Acoustic regularities in infant-directed speech and song across cultures

Abstract

Humans often produce vocalizations for infants that differ from vocalizations for adults. Is this property common across societies? The forms of infant-directed vocalizations may be shaped by their function in parent-infant communication. If so, infant-directed song and speech should be differentiable from adult-directed song and speech on the basis of their acoustic features, and this property should be relatively invariant across cultures. To test this hypothesis, we built a corpus of 1,614 recordings of infant-and adult-directed singing and speech produced by 411 people living in 21 urban, rural, and small-scale societies. We studied the corpus in a massive online experiment and in a series of acoustic analyses. Naïve listeners (N = 13,218) reliably identified infant-directed vocalizations as infant-directed, and adult-directed speech (but not songs) as adult-directed, at rates far higher than chance. Ratings of infant-directed song were the most accurate and the most consistent across all societies; infant-directed speech was accurately identified on average, but inconsistently across societies. To determine the mechanisms underlying these results, we extracted many acoustic features from each recording and identified those that most reliably characterize infant-directed song and speech across cultures, via preregistered exploratory-confirmatory analyses and machine classification. The features distinguishing infant-and adult-directed song and speech concerned pitch, rhythmic, phonetic, and timbral attributes; a hypothesis-free classifier with cross-validation across societies reliably identified all vocalization types, with highest accuracy for infant-directed song. Last, we isolated 12 acoustic features that were predictive of perceived infant-directedness; of these, two pitch attributes (median F0 and its variability) were by far the most explanatory. These findings demonstrate cross-cultural regularities in infant-directed vocalizations that are suggestive of universality; moreover, infant-directed song appears to be more cross-culturally stereotyped than infant-directed speech, informing hypotheses of the functions and evolution of both.

Children infer the behavioral contexts of unfamiliar foreign songs.

“…This research builds on previous studies from our group (Hilton, Moser, et al, 2022;Mehr et al, 2019, demonstrating (a) cross-cultural regularities in the associations between acoustical forms of music and particular behavioral contexts in which music appears; and (b) the sensitivity of adults to these associations. Here, we explored how musical experience in childhood shapes that sensitivity, using the same stimuli as previous work.…”

Globally, songs and instrumental melodies are slower and higher and use more stable pitches than speech: A Registered Report

“…We selected these features by reviewing what past studies focused on for the analysis of song-speech comparison and prominently observed features in music (e.g. Fitch, 2006;Hansen et al, 2020;Hilton et al, 2022;Savage et al, 2015;Sharma et al, 2021, see the Supplementary Discussion section S1.1 for a more comprehensive literature review). Here, f 0 , rate of change of f 0 , and spectral centroid are extracted purely from acoustic signals, while IOI rate is based purely on manual annotations.…”

“…In music, these include regular rhythms, discrete pitches, small melodic intervals, and a predominance of songs with words (rather than instrumental music or wordless songs) (Mehr et al, 2019;Savage et al, 2015). However, non-signed languages also use the voice to produce words, and other proposed musical universals may also be shared with language (e.g., discrete pitch in tone languages; regular rhythms in "syllable-timed" / "stress-timed" languages; use of higher pitch when vocalizing to infants) (Haiduk & Fitch, 2022;Hilton et al, 2022;Patel, 2008;Tierney et al, 2011). Moreover, vocal parameters of speech and singing, such as fundamental frequency and vocal tract length as estimated from formant frequencies, are strongly intercorrelated in both men and women (Valentova et al, 2019).…”

“…Blasi et al, 2022). The available dataset closest to our study is Hilton, Moser, et al's (2022) recordings sampled from 21 societies. Their dataset covers 11 language families and each participant produced a set of adult-directed and infant-directed song and speech.…”

“…S1). For example, Mertens' (2022) automated segmentation algorithm used by Hilton et al (2022) mis-segmented two out of the first three words "by a lonely" from the English song used in our pilot analyses ("The Fields of Athenry"), over-segmenting "by" into "b-y", and under-segmenting "lonely" by failing to divide it into "lone-ly" (cf. Fig.…”