Frequency Effects in Cross-linguistic Stop Place Perception: A Case of /t/-/k/ in Japanese and English

Abstract

UC Berkeley Phonology Lab Annual Report (2007) Frequency Effects in Cross-Linguistic Stop Place Perception: A Case of /t/ - /k/ in Japanese and English Reiko Kataoka and Keith Johnson 1. Introduction The study described in this paper is an attempt to answer the question ‘Which aspect of speech perception is altered by linguistic experience and how this alteration is done?’ As Strange and Jenkins stated, “[t]he knowledge of a language possessed by a normal adult is a product of many years of exposure to a specific language environment” (1978: 125). Therefore, if linguistic knowledge influences speech perception, then adult listeners would perceive speech sounds in language-specific ways. Evidence of experience-based speech perception has been accumulated from both cross-linguistic and within language studies. One source of evidence is the phenomena called categorical perception (Liberman, et al. 1957, 1961a, 1961b), which is characterized by difficulty in discriminating acoustically similar patterns in the single phonemic category and near perfect discrimination for acoustic patterns that straddle different categories. Categorical perception has been replicated with listeners of various language groups, and it seems to be an undeniable fact that linguistic experience influences listeners’ ability to discriminate speech stimuli. However, exactly which aspect of linguistic experience is responsible for language-specific speech perception is not well understood. This is the primary question addressed in this paper. Also, it is often reported that listeners react to speech sounds differently depending on whether they use a continuous mode or a categorical mode of memory (see, for example, Pisoni 1973). Thus, it is of interest to test whether experience-based perception is elicited when the listening task calls for auditory acoustic perception. The goal of this study is to obtain answers of these two questions. The first question was treated by testing whether a particular aspect of linguistic structure—phoneme frequency— influences speech perception; and the second question was treated by testing whether the linguistic knowledge has any effect in an auditory acoustic perception task. The rest of the paper is organized as follows. Section 2 provides reviews of previous studies on the experience-based auditory perception, and gives rationales for the particular ways current experiment is designed. Section 3 describes the experimental study, provides the results, and discusses the results. Finally, Section 4 discusses the implications of some of the findings from current study for theories of speech perception. 2. Background 2.1. Experience-based speech perception One of the prime examples of experience-based speech perception is categorical perception, a phenomena originally defined by Liberman, Harris, Hoffman, and Griffith (1957). Categorical perception can be demonstrated in an experiment that uses a series of synthetic consonant-vowel stimuli, ranging across two or more initial consonant categories (e.g. ten-equal- step /da/-/ga/ continuum), and involves identification and discrimination tasks. One of the defining characteristics of categorical perception is predictability of discrimination accuracy from the identification result. In its strongest form, categorical perception predicts that listeners