Acoustic parameters of infant-directed singing in mothers of infants with down syndrome

Infant Behavior and Development 49 (2017) 151–160

Contents lists available at ScienceDirect

Infant Behavior and Development journal homepage: www.elsevier.com/locate/inbede

Full length article

Acoustic parameters of infant-directed singing in mothers of infants with down syndrome

MARK

⁎

Shannon de l’Etoilea, , Samarth Behuraa, Cengiz Zopluoglub a b

University of Miami, Frost School of Music, United States Educational and Psychological Studies University of Miami, School of Education and Human Development, United States

AR TI CLE I NF O

AB S T R A CT

Keywords: Infant-directed singing Infant self-regulation Infants with down syndrome

This study compared the acoustic parameters and degree of perceived warmth in two types of infant-directed (ID) songs − the lullaby and the playsong − between mothers of infants with Down syndrome (DS) and mothers of typically-developing (TD) infants. Participants included mothers of 15 DS infants and 15 TD infants between 3 and 9 months of age. Each mother’s singing voice was digitally recorded while singing to her infant and subjected to feature extraction and data mining. Mothers of DS infants and TD infants sang both lullabies and playsongs with similar frequency. In comparison with mothers of TD infants, mothers of DS infants used a higher maximum pitch and more key changes during playsong. Mothers of DS infants also took more time to establish a rhythmic structure in their singing. These differences suggest mothers are sensitive to the attentional and arousal needs of their DS infants. Mothers of TD infants sang with a higher degree of perceived warmth which does not agree with previous observations of “forceful warmth” in mothers of DS infants. In comparison with lullaby, all mothers sang playsong with higher overall pitch and slower tempo. Playsongs were also distinguished by higher levels of spectral centroid properties related to emotional expressivity, as well as higher degrees of perceived warmth. These similarities help to define specific song types, and suggest that all mothers sing in an expressive manner that can modulate infant arousal, including mothers of DS infants.

Infant-directed (ID) singing is the distinct way mothers sing to young infants as part of caregiving (Dissanayake, 2000; Trehub, Becker, & Morley, 2015; Trehub & Gudmundsdottir, 2015). In an effort to either match or alter infant state, mothers instinctively manipulate the acoustic features of their singing, thus establishing a dynamic interaction through which infants gain valuable experience in self-regulation. To explain, ID singing is characterized by salient rhythmic components, such as regular beat, metrical structure, and small-scale temporal deviations; as well as high levels of repetition (Corbeil, Trehub, & Peretz, 2013a; Delavenne, Gratier, & Devouche, 2013; Longhi, 2009; Margulis, 2013; Nakata & Mitani, 2005). Such qualities create perceptual expectancies, thereby effectively attracting and directing infant attention to the caregiver, and maintaining infant attention over time. In addition to these temporal features, mothers commonly adjust the pitch, dynamics, and timbre of their singing, thus providing a rich affective experience for infants (Longhi, 2009; Trainor, 1996; Trainor, Clark, Huntley, & Adams, 1997; Trehub, Ghazban, & Corbeil, 2015; Tsang & Conrad, 2010). As an example, mothers commonly smile when singing to infants which alters the mouth opening and changes vocal tract shape, thus elevating pitch and increasing amplitude (Tartter, 1980). Singing in this manner impacts vocal timbre, thereby producing high positive vocal affect which serves to regulate infant attention (Corbeil et al., 2013a). Consequently, infants undergo affect regulation that is reflected through changes in both behavioral and physiological markers of

⁎

Corresponding author at: University of Miami, Frost, School of Music, P.O. Box 248165, Coral Gables, FL 33124, United States. E-mail address: [email protected] (S. de l’Etoile).

http://dx.doi.org/10.1016/j.infbeh.2017.09.001 Received 23 April 2017; Received in revised form 25 July 2017; Accepted 1 September 2017 0163-6383/ © 2017 Elsevier Inc. All rights reserved.

Infant Behavior and Development 49 (2017) 151–160

S. de l’Etoile et al.

arousal (Corbeil et al., 2013b; Ghazban, 2013; Shenfield et al., 2003). Ultimately, mothers use ID singing to guide infants toward focused attention and optimal arousal; two ingredients considered essential for self-regulation (Berger, 2011a; Thompson, 1994). By modifying these acoustic features, mothers create specific types of ID songs, including lullabies that soothe and calm infants (Trehub & Trainor, 1998), as well as playsongs which increase stimulation and promote more lively interactions (Rock et al., 1999; Trainor, 1996). While the two song types share certain acoustic features, each one is characterized by unique elements. For instance, both lullabies and playsongs are marked by high mean pitch and high levels of jitter (e.g., variation in fundamental frequency). At the same time, playsongs are known to be more rhythmic than lullabies, and to convey feelings of joy and happiness through greater pitch variability and increased dynamic range (Trainor et al., 1997). Meanwhile, lullabies are perceived as tender and soothing due to having a slower tempo, more regular rhythm, lower mean pitch and pitch variability, and being sung with a full or resonant timbre (Juslin & Laukka, 2003; Trainor et al., 1997). A mother’s choice of song type is largely dictated by her infant’s self-regulatory needs and capacities in the moment. As an example, Ghazban (2013) found that when 10-month-old infants were upset, most mothers sang lively playsongs which effectively captured infant attention, promoted their engagement, and reduced physiological distress. In essence, mothers appeared to choose a song type associated with the desired infant state, as opposed to selecting a song type that reflected current infant state. Ghazban (2013) suggests that infant response to song type may be further mediated by age and familiarity, as mothers tend to sing lullabies more frequently to very young infants and less often with older (i.e., 10-month-old) infants. Additionally, lullabies may be less familiar to infants as they are typically limited to being sung at bedtime, and not repeatedly throughout the day, like playsongs. Thus, playsongs may be more familiar to infants and thereby, comforting in a stressful situation. Through providing the appropriate song type, mothers help infants practice critical skills of self-regulation. By definition, selfregulation encompasses all efforts designed to modify ones’ inner state or responses and is essential for emotional capacity across the life span, having significant ramifications for social and cognitive functioning (Calkins & Hill, 2007; McCabe et al., 2004; Vohs & Baumeister, 2004). The ability to self-regulate begins to develop during infancy and is largely dependent upon the consistent provision of sensitive caregiving early in life (Berger, 2011b; Feldman, 2007; McCabe, Cunnington, & Brooks-Gunn, 2004; Sroufe, 2000). Infants with Down syndrome may be at a disadvantage for acquiring self-regulation due to various neurological anomalies that impact attention and arousal mechanisms, and that subsequently may alter the nature of caregiver interactions. For infants with Down syndrome (DS), delayed maturation of neuroregulatory systems can manifest as diminished emotionality and responsivity (Carvajal & Iglesias, 2002; Cicchetti et al., 1991). These infants may therefore be slower to respond and orient during caregiver interactions and appear to be passive or disengaged. Such cues may be more challenging or less rewarding for mothers to read, who then respond with an adapted interaction style described as warm, yet directive (Cebula et al., 2010; Spiker, 2006). Through “forceful warmth,” mothers may effectively engage infants with DS, but may not recognize their infants’ social bids, thus their interaction style may lack the sensitivity needed for self-regulation (Moore et al., 2002). Over time, a forcefully-warm interaction style may also result in DS infants becoming dependent on their mothers for attention regulation, thereby limiting opportunities for triadic engagement as needed for language development and various cognitive functions, such as problem-solving (Cebula, Moore, & Wishart, 2010; Moore, Oates, Hobson, & Goodwin, 2002). To summarize, the unique acoustic features of ID singing appear to promote infant self-regulation. An important developmental milestone, self-regulation supports social and cognitive functioning, yet may be delayed or impaired in infants with DS due to neurological differences inherent in the diagnosis as well as subsequent changes in caregiver interaction style. The impact of Down syndrome on the acoustic parameters of maternal singing has not yet been articulated, nor has the construct of warmth been explored in relation to ID singing. Thus, evidence is lacking to explain whether or not mothers’ singing toward DS infants includes the acoustic and musical features that effectively direct infant attention and modulate infant arousal as needed for self-regulation. The purpose of this exploratory study was to compare the acoustic parameters and degree of perceived warmth in two types of ID songs − the lullaby and the playsong − between mothers of DS infants and mothers of typically-developing (TD) infants. Findings will help clarify whether or not a modified interaction style transfers to ID singing, to the extent that self-regulation could be supported or compromised in infants with DS. Specifically, the following research questions were addressed: what effect does the diagnosis of DS and/or song type have on the way a mother manipulates the musical and vocal qualities of her singing? Further, what is the effect of the DS diagnosis and/or song type on perceived warmth in ID singing? Finally, which acoustic parameters are common to or specific to song type (i.e., lullaby vs. playsong) in mothers’ singing? 1. Method 1.1. Participants Participants included 16 mothers and their infants with Down syndrome (DS), as well as 15 mothers and their typically-developing (TD) infants. Recruiting occurred in community organizations serving mothers and infants in a culturally diverse, major metropolitan area within the southeastern United States, following approval from a university human subject committee. Data were discarded for the mother of one infant with DS because the mother did not sing during the testing procedure. The subsequent data analyses pertain to the remaining 30 mothers of infants who ranged in age from 3 to 9 months and were matched by chronological age across the two groups. Infants were matched by chronological age due to the fact that developmental differences between DS and TD infants become more pronounced during the second post-natal year (Carvajal & Iglesias, 2002). Thus, most research pertaining to DS and TD infants within the first post-natal year involves matching by chronological age, as opposed to matching by verbal or mental age. 152

Infant Behavior and Development 49 (2017) 151–160

S. de l’Etoile et al.

Table 1 Demographic Information on Infant Participants According to Infant Status.

Average age Gender Male Female Birth order First-born Second-born Third-born Other

Down Syndrome Infants

Typically-Developing Infants

Total Infants

N = 15

N = 15

N = 30

5 months, 28 days

5 months, 27 days

8 7

8 7

16 14

5 5 2 3

8 7 0 0

13 12 2 3

Infants born prematurely were matched according to their adjusted age. To meet inclusionary criteria, infants were born after the 32nd week of gestation; the age at which the human auditory system is typically functional (Graven, 2000). Infants meeting this criterion were believed to have had intact sensory and perceptual systems, and while infants were not screened for sensory impairment, all mothers reported typical vision and hearing for their infants. Using cross-tabulations, no significant differences existed between infant groups in regard to infant age, gestation length, or birth order. Further, no significant differences emerged between infant groups pertaining to mothers’ age, educational background, or ethnicity. See Tables 1 and 2 for participants’ demographic information. 1.2. Procedure Mother-infant pairs met individually with the researcher for one data collection appointment. Mothers were instructed to sing any song(s) of their choosing and to sing to their infant as they normally would at home, in any language they preferred. Thus, mothers were not instructed to specifically sing a lullaby or playsong, or to sing in any particular style. While such an approach may introduce variability, this approach was deemed appropriate for the following reasons: 1) this study was exploratory in nature, in that no previously-published research could be located in regard to ID singing and infants with Down syndrome; 2) in previous research, TD infants responded more positively to ID singing when mothers selected the song (de l’Etoile, 2006); and 3) because the intent of the current study was to capture the natural, authentic singing of mothers as they intuitively modified vocal and musical features to best accommodate infant state at the time of data collection. During the two-minute data collection period, mothers either sang one song repeatedly, or sang several different songs. Each infant was secured in an infant car seat placed on a table approximately 18 in. from the mother’s face while she was seated at the table. Each mother’s singing voice was recorded using a hand-held digital recorder (H4 Zoom) which was placed approximately 24 in. away from the mother. This procedure was part of a larger study that examined infant response to ID singing (e.g., specifically the mother’s face and voice); therefore, mothers were discouraged from using hand gestures or making physical contact with infants while singing. A researcher was present while mothers sang, but stood behind a screen so as not to distract infants. Following data collection, each mother received a $25.00 gift card. Additionally, the researcher gave each mother the opportunity to watch a video recording of her infant during the ID singing recording. The researcher offered feedback to each mother regarding her infant’s behavior, in an effort to help the mother recognize the importance of infant responses and to promote the use of ID Table 2 Demographic Information on Mother Participants According to Infant Status.

Age range, in years Average age Educational background Did not complete high school High school Some college coursework Associate’s degree Bachelors degree Graduate degree Ethnicity African-American Hispanic Caucasian Other

Mothers of Infants with Down syndrome N = 15

Mothers of Typically-Developing Infants N = 15

Total Mothers N = 30

20–42 33.9

29–39 33.27

20–42 33.60

2 1 2 2 3 5

(13.3%) (6.7%) (13.3%) (13.3%) (20%) (33.3%)

0 0 0 0 5 (33%) 10 (66.7%)

2 (6.7%) 1 (3.3%) 2 (6.7%) 2 (6.7%) 8 (26.7%) 15 (50%)

2 9 2 2

(13.3%) (60%) (13.3%) (13.3%)

0 12 (80%) 1(6.7%) 2 (13.3%)

2 (6.7%) 21 (70%) 3 (10%) 4 (13.3%)

153

Infant Behavior and Development 49 (2017) 151–160

S. de l’Etoile et al.

singing in future situations.

1.3. Preparation of acoustic data Audacity software was used to prepare and edit the audio files for subsequent feature extraction, data mining, and statistical analysis. The audio files were first recorded to stereo 44.1-kHz, 16-bit uncompressed WAV files, and converted to monaural by merging the left channel with the right channel in order to ensure that sounds in both channels would be equally represented. To remove non-ID singing elements, a two-stage process was used. First, each file was high-pass filtered at 150 Hz with a gentle slope of 6db per octave to remove any noise due to floor thumps, HVAC rumble, and other low frequencies not typically emitted by the female singing voice. Second, each file was down-sampled to 16 kHz, which allowed accurate representation of audio frequencies up to 8 kHz; well above the most prominent vocal formants of the typical female singing voice. The two-minute audio file for each mother was then subjected to a careful editing process that produced “sub-clips” of her singing according to the following criteria: 1. The beginning of the sub-clip aligned with the identifiable beginning section of an ID song, typically defined by the clear establishment of a tonal key in the mother’s singing. 2. The end of the sub-clip aligned with the identifiable ending section of an ID song, typically defined by a clear, harmonic cadence in which the mother’s voice returned to the tonal key. Each sub-clip was on average 20 s in length and no sub-clip overlapped with any other sub-clip in the mother’s recorded singing. From the sub-clips generated for each of the 30 mothers, three sub-clips having minimal noise profiles were selected for analysis, thus producing a total of 90 audio files. Acoustic feature extraction involved the use of MATLAB; a popular and widely-used programming interface for signal analysis, data visualization, etc. The MIRToolBox is a set of integrated features written in MATLAB to extract musical and audio related features most commonly used in the field of Musical Information Retrieval (Lartillot & Toiviainen, 2007). These programs enable the extraction of large sets of musical features from audio databases, thus providing a workable foundation for data analysis. All data processing and feature extraction tasks were carried out on a MacBook Pro 2.5 GHz Intel Core i7 with 8 GB DDRAM. The down sampling and the length of the sub-clips helped to reduce the amount of time required for processing and feature extraction. The extraction process yielded 609 acoustic features which next underwent data mining using the Waikato Environment for Knowledge Analysis (WEKA) (Frank et al., 2004). From the data mining process, the set of acoustic features was narrowed to 18 items within six categories which were further analyzed using Praat software (Boersma & Weenink, 2016): 1. Pitch: starting pitch, mean pitch (including overall average pitch, as well as average maximum and average minimum pitch), pitch variability, and jitter 2. Intensity: mean intensity (including overall average intensity, as well as average maximum and average minimum intensity), variability, and shimmer 3. Tempo: mean and variability 4. Tonal key clarity: mean and variability 5. Rhythmicity: rhythm attack time and degree of voice breaks 6. Vocal timbre: formants, mean harmonics-to-noise ratio, spectral centroid, spectral brightness, and spectral spread Although not a comprehensive list, previous research on ID singing has identified these categories of acoustic features as critical for garnering infant attention, and for communicating emotion (Coutinho & Dibben, 2013; Trainor, 1996; Trainor et al., 1997). Additionally, these features have been utilized to distinguish ID from other types of songs, and to further classify them as lullabies and/or playsongs (Bergeson, & Trehub, 1999; Milligan et al., 2003; Trehub, 2000; Trehub et al., 1993).

1.4. Data coding for song type and maternal warmth To determine song type as lullaby or playsong for each of the 90 sub-clips, 10 college students who were blind to the purpose of the study listened to all 90 clips in random order. Student coders were instructed to identify each sub-clip as either a lullaby, that is, a song designed to soothe and calm an infant or perhaps help an infant to fall asleep; or as a playsong, meaning a song intended to stimulate and arouse the infant or to engage the infant in play. Coders were further instructed to not use lyric content in making these determinations; rather, coders needed to attend to how each mother was singing. To identify differences in perceived warmth of each mothers’ singing, two other college students who were also blind to the purpose of the study listened to all 90 sub-clips in random order and rated degree of warmth. Coders were told: “Mothers may sing to their infants with varying degrees of warmth. Greater amounts of warmth may indicate the mother is making a greater effort to engage her infant, or displaying a higher level of absorption in her infant.” They then rated perceived degree of warmth in each subclip using a 5-point Likert scale ranging from “1” meaning “no perceived warmth” to “5” indicating “extremely high amount of warmth.” 154

Infant Behavior and Development 49 (2017) 151–160

S. de l’Etoile et al.

1.5. Statistical analysis A linear regression model was used to explore the main effect of infant status (i.e., DS or TD) as well as main effect of song type (i.e., lullaby or playsong) on the acoustic parameters and degree of perceived warmth. Interaction effects were also explored between infant status and song type, while controlling for infant age, gender, and sub-clip order (i.e., first, second, or third sub-clip in time during the 2 min of singing). A generalized estimating equation (GEE) with a first order autoregressive (AR-1) structure imposed to the correlation matrix was used to fit the model. Based on student coder ratings, song type was considered a continuous variable with 0 indicating lullaby and 1 designating playsong. As an example, if 2 students rated a sub-clip as playsong, while 8 students rated the same sub-clip as lullaby, the value of the sub-clip was 0.2. For statistical significance, the level was set at p ≤ 0.05, and for practical significance, Cohen’s d was utilized with the following effect size determinations: 0.2 for small effect, 0.5 for medium effect, and 0.8 for large effect (Cohen, 1988). All statistical analysis was run using R, an open source statistical software (R Core Team, 2016). 2. Results 2.1. Results for song type Seven out of the 90 sub-clips received poor reliability ratings (i.e. < 70%) and were not clearly identified as either a lullaby or playsong. Thus, only the remaining 83 sub-clips were included in all subsequent analyses. Of the 7 sub-clips that were excluded, only 2 sub-clips pertained to the same mother; therefore, no participants had to be eliminated from the study. Average inter-rater agreement among the student data coders for song type of the 83 selected sub-clips was 92%, which is an acceptable level of agreement (Graham, Milanowski, & Miller, 2012). Further, 44 of the 83 sub-clips received 100% agreement across all 10 data coders. Overall, mothers of DS infants sang 17 lullabies and 25 playsongs, while mothers of TD infants sang 15 lullabies and 26 playsongs; thus the two infant groups did not differ significantly in terms of song type prevalence. 2.2. Results for acoustic parameters Interaction effects were explored to determine the combined effect of Down syndrome and song type on the musical and vocal qualities of mothers’ singing. Of the 18 acoustic parameters included in the analysis, 3 features showed a significant interaction effect between infant status and song type, including maximum pitch, tonal key clarity, and timbre (formants 2 and 4). For the purposes of this study, the variable of mean pitch was parsed out to include overall average pitch, as well as average maximum and average minimum pitch. Of these three features, average maximum pitch, defined as the highest mean pitch across sub-clips, was higher for playsong (447.48 Hz) than lullaby (394.20 Hz) in the singing of mothers with DS infants. Meanwhile, mothers with TD infants sang with higher maximum pitch during lullaby (436.82 Hz) than playsong (393.71 Hz), thus showing an inverse relationship, by comparison. The interaction between infant status and song type for maximum pitch was significant (b = 96.38, SE = 41.621, p = 0.021). While maximum pitch in lullaby was not significantly different between infant groups (b = 42.6, SE = 33.18, p = 0.199, d = 0.64), maximum pitch in the playsongs of mothers with DS infants was significantly higher than that of mothers with TD infants (b = 53.3, SE = 26.98, p = 0.048, d = 0.80). Differences also emerged for average tonal key clarity which indicates how well a mother maintains a tonal center or “keyness” when singing to her infant. Mothers of DS infants sang with greater tonal key clarity during lullaby (0.49) than playsong (0.47) whereas mothers of TD infants showed the opposite effect, singing with greater tonal key clarity for playsong (0.48) than lullaby (0.47). The interaction between infant status and song type for tonal key clarity was significant (b = 0.038, SE = 0.015, p = 0.012). While no significant difference emerged between groups for lullaby (b = 0.022, SE = 0.013, p = 0.10, d = 0.89), tonal key clarity was significantly lower in the playsongs of mothers with DS infants (b = 0.016, SE = 0.007, p = 0.0.17, d = 0.70). Interaction effects also emerged for certain parameters of vocal timbre, namely Formant 2 (F2) and Formant 4 (F4). Formants represent the presence of acoustic energy around certain frequencies in the speech wave (Wood, 2005). In the singing of mothers with DS infants, F2 values for playsong (1933.19 Hz) were higher than F2 values for lullaby (1786.09 Hz). By contrast, mothers of TD infants sang with higher F2 during lullaby (1993.99 Hz) than playsong (1893.01 Hz). The interaction between infant status and song type for F2 was significant (b = 248.07, SE = 59.09, p < 0.001). F2 values for lullaby from mothers of DS infants were significantly lower than those of mothers of TD infants (b = 207.9, SE = 53.1, p < 0.001, d = 1.7), while differences for playsong were not significant (b = 40.2, SE = 49.6, p = 0.42, d = 0.33). In regard to F4, mothers of DS infants had higher values for playsong (4143.14 Hz) than for lullaby (4065.74 Hz). The effect was reversed in mothers of TD infants, for whom F4 values were higher for lullaby (4154.97 Hz) than for playsong (4062 Hz). For F4, the interaction between infant status and song type was statistically significant (b = 170.6, SE = 58.9, p = 0.004). Between groups, F4 values for lullaby from mothers of DS infants were significantly lower (b = 89.2, SE = 43.1, p = 0.038, d = 0.83), but F4 differences for playsong were not significant (b = 81.4, SE = 43.1, p = 0.090, d = 0.76). Main effects also emerged from the analysis, thus demonstrating the individual impacts of infant status or song type on acoustic features. One feature, rhythm attack time, demonstrated a main effect of both infant status and song type. Defined as the average amount of time until a mother’s singing achieves a perceivable rhythmic structure (i.e., underlying pulse or meter), rhythm attack time was significantly higher in the singing of mothers with DS infants (.03195 s) than in the singing of mothers with TD infants (.03155 s), regardless of song type (b = 0.000403, SE = 0.000126, p = 0.001, d = 0.47). Additionally, for both infant groups, average rhythm attack time was significantly higher for lullaby (.0320 s) than for playsong (.0315) (b = 0.000534, SE = 0.000147, 155

Infant Behavior and Development 49 (2017) 151–160

S. de l’Etoile et al.

p = 0.0003, d = 0.63). Nine other acoustic parameters demonstrated a main effect of song type, regardless of infant status, within the categories of pitch, tempo, and timbre. For all mothers, average pitch for playsong was 264.94 Hz, and 243.83 Hz for lullaby, revealing a statistically significant difference (b = 21.1, SE = 10.2, p < 0.05, d = 0.61). Average minimum pitch for both infant groups was also significantly higher for playsong (89.68 Hz) than for lullaby (74.63 Hz) (b = 15.1, SE = 4.1, p = 0.0003, d = 0.87). The analysis further indicated that average tempo of mothers’ singing, as expressed in beats per minute (bpm), was significantly lower for playsong (112.24 bpm) than for lullaby (118.57 bpm), irrespective of infant status (b = 6.3, SE 2.9, p = 0.028, d = 0.73). In regard to vocal timbre, values for Formant 1 (F1) were significantly higher during playsong (620.31 Hz) than lullaby (561 Hz) for all mothers’ singing (b = 59.3, SE = 15.9, p = 0.0002, d = 1.09). Significant main effects of song type were also noted for various aspects of spectral centroid; a multi-dimensional acoustic parameter computed as the weighted mean of frequencies in the sound signal (Coutinho & Dibben, 2013). For all mothers, spectral centroid mean was significantly higher during playsong (1762.43 Hz) than lullaby (1554.45 Hz) (b = 208.0, SE = 60.8, p = 0.006, d = 1.08). Spectral centroid standard deviation was also significantly higher for playsong (807.50 Hz) than for lullaby (646.74 Hz) (b = 160.8, SE = 41.01, p < 0.0001, d = 1.13). Likewise, spectral spread standard deviation was significantly greater for playsong (478.10 Hz) than for lullaby (402.1 Hz) (b = 76.0, SE = 18.8, p < 0.0001, d = 1.13). Finally, spectral brightness mean showed significantly higher values for playsong (0.3413 Hz) than lullaby (0.2773 Hz) (b = 0.064, SE = 0.015, p < 0.0001, d = 1.41), as did spectral brightness standard deviation (playsong = 0.1727 Hz, lullaby = 0.1423 Hz) (b = 0.031, SE = 0.011, p = 0.004, d = 1.02). 2.3. Results for degree of perceived warmth Average inter-rater agreement between the two student data coders for degree of perceived warmth in mothers’ singing for the 83 selected sub-clips was 86%, which is an acceptable level of agreement (Graham, Milanowski, & Miller, 2012). The linear model revealed main effects of infant status as well as song type on degree of perceived warmth. Average warmth, calculated as the mean rating of the two coders, was significantly lower in the singing of mothers with DS infants (2.39 on a 5-point Likert scale) than in the singing of mothers with TD infants (3.12) (b = 0.74, SE = 0.27, p = 0.006, d = 0.81). Additionally, in the singing of all mothers, average degree of perceived warmth was significantly lower during lullaby (2.54) than playsong (2.97) (b = 0.43, SE = 0.21, p = 0.04, d = 0.47). No interaction effect was observed between infant status or song type for degree of perceived warmth. 3. Discussion This exploratory study compared the acoustic parameters and degree of perceived warmth in two types of ID songs; the lullaby and the playsong, between mothers of DS infants and mothers of TD infants. Any noted differences could reflect a modified interaction style in mothers of DS infants, potentially impacting infant self-regulation. Ratings of ID songs as either lullaby or playsong revealed a high level of consistency, thereby agreeing with previous findings (Trainor, 1996) and providing further support for distinct song types that serve specific functions. Significant interaction effects between song type and infant status for certain acoustic parameters in the current study suggest that, in some respects, mothers of DS infants sing differently than mothers of TD infants. Such distinctions; however, may reflect mothers’ awareness of the unique attention and arousal needs of infants with DS. As an example, mothers of DS infants displayed a higher maximum pitch, especially during playsong, in comparison with mothers of TD infants. In previous research involving TD infants, Trainor et al. (1997) also observed higher pitch in playsong than lullaby, and suggested that greater pitch variation is related to emotional engagement and thus more effective for garnering attention. Consequently, mothers of DS infants may elevate playsong pitch in an effort to increase emotional engagement and capture infant attention. Mothers of DS infants also showed differences in tonal key clarity compared to mothers of TD infants. Tonal key clarity reflects the salience of the tonal center or perceived key of a musical piece, which helps to determine its emotional meaning or expressivity (Fonari & Eerola, 2010). While music in a major key typically reflects joy or happiness, minor key music may be perceived as dreamy or sad, or even tense and angry (Gabrielsson, 2009). Consequently, tonality of ID singing contributes to the mother’s emotional message and may indicate her sensitivity to infant state. Previous research revealed low tonal key clarity in ID singing for TD infants, perhaps reflecting mothers’ intuitive modifications for maintaining infant attention or modulating infant arousal (de l’Etoile & Leider, 2011). Presently, mothers of DS infants sang with lower tonal key clarity during playsong, meaning they made more frequent key changes than did mothers of TD infants. This noted effect could reflect sensitivity to infant state, and thus represent a positive aspect of ID singing. That is, the lack of tonal center in playsong suggests mothers were modulating key to accommodate their DS infants’ fluctuating attention or arousal levels. Further interaction effects emerged for aspects of timbre known as vocal formants, which are influenced by changes in the articulators, such as the jaw, tongue, lips, velum, and larynx (Forming the vowel sounds, 2017; Sundberg, 1987a). Across languages, mothers commonly modify formants during infant-directed speech in order to provide clear linguistic input (Kuhl et al., 1997). Consequently, the fact that mothers in the present study sang in different languages should have little to no bearing on the findings that emerged for formants. Formant 2 (F2) is largely determined by changes in tongue shape and placement, thus helping to determine vowel clarity and color (Sundberg, 1987b). For mothers of DS infants, F2 values were significantly lower during lullaby, in comparison with mothers of TD infants. This finding suggests that mothers of DS infants produced vowel sounds with less clarity during lullaby, perhaps with the intention of creating a smooth, soothing voice. Yet, because all mothers were encouraged to select their own songs, mothers of DS 156

Infant Behavior and Development 49 (2017) 151–160

S. de l’Etoile et al.

infants may have sung different vowels than mothers of TD infants, or in different combinations or frequencies. Without visual imaging of mothers’ tongue shape and placement during singing or requiring all mothers to sing the same songs with the same vowels, definitive conclusions do not easily emerge from these findings. An interaction effect was also detected for formant 4 (F4), the acoustic parameter considered most relevant to vocal timbre, or distinct sound of the voice (Sundberg, 1987b). F4 is influenced by vocal tract length, defined as the distance from the top of the vocal folds to the front of the lips (Detweiler & Detweiler, 1995; Sundberg, 1987b). Smiling, which requires retraction of the mouth corners, reduces vocal tract length and subsequently increases F4, as well as pitch and amplitude (Ohala, 1980; Sundberg, 1987b; Tartter, 1980). These various outcomes of smiling create a highly-positive vocal affect which effectively garners infant attention (Corbeil et al., 2013a). In the present study, mothers of DS infants sang with low F4 values during lullaby, suggesting that they smiled less than mothers of TD infants. Lacking visual confirmation of mothers’ facial expressions during data collection, however, such conclusions at this point are merely conjecture. One musical feature that demonstrated main effects of both infant status and song type was average rhythm attack time; the amount of time until a perceivable rhythmic structure, such as a consistent pulse, emerges in a mother’s singing. Mothers of DS infants took significantly more time to achieve rhythmic structure than did mothers of TD infants, across both song types. This finding may reflect the slow response time that has been observed in DS infants during caregiver interactions (Carvajal & Iglesias, 2002; Cicchetti, Ganiban, & Barnett, 1991). Subsequently, mothers may have needed more time to read and recognize infant responses to their singing, in order to determine how to best accommodate infant needs. Higher rhythm attack time values for mothers of DS infants could therefore represent sensitivity to infants’ slower information processing rates. In addition, rhythm attack time was higher for lullaby than playsong within both infant groups. This finding agrees with previous research in which playsongs were perceived as more rhythmic than lullabies (Trainor, 1996). Rhythmicity appears to serve as one of the defining features of playsong and should thus be evident more quickly during initial song presentation. By contrast, the underlying beat or meter of a lullaby may develop more slowly, thereby reflecting its tender, soothing nature. Several other acoustic parameters showed a main effect of song type, regardless of infant status, thus indicating similarities between the two infant groups and providing further confirmation of two distinct forms of ID singing. For example, average overall pitch was consistently higher for playsong than lullaby, as was average minimum pitch. These findings align with previous research in which pitch was not only higher during playsong, but also preferred by infants and more effective at reducing infant distress (Ghazban 2013; Trainor et al., 1997; Tsang & Conrad, 2010). Overall, pitch seems to effectively capture and maintain infant attention, as well as communicate high levels of emotional expressivity, specifically during playsong. Surprisingly, tempo was slower in playsong than lullaby across both infant groups. This finding contradicts previous reports, in which tempo was slower in lullabies in order to achieve a soothing effect (Juslin & Laukka, 2003; Trainor et al., 1997; Trehub & Trainor, 1998), and faster during playsong as needed to arouse and stimulate infants (Conrad et al., 2011; Rock, Trainor, & Addison, 1999). Additionally, the tempo results seem to conflict with prior research in which “fast” playsongs were presented at 76 and 72 bpm, and “slow” lullabies sung at 58 and 50 bpm (Conrad, Walsh, Allen, & Tsang, 2011). In the present study, average tempo for playsong was 112.29 bpm and average tempo for lullaby was 119.17 bpm; both values being substantially higher than any tempi provided by Conrad et al. (2011). These discrepancies may be partly explained by differences in tempo calculation. In Conrad et al. (2011) study, a vocalist sang at the prescribed tempi for the duration of each song. In the current study, mothers were likely making frequent tempo changes to accommodate infant state. Thus, rather than reporting a single, dominant tempo for each sub-clip, tempo values were averaged across each sub-clip. Recognizing these different approaches to tempo calculation, making comparisons across studies becomes challenging, at best. Additionally, perhaps faster singing does not automatically equate to more stimulating singing. In the current sample, mothers may have provided an adequate amount of vocal energy and rhythmicity through playsong as needed for infant arousal, while still singing at moderate tempi. Anecdotal listenings of the sub-clips used in data analysis further revealed that mothers tended to sing in a more exaggerated manner during playsong than lullaby, an approach that would be supported by an overall slowing down of tempo. Such discrepancies make it complicated to draw definitive conclusions about tempo differences between playsong versus lullaby within the current sample. A final set of acoustic parameters demonstrated a main effect of song type; all of which pertained to vocal timbre. Formant 1 (F1) is determined by jaw movement, as opening the jaw more widely constricts the vocal tract at the glottal end and expands it at the lip end, thus impacting vowel production and perception, and increasing F1 values (Sundberg, 1987b). In the present study, F1 was higher for playsong than lullaby regardless of infant status, suggesting that mothers tended to sing with a consistently wider jaw opening for playsong in comparison with lullaby. Again, without visual displays of mothers’ facial orientation, such conclusions are difficult to support. Song types were further distinguished by various components of spectral centroid; acoustic features that determine sound brightness and thus represent a fundamental component of musical timbre (Grey & Gordon, 1978; Schubert et al., 2004). In the current study, five out of six measures of spectral centroid were substantially higher during playsong than lullaby, including spectral centroid mean, spectral centroid standard deviation, spectral spread, spectral brightness, and spectral brightness standard deviation. These features play a critical role in determining the emotional content or emotional expressivity of music (Coutinho & Dibben, 2013), thus, mothers in the present study appeared to use a high degree of emotional expressivity when singing playsong. The importance of this finding becomes evident in light of previous research which demonstrated that playsongs are more effective than lullabies for soothing distressed infants (Ghazban, 2013; Trehub, Becker et al., 2015; Trehub, Ghazban et al., 2015). 157

Infant Behavior and Development 49 (2017) 151–160

S. de l’Etoile et al.

Essentially, the lively, rhythmic, and expressive nature of playsongs serves as an effective distraction, thereby reducing undesired infant arousal. The data reported here indicate that mothers of both DS infants and TD infants sing playsong with high emotional content and a high level of emotional expressivity, thereby modulating infant arousal levels, and ultimately supporting self-regulation. In addition to effects noted for acoustic parameters, a main effect of song type emerged for degree of perceived warmth, with playsong evidencing greater warmth than lullaby, from all mothers. This result aligns with spectral centroid findings, in that the playsongs produced in this study seemed to incorporate high emotional content. ID singing that is emotionally expressive is likely to also be perceived as warm. Degree of perceived warmth also demonstrated a main effect of infant status, as it was consistently higher in the singing of mothers with TD infants, compared to mothers of DS infants. This finding suggests that an overly-warm interaction style (i.e., “forceful warmth”) is not evident during the ID singing of mothers with DS infants. Only comparisons with other, nonsinging interactions (i.e., talking to or playing with infants); however, will determine if the lack of forceful warmth can be attributed specifically to singing. Further, the influence of artificiality must be recognized, as data collection occurred in a laboratory setting which may have impacted mothers’ song choice and delivery. Observations of ID singing in natural home environments may reveal different outcomes for degree of perceived warmth. Finally, no previously-published studies have explored the construct of “perceived warmth” in relation to ID singing, thus the term may not yet be clearly-defined. 4. Conclusions The findings gleaned from this exploratory study suggest that mothers of DS infants and TD infants both create songs that can be reliably identified as either lullaby or playsong, with similar frequency. Further, while certain acoustic and musical features of ID singing differ between mothers of DS infants and TD infants, some qualities are shared between the two groups, thus providing additional confirmation of two distinct song types. The differences noted in acoustic parameters suggest that mothers of DS infants sing in a way that accommodates their infants’ unique attentional and arousal needs, thus supporting self-regulation. For example, mothers of DS infants use a higher maximum pitch and more frequent key changes during playsong, in comparison with mothers of TD infants. These acoustic contrasts may reflect mothers’ awareness of infant state and heightened efforts to engage their infants with DS. Mothers of DS infants also take more time to establish rhythmic structure in both song types, thus demonstrating sensitivity to their infants’ potentially slower information processing rates. Beyond acoustic parameters, a noted discrepancy between groups was the higher degree of perceived warmth detected in the singing of mothers with TD infants. These findings do not suggest the presence of a forceful warmth interaction style from mothers toward their DS infants, at least not during ID singing. Ultimately, the observed differences in acoustic features may be entirely appropriate considering the varying needs of the two infant groups, and may reflect mothers’ ability to read and respond to infant cues. In previous research, both DS infants and TD infants responded to ID singing with high levels of sustained gaze toward their mothers, indicating infants were attentive to and interested in the interaction (de l’Etoile, 2015). Thus, in both infant groups, ID singing helps infants to maintain focus on the mother during face-to-face interaction, as needed for self-regulation. Additionally, both DS and TD infants tended to display neutral affect during ID singing; a typical response of young infants during face-to-face interaction (Van Puyvelde et al., 2013) that suggests a calm, curious arousal state, thus further supporting self-regulation (de l’Etoile, 2015). In the present study, mothers of the two infant groups may have effectively promoted self-regulation in different ways by making unique adjustments to the musical and vocal elements of their singing. Regarding similarities in acoustic features between the two infant groups, all mothers appear to sing in ways that clearly define two different song types; lullaby and playsong. In comparison with lullaby, all mothers tend to sing playsong at a higher pitch and slower tempo. Further, all mothers take more time to establish rhythmic structure for lullaby than playsong. Finally, all mothers display timbral qualities (i.e., spectral centroid features) during playsong that are associated with emotional expressivity. Such features effectively soothe and distract distressed infants, thus lowering infant arousal and supporting infant self-regulation. Collectively, these findings suggest that mothers of DS infants are sensitive to infant cues, and can provide singing that appropriately reflects and modifies infant state. Future research with larger sample sizes will provide the statistical power necessary to confirm differences and similarities between infant groups. Additionally, refined definitions of certain parameters (i.e., perceived warmth, song tempo markings) as well as data collection in more natural settings will contribute to reliable, generalizable findings. Further, all infants should be screened for sensory impairments prior to data collection, as individuals with DS often have hearing loss which could impact their response to ID singing (Shott, 2000). Moreover, capturing visual images of mothers’ articulating structures and facial expressions during ID singing will allow for greater confidence in interpreting the effect of infant status on timbral features of ID singing, such as vocal formants. On a final note, the current study was part of a larger project that also explored infant response specifically to the mother’s face and voice; thus, mothers were discouraged from using gestures or physical contact while singing. However, several studies have demonstrated that multi-modal communication (i.e., visual, auditory, and tactile interaction) can impact infant response (Ferber et al., 2008; Gogate et al., 2000; Liebal, Behne, Carpenter, & Tomasello, 2009). Consequently, subsequent studies involving ID singing for infants with DS may incorporate and explore such variables, especially as they may relate to infant self-regulation. Funding This research was supported by a Max Orovitz Award in the Arts and Humanities and a General Research Support Award. Both 158

Infant Behavior and Development 49 (2017) 151–160

S. de l’Etoile et al.

awards were granted by the Research Council of the University of Miami. Acknowledgements The authors wishes to acknowledge the mothers and infants who participated in the study as well as the many agencies and individuals who assisted with recruitment, in particular the Children’s Diagnostic & Treatment Center. References Berger, A. (2011a). Introduction. Self-regulation: Brain, cognition and development. Washington, D.C: American Psychological Association3–17. Berger, A. (2011b). Individual differences in self-regulation. Self-regulation: Brain, cognition and development. Washington, D.C: American Psychological Association61–90. Bergeson, T. R., & Trehub, S. E. (1999). Mothers’ singing to infants and preschool children. Infant Behavior and Development, 22, 51–64. http://dx.doi.org/10.1016/ S0163-6383(99)80005-8. Boersma, P., & Weenink, D. (2016). Praat: Doing phonetics by computer [Computer program]. version 6.0.19. [retrieved 13 June 2016 from http://www.praat.org]. Calkins, S. D., & Hill, A. (2007). Caregiver influences on biological and behavioral aspects of emotion regulation. In J. J. Gross (Ed.). Handbook of emotion regulation (pp. 229–248). New York: The Guilford Press. Carvajal, F., & Iglesias, J. (2002). Face-to-face emotion interaction studies in Down Syndrome infants. International Journal of Behavioral Development, 26, 104–112. http://dx.doi.org/10.1080/01650250042000609. Cebula, K. R., Moore, D. G., & Wishart, J. B. (2010). Social cognition in children with Down’s syndrome: Challenges to research and theory building. Journal of Intellectual Disabilities Research, 54(2), 113–134. http://dx.doi.org/10.1111/j.1365-2788.2009.01215.x. Cicchetti, D., Ganiban, J., & Barnett, D. (1991). Contributions from the study of high-risk populations to understanding the development of emotion regulation. The development of emotion regulation and dysregulation. New York: Cambridge University Press15–48. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Conrad, N. J., Walsh, J., Allen, J. M., & Tsang, C. D. (2011). Examining infants’ preferences for tempo in lullabies and playsongs. Canadian Journal of Experimental Psychology, 65, 168–172. http://dx.doi.org/10.1037/a0023296. Corbeil, M., Trehub, S. E., & Peretz, I. (2013a). Speech vs. singing: Infants choose happier sounds. Frontiers in Psychology, 4. http://dx.doi.org/10.3389/fpsyg.2013. 00372. Corbeil, M., Trehub, S. E., & Peretz, I. (2013b). Singing delays the onset of infant distress. Infancy, 21(3), 373–391. http://dx.doi.org/10.1111/infa.12114. Coutinho, E., & Dibben, N. (2013). Psychoacoustic cues to emotion in speech prosody and music. Cognition and Emotion, 27, 658–684. http://dx.doi.org/10.1080/ 02699931.2012.732559. Delavenne, A., Gratier, M., & Devouche, E. (2013). Expressive timing in infant-directed singing between 3 and 6 months. Infant Behavior and Development, 36, 1–13. http://dx.doi.org/10.1016/j.infbeh.2012.10.004. de l’Etoile, S. K. (2006). Infant behavioral responses to infant-directed singing and other maternal interactions. Infant Behavior and Development, 29, 456–470. http:// dx.doi.org/10.1016/j.infbeh.2006.03.002. de l’Etoile, S. K. (2015). Self-regulation and infant-directed singing in infants with Down syndrome. Journal of Music Therapy, 52, 195–220. http://dx.doi.org/10.1093/ jmt/thv003. de l’Etoile, S. K., & Leider, C. N. (2011). Acoustic parameters of infant-directed singing in mothers with depressive symptoms. Infant Behavior and Development, 34, 248–256. http://dx.doi.org/10.1016/j.infbeh2010.12.013. Detweiler, R., & Detweiler, G. (1995). Investigation of the laryngeal system as the resonance source of the singer’s formant: Data from a pedagogically heterogeneous population. Paper given at the 24th annual symposium care of the professional voice. Dissanayake, E. (2000). Antecedents of the temporal arts in early mother-infant interaction. In N. L. Wallins, B. Merker, & S. Brown (Eds.). The origins of music (pp. 389– 410). Cambridge, MA: MIT Press. Feldman, R. (2007). Parent-infant synchrony and the construction of shared timing; physiological precursors, developmental outcomes, and risk conditions. Journal of Child Psychology and Psychiatry, 48, 329–354. http://dx.doi.org/10.1111/j.1469-7610.2006.01701.x. Ferber, S. G., Feldman, R., & Makhoul, I. R. (2008). The development of maternal touch across the first year of life. Early Human Development, 84, 363–370. http://dx. doi.org/10.1016/j.earlhumdev.2007.09.019. Fonari, J., & Eerola, T. (2010). The pursuit of happiness in music: retrieving valence with contextual music descriptors. Computer music modeling and retrieval. Genesis of meaning in sound and music. Fifth international symposium, CMMR 2008 copenhagen, Denmark, may 19–23, 2008 revised papers: vol. 5493, (pp. 119–133). . http://dx.doi.org/10.1007/978-3-642r-r02518-1. Forming the vowel sounds. (n.d.). Retrieved from http://hyperphysics.phy-astr.gsu.edu/hbase/Music/vowel.html. Frank, E., Hall, M. A., Trigg, L., Holmes, G., & Witten, I. H. (2004). Data mining in bioinformatics using Weka. Bioinformatics, 20(15), 2479–2481. http://dx.doi.org/ 10.1093/bioinformatics/bth261. Gabrielsson, A. (2009). The relationship between musical structure and perceived expression. In S. Hallam, I. Cross, & M. Thaut (Eds.). The Oxford handbook of music psychology (pp. 141–150). New York: Oxford University Press. Ghazban, N. (2013). Emotion regulation in infants using maternal singing and speech. Toronto: Ryerson University [Unpublished doctoral dissertation]. Gogate, L. J., Bahrick, L. E., & Watson, J. D. (2000). A study of multimodal motherese: The role of temporal synchrony between verbal labels and gestures. Child Development, 71, 878–894. http://dx.doi.org/10.1111/1467-8624.00197. Graham, M., Milanowski, A., & Miller, J. (2012). Center for educator compensation reform; U.S. department of education. Measuring and promoting inter-rater agreement of teacher and principal performance ratings (34705.0212.83670507). Rockville, MD: Westat. Graven, S. N. (2000). The full-term and premature newborn: sound and the developing infant in the NICU: Conclusions and recommendations for care. Journal of Perinatology, 20, S88–S93. Grey, J. M., & Gordon, J. W. (1978). Perceptual effects of spectral modifications on musical timbres. Journal of the Acoustical Society of America, 63(5), 1493–1500. http://dx.doi.org/10.1121/1.381843. Juslin, P., & Laukka, P. (2003). Communication of emotions in vocal expression and music performance: Different channels, same code. Psychological Bulletin, 129, 770–814. http://dx.doi.org/10.1037/0033-2909.129.5.770. Kuhl, P. K., Andruski, J. E., Chistovich, I. A., Chistovich, L. A., Kozhevnikova, E. V., Ryskina, V. L., et al. (1997). Cross-language analysis of phonetic units in language addressed to infants. Science, 277, 684–696. http://dx.doi.org/10.1126/science.277.5326.684. Lartillot, O., & Toiviainen, P. (2007). A Matlab toolbox for musical feature extraction from audio. International conference on digital audio effects, 237–244. Liebal, K., Behne, T., Carpenter, M., & Tomasello, M. (2009). Infants use shared experience to interpret pointing gestures. Developmental Science, 12, 264–271. http:// dx.doi.org/10.1111/j.1467-7687.2008.00758.x. Longhi, E. (2009). ‘Songese’: Maternal structuring of musical interaction with infants. Psychology of Music, 37, 195–213. http://dx.doi.org/10.1177/ 0305735608097042. Margulis, E. H. (2013). Repetition and emotive communication in music versus speech. Frontiers in Psychology, 4. http://dx.doi.org/10.3389/fpsyg.2013.00167. McCabe, L. A., Cunnington, M., & Brooks-Gunn, J. (2004). The development of self-regulation in young children. In R. F. Baumeister, & K. D. Vohs (Eds.). Handbook of self-regulation: Research, theory, and applications (pp. 340–356). New York: The Guilford Press. Milligan, K., Atkinson, L., Trehub, S. E., Benoit, D., & Poulton, L. (2003). Maternal attachment and the communication of emotion through song. Infant Behavior and

159

Infant Behavior and Development 49 (2017) 151–160

S. de l’Etoile et al.

Development, 26, 1–13. http://dx.doi.org/10.1016/s0163-6383(02)00165-0. Moore, D., Oates, J. M., Hobson, R. P., & Goodwin, J. (2002). Cognitive and social factors in the development of infants with Down Syndrome. Down Syndrome Research and Practice, 8, 43–52. http://dx.doi.org/10.3104/reviews.129. Nakata, T., & Mitani, C. (2005). Influences of temporal fluctuation on infant attention. Music Perception, 22, 401–409. http://dx.doi.org/10.1525/mp.2005.22.3.401. Ohala, J. J. (1980). The acoustic origin of the smile (Abstract). Journal of the Acoustical Society of America, 1(Suppl. 68), S33. R. Core Team (2016). R: A language and environment for statistical computing. R Foundation for Statistical Computing. [Retrieved from https://www.R-project.org/]. Rock, A. M., Trainor, L. J., & Addison, T. (1999). Distinctive messages in infant-directed lullabies and play songs. Developmental Psychology, 35, 527–534. http://dx.doi. org/10.1037/0012-1649.35.2.527. Schubert, E., Wolfe, J., & Tarnopolsky, A. (2004). Spectral centroid and timbre in complex, multiple instrumental textures. In S. D. Lipscomb, R. Ashley, R. O. Gjerdingen, & P. Webster (Eds.). Proceedings of the 8th international conference on music perception & cognition, northwestern university, Evanston, IL (pp. 654–657). Adelaide, Australia: Causal Productions. Shenfield, T., Trehub, S. E., & Nakata, T. (2003). Maternal singing modulates infant arousal. Psychology of Music, 31, 365–375. http://dx.doi.org/10.1177/ 03057356030314002. Shott, S. R. (2000). Down syndrome: Common paediatric ear, nose and throat problems. Down Syndrome Quarterly, 5(2), 1–6. Spiker, D. (2006). Off to a good start: Early intervention for infants and young children with Down syndrome and their families. Down syndrome: Neurobehavioral specificity. Somerset, NJ: John Wiley & Sons, Ltd175–190. Sroufe, L. A. (2000). Early relationships and the development of children. Infant Mental Health Journal, 21(1–2), 67–74. http://dx.doi.org/10.1002/(SICI)10970355(200001/04)21:1/2<67:AID-IMHJ8 >3.0.CO ;2-2. Sundberg, J. (1987a). The voice organ. The science of the singing voice. DeKalb,IL: Northern Illinois University Press6–24. Sundberg, J. (1987b). Articulation. The science of the singing voice. DeKalb,IL: Northern Illinois University Press93–133. Tartter, V. C. (1980). Happy talk: Perceptual and acoustic effects of smiling on speech. Perception and Psychophysics, 27, 24–27. http://dx.doi.org/10.3758/ BF03199901. Thompson, R. A. (1994). Emotion regulation: Atheme in search of definition. Monographs for the Society of Research in Child Development, 59, 259–283 [2–3, Serial No. 240]. Trainor, L. J., Clark, E. D., Huntley, A., & Adams, B. A. (1997). The acoustic basis of preferences for infant-directed singing. Infant Behavior and Development, 20, 383–396. http://dx.doi.org/10.1016/S0163-6383(97)90009-6. Trainor, L. J. (1996). Infant preferences for infant-directed versus noninfant-directed playsongs and lullabies. Infant Behavior and Development, 19, 83–92. http://dx. doi.org/10.1016/S0163-6383(96)90046-6. Trehub, S. E., & Gudmundsdottir, H. R. (2015). Mothers as singing mentors for infants. In G. F. Welch, & J. Nix (Eds.). Oxford handbook of singinghttp://dx.doi.org/10. 1093/oxfordhb/9780199660773.013.25. Trehub, S. E., & Trainor, L. J. (1998). Singing to infants: Lullabies and play songs. Advances in Infancy Research, 12, 43–77. Trehub, S. E., Unyk, A. M., & Trainor, L. J. (1993). Adults identify infant-directed music across cultures. Infant Behavior and Development, 16, 285–295. Trehub, S. E. (2000). Human processing predispositions and musical universals. In N. L. Walling, B. Merker, & S. Brown (Eds.). The origins of music (pp. 427–448). Cambridge, MA: The MIT Press. Trehub, S. E., Becker, J., & Morley, I. (2015). Cross-cultural perspectives on music and musicality. Philosophical Transactions of the Royal Society B, 370, 20140096. http://dx.doi.org/10.1098/rstb.2014.0096. Trehub, S. E., Ghazban, N., & Corbeil, M. (2015). Musical affect regulation in infancy. Annals of the New York Academy of Sciences, 1337, 186–192. http://dx.doi.org/ 10.1111/nyas.12622. Tsang, C. D., & Conrad, N. J. (2010). Does the message matter? The effect of song type on infants’ pitch preferences for lullabies and playsongs. Infant Behavior and Development, 33, 96–100. http://dx.doi.org/10.1016/j.infbeh.2009.11.006. Van Puyvelde, M., Loots, G., Vinck, B., de Coster, L., Matthijs, L., & Mouvet, K. (2013). The interplay between tonal synchrony and social engagement in mother-infant interaction. Infancy, 18(3), 1–24. http://dx.doi.org/10.1111/infa.12007. Vohs, K. D., & Baumeister, R. F. (2004). Understanding self-regulation: An introduction. In R. F. Baumeister, & K. D. Vohs (Eds.). Handbook of self-regulation: Research, theory, and applications (pp. 1–9). New York: The Guilford Press. Wood, S. (2005). What are formants? Beginners guide to praat [January 15 Retrieved from http://person2.sol.lu.se/SidneyWood-praate/whatform.html].

160