Bilingual Speech Processing in Infants and Adults

JANET F. WERKER, WHITNEY M. WEIKUM and KATHERINE A. YOSHIDA

Adults are highly adept at processing their native language. They use phonological (sound) properties to effortlessly segment sentences, phrases and individual words, correctly detect morphological markers, and distinguish words that are phonetically similar such as bad vs. dad. All of these functions are performed rapidly and automatically when listening to speech. Furthermore, viewing speech has also been shown to play an important role, as adults are adept at extracting and processing the visible concomitants of spoken language that are available in the movements of the lips, tongue, jaw, face, head and neck. Considerable research has shown that the ease with which adults process their native language has its roots in listening to and observing speech in the infancy period (Werker & Tees, 1992). The resulting efficiency of phonological processing is essential for successfully understanding the meaning of the language being spoken and for recovering the underlying speaker intent.

As effortlessly as adults are able to process a native language, attempting to learn a second language (L2) is difficult. Finding boundaries between significant phonological units, pulling out and distinguishing individual words, and properly encoding morphological markers can all be very challenging tasks. Together, these difficulties in phonological perception of an unfamiliar language, or an L2 acquired late in life, summate to make the task of listening for meaning in the L2 considerably more taxing than it is in the first language (L1; for a review, see Sebastián-Gallés & Bosch, in press).

Our present understanding of how speech develops and is processed is largely based on the results of studies examining monolingual adults and infants. However, as the populations of bilinguals in North America grow, the study of bilingual language acquisition has become an increasingly important focus. As noted, individuals may face unique challenges in acquiring their L2, and such challenges will differ depending on whether the L2 was learned in infancy or in adulthood. Any differences found between early and late bilinguals are also theoretically interesting as they inform on a number of issues, including the question of whether or not there are sensitive periods for learning of the sound properties, and whether or not the L2 acquisition process is the same or different later in life.

In this chapter, the findings of several studies that are beginning to address the question of phonological processing in bilinguals are reviewed. We begin with an examination of phonetic categories in adult and infant bilinguals, and examine the role that perception of these phonetic categories plays in associative word–object learning. This is followed by an overview of the ability to recognize languages on the basis of factors such as rhythmicity and visual information. We conclude with a reconsideration of what it means to grow up bilingual, and how studies of bilingual speech processing can not only help inform linguistic and psychological theory, but also inform policy makers.

Throughout this chapter, we will refer to those who acquired their two languages from an early age as 'bilingual first language acquisition' (BFLA; e.g., de Houwer, 1995). In all cases, when we talk about bilingual infants we, of course, are using a shorthand for bilingual-learning infants. In studies reported from our lab, we assessed adult language dominance (proficiency in each language) using a questionnaire developed by Alain Desrochers (2003). To ensure that the bilingual infants in our studies have had relatively equal exposure to each of their languages, we required that they have a minimum of 30% and a maximum of 70% input in each language as determined by a parent report scale designed by Bosch and Sebastián-Gallés (1997a).

Adult Phonetic Perception: Monolingual and Cross-linguistic Studies

Among the hallmarks of adult L1 speech perception is the ability to easily perceive just those acoustic/phonetic differences, in individual consonants or vowels, that signal a potentially meaningful distinction in the native language. For example, when presented with equally spaced stimuli on /ba/ to /pa/ voicing continuum, adult English speakers show a very sharp perceptual category boundary. They label all of those stimuli up to approximately +25 msec in voice onset time (VOT) as /ba/ and all of those with longer VOTs as /pa/. Their ability to discriminate stimuli from along the continuum is predicted by their labelling performance (Liberman et al., 1967; Repp, 1984). Perception of phonetic continua is language-specific: adults have difficulty discriminating those phonetic differences that are not used to convey meaningful distinctions in their native language. For example, the Japanese language does not contrast the consonants /r/ and /l/, and Japanese speakers have difficulty perceiving the difference between them (Goto, 1971; Miyawaki et al., 1975). Similarly, Hindi has a contrast between a front (dental) /d/ and a /D/ that is produced with the tongue curled over with the tip pressed against the roof of the mouth (retroflex), whereas in English a /d/ is produced at the alveolar ridge, between the two Hindi phones. Adult English speakers have difficulty discriminating the two Hindi sounds (Werker et al., 1981). Although performance on non-native phonetic distinctions improves with training (Lively et al., 1993; McClelland et al., 2002), there is conflicting evidence as to...