Acoustics'08 Paris

[ Lay Language Paper Index | Press Room ]

Babies’ brain waves and cognitive control skills are linked to learning of second-language speech sound information


Barbara T. Conboy --
Jessica A. Sommerville --
Patricia K. Kuhl --
Institute for Learning and Brain Sciences
University of Washington, Box 357988
Seattle, WA 98195-7988

Popular version of paper 4aSCb9
"Cognitive control skills and speech perception after short-term second language experience during infancy"
Presented Thursday July 3, 2008 at 11:40 a.m. in Room 250B of the Palais des Congres, Paris

Babies display remarkable language learning abilities. Early in life, babies already possess the ability to tell apart speech sounds, whether or not they are sounds that are used in the language the baby is learning. However, by the end of the first year, babies learn to narrow their perceptual abilities to match the language they are learning. By 10-12 months of age, infants can detect differences between sounds that change the meaning of a word in their own language, but not in another language. Yet it is well known that babies this age are good second language learners. Previous research showed that when 10-month-old infants were provided with naturalistic experience in a second language, as little as 5 hours spread over a period of one month, they learned to tell apart speech sounds in that new language in a behavioral task.

Although they are quick learners, babies, like adults, might vary in how well they learn a second language, and such variation could shed light on the processes that determine successful learning. No previous research has examined individual differences in second language learning during infancy. We investigated whether infants’ learning of second-language phonetic information is linked to nonlinguistic cognitive control skills that are also developing in this time period. We hypothesized that infants who had better cognitive control skills would also show better discrimination of speech sounds from the second language after experience with that language, compared to infants with less well developed cognitive abilities at that age.

We examined the performance of infants from monolingual English-speaking homes on three reaching tasks that required them to attend to task-relevant information and to resist distraction in the face of competing cues. Infants were tested at the age of 9 months and then again at 11 months. The tasks were selected so that they would be challenging enough at these ages that we would see differences across infants. The tasks required infants to a) search for an object hidden under a cloth after a switch in location, b) reach for an object around a detour rather than reaching directly for the object, and c) use a tool to reach for an object rather than reaching directly for the object. Thus in each of these tasks, babies had to selectively attend to the relevant cues and resist distraction from irrelevant information.

From 9.5 – 10.5 months, the same infants attended a series of 12 play sessions conducted in a second language (Spanish). During these sessions, infants sat in a small room with a parent while an adult “tutor” spoke to them in Spanish in a naturalistic way, showing them toys and picture books that contained the Spanish consonants “d” and “t”. The Spanish “d” and “t” are difficult for monolingual English speakers to tell apart, and by 11 months infants from monolingual English-speaking environments already show reduced sensitivity to the differences between these sounds. The critical difference between these Spanish sounds is the timing of vocal fold vibration, known as “Voice onset time”. English-learning infants become perceptually attuned to the difference in this timing that is important for English, but disregard the differences that are important for other languages. Thus they have difficulty discriminating the Spanish word “dos” (two) from “tos” (cough) because of the subtle differences in how the initial consonants are produced in these words.

We tested infants’ discrimination of the Spanish sounds “da” vs. “ta” by examining differences in their brain activity to these sounds. This brainwave test, known as “Event-related potentials”, is performed by fitting infants with an elastic cap with an array of metal electrodes sewn into it. The cap rests comfortably on the infant’s scalp, and is connected to a battery-powered amplifier and then to a computer. The electrodes capture moment-by-moment fluctuations in the electrical activity produced when thousands of neurons fire in synchrony after the infant hears a particular speech sound. Thus, if different patterns of brain activity are observed to two different sounds, it can be concluded that the brain has discriminated those sounds. Previous research has shown that infants’ brain activity differs for two sounds they discriminate by around 250 milliseconds (around of quarter of a second) after infants hear the sounds.

As predicted, babies did indeed show learning of the Spanish speech sound contrast after experience listening to Spanish. At 9 months, before they had any exposure to the Spanish language, infants’ brain activity differed for two English sounds (the English “d” vs. “t” sounds) by around 250-450 milliseconds after infants heard the sounds, but did not differ for the two Spanish sounds. However, at 11 months, after 12 play sessions in Spanish, infants’ brain activity differed for the two Spanish sounds as well as for the English sounds by 250-450 milliseconds. In other words, babies learned to discriminate Spanish sounds after listening to Spanish for only 5 hours from live speakers.

However, there was variability across babies in the strength of this response. While some showed a large discriminatory response to the change from the Spanish “t” to Spanish “d”, other infants showed a smaller or an absent response, suggesting that their brains treated those two sounds as the same speech sound.

As predicted, we found that the babies who performed better on the nonlinguistic cognitive tasks had larger brain responses to the Spanish speech sound contrast than those with less advanced performance on the cognitive tasks. However, there was no link between their discrimination of the English speech sound contrast and their nonlinguistic cognitive skills. Also, there was no difference between their performance on the nonlinguistic tasks at 9 months and their brain responses to the speech stimuli at 11 months. The specific link between Spanish speech sound discrimination and the cognitive skills after exposure to Spanish suggests that language learning was linked to cognitive abilities.

These findings suggest that babies of the same age may vary in their readiness to learn from particular experiences with a language, depending on how well they attend to relevant information and resist distractions. These abilities may be linked across language and nonlinguistic domains of learning. All of the infants in this study were typically developing. All are expected to learn their native language, English, normally. And all will eventually be able to perform successfully on the nonlinguistic cognitive tasks. However, when variations in these abilities are linked across infants of a particular age, they suggest that common underlying mechanisms.

Our results add to a growing body of evidence suggesting that children’s ability to learn two languages is linked to their ability to deal with conflicting cues in nonlinguistic domains. While we cannot tell from these results whether learning a second language produced improvement in infants’ nonlinguistic cognitive skills or vice versa, we have shown the first link between these abilities in infants this age. Our results suggest that infants actively learn from the language input provided to them, and that they bring other abilities to aid in this learning.

[ Lay Language Paper Index | Press Room ]