Development of infants' segmentation of words from native speech: a meta‐analytic approach

Abstract

Infants start learning words, the building blocks of language, at least by 6 months. To do so, they must be able to extract the phonological form of words from running speech. A rich literature has investigated this process, termed word segmentation. We addressed the fundamental question of how infants of different ages segment words from their native language using a meta-analytic approach. Based on previous popular theoretical and experimental work, we expected infants to display familiarity preferences early on, with a switch to novelty preferences as infants become more proficient at processing and segmenting native speech. We also considered the possibility that this switch may occur at different points in time as a function of infants' native language and took into account the impact of various task- and stimulus-related factors that might affect difficulty. The combined results from 168 experiments reporting on data gathered from 3774 infants revealed a persistent familiarity preference across all ages. There was no significant effect of additional factors, including native language and experiment design. Further analyses revealed no sign of selective data collection or reporting. We conclude that models of infant information processing that are frequently cited in this domain may not, in fact, apply in the case of segmenting words from native speech.

Infants’ sensitivity to rhyme in songs

“…As rhymes are particularly frequent in songs, the preference for the non-rhyming pattern may be specific to the sung register, and, for example, not generalize to spoken single words (Jusczyk et al, 1999, experiment 1). This stimulus specific interpretation of the preference direction has been suggested earlier (Bergmann & Cristia, 2016). Yet, given the fact that half of the sample displayed a preference for rhyming songs, infant age and cognitive/linguistic maturation may play a role as well (Hunter & Ames, 1988).…”

What’s in a looking time preference?

“…For example, in important work on how infants segment words from fluent speech, Jusczyk and Aslin (1995) familiarized infants to two words spoken in isolation, and then assessed if they preferred to listen to sentences that contained these now-familiar words, or control sentences that, instead, contained two different words. In a design like this, it is not clear which of the two types of test sentences should have been more interesting for the participants; in this study, infants preferred the sentences containing familiar words, a finding that has frequently been replicated (see Bergmann & Cristia, 2016). Conversely, other studies have found consistent preferences for more novel stimuli.…”

“…The evidence is less consistent for studies that use more complex stimuli, or that use habituation to assess cognitive processes other than visual attention and memory. A recent meta-analysis of infant word segmentation (Bergmann & Cristia, 2016) did not find evidence that novelty versus familiarity preferences could be predicted by factors such as cognitive load or age. In particular, their analysis of 168 experiments from 51 articles and with data gathered from 3774 infants, did not find any evidence that a switch can be predicted based on age, native language, or various task-and stimulus-related factors that might affect difficulty.…”

The profile of abstract rule learning in infancy: Meta‐analytic and experimental evidence

“…In future work, it will be important to generalize this method, to see if the conclusion that only the more common behavioural response provides evidence also holds true for other phenomena. For example, most studies of infant word segmentation show a familiarity preference (Bergmann & Cristia, 2016); our finding would suggest that novelty preferences in that paradigm are likely to contain less evidential value.…”

“…We would argue that this combination of approaches provides important benefits. Meta-analyses, like big data approaches, allow conclusions to be evaluated at scales that cannot be reached in typical lab experiments (Bergmann & Cristia, 2016;Cristia, 2018;Tsuji, Bergmann, & Cristia, 2014), while lab experiments complement the meta-analyses by permitting direct causal tests of their conclusions.…”