Page 39 - 2017Spring

up to the age of 5 years. It is beyond the scope of this ar- ticle to provide a summary of this vast research literature so only a few key findings are mentioned here. Recent work has particularly highlighted the influence of the language environment experienced by an infant on his/her develop- ment of speech. To investigate the impact of language input on word learning, a researcher from the MIT Media Lab fa- mously recorded about 70% of his son’s speech exposure and interaction with his caregivers between birth and the age of 3 years via an array of microphones and video cameras in- stalled in his house, resulting in over 250,000 hours of audio and video (Roy, 2009; Vosoughi and Roy, 2012). Even a very partial analysis of this unique corpus allowed Roy’s research team to examine “word births,” that is, early utterances of lexical items related to the exposure that his son had to these words and the physical context in which they were heard. This work also showed how attuned caregivers were to the stage of development of his infant, adjusting the length of their utterances and the diversity of their lexical items ac- cordingly.
Even though this unique study is unlikely to be replicated, the move toward “big data” has been facilitated by the de- velopment of devices such as LENA Pro (e.g., Oller et al., 2010). This device can record up to a full day of an infant’s sound environment and provides an automatic classification of the speech and environmental sounds that the infant is exposed to as well as of the infant’s own vocalizations. Stud- ies based on such data can show links between the amount of conversational turns and later linguistic outcomes (e.g., Ambrose et al., 2014). The importance of social interaction and joint attention for speech development has also been shown in more controlled laboratory studies. For example, in a study of interactions between 8 month olds and their mothers, infant vocalizations were significantly higher and more mature when the mother’s social response (smiling, moving closer to the infant) was synchronized to the infant’s vocalization than when a similar amount of feedback was present but manipulated to be desynchronized from the child’s vocalization (Goldstein et al., 2003).
Later Stages of Speech Development
in Childhood
Even though children are efficient communicators by the age of 5 years, their development of speech production is still far from complete. The differences that can still be measured relative to an adult norm may be subtle and only identifiable using analytic tests, but they are still likely to have an impact
on everyday communication. The fact that there are ongoing changes in speech production throughout childhood is per- haps not surprising when one considers the very significant physical and cognitive changes that occur in this period and especially around puberty. The dimensions of the vocal tract increase with body size, as shown by a study involving MRI scans of individuals aged 2 to 25 years (Fitch and Giedd, 1999), with differentiation according to talker sex appear- ing around puberty and marked by a particularly marked growth in pharyngeal length in males. There are also physi- cal changes to the larynx that occur for individuals of both sexes but are especially marked in males during puberty, re- sulting in a marked increase in the size and thickness of the vocal folds (e.g., Hollien et al., 1994).
There are many acoustic consequences of these physical de- velopments. In terms of the voice source, physical changes in the larynx lead to a decrease in the frequency of vocal fold vibration, which determines perceived pitch. This decrease occurs in both sexes but is particularly marked in males, thus leading to a greater differentiation according to talker sex at puberty (e.g., Hollien et al., 1994). Due to changes in vocal tract size, the acoustic patterns characterizing vowels (e.g., Lee et al., 1999; Perry et al., 2001) and consonants (e.g., Mc- Gowan and Nittrouer, 1988; Romeo et al., 2013) that result from resonances in the vocal tract also decrease throughout childhood as the vocal tract size increases.
Some changes that occur in children’s speech are not due to physical changes but to the increasing use of gender or social identity markers. An article on the acoustics of regional ac- cents, one such marker, recently appeared in Acoustics Today (Jacewicz and Fox, 2016). As an example of a gender marker, the sound “s” as in “Sue” is produced with a higher frequen- cy of frication for girls than for boys to a degree beyond what would be expected from differences in the physical charac- teristics alone (Flipsen et al., 1999). Sociophonetic studies have shown that variants signaling gender or social identity are already documented by the age of 3 years, and there is also evidence that mothers may use different phonetic vari- ants when speaking to girls and boys (Foulkes et al., 2005). In a recent study (Munson et al., 2015), boys in the 5- to 13-year age range diagnosed as experiencing gender iden- tity disorder, that is, who were distressed or uncomfortable with their biological status as male, produced vowels and consonant sounds such as “s” that differed acoustically and were perceived as less male-like than for boys with typical gender identity. This further suggests that phonetic markers of gender identity can be established early as well as appear-
Spring 2017 | Acoustics Today | 37