Vowel-hyperarticulation in Infant-directed Speech: A Multi-method Meta-analysis

Abstract

There is a long-standing debate about the extent to which vowel hyperarticulation, the production of acoustically exaggerated vowels, occurs in Infant-Directed Speech (IDS). This exaggeration has been argued to result in clearer speech sounds that are easier for infants to process and might be positively related to infants’ linguistic outcomes. However, previous findings regarding the presence of vowel hyperarticulation in IDS are seemingly inconsistent and contradictory, making it hard to advance our understanding of the role of hyperarticulation in language development and consequently the potential functions/roles of IDS in language acquisition. Thus, we adopted a systematic review and meta-analytic approach to investigate the robustness of vowel hyperarticulation in IDS and identify sources of heterogeneity in the literature. We employed four complementary meta-analytic approaches and evaluated the robustness of results across the different choices. We performed both traditional (Frequentist) and Bayesian meta-analyses on, first, methodologically consistent studies (17 studies, 36 effect sizes), and then, all studies of vowel hyperarticulation in IDS irrespective of the method (33 studies, 75 effect sizes). Findings indicate the presence of vowel hyperarticulation in IDS compared to ADS (effect sizes ranging from 0.35 to 0.62), as well as systematic and unsystematic variability due to e.g., cross-linguistic variability and methods employed, making it difficult to identify specific factors associated with stronger vowel hyperarticulation. The quantitative results combined with systematic review of the literature also enables important methodological insights, which we summarize into recommended practices such as enlarging sample sizes and explicitly incorporating sources of heterogeneity in analyses.

A longitudinal investigation of the acoustic properties of infant-directed speech from 6 to 18 months

“…Finally, with respect to the segmental properties of vowels, our results echo previous reports of between-register differences in the literature, albeit with a disclaimer regarding vowel space expansion (Cox et al, 2022;Lovcevic et al, 2023). Namely, only fathers in our sample had larger vowel spaces in IDS, while mothers' vowel spaces, overall, did not differ between IDS and ADS.…”

“…In line with previous above-reviewed research and meta-analytic evidence (where available), and per our detailed preregistration (https://osf.io/8k3eh), we expected that parents' IDS, compared to ADS, would be characterised by overall higher pitch, wider pitch range, slower articulation rate, longer vowel duration, increased vowel spaces, and more variable and less distinct vowel categories (Cox et al, 2022;Lovcevic et al, 2023;McMurray et al, 2013;Miyazawa et al, 2017;Rosslund et al, 2022Rosslund et al, , 2023. Crucially, we expected that the predicted IDS-ADS differences in pitch, articulation rate, vowel duration, vowel category variability, and vowel distinctiveness would decrease with infants' age, attributed to IDS becoming more similar to ADS (Bernstein Ratner, 1984;Cox et al, 2022;Rosslund et al, 2022Rosslund et al, , 2023.…”

“…Exaggeration of the most peripheral vowels (such as /i/-/a/-/u/) expands the underlying vowel space and increases the distance between vowels (Kuhl et al, 1997;Liu et al, 2003). Age-related changes in relative vowel space area in IDS were not found consistently, with studies showing increases (British English, Japanese, French: Dodane & Al-Tamimi, 2007), decreases (Australian English: Xu Rattanasone et al, 2013), and stability (Dutch: Benders, 2013; Australian English: Burnham et al, 2015;Kalashnikova & Burnham, 2018;German: Weirich & Simpson, 2019) across infants' ages -the latter is also the conclusion of two separate meta-analyses (Cox et al, 2022;Lovcevic et al, 2023).…”

Longitudinal changes in consonant production in infant-directed speech and infants’ early speech production from 6 to 12 months

“…Parents might be focusing on the vowel, as a more salient acoustic unit that is easy to act on by increasing its length and modulating its pitch, which can go handin-hand with increased VOT duration (Fish et al, 2017;Port & Rotunno, 1979). Vowels in IDS are typically articulated with more exaggerated formants (Cox et al, 2022;Lovcevic et al, 2023). Future research needs to examine whether changes in parents' consonant production are driven by the exaggeration of the following vowel.…”

“…In the case of vowels, perceptual distinctions between different categories (e.g., /i, a, u/) are primarily based on formant frequencies (where the first formant corresponds to mouth openness and the second formant corresponds, mostly, to tongue back-/frontness). Fittingly, parents across several cultures and languages tend to exaggerate vowel articulation in IDS as compared to ADS, increasing the formant (hence perceptual) distance between vowel categories (Kuhl et al, 1997;Lovcevic et al, 2023; but see also Miyazawa et al, 2017).…”