Search

日本人大学生の日本語作文技術を向上させる訓練プログラムを開発するための予備的研究を行った。刺激文として多義文を30文作成し，それぞれを2つの一義文として書き直す推敲課題を用いた。実験条件として推敲後に正答例を示す条件と正答例を読んだ後に自らの回答を自己採点させる条件を用意し，これを独立変数とした。参加者24人はいずれかの実験条件に無作為に割り当てた。参加者の回答を正答例との一致度で採点し，これを従属変数とした。実験条件の効果は参加者間多層ベースライン法を用いて参加者内で比較した。実験の結果，訓練効果は小さく，刺激文の効果が大きかった。刺激文と参加者の回答を元に推敲行動に及ぼす要因について質的分析を行った。その結果，今後の課題として，多義文の定義を見直すこと，推敲に要する語彙などの下位行動を統制すること，刺激文の理解と一義文の作文を独立して測定すること，刺激文から読み取れる5W1Hについて尋ねるプロンプトを用いた推敲訓練の効果を検証することを提案した。, A preliminary study was conducted to develop a training program to improve Japanese university students’ writing skills. Thirty ambiguous sentences were prepared for proofreading. The task required reading the ambiguous sentences and writing two single-meaning sentences that can be interpreted from each ambiguous sentence. In one experimental condition, the correct rewritten examples were presented just after the proofreading task. In the other condition, participants were asked to score their own rewritten versions by comparing their answers to the correct ones. Twenty-four university students participated in the study. They were randomly assigned to one of two experimental conditions. The dependent variable was the proofreading task score, evaluated by the experimenter, using the same correct examples. The effects of the experimental conditions were examined using a multiple-baseline across subjects design. Both training types showed statistically significant yet educationally small effects. The difficulty of the ambiguous sentences explained a large proportion of the variance in the score. The results suggest that future research should employ a revised definition of an ambiguous sentence, independent measurement of reading comprehension and rewriting, and prompt participants to identify 5W1H (what, why, when, where, who, and how) in ambiguous sentences.

application/pdf, In this article, we propose a new semantic analyzing method of adjective parts in Japanese documents. The method uses a recursive structure, as a semantic representational structure, to recognize adjective parts in a sentence. The data structure, which was defined by us, has some attributes of semantic features and syntactic features of adjective parts. One of these is called 'Semantic ID', and the other is called 'Grammar ID'. With the method, we developed an analyzing system called 'Modifier Analyzer', which can detect modifiers, their target words, and relations of them in Japanese sentences. And we implemented this system on a web server as a CGI application, which uses HTML tags for user-interface, therefore the 'Modifier Analyzer' can be used with a general web browser on any platforms. Then we attempted evaluating the performance of the 'Modifier Analyzer'. As a result, it was confirmed that the method has a high reliability for semantic and syntactic analyzing.

application/pdf, In this article, we propose a new method of partial correcting in Japanese documents. This method uses an information of co-occurred cases, which determines a construction of a sentence. In our method, we extended the information of co-occurred cases to handle an error-detection and an error-correction for proofreading. In this extension, we designed a data structure of 'Case Frame' of predicates as a 'Case Unit'. Case unit is a data structure which is composed of 'Semantic Feature', 'Case Tail', and 'Case Type'. With this method, we developed a proofreading system for Japanese which is called 'CaseCorrect'. And we implemented this system on a Web Server as a CGI application which uses HTML tags for user-interface, therefore the 'CaseCorrect' can be used with a general Web browser on any platforms. Then we attempted evaluating a performance of the 'CaseCorrect'. As a result, it was confirmed that the performance of our method is high for error-detection and error-correction for proofreading.