The study of foreign accent observed in the speech generated by learners of Japanese is one of the most important challenges in the phonetic study of the teaching of Japanese as a second language. The current author believes that a large-scale database of learners' speech is indispensable for the full-fledged scientific study of foreign accent phenomena. The main reason for this belief stems from the failure of past studies of contrastive phonology, which revealed the sterility of rule-based, top-down analysis in this field. In addition to this, recent progress in automatic speech recognition of mother-tongue speech convincingly demonstrates the effectiveness of a statistical, bottom-up approach as represented by the corpus-based construction of acoustic models. In the last half of the paper, issues regarding the corpus design, construction, and analysis are discussed based upon the author's personal experiences in corpus compilation. The paper ends with the clarification of the role that should be played by the Society for Teaching Japanese as a Foreign Language in the infrastructure development for the corpus-based analysis of the teaching of Japanese as a foreign language.

This paper examines the behavior and introspection of the learners and teacher during prosody teaching, based on data provided by the participants via email, grappling with such issues as: the utility of providing a model for pronunciation; discrepancies in evaluation; choosing methods for pointing out errors; timing of the teacher's error judgements and replies; the teacher's evaluation of the learners' pronunciation. Quite apart from the long-range goal of improving classroom practice, simply to engage in participant observation, it is essential to get away from an idealized view of education, and to learn how to be critical of one's own value judgements: constructive debate is otherwise impossible. Further research is necessary to shed more light on how to contribute to improvements in teaching, and how to facilitate the sharing of introspective data.

In this paper I analyze and compare two cases of classroom discourse in the teaching of pronunciation, in which one may observe the process by which the learners' pronunciation approaches the target pronunciation, attempting to establish just what it is that learners do in the course of this process. It has been suggested that since learners do not merely learn by following the instructions and information imparted by their instructors, instructors have a responsibility to go beyond a focus on quantity and quality of information, and to consider how learners actually use such information as a resource, and how best to set up the information in a useful format. However, even with the same goals, the process of attaining these goals is likely to vary, and it is essential to collect detailed analyses of how the various components of the learning process function in actual practice. As one methodology for this purpose, the analysis of classroom discourse makes it possible to follow the learning process to its result, and the accumulation and shared use of the data thus produced will make it possible for instructors to reconsider, not just their own beliefs and practices, but the very nature of phonetic instruction itself.

In this paper, I will introduce recent research findings pertaining to phonological acquisition by L2 learners of Japanese, and discuss the application of these findings to teaching Japanese pronunciation.

The results of three studies show that: 1) Japanese learners realize that their pronunciation problems would cause miscommunication; 2) adult learners of Japanese can attain native-level pronunciation; and 3) good learners have achieved the ultimate level in Japanese pronunciation because of their attitude towards learning pronunciation, their study methods and the amount of input.

Following these research findings, it is suggested that we facilitate autonomous learning by establishing learning environments inside/outside the classroom. This allows learners to practice pronunciation and provides them with learning opportunities. Three examples are presented: 1) DVD for shadowing practice, 2) on-demand Japanese pronunciation course, and 3) software for practicing Japanese pronunciation.

Deshō has two meanings: conjecture and confirmation. By using corpus data, this paper demonstrates that the former is used much less than the latter in everyday conversation. Usage of deshō in "bare" form (i.e. without sentence-final particles) is especially rare. Deshō can be used in bare form only when the speaker is a "specialist". However, the conjectural deshō appears in most basic-level textbooks. We can presume that such frequent appearance results from descriptive grammar, where the conjectural usage of deshō/darō is regarded as fundamental. Such influence from descriptive grammar is harmful and must be eliminated. Instead, this paper proposes a natural order for the introduction of expressions containing deshō/darō from the viewpoint of pedagogical grammar, the latter being independent of descriptive grammar.

Lexical or grammatical syllabi for Japanese language learners have mainly been created drawing on the intuition and experience of teachers and editors. Information concerning the collocations between words or words and grammatical patterns has largely been neglected. In this paper, we propose the usage of corpora within Japanese language education by exploring the possibilities and appropriateness of creating Japanese language lexical syllabus based on various kinds of Japanese language corpora. We concentrate on a set of suppositional adverbs and clause-final modality forms, exploring their collocational behaviors across various types of corpora, and based on analyses of frequency and salience within the corpora, we propose improvements to the coverage of lexical or collocational items within existing Japanese language syllabi. Additionally, we differentiate between the various corpora according to adverb distributions, and based on those analysis results, we discuss the relation between learning goals and corpus type, as well as the relation between introduction order for items within a syllabus and item frequency within the relevant corpora.

This study examined Chinese speakers' knowledge of the distinction between Japanese purpose clauses with tame ni and those with yō ni (e.g. Nihongo kyōsi-ni [naru tame ni / nareru yō ni] benkyō site iru ‘I am studying Japanese [in order to become / so as to be able to become] a Japanese teacher'). Tame ni is used when the event denoted by the purpose clause is controllable by the subject of the main clause, whereas yō ni is used when the event expressed is uncontrollable by the main-clause subject. Based on a comparison between Japanese and Chinese and the nature of positive evidence for transfer, it is hypothesized that Chinese-speaking learners of Japanese will overgeneralize tame ni, and that this overgeneralization will persist until advanced levels. Advanced Chinese learners of Japanese and Japanese native speakers were tested using a preference task in which sentences with tame ni and yō ni were compared. Results supported the predictions, indicating that the transfer-and-learnability account of second language acquisition, originally proposed to explain data from Indo-European languages, can be extended to data involving a non-Indo-European target language (Japanese) and first language (Chinese).

This paper examines the causes for misusage of the compound particle ni totte by learners by comparing their usage with that of native speakers. The main causes established in this research are as follows:

1. In the sentence X ni totte A wa B, some learners do not understand that X is not the subject of B but the recipient for B.

2. Some learners improperly extend the use of ni totte to cases in which subjective interpretation is not required in the statement A wa B.

3. Some learners do not understand that X has to have commitment to A.

With reference to the first point, it was found that misusage was often accompanied with a verb-predicate, and therefore A was missing in many cases. The third point confirmed that learners making such misusage did not understand that X (watashi) and A should have a direct relationship in order to use ni totte correctly. This unawareness was observed in the confusion of watashi ni totte and watashi wa ... to omou by learners.

This report investigates and analyzes learner perceptions of mutual self-evaluation from a Grounded Theory Approach. From the analysis come the following points: First, even though students felt some inner resistance to mutual self-evaluation, they overcame their resistance and carried out the process for reviewing the role of evaluation. Secondly, they found meaning in being able to change their sense of values. Thirdly, they regard these changes as motivation for studying Japanese in the future. On the other hand, they also felt some dissatisfaction with mutual self-evaluation. From these examples, one can improve action research in the teaching Japanese as a second language.

Recently, attention has been focused on issues in Japanese Language Education for foreign children who were born and brought up in Japan. It is believed that understanding the characteristics of these children's language development in Japanese is essential for ensuring the quality of their education. In this article, language data from interview tasks was analyzed by comparing foreign children and Japanese native speakers in the same year group. The interview tasks were conducted in June (two months after entering elementary school). According to the results, foreign children got lower scores than the native speakers in task achievement. There were distinct differences between them in fluency and complexity of the sentences they produced. The results show that the foreign children seem to have weaknesses in the following areas: 1) producing sentences; 2) understanding utterances within conversation; 3) giving additional information or reconstructing information in responding to the other person's utterance. However, this analysis was based on conversation within a fairly restricted context. More respondents and a more refined method of analysis will be needed in order to raise the validity of this approach.