Auditory processing and spoken word recognition difficulties have been observed in Specific Language Impairment (SLI), raising the possibility that auditory perceptual deficits disrupt word recognition and, in turn, phonological processing and oral language. In this study, fifty-seven kindergarten children with SLI and fifty-three language-typical age-matched controls were assessed with a speech-gating task to measure spoken word recognition, psychophysical tasks to measure auditory Frequency Modulation (FM) detection and Frequency Discrimination (FD), and standardized psychometric tests of phonological processing and oral language. As a group, children with SLI took significantly longer than language-typical controls to recognize words with high neighborhood density, perhaps reflecting subpar phonological representations. FM, but not FD, was significantly worse in SLI. However, while both poorer speech-gating performance and poorer auditory thresholds (FM) were evident in SLI, spoken word recognition did not mediate any relation between auditory perception and either phonological processing or oral language.

An in-depth exploration of the associations of two aspects of morphological awareness in Chinese—homophone awareness and lexical compounding awareness—to Chinese word reading and vocabulary knowledge was the primary focus of the present study. Among 154 9-year-old Hong Kong Chinese children, both lexical compounding and homophone awareness were significantly associated with word reading (r = .54 for compounding, r = .38 for homophones) and vocabulary knowledge (r = .41 for compounding, r = .53 for homophones). However, with autoregressors additionally statistically controlled, homophone awareness remained uniquely associated with vocabulary but not word reading; lexical compounding was uniquely associated with both word reading and vocabulary. Path analyses best illustrated this pattern. Both morphological awareness constructs are likely bidirectionally associated with word reading and vocabulary knowledge. However, homophone awareness is more centrally associated with vocabulary knowledge because it taps specific, existing morpheme knowledge. In contrast, lexical compounding requires structural understanding of one's language, which seems to be helpful for both learning to read and vocabulary acquisition in Chinese.

To express object transfer, Cantonese-speakers use a ‘ditransitive’ ([V–R–T] or [V–T–R] where V=Verb, T=Theme, R=Recipient), or a more complex prepositional/serial-verb (P/SV) construction. Clausal elements in Cantonese datives can be optional (resulting in ‘full’ versus ‘non-full’ forms) or appear in variant orders (full non-canonical and full canonical). We report on usage of dative constructions with the word bei2 ‘to give’ in 86 parents and 53 three-year-old children during conversations. The parents used more P/SV than ditransitive bei2-datives, and vice versa for the children. Both groups showed a similar usage pattern of optional elements and variant structures in their ditransitive and P/SV bei2-datives. The roles of multiple construction types, optional elements and variant structures in children's learning of bei2-dative constructions are described.

The production of passive sentences by children with specific language impairment (SLI) was studied in two languages, English and Cantonese. In both languages, the word order required for passive sentences differs from the word order used for active sentences. However, English and Cantonese passive sentences are quite different in other respects. We found that English-speaking children with SLI were less proficient than both same-age and younger typically developing peers in the use of passives, although difficulty could not be attributed to word order or a reliance on active sentences. Cantonese-speaking children with SLI proved less capable than same-age peers in their use of passive sentences but at least as proficient as younger peers. The implications of these cross-linguistic differences are discussed.

The ability to make clear reference in connected discourse was examined in children learning Cantonese, a Chinese language where noun phrase constituents, whatever their grammatical role, are omissible from sentences under discourse conditions that are not well-understood. Forty-three typically developing children aged 3;0, 5;0, 7;0 and 12;0 told 16 stories based on picture sequences. A panel of adult native Cantonese speakers was asked to judge the referential adequacy of each child's stories by identifying the character the child was talking about in 32 targeted referential acts. The targeted acts were of three sorts: MAINTENANCE of a known character, INTRODUCTION of a second new character, and REINTRODUCTION of a known character. Reference was judged to be adequate when 3 out of 4 ‘listeners’ could successfully identify the character. Children's referential expressions were most adequate for Maintenance, less adequate for Introduction, and least adequate for Reintroduction. The twelve- and seven-year-olds approached ceiling on all three functions. The five-year-olds scored poorly on Reintroduction, and the three-year-olds failed both Introduction and Reintroduction, despite knowledge of at least one of the possible linguistic forms required for these acts as evidenced in a sentence imitation task. Viewed within the framework of Levelt's (1989) discourse model, the data improve our understanding of the developmental period during which children learn to make appropriate presuppositions about the listener's knowledge and attentional states.