Page 29 - Winter 2020

The finding that the vowel space is larger, with the corner vowels spaced further apart in IDS compared with ADS, is also illustrated schematically in Figure 2.
Increasing the acoustic distance between vowels enhances recognition of distinct vowel sounds. Moreover, even in ADS, a larger vowel space is typically associated with more intelligible speech (Bradlow et al., 1996). An expanded vowel space was observed for IDS vowels pro- duced in several languages, suggesting this is a universal feature of IDS (Kuhl et al., 1997). Moreover, infants with a mother who expanded her vowel space when producing IDS also performed better in a speech sound discrimi- nation task (Liu et al., 2009). Other work suggests that early exposure to vowel expansion in IDS is associated with better expressive and receptive language skills at two years (Hartman et al., 2016).
The idea that caregivers expand their vowel space to make speech clearer is consistent with the finding that adults do not expand their vowel space when speaking to a pet with little or no capacity for acquiring language skills even though pet-directed speech typically contains the characteristic pitch properties of IDS that convey affect as outlined above (Burnham et al., 2002). IDS appears to be a form of hyperarticulated speech that promotes lan- guage development by clarifying and enhancing speech segments (e.g., vowels and consonants).
We now see this view as incomplete. As work advanced, an expanded vowel space in IDS has not been found in all languages, in all interactions, or at all infant ages (see Hartman et al., 2016). Vowel expansion has also been absent when studies relied on samples of natural spontaneous speech instead of the structured laboratory-recorded samples used in earlier studies (Martin et al., 2015). In a study of IDS in Japanese, Miyazawa and colleagues (2017) observed vowel space expansion when average formant values were consid- ered. However, the IDS vowel sounds were actually not more distinct because there was much more acoustic variability and overlap among different vowels pro- duced in IDS, which makes recognizing distinct vowel sounds more, not less, difficult.
Actually, some researchers suggested that vowel space expansion in IDS is an unintended side effect of the increased pitch (which also raises the larynx and
shortens the vocal tract) and slower speaking rate that characterizes this speech style rather than being shaped by the parents’ direct effort to clarify speech patterns (McMurray et al., 2013). For example, in her study of IDS in Dutch, Benders (2013) observed a reduced rather than an expanded vowel space, and she also observed an overall rise in all formant frequencies for IDS vowels relative to ADS vowels. This pattern is noteworthy given that smiling, which shortens the vocal tract length, is known to shift formants upward and further apart in frequency (Shor, 1978). Thus, Benders claimed that care- givers modify their vowel sounds in IDS, especially with young infants, to communicate positive emotion rather than to provide clearer speech. Meanwhile, Miyazawa and colleagues (2017) observed that IDS had a breathier voice quality compared with ADS vowels, an effect also associated with communicating emotion.
Yet another perspective emerged from a study by Kalash- nikova and colleagues (2017) that directly examined the tongue and lip movements that caregivers make when they produce IDS and ADS, which are the source of these formant patterns. In this study, special sensors were strategically placed on eight moms to track their lip and tongue movements while they produced IDS with their 11-month-old infants and ADS with the experimenter. Surprisingly, when the moms produced the corner vowels in IDS, their tongue and lips movements were not exaggerated or hyperarticulated as the researchers expected, even though simultaneous acoustic recordings showed expansion of the IDS vowel space.
Through an analysis that combined lip movement and vowel formant measures, Kalashnikova et al. (2017) inferred that the IDS speech was produced with a shorter vocal tract and higher pitch, which mothers can create by raising their larynx. They concluded that expansion of the F1/F2 acoustic vowel space in IDS is not created by the mother’s intentional efforts to produce clearer vowel sounds using exaggerated articulatory movements. Instead, mothers are speaking with the unintentional purpose of sounding smaller and thus unthreatening and nonaggressive.
This speaking style can also be viewed as the mother trying to imitate her infant. Moreover, this form of vocal social convergence is observed in other species and is regarded as amechanismforcreatingacloseemotionalbondbetween
Spring 2021 • Acoustics Today 29