Page 28 - Winter 2020

BABY TALK
It is widely held that the primary goal or intention guiding these characteristic voice pitch properties is conveying emotion to the young infant (Saint-Georges et al., 2013). Understanding the emotional expression in IDS led researchers to explore the pitch contours found in IDS. Fernald and Simon (1984) observed that most utterances in IDS had either rising or falling pitch con- tours. Stern and colleagues (1982) identified the social and linguistic context where these pitch contours were used. For example, a rising contour was frequently used when mothers tried to engage in eye contact with an inat- tentive baby. Studies also show that creating “happy talk” is the fundamental goal of IDS and that positive affect is what drives infant preference (Singh et al., 2002). Thus, understanding pitch contours in IDS can help us decode the affective function of IDS.
In terms of rhythmic features, IDS universally contains shorter utterances and longer pauses between words; in some languages, including English and Japanese, there is also an enhanced lengthening of words or syllables at the end of a phrase or utterance (Fernald et al., 1989; Martin et al., 2016). This is helpful because natural fluent speech typically lacks pauses between words, something you notice when encountering an entirely foreign lan- guage. This also highlights an initial challenge for babies, learning which speech patterns are reoccurring words, aka word segmentation. Infants begin to acquire word segmentation skills at around 6 months, through expe- rience listening to a specific language and before they attach meaning to each word they hear (Jusczyk, 1999).
Overall, the tempo of IDS provides the infant with a speech stream that is easier to track with clearer cues marking word boundaries and other syntactic units. Consistent with this, the most prominent rhythm in the acoustic speech signal, which matches the timing of stressed syllables, was observed to be stronger in IDS compared with ADS (Leong et al., 2017). This speech rhythm was also prominent (and synchronized) in mother and infant brain patterns when they watched a nursery rhyme video together (Santamaria et al., 2020).
The enhanced temporal properties of IDS likely explains the positive effects of IDS on infant speech processing. For example, IDS facilitates infant word segmenta- tion performance (Thiessen et al., 2005), an important language-specific speech-processing skill that emerges
in infancy and supports early word learning. The list of the positive effects of IDS rhythm on speech processing, which includes supporting better discrimination and tracking of syllable patterns and detection of speech in noise, continues to grow (Soderstrom, 2007).
Vocal Resonance Properties of Infant- Directed Speech
Research on IDS has also considered the other funda- mental component of speech, the filter or resonance properties. The focus here has been on vowel sounds. Early research by Kuhl and colleagues (1997) reported that vowels are produced in an exaggerated form in IDS; this hyperarticulation of vowels expands the vowel space, a standard graphic display that captures how vowel artic- ulation and formant patterns are related.
In the classic vowel space (Figure 2), F1 increases as the tongue/jaw height decreases and F2 increases as the tongue constriction moves to the front of the mouth. Importantly, three vowel sounds found in every spoken language, “ee,” “aw,” and “oo,” form the corners of this F1/F2 vowel space. These corner vowels are associated with gestural extremes that define the full range of movements that we use to create vowel sounds: "ee" has the most high and front constriction of the vocal tract,
"oo" has the most high and back constriction of the vocal tract, and "aw" has the most open and unconstricted posture of the vocal tract. All other vowel sounds fall within the limits defined by these corner vowel sounds.
Figure 2. The articulatory/acoustic vowel space corresponding to vowels produced by an adult female in ADS (squares) and in IDS (circles), and by an infant (triangles). F1 and F2, 1st and 2nd formants, respectively.
28 Acoustics Today • Spring 2021