Infant Behavior & Development 33 (2010) 361–364

Contents lists available at ScienceDirect

Infant Behavior and Development

Short communication

Sadness in mothers’ ‘baby-talk’ predicts affective disorder inadolescent offspring

Lynne Murraya,∗, Helen Marwickb, Adriane Artechea

a University of Reading, UKb University of Strathclyde, UK

a r t i c l e i n f o

Article history:Received 13 August 2009Received in revised form24 December 2009Accepted 19 March 2010

Keywords:SpeechPostnatal depressionBaby-talkParenteseAffective disorderDepression

a b s t r a c t

‘Baby-talk’ is common across cultures. It underpins infant vocal preferences, and helpsregulate infant engagement. Its longer-term significance is unclear. In a longitudinal study,we found indications of ‘sadness’ in postnatally depressed mothers’ baby-talk statisticallymediated effects of maternal depression on offspring adolescent affective disorder.

© 2010 Elsevier Inc. All rights reserved.

When adults speak to infants, they do so in specially adapted ways, encapsulated in the term ‘baby-talk’, or ‘parentese’.Compared to adult-directed speech, baby-talk is reduced in rate, has short, regularly spaced utterances, and has higher andmore modulated and expanded pitch contours. Voice timbre is also distinctive, with more whispered speech in talk to infants.These adaptations occur in cultures differing widely in adult-directed speech characteristics (Fernald et al., 1989), suggestingthey might be of functional significance for the infant. Indeed, infant-directed speech characteristics, and particularly theprosodic and timbral features, underpin infant vocal expression preferences; and they facilitate infant participation in socialexchanges by prolonging attention, and regulating infant mood and state (see reviews of Kuhl, 2004; Saffron, Werker, &Werner, 2006). An important question is whether such adjustments are of longer-term significance for infant and childdevelopment. One way to address this is by studying the development of populations exposed to speech characteristics thatdeviate from the norm. A good candidate population comprises children exposed to maternal depression in the postnatalmonths. Depression affects both prosodic and voice quality characteristics of adult-directed speech, including a reductionin pitch range, a preponderance of falling pitch contours, and the presence of abnormal voicing quality, such as ‘creak’phonation, where tense larynx creaky pulses occur on voiced utterances (Laver, 1980). Together, these features give theimpression of flat and monotonous speech; and they are associated with non-depressed individuals’ perception of sorrow orsadness (Scherer, 2003). More restricted evidence indicates that depressed mothers’ speech to their infants is also affected,such that it lacks many of the adjustments typical of baby-talk, being both less responsive, and having fewer prosodically‘exaggerated’ intonation contours and fundamental frequency modulations (Bettes, 1988).

To date, the question of whether the speech of depressed mothers is associated with the adverse child psychologicaloutcomes associated with the disorder has principally been examined in relation to infant cognitive functioning. Here,

∗ Corresponding author at: School of Psychology and Clinical Language Sciences, University of Reading, RG6 6AL, UK. Tel.: +44 0118 378 6667.E-mail address: lynne.murray@reading.ac.uk (L. Murray).

0163-6383/$ – see front matter © 2010 Elsevier Inc. All rights reserved.doi:10.1016/j.infbeh.2010.03.009

362 L. Murray et al. / Infant Behavior & Development 33 (2010) 361–364

disturbances in the attention-regulatory characteristics of baby-talk have been found to mediate the adverse effects ofmaternal depression on infant learning (Kaplan, Bachorowski, Smoski, & Hudenko, 2002). In contrast, there have been nostudies on the implications of postnatally depressed mothers’ speech quality for child emotional functioning. Offspring ofdepressed parents are at raised risk for affective disorder themselves, but since such disorders (and especially depression)are not normally manifest before the teenage years, long-term follow-up studies are required to investigate any associationswith speech quality in early mother–infant interactions.

We investigated this question in the context of a prospective longitudinal study of the development of offspring of post-natally depressed and well mothers (Murray, 1992). As part of our larger study, we audio-recorded a random sub-sampleof the mothers as they spoke to their infants in a 5-min face-to-face interaction at 2–3 months postpartum. Twenty-nineof these mothers had DSM-IV depressive disorder, and 20 were non-depressed controls. The groups were demographicallycomparable. We have previously reported that structural aspects of maternal speech during these interactions, includ-ing the number and length of utterances, did not distinguish depressed and well women (Murray, Kempton, Woolgar, &Hooper, 1993). Here, we report on mothers’ speech quality in these same interactions, and its relation to the occurrence ofDSM-IV affective disorder (anxiety, depression, dysthymia) in offspring, when followed up at age 13 years and assessed bystandardized interview. All participants gave written informed consent.

Maternal utterances (identified by contour-marking pauses of ≥0.5 s) were coded by trained listeners who were blind tomaternal and offspring psychiatric state, using Laver’s scheme (Laver, 1980) for the presence of the principal speech featuressignaling sadness, i.e., falling intonation and ‘creak’ voice, together with their repetitiveness. The frequency of occurrence ofeach speech feature was expressed as the percentage of the total number of both voiced and unvoiced (e.g., tongue clicks)utterances; and the degree of repetition as the percentage of utterances within each of these categories that comprised‘identical runs’, i.e., where at least two consecutive instances occurred.

First, we compared the depressed and well mothers’ speech on these dimensions; then we examined maternal speechin relation to adolescent disorder; and finally, since we had already shown the rate of affective disorder to be raised inthe postnatally depressed mothers’ children (Halligan, Murray, Martins, & Cooper, 2007), we examined whether particularspeech characteristics statistically accounted for this association, having first checked for potentially confounding effects ofmaternal depression occurring beyond the postnatal period.

Analyses of variance showed, as predicted, that compared to non-depressed mothers, those with depression used a higherproportion of falling intonation utterances [M = 57.50 (SD = 9.94) vs. 50.50(9.13), F(1, 45) = 5.98, p = .02], as well as a morerepetitive use of falling intonation [46.17(10.39) vs. 39.65(10.65), F(1, 45) = 4.36, p = .04]. Depressed mothers’ speech alsoshowed more overall ‘creak’ than that of controls [13.90(8.75) vs. 7.82(9.92) F(1, 45) = 4.90, p = .03]; although repetitivenessof ‘creak’ was somewhat greater in the index group [40.10(27.19) vs. 28.44(30.04), F(1, 45) = 1.91], the difference was notsignificant p = .17.

Chi-square analysis showed that just over a third (34.5%) of the postnatally depressed mothers’ children had experiencedan episode of affective disorder by age 13 years, compared to 10.0% of control adolescents �2 (df = 1, N = 49) = 3.83, p = .05 one-tailed. We examined the relationship between mothers’ speech to their infants and the occurrence of adolescent disorder viacorrelational analysis. All four speech measures were associated with adolescent affective disorder in the predicted direction,and this was significant for repeated falling intonation contours r = .34, p = .02.

Because postnatally depressed mothers are known to be at increased risk for subsequent depression, and effects ofsuch subsequent disorder could confound interpretation of associations with PND, we also examined whether chronicityof maternal depression beyond 4 months postpartum was related to child diagnosis, as well as to maternal speech at 2–3months. Results showed, as expected that, compared to controls, PND mothers experienced significantly more depression inthe 13 years after the postpartum period [PND group M = 18.64 (SD = 18.05) months vs. control M = 3.65 (SD = 4.65); df = 31.92,t = −4.20, p < .001]. Nevertheless, the chronicity of maternal depression occurring beyond the fourth PN month showed only atrend association with offspring diagnosis within the sample considered as a whole (B(S.E.) = .04(.02), Wald = 3.59, OR = 1.04,confidence interval = .99–1.08, p = .06), and no association within the PND group (p = .30). Finally, correlation analyses showedthat the chronicity of maternal depression beyond the PN months was not significantly related to any of the speech variablesassessed at 2–3 months, either in the total sample, or within PND mothers only. The association with repeated fallingintonation was r = .16, p = .28 for the total sample, and r = .06, p = .77 for the PND group only. In the light of these results, wethen used regression analyses to determine whether use of repetitive falling intonation statistically mediated the effects ofthe mother’s PND on adolescent outcome. These showed that, once this speech measure was taken into account, the effectof maternal postnatal depression on adolescent disorder was no longer significant B(S.E.) = 1.00(.88), Wald = 1.52, OR = 2.97,confidence interval = .52–16.82, p = .21, and Sobel’s test confirmed a mediational effect (Z = 2.31, p = 0.02). Moreover, theexplained variance in adolescent disorder was increased by 6% when repeated falling intonation was included, highlightingits unique contribution over maternal depression. Thus, our findings suggest that the effects of maternal depression onoffspring affective disorder are at least partially mediated by exposure to a specific speech environment (i.e., monotonousspeech signaling sadness).

As young infants do not understand the speech lexicon, the role of the distinctive non-verbal features of speech directedto them is of particular interest. It is notable that infant-directed speech in the early months, in contrast to that in late infancy,is primarily affective in nature. Moreover, since infant vision and prehension are limited, it is also largely non-referential,apart from reference to interpersonal or ‘relational’ affects (Bornstein et al., 1992). As such, early baby-talk, and particularlyits prosodic features, shares characteristics with vocalizations between non-human primates which also serve important

L. Murray et al. / Infant Behavior & Development 33 (2010) 361–364 363

relational, social affective functions (e.g., in regulating group response to threat, or reinforcing social bonds (Seyfarth &Cheney, 2003)). This parallel is underscored by evidence for specialized regions in the monkey brain for conspecific vocal-izations that suggest a close functional correspondence between the monkey and human voice regions (Petkov et al., 2008).Intriguingly, our findings suggest that emotions expressed in human parents’ early baby-talk might not only help regulateconcurrent infant affect and behavior, but might also have longer-lasting implications for child emotional development.

Our findings also raise some questions. First, it is important to note that maternal depression is associated with a range ofdisturbances in contacts with the infant, including a reduction in behavioural responsiveness and change in facial expression.The extent to which the effects of speech quality we identified are privileged, or else depend on behavioural aspects ofmaternal functioning (Kuhl, Tsao, & Lui, 2003), is unclear, and this requires further investigation. Further questions also ariseconcerning timing of effects, namely, by what age do infants first distinguish cues signaling depressed affect; and is there aperiod in development when exposure to such cues is required for any effects to endure. With regard to the former, whileour study is consistent with the suggestion that distinctions concerning maternal affect might be detected by 3 months,our evidence is only indirect and cannot speak to the time scale of acquisition up to this point. Experimental studies couldusefully inform this issue. With regard to the second question concerning timing of exposure, it is notable that, in our study,maternal depression generally remitted by 4 months postpartum (Murray, 1992). Furthermore, our analyses showed thatthe speech changes associated with the postnatal episode were not simply a marker for subsequent maternal depression thatmight, in principle, have better accounted for offspring disorder. While this suggests the possibility of an early sensitizationto affective dimensions of speech input, previous research has generally identified a rather later period (i.e., the secondhalf of the first year) as one when speech sensitivities are established (e.g., for consonant distinctions, stress, or language-specific sound combinations (Kuhl, 2004; Saffron et al., 2006)). Having said this, there is some evidence that sensitizationto other dimensions, such as vowels, might occur somewhat earlier (Kuhl, Williams, Lacerda, Stevens, & Lindblom, 1992;Polka & Werker, 1994). Together, therefore, our own and others’ findings stress the need for further research to refine theunderstanding of timing effects in relation to different aspects of speech directed to infants.

Comment is also required concerning the fact that the principal feature of maternal speech predicting adolescent affectivedisorder was not the absolute frequency of utterances with falling intonation, but its repetitiveness. In fact, falling intonationcontours are common (e.g., Stern, Spieker, & Mackain, 1982), but their use is normally balanced and contrasted with rising andother contours to regulate infant attention and state. When organized in repeated runs, however, falling intonation producesthe flat, monotonous quality characteristic of depressed individuals’ speech. Just as exposure in infancy to the statistical andprosodic features of one’s native language promotes dedicated neural networks that encode for its particular patterns (Kuhl,2004; Saffran, 2003), it is possible that the preponderance of repetitive falling intonation contours in depressed mothers’baby-talk may similarly sensitize the child to the expression of negative affect. This could be important since, in line withevidence for auditory-vocal mirroring in neurons for learned vocal communication in song-birds (Prather, Peters, Nowicki,& Mooney, 2008), consistent exposure to sadness features of baby-talk during the early months might stimulate infant brainareas concerned with the subjective experience of negative affect (Gallese, 2005); and their increased repetitiveness mightalso enhance the perception of their salience, making activation of such neural responses more likely (Liepelt, von Cramon,& Brass, 2008). Either of these mechanisms might contribute to the increased risk for affective disorder in offspring.

Acknowledgements

The research was funded by the MRC (UK). We thank Carla Martins and Sheelah Seeley for assistance with data collectionand analysis.

References

Bettes, B. A. (1988). Maternal depression and motherese: Temporal and intonational features. Child Development, 59, 1089–1096.Bornstein, J., Tal, J., Rahn, C., Galperin, C., & Lamour, M. (1992). Functional analysis of the contents of maternal speech to infants of 5 and 13 months in four

cultures: Argentina, France, Japan, and the United States. Developmental Psychology, 28, 593–603.Fernald, A., Taeschner, T., Dunn, J., Papousek, M., de Boysson-Bardies, B., & Fukui, I. (1989). A cross-language study of prosodic modifications in mothers’

and fathers’ speech to preverbal infants. Journal of Child Language, 16, 477–501.Gallese, V. (2005). Embodied simulation: From neurons to phenomenal experience. Phenomenology and the Cognitive Sciences, 4, 23–48.Halligan, S. H., Murray, L., Martins, C., & Cooper, P. J. (2007). Maternal depression and psychiatric outcomes in adolescent offspring: A 13-year longitudinal

study. Journal of Affecive. Disorders, 97, 145–154.Kaplan, P. S., Bachorowski, J.-A., Smoski, M. J., & Hudenko, W. J. (2002). Infants of depressed mothers, although competent learners, fail to learn in response

to their own mothers’ infant-directed speech. Psychological Science, 13, 268–271.Kuhl, P. (2004). Early language acquisition: Cracking the speech code. Nature Neuroscience, 5, 831–843.Kuhl, P., Williams, K., Lacerda, F., Stevens, K., & Lindblom, B. (1992). Linguistic experience alters phonetic perception. Science, 255, 606–608.Kuhl, P., Tsao, F., & Lui, H. (2003). Foreign language experience in infancy: Effects of short term exposure and social interaction on phonetic learning.

Proceedings of the National Academy of Sciences, USA, 100, 9096–9101.Laver, J. (1980). The phonetic description of voice quality. Cambridge: CUP.Liepelt, R., von Cramon, D. Y., & Brass, M. (2008). What is matched in direct matching? Intention attribution modulates motor priming. Journal of Experimental

Psychology: Human Perception and Performance, 34, 578–591.Murray, L. (1992). The impact of postnatal depression on infant development. Journal of Child Psychology and Psychiatry, 33, 543–561.Murray, L., Kempton, C., Woolgar, M., & Hooper, R. (1993). Depressed mothers’ speech to their infants and its relation to infant gender and cognitive

development. Journal of Child Psychology and Psychiatry, 34, 1083–1101.Petkov, C., Kayser, C., Steudel, T., Whittingstall, T., Augath, M., & Logothetis, N. (2008). A voice region in the monkey brain. Nature Neuroscience, 11, 367–374.

364 L. Murray et al. / Infant Behavior & Development 33 (2010) 361–364

Polka, L., & Werker, J. (1994). Developmental change in perception of nonnative vowel contrasts. Journal Experimental Psychology: Human Perception andPerformance, 20, 421–435.

Prather, J. F., Peters, S., Nowicki, S., & Mooney, R. (2008). Precise auditory-vocal mirroring in neurons for learned vocal communication. Nature, 451, 305–310.Saffran, J. R. (2003). Statistical language learning: Mechanisms and constraints. Current Directions in Psychological Science, 12, 110–114.Saffron, J., Werker, J., & Werner, L. (2006). The infant’s auditory world: Hearing, speech, and the beginnings of language. In R. Siegler, & D. Kuhn (Eds.),

Handbook of child psychology, Vol. 2, cognition, perception and language (6th edition, pp. 58–108). New York: Wiley.Scherer, K. R. (2003). Vocal communication of emotion: A review of research paradigms. Speech Communication, 40, 227–256.Seyfarth, R. M., & Cheney, D. L. (2003). Meaning and emotion in animal vocalizations. Annals of the New York Academy of Sciences, 1000, 32–55.Stern, D., Spieker, S., & Mackain, K. (1982). Intonation contours as signals in maternal speech to prelinguistic infants. Developmental Psychology, 18, 727–735.