Phonotactic Constraints on Infant Word Learning

How do infants use their knowledge of native language sound patterns when learning words? There is ample evidence of infants’ precocious acquisition of native language sound structure during the first year of life, but much less evidence concerning how they apply this knowledge to the task of associating sounds with meanings in word learning. To address this question, 18‐month‐olds were presented with two phonotactically legal object labels (containing sound sequences that occur frequently in English) or two phonotactically illegal object labels (containing sound sequences that never occur in English), paired with novel objects. Infants were then tested using a looking‐while‐listening measure. The results revealed that infants looked at the correct objects after hearing the legal labels, but not the illegal labels. Furthermore, vocabulary size was related to performance. Infants with larger receptive vocabularies displayed greater differences between learning of legal and illegal labels than infants with smaller vocabularies. These findings provide evidence that infants’ knowledge of native language sound patterns influences their word learning.     

Modeling Age Differences in Infant Category Learning

We used an encoder version of cascade correlation to simulate Younger and Cohen's (1983, 1986) finding that 10‐month‐olds recover attention on the basis of correlations among stimulus features, but 4‐ and 7‐month‐olds recover attention on the basis of stimulus features. We captured these effects by varying the score threshold parameter in cascade correlation, which controls how deeply training patterns are learned. When networks learned deeply, they showed more error to uncorrelated than to correlated test patterns, indicating that they abstracted correlations during familiarization. When prevented from learning deeply, networks decreased error during familiarization and showed as much error to correlated as to uncorrelated tests but less than to test items with novel features, indicating that they learned features but not correlations among features. Our explanation is that older infants learn more from the same exposure than do younger infants. Unlike previous explanations that postulate unspecified qualitative shifts in processing with age, our explanation focuses on quantitatively deeper learning with increasing age. Finally, we provide some new empirical evidence to support this explanation.     

Common Mechanisms in Infant and Adult Category Learning

Computational models of infant categorization often fail to elaborate the transitional mechanisms that allow infants to achieve adult performance. In this article, we apply a successful connectionist model of adult category learning to developmental data. The Supervised and Unsupervised Stratified Adaptive Incremental Network (SUSTAIN) model is able to account for the emergence of infants' sensitivity to correlated attributes (e.g., has wings and can fly). SUSTAIN offers 2 complimentary explanations of the developmental trend. One explanation centers on memory storage limitations, whereas the other focuses on limitations in perceptual systems. Both explanations parallel published findings concerning the cognitive and sensory limitations of infants. SUSTAIN's simulations suggest that conceptual development follows a continuous and smooth trajectory despite qualitative changes in behavior and that the mechanisms that underlie infant and adult categorization might not differ significantly.     

Infant Side Biases and Familiarity‐Novelty Preferences During a Serial Paired‐Comparison Task

We monitored changes in looking that emerged when 3‐ to 6‐month‐old infants were presented with 48 trials pairing familiar and novel faces. Graphic displays were used to identify changes in looking throughout the task. Many infants exhibited strong side biases produced by infants looking repeatedly in the same direction. Although an overall novelty preference was found for the group, individual infants exhibited brief novelty runs. Few infants began with a familiarity preference. We argue that variable looking patterns emerged during the task from competition between the infants' preference to look for something novel versus their tendency to look back to previous locations. Our data suggest that looking during paired‐comparison tasks is a dynamic process dependent on perceptual‐motor events happening during the task itself.     

Infant Learning: Historical,Conceptual, and Methodological Challenges

Acquiring knowledge about the underlying structures of the environment presents a number of challenges for a naive learner. These challenges include the absence of reinforcement to guide learning, the presence of numerous information sources from which only a select few are relevant, and the uncertainty about when an underlying structure may have undergone a change. A crucial implication of these challenges is that the naive learner must make implicit decisions about when to generalize to novel inputs and when to restrict generalization because there are multiple underlying structures. An historical perspective on these challenges is presented, and some potential solutions are proposed.     

The Effect of Learning Experiences and Context on Infant Imitation and Generalization

Over the first years of life, infants gradually develop the ability to retrieve their memories across cue and contextual changes. Whereas maturational factors drive some of these developments in memory ability, experiences occurring within the learning event may also impact infants' ability to retrieve memories in new situations. In 2 experiments we examined whether it was possible to facilitate 12‐month‐old infants' generalization of learning in the deferred imitation paradigm by varying experiences before or during the demonstration session, or during the retention interval. In Experiment 1, altering the length, timing, or variability of training had no impact on generalization; infants showed a low, but consistent level of memory retrieval. In Experiment 2, infants who experienced a unique context for encoding and retrieval exhibited generalization; infants who experienced the context prior to the demonstration session, or during the retention interval, did not. Specificity is a robust feature of infant memory and is not substantially altered by encoding experiences in an observational learning paradigm. Previous history with a learning environment can, however, impact the flexibility of memory retrieval.     

Exploring the Emergence of Side Biases and Familiarity–Novelty Preferences from the Real‐Time Dynamics of Infant Looking

This study examined how look dynamics contribute to infants’ emerging novelty preferences. Time‐series analyses were used to study the temporal nature of looking displayed by 3‐ to 5‐month‐old infants during a serial paired‐comparison task. Evidence was found only for short‐term stability: Novelty preferences and side biases were not stable from one visit to the next, but looking was consistent from one moment to the next producing stability within trials and temporarily across trials leading to the formation of behavioral runs. Persistence in looking left or right across multiple trials did not change from one visit to the next, but persistence in looking at familiar stimuli declined with age. By Visit 3, familiarity runs occurred less often than did novelty runs. Frequent but highly variable runs, including surprisingly late familiarity preferences, suggest that overall side biases and novelty preferences found during visual preference tasks are emergent phenomena affected by moment‐to‐moment changes in looking.     

Sample Size,Statistical Power,and False Conclusions in Infant Looking‐Time Research

Infant research is hard. It is difficult, expensive, and time‐consuming to identify, recruit, and test infants. As a result, ours is a field of small sample sizes. Many studies using infant looking time as a measure have samples of 8–12 infants per cell, and studies with more than 24 infants per cell are uncommon. This paper examines the effect of such sample sizes on statistical power and the conclusions drawn from infant looking‐time research. An examination of the state of the current literature suggests that most published looking‐time studies have low power, which leads in the long run to an increase in both false positive and false negative results. Three data sets with relatively large samples (>30 infants) were used to simulate experiments with smaller sample sizes; 1,000 random subsamples of 8, 12, 16, 20, and 24 infants from the overall samples were selected, making it possible to examine the systematic effect of sample size on the results. This approach revealed that despite clear results with the original large samples, the results with smaller subsamples were highly variable, yielding both false positive and false negative outcomes. Finally, a number of emerging possible solutions are discussed.     

Computational Modeling of Statistical Learning: Effects of Transitional Probability Versus Frequency and Links to Word Learning

Statistical learning mechanisms play an important role in theories of language acquisition and processing. Recurrent neural network models have provided important insights into how these mechanisms might operate. We examined whether such networks capture two key findings in human statistical learning. In Simulation 1, a simple recurrent network (SRN) performed much like human learners: it was sensitive to both transitional probability and frequency, with frequency dominating early in learning and probability emerging as the dominant cue later in learning. In Simulation 2, an SRN captured links between statistical segmentation and word learning in infants and adults, and suggested that these links arise because phonological representations are more distinctive for syllables with higher transitional probability. Beyond simply simulating general phenomena, these models provide new insights into underlying mechanisms and generate novel behavioral predictions.     

Changing Opportunities for Learning in Everyday Life: Infant Body Position Over the First Year

Developmental theories depend on characterizing the input to potential learning mechanisms—infants’ everyday experiences. The current study employed a novel ecological momentary assessment (EMA) to measure two aspects of the physical context of those experiences: body position and location. Infant body position was selected because it relates to the development of a variety of other skills. Caregivers of 3‐, 6‐, 9‐, and 12‐month‐olds reported infants’ body position—held, supine, reclined, prone, sitting, or upright—in response to text message notifications over a week to capture infants’ experiences across the entire range of their daily activities. Findings revealed a tremendous disparity in the distribution of body position experiences over the first year. Younger infants spend more time held, supine, and reclined, whereas older infants spend more time sitting and upright. Body position experiences differed substantially between same‐age infants who possess a motor skill (e.g., ability to sit or walk) compared with those who did not, suggesting that developing motor skills change infants’ everyday experiences. Finally, the success of the methodology suggests that similar EMAs might be used to study a wide range of infants’ naturalistic experiences.     

From Syllables to Syntax: Multilevel Statistical Learning by 12‐Month‐Old Infants

To successfully acquire language, infants must be able to track multiple levels of regularities in the input. In many cases, regularities only emerge after some learning has already occurred. For example, the grammatical relationships between words are only evident once the words have been segmented from continuous speech. To ask whether infants can engage in this type of learning process, 12‐month‐old infants in 2 experiments were familiarized with multiword utterances synthesized as continuous speech. The words in the utterances were ordered based on a simple finite‐state grammar. Following exposure, infants were tested on novel grammatical and ungrammatical sentences. The results indicate that the infants were able to perform 2 statistical learning tasks in sequence: first segmenting the words from continuous speech, and subsequently discovering the permissible orderings of the words. Given a single set of input, infants were able to acquire multiple levels of structure, suggesting that multiple levels of representation (initially syllable‐level combinations, subsequently word‐level combinations) can emerge during the course of learning.