspect of the current growing interest in testing claims about the psychological reality of various linguistic phenomena is the exploration of new experimental paradigms for accomplishing this testing. This paper explores the use of the standard classical conditioning and concept-formation paradigms, using the notion of the phoneme (in particular the claim that naïve speakers perceive allophones of a phoneme as being “the same”), as the focus of the inquiry. In Experiment 1, subjects were trained by classical conditioning (shock-GSR) to respond to words containing what is analyzed as one allophone of the phoneme /k/ in English (the aspirated allophone), and then were played words containing other allophones ([k], [k°]) as well as non-/k/-words, to see if there was generalization of response to these phonetically different but phonologically “same” sounds. In Experiment 2, subjects were asked to perform a category-formation task: They were trained that the correct category was words containing the aspirated allophone of /k/; they were then tested to see if they would include other allophones in the category. (Particular attention was paid to the categorization of the [k] after word-initial /s/.) The subjects in both experiments made phoneme-like responses. Implications of these results for the theory of the phoneme and for future experimentation are discussed.