Exploiting Pitch Accent Information in Compound Processing: A Comparison between Adults and 6- to 7-Year-Old Children

ABSTRACT

A noun can be potentially ambiguous as to whether it is a head on its own, or is a modifier of a Noun + Noun compound waiting for its head. This study investigates whether young children can exploit the prosodic information on a modifier constituent preceding the head to facilitate resolution of such ambiguity in Japanese. Evidence from English suggests that young speakers are not sensitive to compound stress in distinguishing between compounds and syntactic phrases unless the compound is very familiar (Good, 2008; Vogel & Raimy, 2002). This study concerns whether children in general have such limited capability to use prosodic cues to promptly compute a compound representation without the lexical boost, or whether they might show greater sensitivity to more categorical compound prosody such as that associated with the Compound Accent Rule (CAR) in Japanese. A previous study (Hirose & Mazuka, 2015) demonstrated that adult Japanese speakers can predict the compound structure prior to the head if the prosodic information on the modifier unambiguously signals that the CAR is being applied. The present study conducted the same on-line experiment with children (6- to 7-year-olds) and compared the time course of the effects with that of adults using permutation-based analysis (Maris & Oosternveld, 2007). The results reveal that children are sensitive to pitch accent information that facilitates the quicker processing of the compound or the single head noun representation compared to when such prosodic signals are less apparent, depending on the type of the lexical accent of the noun in question.

Acknowledgments

We are greatly indebted to Franklin Chang, Douglas Roland and Sho Tsuji for their various comments and advice. We deeply appreciate various valuable suggestions from the anonymous reviewers and the editor of the journal.

Funding

This research was supported in part by MEXT/JSPS KAKENHI Grant Number 25580086 to the first author.

Notes

¹ An apostrophe (’) indicates that the lexical accent falls on the following mora.

² A mora (plural morae) is a phonological unit, usually comprising a vowel or a consonant-vowel combination, which is the basis of the sound/rhythm system of Japanese (in contrast to the syllable).

³ Including a Competitor Compound object was considered crucial in the original Hirose and Mazuka (2015) experiment to test whether the processing advantage of the Unambiguous prosody found in adults was anticipatory in nature, which could take place before the segmental information of the head noun. In the present study, direct comparisons between Target Compounds and Competitor Compounds are not reported: such comparisons would not be very informative, given the size of the latency for the prosodic unambiguity effect in children.

⁴ A t value reported here is the sum of the t-statistics for the individual time windows within the entire cluster in consideration. A p value reported here refers to the proportion of the overall permutation distribution within the cluster that has a greater test statistics than the observed value.

Additional information

Funding

This research was supported in part by MEXT/JSPS KAKENHI Grant Number 25580086 to the first author.