In 4 studies, 7.5‐month‐olds used synchronized visual–auditory correlations to separate a target speech stream when a distractor passage was presented at equal loudness. Infants succeeded in a segmentation task (using the head‐turn preference procedure with video familiarization) when a video of the talker's face was synchronized with the target passage (Experiment 1, N=30). Infants did not succeed in this task when an unsynchronized (Experiment 2, N=30) or static (Experiment 3, N=30) face was presented during familiarization. Infants also succeeded when viewing a synchronized oscilloscope pattern (Experiment 4, N=26), suggesting that their ability to use visual information is related to domain‐general sensitivities to any synchronized auditory–visual correspondence.

Word-learning skills were tested in normal-hearing 12- to 40-month-olds and in deaf 22- to 40-month-olds 12 to 18 months after cochlear implantation. Using the Intermodal Preferential Looking Paradigm (IPLP), children were tested for their ability to learn two novel-word/novel-object pairings. Normal-hearing children demonstrated learning on this task at approximately 18 months of age and older. For deaf children, performance on this task was significantly correlated with early auditory experience: Children whose cochlear implants were switched on by 14 months of age or who had relatively more hearing before implantation demonstrated learning in this task, but later implanted profoundly deaf children did not. Performance on this task also correlated with later measures of vocabulary size. Taken together, these findings suggest that early auditory experience facilitates word learning and that the IPLP may be useful for identifying children who may be at high risk for poor vocabulary development.

In this article, we hypothesize that invariance detection, a general perceptual phenomenon whereby organisms attend to relatively stable patterns or regularities, is an important means by which infants tune in to various aspects of spoken language. In so doing, we synthesize a substantial body of research on detection of regularities across the domains of speech perception, word segmentation, word-referent mapping, and grammar learning. In addition, we outline our framework for how invariance detection might serve as a perceptual gateway to more sophisticated communication by providing a foundation for subsequent emergent capacities. We test our hypothesis using the domain of word mapping as a case in point, emphasizing its epigenetic nature: Word mapping is rooted in the real-time interactions between the infant and the physical world. The present account offers an alternative to prior theories of early language development and helps to link the field of early language development with more general perceptual processes.

In this study, we used the splitscreen preferential looking paradigm to test 13‐, 15‐, and 20‐month‐olds' developing understanding of simple matrix what‐questions of the forms “What hit the X?” (subject‐question) and “What did the X hit?” (object‐question). Infants responded appropriately to subject‐questions by 15 months of age, and to both subject‐ and object‐questions by 20 months. At no age did infants look longer toward the object overtly mentioned in the question, as might be expected based on a surface account of early language acquisition. This suggests that infants may have some understanding of these complex structures long before they are produced.

By their second birthday, children are beginning to map meaning to form with relative ease. One challenge for these developing abilities is separating information relevant to word identity (i.e. phonemic information) from irrelevant information (e.g. voice and foreign accent). Nevertheless, little is known about toddlers' abilities to ignore irrelevant phonetic detail when faced with the demanding task of word learning. In an experiment with English-learning toddlers, we examined the impact of foreign accent on word learning. Findings revealed that while toddlers aged 2 ; 6 successfully generalized newly learned words spoken by a Spanish-accented speaker and a native English speaker, success of those aged 2 ; 0 was restricted. Specifically, toddlers aged 2 ; 0 failed to generalize words when trained by the native English speaker and tested by the Spanish-accented speaker. Data suggest that exposure to foreign accent in training may promote generalization of newly learned forms. These findings are considered in the context of developmental changes in early word representations.