Native listening: Language experience and the recognition of spoken words by Anne Cutler

Reviewed by: Native listening: Language experience and the recognition of spoken words by Anne Cutler Annie Tremblay Native listening: Language experience and the recognition of spoken words. By Anne Cutler. Cambridge, MA: MIT Press, 2012. Pp. xvii, 555. ISBN 9780262017565. $50 (Hb). Listening to one’s native language (L1) seems effortless. Native listeners recognize words rapidly and efficiently using a wide range of linguistic information, from subsegmental (acoustic-phonetic, allophonic) cues to segmental (transitional, phonotactic) cues, rhythmic cues, suprasegmental (stress, prosodic) cues, and contextual (syntactic, semantic) cues. Native listeners are also adept at recognizing words under more difficult listening conditions. This efficiency, however, comes at a cost when listening to a second/foreign language (L2). The cues that make native listening efficient are often not useful for recognizing L2 words, and those that would be are typically not used by nonnative listeners. In her book, Anne Cutler provides a comprehensive review of native and nonnative speech-processing research over the past four decades, much of which she has conducted. She argues, quite convincingly, that speech processing is shaped almost exclusively by native listening experience. Through studies that employ a variety of experimental paradigms, she demonstrates that native listeners use all information available in the signal to maximize efficiency and minimize interference. C argues that it is this efficiency that results in less-than-optimal nonnative listening. Nonnative listeners experience more lexical activation and competition than native listeners due to asymmetrical mappings between the L1 and L2 sound systems. C assumes that the processing system remains adaptive throughout life, as evidenced by native listeners’ adaptation to regional-and foreign-accented speech and their adjustment of phonetic categories. It is thus the native listening experience, and not a biologically determined time-sensitive period for speech learning, that C holds responsible for nonnative listeners’ inefficient processing. Universal to native listeners is the ability to segment continuous speech into discrete units such as words and phonemes. By the end of the first year of life, infants have already tuned in to native phonemes and can locate words in speech. Although this ability is universal, the knowledge that results from it is language-specific. It is only by conducting crosslinguistic research that psycholinguists can specify what aspects of speech processing are universal and what aspects are language-specific. This argument, made explicit in Ch. 1, is central to C’s book and to her extensive research, which has focused on no less than twelve languages (Cantonese, Dutch, English, Finnish, French, German, Italian, Japanese, Korean, Sesotho, Spanish, and Telugu). Understanding language-specific aspects of speech processing begins with an examination of the types of information that distinguish words from one another in different languages. C exemplifies this in Ch. 2. She shows that languages do not have enough segments for words to be sufficiently distinct, in part because short words can be embedded segmentally within longer words (e.g. can in candle). However, with the addition of stress information, for example, the number of lexical embeddings decreases considerably. As C further illustrates, phonetic differences in how languages instantiate phonemic distinctions can also impact how listeners categorize speech, and so will the nature of the segment (consonant vs. vowel) that carries the distinguishing information. [End Page 294] Lexical characteristics that may also impact speech processing include morphological structure, word type (open vs. closed class), and frequency. One universal aspect of speech processing is that words that completely or partially match the signal are activated and compete for selection. C discusses this in Ch. 3. She demonstrates how these lexical processes are modeled in auditory word-recognition models, including Norris’s (1994) connectionist shortlist a model and Norris and McQueen’s (2008) Bayesian shortlist b model. Based on some of her research, C provides evidence that phonological and conceptual representations are separate, with lexical competition involving only phonological representations but with lexical activation passing rapidly to conceptual representations. She further shows that lexical activation can be affected by suprasegmental information, morphological structure, gender, and word type (open vs. closed class). Lexical activation and competition do not, on their own, provide a satisfactory explanation of how listeners recognize words in continuous speech, however. These two processes imply a lexicon...