In order to learn words, infants must determine the range of sequences of speech-sounds that get to ?count? as a production of a given word. Decades of research have shown that over the first year of life infants learn to identify the sounds that are present in their language, and subsequently begin to learn the meanings of words, before production of these words begins. While infants learn to distinguish native language sounds from foreign language sounds over the first 6-12 months of life, their representations of the sounds of known words are often underspecified, or ?fuzzy?. At the same time, early representations are in some ways too specific, e.g. when infants fail to recognize the same word said by a new talker. Thus, a critical challenge for language learners is forming appropriately specific representations of words, ones that are narrow enough to disregard incorrect pronunciations but broad enough to recognize new instances. What factors influence the formation of appropriately-specific word representations? Previous word-learning research suggests that acoustic variability stemming from multiple talkers, or from infant directed speech, can help learners acquire novel words in short lab studies. Extending previous work to both lab and ecologically-valid environments, the current proposal?s specific aims are to: 1. Quantify acoustic variability in naturalistic settings and test its links to word-learning 2. Test the role of acoustic variability in the formation of word representations in the lab Using both corpus methods and experimental approaches, the current proposal seeks to understand how acoustic variability that is inherent to an infants? linguistic input influences the formation of appropriately- specific word representations. This combined approach permits an investigation of this question both ?in the wild? and ?in the lab?, allowing for a complementary and more thorough understanding of the role of acoustic variability than previous research has attained. The results of this work will contribute to an understanding of how properties of the input can impact word learning. A better understanding of factors that influence word learning is beneficial for developmental psychology, particularly as it considers the best ways to support early learners, both with typical and atypical trajectories. The findings will have implications for educational and clinical settings, as the amount of acoustic variability in the input can be manipulated to optimize word learning.

Public Health Relevance

The current proposal asks how a specific aspect of infants? daily linguistic input?the variability in the way words sound when they?re produced by the same talker or by different talkers?influences word learning, both in the lab and in real life. Using naturalistic observational data alongside experimental methods, the proposed work seeks to address how these sources of talker variability aid infant learners in forming appropriately specific representations of words: representations that are broad enough to recognize new instances of words, but also narrow enough to disregard incorrect pronunciations. The results from this two-part approach will provide insights that can be readily applied to educational or clinical settings.

National Institute of Health (NIH)
Eunice Kennedy Shriver National Institute of Child Health & Human Development (NICHD)
Postdoctoral Individual National Research Service Award (F32)
Project #
Application #
Study Section
Special Emphasis Panel (ZRG1)
Program Officer
Griffin, James
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Duke University
Schools of Arts and Sciences
United States
Zip Code