Cerebral responses to infant-directed speech and the effect of talker familiarity

Abstract

A number of behavioral studies suggest that infant-directed speech (IDS) plays a more important role in facilitating both: a) speech perception, and b) adult–infant social interactions than does adult-directed speech (ADS), and hence that IDS contributes to subsequent social and language development. However neural substrates that may underlie these IDS functions have not been examined. The present study examined cerebral hemodynamic responses to IDS in 48 infants (4–13 months of age) using near-infrared spectroscopy (NIRS). Japanese sentences uttered by the infants' own mothers and by unfamiliar mothers were used to record activations in temporal and frontal area separately. Increased activations were observed predominantly in infants' left and right temporal areas when they listened to IDS rather than to ADS when both involved voices of their own and unfamiliar mothers. In contrast, significantly greater activations were observed in the frontal area when infants listened to IDS produced by their own mothers, not when IDS arose from unfamiliar mothers. Furthermore, the present results indicate that responses to IDS do vary as a function of the infant's age and the talker familiarity. These findings suggest a differential function for frontal and temporal areas in processing infant-directed speech by the different speakers.

Introduction

Adults who interact with young infants tend to modify their speech in certain characteristic ways and this type of speech is known as infant-directed speech (IDS) (Cooper and Aslin, 1990, Cooper and Aslin, 1994, Pegg et al., 1992, Werker and McLeod, 1989). Many studies have found that, relative to normal adult-directed speech (ADS), IDS is higher in pitch, has a wider pitch range, and exhibits exaggerated pitch contours; in addition, although IDS tend to be shorter than normal speech, they are slower and separated by longer pauses (Fernald and Simon, 1984, Snow and Ferguson, 1977, Stern et al., 1982, Stern et al., 1983). Moreover, this manner of producing speech may be universal in that IDS has been observed cross-linguistically (Ferguson, 1977, Fernald, 1989, Fernald and Simon, 1984, Grieser and Kuhl, 1988, Kitamura et al., 2002, Papousek and Hwang, 1991, Shute and Wheldall, 1989, Stern et al., 1983). IDS is also called “motherese” (Ferguson, 1977, Grieser and Kuhl, 1988, Kemler Nelson et al., 1989, Masataka, 1992) because it is frequently observed in mother–infant interactions. However, IDS has also been reported in fathers (Fernald, 1989, Jacobson et al., 1983), as well as in individuals who have never borne or raised a child (Nakagawa and Matsumura, 2006). It is even observed in preschool children (Tomasello and Mannle, 1985, Weppelman et al., 2003).

An adult's strategic use of voice in IDS may play an important role in a child's development. In fact, the role of IDS in language acquisition has recently become an issue of great debate. Infants have been shown to be sensitive to word or clause boundaries in IDS, but not to those in ADS (Kemler Nelson et al., 1989, Thiessen et al., 2005). In addition, the degree of exaggeration of vowel sounds in maternal IDS is significantly correlated with performance in phonetic discrimination in 6 to 12 months infants (Liu et al., 2003). These findings suggest that IDS in general may enhance an infant's ability to extract syntactic units from fluent speech.

A second issue concerns the role of IDS in eliciting infant affect and attention. A number of behavioral studies suggest infants prefer IDS to ADS and that they are more likely to attend to IDS than to ADS (e.g. Fernald, 1989, Pegg et al., 1992). Consistent with this, 4- to 5-month-old infants expressed greater positive affect, such as smiling, to IDS than to ADS produced by unfamiliar male and female actor (Werker and McLeod, 1989). This preference for IDS emerges early in development; it appears even in 2-day-old neonates (Cooper and Aslin, 1990). Although this preference for IDS appears early, infants' preference for IDS changes as they age, Hayashi et al. (2001) demonstrated that infants between 4 and 6 months of age prefer IDS of an unfamiliar female but that the preference attenuates between the ages of 7 to 9 months. However, for still older infants the preference for IDS versus ADS re-appears. Because infants show greater attention and more positive affect to IDS versus ADS, it is possible that IDS also contributes to positive adult–infant interactions. If so, this could lead to further opportunities for incidental verbal and social learning for children to occur. Therefore, IDS in such indirect ways, may also contribute more broadly to a child's subsequent development in language and social abilities later in life.

Taken together, this account leads to the identification of two potential functions of IDS for a developing child. First, IDS may simply facilitate speech perception by highlighting linguistic structure. Second, IDS may contribute to positive interactions between caregiver and infant by regulating infant arousal and affect.

To date, few neurological studies have addressed these IDS functions. With respect to infant speech perception, one fMRI study has shown that IDS results in larger activation in left temporal brain regions when 2–3 months old infants listen to IDS played forward and backward, compared to silence (Dehaene-Lambertz et al., 2002). Using younger infants aged 2 to 5 days old, Penã et al. (2003) applied near infrared spectroscopy (NIRS) to show significantly larger activation in left temporal area when these infants listened to normal IDS relative to backward IDS. Although these studies converge to indicate an involvement of left temporal area in infant-directed speech perception, relatively few neural studies have directly compared infants' responses to IDS with those to ADS. Using event-related potential (ERP), Zangl and Mills (2007) found that ERP responses between 600 and 800 ms (N600-800), which is linked to attentional processing, was larger to IDS than to ADS for familiar words but not for unfamiliar words in 6-month-old infants, whereas, N600–800 was larger to IDS than to ADS for both familiar and unfamiliar words in infants 13 months of age. Infants of both ages (6 and 13 months) showed increased brain activity to IDS for familiar words only in the left hemisphere. At 13 months, differences between unfamiliar words in IDS versus ADS were significant both in the left and right hemispheres. In addition, ERP responses to IDS from 200 to 400 ms (N200–400), which is linked to word meaning, was larger to IDS than to ADS only for the familiar words in the left temporal and parietal regions of 13-month-old infants. The findings of Zangl and Mills (2007) suggest that the effect of IDS on cerebral function may change with age and experience.

The second function of IDS involves facilitating positive adult–infant interaction. Trainor et al. (2000) suggested that IDS reflects the widespread vocal expression of emotion and IDS is more emotional than ADS. There is some evidence that frontal brain regions are involved in processing emotions. Santesso et al. (2007) examined the electroencephalographic (EEG) responses to emotions including love/comfort, fear, and surprise expressed in IDS and found the greatest activation in response to fear in the frontal region, however, they did not directly compare the response to IDS with that to ADS. Using a two-channel NIRS, Saito et al. (2007) compared neonatal infants' hemodynamic responses in frontal lobe areas while listening to the same stories read by their own mothers in IDS and in ADS. They found that these neonates showed larger frontal activation when they listened to IDS than to ADS. Saito et al. (2007) provides some support for IDS's contribution to infants' frontal activation, however, it is possible that not only the manner of speaking (i.e. adult- or infant-directed) but also the familiarity of a speaker may contribute to an infant's cortical response. Saito et al. (2007) did not compare the cerebral responses to IDS samples read by the infants’ own mothers with ones read by unfamiliar mothers, therefore, it remains possible that the effect of IDS on the frontal activations in infants is limited to IDS of infants' own mothers' voice.

Only a few studies have examined infants' responses to maternal stimuli using brain imaging techniques. Using fMRI, Nitschke et al. (2004) found bilateral brain activations in mothers' orbito-frontal cortex (OFC) when they viewed pictures of their own 3- to 5-month-old infants, but not when they viewed photos of unfamiliar infants. In Leibenluft et al. (2004), viewing pictures of one's own child (ages 5 to 12 years) activated various emotion-related brain areas such as the amygdala, insula, anterior paracingulate cortex, and inferior frontal area near the OFC when compared to familiar infants who were not their own, i.e. their children's friends. Furthermore, using NIRS, Minagawa-Kawai et al. (2009) have examined cerebral activations not only in mothers while viewing movies of both their own infants' and of unfamiliar infants' smiling. They also examined cerebral activations in infants while they viewed movies of their own mothers smiling and movies of unfamiliar mothers smiling. The results indicated that viewing one's own infant's smile elicited increased activations around the anterior part of OFC in the mothers. Furthermore, in 9–13 months infants, the prefrontal activation around the anterior OFC was shown to be specific to viewing their mother's smile but not to viewing an unfamiliar mother' smile.

Infants' brain responses to auditory events (e.g., speech, music, etc.) has been rarely examined in cortical responses of the infant member of mother–child interactions. In infants, a preference for the maternal voice appears soon after birth (DeCasper and Fifer, 1980, Hepper et al., 1993). Barker and Newman (2004) found that speech perception in 6- to 8-month infants was facilitated when it was produced by the infant's mother. Using ERP, Purhonen et al. (2004) demonstrated that the latency of earlier ERP components, i.e., those occurring before 350 ms, which are associated with an orienting response, is significantly shorter to infants' own mothers' voice than to the voice of an unfamiliar female in infants at 4 months of age. The latency of the later ERP components (N450 and P600) was longer to the mother's voice than to the unfamiliar voice. In addition, in these infants, the amplitudes of later ERP components (P600) were lower in response to their own mothers' voice than to an unfamiliar voice (Purhonen et al., 2005). Purhonen et al. (2005) interpreted these later ERP components as reflecting cognitive processing; they suggest that functional cortical specialization involved in the familiar maternal voice is already initiated by the age of 4 months. Furthermore, using fMRI, Dehaene-Lambertz et al. (2010) found significantly stronger responses were elicited in the left posterior temporal regions and in the left and right anterior prefrontal cortex when 2-month-old infants listening to their own mothers' voice than to the voice of an unfamiliar female.

In the present study, we examined cerebral responses of infants to IDS and ADS using near-infrared spectroscopy (NIRS). In order to examine the effects of IDS on cortical activation, we independently measured activations in the temporal and frontal brain areas. If an infant's speech perception was generally affected by IDS, we expected to find higher levels of activation for IDS than for ADS in the temporal brain regions.

Furthermore, to determine the role of speaker familiarity, we compared the infants' cerebral responses when listening to IDS of their own mothers with corresponding responses to IDS given by unfamiliar mothers (strangers). If familiarity contributed to an infant's cerebral response, we anticipated heightened activity associated with this variable.

Moreover, we examined whether speaker familiarity contributed to the differential activities in the temporal and frontal cortical areas. If familiarity was primarily linked to arousal and affect in infants, we expected to find heightened activity in response to the mother's familiar voice largely in the arousal regions such as the frontal cortical areas.

Furthermore, we compared the infant hemodynamic responses in 3 different infant age groups (4–6 months, 7–9 months, and 10–13 months) to examine age-related changes in response to IDS. If the infants’ brain responses were consistent with Hayashi et al. (2001), who showed that the behavioral preference for IDS attenuates between the ages of 4–6 months to 7–9 months and re-appears after 10–11 months of age, we expected decreased levels of activation to IDS in 7–9-month-olds.