1. The causal relation between the Japanese language proficiency of 850 foreign learners and productive frequencies of lexical categories in expository text
- Author
-
TAMAOKA, Katsuo
- Subjects
日本語能力 ,J-CAT (Japanese Computerized Adaptive Test) ,Japanese language proficiency ,SEM (structural equation modeling) ,I-JAS ,productive frequencies ,説明力 ,構造方程式モデリング ,J-CAT ,産出頻度 ,I-JAS (International Corpus of Japanese as a Second Language) ,explanatory ability - Abstract
SNSの普及にともない,書き言葉によるコミュニケーションが頻繁に行われるようになってきた。とりわけ,事実に即した情報や依頼などをわかり易く書き言葉で伝達するための「説明文(expository writing)」を書く能力が要求されるようになってきた。この能力を「説明力(explanatory ability)」と呼ぶ。現代社会において,説明力は日本語母語話者ばかりでなく外国人日本語学習者の実生活でも,なくてはならない能力になっている。そこで,『多言語母語の日本語学習者横断コーパス: I-JAS』に収録された2つのストーリーライティング課題が説明文を反映していると仮定し,海外で学習する日本語学習者850名が書いたテキストデータを基に,(1)動詞,助動詞,副詞,助詞の4つの品詞を「述部」とし,(2)名詞,形容詞,連体詞の3つの品詞を「名詞句」として,品詞別の産出頻度を算出した。そして,J-CAT(Japanese Computerized Adaptive Test)で測定された日本語能力からI-JASのテキストの産出頻度への因果関係モデル(図1を参照)を5つ想定して,構造方程式モデリング(structural equation modelling, SEM)の解析法で日本語能力と産出頻度データとの適合度を検討した。適合度指標に照らして,最適の因果関係モデル(図2を参照)をみいだした。このモデルは,次のような連続的な因果関係を示した。(1)語彙と文法が「基礎力」を構成し,(2)読解と聴解の「理解力」を促進し,(3)「述部」の語彙産出を豊かにし,(4)「名詞句」の語彙産出に大きく貢献する。, With the widespread use of SNS (social networking services), people are communicating more frequently by written text than ever before. The ability to express content plainly is in great demand. In the present study, this is defined as “explanatory ability”. In contemporary society, this ability has become essential to daily life not only for native Japanese speakers but also for foreign learners of Japanese. Assuming the two story-writing tasks recorded in the International Corpus of Japanese as a Second Language (I-JAS) reflect explanatory ability, the productive frequencies of lexical categories found in the text written by 850 Japanese learners outside Japan were divided into two groups: (1) four categories of verbs, auxiliary verbs, adverbs, and auxiliaries as “predicates” and (2) three categories of nouns, adjectives, and conjoined words as “noun phrases”. Using the statistical method of structural equation modelling (SEM), the five causal relation models linking Japanese proficiency as measured by the Japanese Computerized Adaptive Test (J-CAT) and text productive frequencies (see Figure 1) were examined to see how these models fit with data from proficiency scores and productive frequencies. Based on the fitting indexes, the study found the best causal model to be that identified in Model 2 (see Figure 2). This model showed causal sequencing in the following order: (1) procession of a basic knowledge of vocabulary and grammar (2) promotes reading and listening comprehension, further (3) enriches the production of predicates and (4) contributes to the production of noun phrases.
- Published
- 2020