Quantifying the role of rhythm in infants' language discrimination abilities: A meta-analysis

More than 30 years have passed since Mehler et al. (1988) proposed that newborns can discriminate between languages belong to different rhythm classes: stress-, syllable- or mora-timed. Thereupon they developed the hypothesis infants are sensitive differences in vowel and consonant interval durations as acoustic correlates of classes. It remains unknown exactly which durational computations use when perceiving speech for purposes distinguishing languages. Here, a meta-analysis studies on infants' language discrimination skills over first year life was conducted, aiming quantify how change with age modulated by classes metrics. A systematic literature search identified 42 tested (birth 12 months) preference two varieties, presenting auditory audio-visual continuous speech. Quantitative data synthesis conducted using multivariate random effects meta-analytic models factors class difference, age, stimulus manipulation, method, metrics operationalising proportions variability durations, explore best account preference. Results revealed smaller (?V) larger successive consonantal (rPVI-C) were associated more successful discrimination, better accounted results factor class. There no but affected age: older get, prefer non-native rhythmically similar their native language, not distinct. These findings inform theories previously focussed class, providing novel way operationalise extent it accounts abilities. • Rhythm is limited accounting discrimination. Infants' explained Discrimination improves differ variability. less Preferences depending rhythmic similarity language.