Research in infant speech perception has demonstrated that young infants are sensitive to prosodic properties of language that are relevant to linguistic rhythm. In the Rhythm-based Prosodic Bootstrapping Hypothesis, it is proposed that infants’ early sensitivity to the rhythmic properties of a language will enable them to adopt a metrical speech segmentation strategy appropriate for their language. Review of the literature shows that while infants are capable of discriminating languages from different rhythm classes, this does not necessarily lead them to adopt metrical segmentation strategies appropriate to each and every particular class. It is argued that the rhythm of a language may be salient for infants in all languages, but how this sensitivity is linked to other aspects of language acquisition needs to be re-considered.

We present children’s non-adult interpretation of sentences that contain the focus particle dake, using the Truth-Value Judgment Task (Crain and Thornton 1998). Our first observation is that children interpreted sentences with dake differently when dake was attached to the subject and to the object (the subject/object asymmetry in the interpretation of dake). We also observed that children interpreted sentences with dake differently when dake was followed by a case particle (the ‘particle/no-particle asymmetry’). Any theory based on the assumption that the Japanese employs an abstract Case feature system fails to capture the systematic pattern shown in children’s non-adult interpretations of dake. Our data provide empirical support for a syntactic theory proposed in Aoyagi (2006), which distinguishes the nature of Nominative and Accusative case particles (ga and o), as well as assumes different derivations for sentences containing dake, depending on whether it is followed by a case particle or not.

The present paper clarified the composition of early vocabulary in Japanese children through the standardized data of the Japanese MacArthur-Bates Communicative Development Inventories (JCDIs), the longitudinal data, and the cross sectional data. We examined especially the issue of noun and verb dominance in early child language. First, the composition of the first 50 words in JCDIs showed that common nouns had the highest proportion. Second, the opportunity scores of nouns, adjectives, verbs, and closed-class words on JCDIs for 158 Japanese children at 20 months of age showed the noun prevalence, and this result was consistent with the results of seven countries in the study by Bornstein et al. (2004). Third, the longitudinal study of two Japanese children showed noun dominance after vocabulary spurt and verb dominance after the emergence of grammar. Caregiver language was verb dominant. Fourth, 31 Japanese children showed the noun dominance in the book-reading context, but in the toy-playing context, there was a shift away as children developed from single words through the presyntactic stage to the syntactic stage. Caregiver language was verb dominant in a number of respects across development in the toy context, and thus was not closely related to child’s lexical balance. We concluded that children have a conceptual disposition to learn nouns in early lexical development. Finally, we discussed the mechanisms of word learning which evoke noun dominance.

This study investigates children’s comprehension of a single-argument sentence that contains either a subject or a direct object of a transitive verb. Adopting the picture-selection task, Experiment 1 reveals 90% accuracy by the oldest group of children (mean age=5;11). This indicates that children have acquired the grammatical knowledge of ga and o by this age and that their competence is reflected in their linguistic performance. In contrast, the results disclose poor performance by the younger groups of children. Their errors were observed on the accusative case-marking more frequently than on the nominative case-marking. This asymmetry seems to reflect the developmental difference between the two case-markers: Nominative ga is acquired earlier than accusative o. In order to explore whether or not this asymmetry is rooted in the children’s grammatical knowledge, Experiment 2 tries to eliminate the effects of agent priority, a possible bias for the children’s agent interpretation of the single argument. For this purpose, test sentences were provided with discourse context in the picture-selection task. The results show that the children’s performance greatly improved with the asymmetry between the nominative and accusative cases removed. These results suggest that the performance factor overrides children’s grammatical knowledge in Experiment 1, and that the elimination of the agent bias discloses their linguistic competence in Experiment 2. However, the percentages of correct responses by the younger groups of children (age range=3;7–5;7) in the second experiment do not reach those of the oldest group in the first experiment, suggesting that they are still in the developmental process of acquiring the nominative and accusative case-markers.

This paper investigates the syntax and semantics of Japanese comparative constructions that make use of words like izyoo(ni) ‘more than’ and gurai ‘roughly equal to’, which are functionally analogous to the English more- and as-comparatives. I argue that they are necessarily interpreted as comparisons of deviation in the sense of Bierwisch (1989) and Kennedy (2001). In explaining why these constructions cannot express comparisons of the absolute projections of two objects on a scale, as the English comparatives do, this paper advocates the line of thinking pursued by Fukui (1986) and Snyder et al. (1994) that AdjP in Japanese is impoverished in such a way that it lacks the position to host a degree variable or constant.

In this paper, I describe the phonological distinctions between word medial consonants in Kadorih. On the basis of distributional facts, I give a strong support to the theory of the phonological changes of these consonants, proposed in Hudson (1967).

As in other Austronesian languages, Kadorih’s canonical word shape is disyllabic CVCCVC where all of the consonants are optional. There are only two types of word medial consonant sequences (-CC-), viz. /h/ followed by a voiceless stop or an affricate (-hT-), and a nasal consonant followed by a voiced stop or an affricate (-ND-). I demonstrate that -hT- and -T- occur more frequently than -ND- and -D- in (i) monomorphemic words, (ii) both of the word medial positions in trisyllabic monomorphemic words, and (iii) frequently used words such as personal pronouns and numerals. Based on these observations, I show that the difference in frequency between -(h)T- and -(N)D- is due to the difference in their functional loads. This claim is corroborated by comparing the quantities of their respective distinctive pairs: CVhTVC – CVTVC on the one hand, and CVNDVC – CVDVC on the other. The comparison presented in this paper provides evidence that -hT- and -T- have a higher functional load than -ND- and -D-. The phonological distinctions attested between these word medial consonants support Hudson’s theory of the phonological changes, *T>hT/V_V and *NT>T, which he assumed in spite of insufficient evidence.