Overriding the Metrical Bias with Lexical Information: English-Learning 7.5-Month-Olds Use Mommy to Segment Iambic Words

ABSTRACT

Previous research shows that English-learning 7.5-month-olds are biased to segment speech at strong syllables consistent with the predominant trochaic (strong-weak) pattern of words in their language (Jusczyk, Houston, & Newsome, 1999). The present study asked whether 7.5-month-olds can use a familiar name to override their metrical bias and segment iambic (weak-strong) words from natural speech. Infants were familiarized to passages containing weak-strong-weak sequences (guiTAR is) preceded by Mama/Mommy or an unfamiliar name. Infants were tested on iambic words (guiTAR in Experiments 1–2) or trochaic nonwords (TARis in Experiment 3). Segmentation occurred only for iambic words that were preceded by Mama/Mommy during familiarization. The presence of Mama/Mommy attenuated infants’ bias to treat strong-syllables as word onsets. We suggest that young infants weight highly familiar words more strongly than metrical information and discuss the implications of this finding for a hierarchical model of segmentation that connects infant and adult segmentation systems.

Acknowledgments

We thank LouAnn Gerken for her comments on an earlier draft of this article and the parents and children who participated in this research.

Funding

This work was supported by NSF BCS-9910203 to RLG.

Notes

^1. In Jusczyk, Houston, et al.’s (1999) passages, the function words her and your and the adjectives old, red, plain, and pink preceded iambic words. We do not know whether the successful segmentation of iambic words by 10.5-month-olds, but not 7.5-month-olds, stemmed from knowledge of some or all of these words and/or sub-lexical cues.

^2. We conducted a post-hoc replication of Experiment 1 with a male sample as visual inspection of the data suggested males had lower discrimination scores than females. We familiarized 16 male infants with the Mama/Mommy sentences and tested them on the two words that occurred during familiarization and two novel control words. A paired samples t-test analyzing the listening times for words heard after the familiar name (M = 9.71 s, SD = 3.35) versus control words (M = 7.66 s, SD = 2.48) revealed significantly longer listening time to words heard after Mama/Mommy, t(15) = 2.56, p = .02, replicating our Experiment 1 findings with a modified procedure.

^3. We thank an anonymous reviewer for raising this caveat.

Additional information

Funding

This work was supported by NSF BCS-9910203 to RLG.