Page 27 - Winter 2020

which infants are presented a choice to listen to samples of IDS and adult-directed speech (ADS), infants (even new- borns) repeatedly show a clear and strong preference to listen to IDS, with few studies deviating from this pattern. A meta-analysis found that the average listening time dif- ference between IDS and ADS (or “the effect size of IDS preference” in statistical terms) was significant and large (Dunst et al., 2012).
Infant preference for IDS, being recognized as one of the most robust behaviors measured in infancy, was selected as the target behavior in a large-scale study designed to under- stand how subject variables and testing methodologies affect the measurement of infant behavior. This study, conducted by the ManyBabies Consortium, involved 67 laboratories across North America, Europe, Australia, and Asia. The findings provided further conclusive evidence of infants’ preference for IDS over ADS (ManyBabies Consortium, 2020). There is no doubt that infants are attracted to IDS.
Acoustic Properties of Infant-
Directed Speech
What is it about IDS that babies like? Studies show that when caregivers talk to their infant, they modify their speech on multiple levels. This includes basic speech pat- terns that play a broad role in communication and can be observed across different languages (conveying emo- tion and talker information and basic units such as vowels, consonants, and word forms) as well as acoustic cues that mark specific lexical, grammatical, and pragmatic features that are important in a specific language. Our focus here is on basic acoustic speech patterns that have a broad impact and are more likely to be universal across languages.
To understand the acoustic properties of IDS, it is useful to know that the acoustic speech signal has two independent components, referred to as the source and the filter. The vocal source component is determined by how fast the vocal folds vibrate, which determines the voice pitch or fundamental frequency (see article on singing by Sund- berg et al. on pages 43-51). The voice pitch of an infant or child is much higher than that of an adult because their short, light vocal folds vibrate faster compared with the longer and thicker vocal folds of an adult. Talkers also vary their voice pitch by adjusting the tension of the vocal folds.
The vocal filter component refers to the effects of the length and shape of the vocal tract, the term used to refer
to the tube formed by the vocal folds on one end and the mouth at the other end. Movements of the tongue, jaw, and lips vary the length and shape of the vocal tract, which determines the resonances of the vocal tract.
The acoustic patterns formed by the vocal resonances cre- ated when we speak are referred to as formants and are numbered in ascending frequency value (the lowest is the 1st formant [F1], next is the 2nd formant [F2], etc.). The formants are essentially narrow frequency regions where acoustic energy is increased because these frequen- cies vibrate most easily within the associated vocal tract space. The first three formants contain critical acoustic information for speech communication.
The vocal resonances and associated formant frequencies are higher for the short vocal tract of an infant or child compared with the longer vocal tract of an adult. Talkers modify the resonance of the vocal tract to create different vowel sounds by moving their articulators to create dif- ferent vocal tract shapes, such as by adjusting the degree and location of constrictions along the vocal tract.
An extensive body of research has concentrated on describ- ing the acoustic structure of IDS. This work has considered each component, typically by comparing samples of IDS with comparable samples of ADS (Soderstrom, 2007).
Voice Pitch and Rhythmic Properties of Infant-Directed Speech
The distinct vocal source properties of IDS are well-estab- lished (see Multimedia 1-5 at acousticstoday.org/polkamedia for audio examples in English and in Turkish; see video example at bit.ly/3m3ecHh). Overall, higher voice pitch, wider voice pitch range and greater pitch variability have been found in IDS (compared with ADS) in a variety of languages, including both nontonal languages (Fernald et al.,
1989) and tonal languages (Liu et al., 2009). Several studies have shown that high voice pitch is the primary acoustic determinant of the infants’ preference for IDS (Fernald and Kuhl, 1987; Leibold and Werner, 2007). Research focused on the speech movements that occur during IDS have observed, as expected, that adults produce faster vocal fold vibrations and also raise their larynx when they talk to young infants (Ogle and Maidment, 1993; Kalashnikova et al., 2017). Larynx raising naturally occurs when vocal fold tension increases (which raises voice pitch) and can also shorten the overall vocal tract length.
Spring 2021 • Acoustics Today 27