Logo image
Computer-Assisted Phrase Learning: From Formulaic Sequences to Paraphrases
Dissertation

Computer-Assisted Phrase Learning: From Formulaic Sequences to Paraphrases

Chen, Mei-Hua
Doctor of Philosophy (PHD), 國立清華大學, 資訊系統與應用研究所
2012

Abstract

詞彙短語 代換片語 詞彙短語輔助學習系統 代換片語建議系統 電腦輔助語言學習 自然語言處理技術 語料庫為本 語用 formulaic expression/sequence paraphrase/paraphrasing computer-assisted language learning natural language processing productive competence phrase learning reference tool corpus-based
English learners have difficulties using appropriate word combinations in language production even after they have devoted considerable time to recognizing and memorizing a large inventory of words. Since primarily this is because they have learned words in isolation rather than in context, this study instead promotes the development of phrasal knowledge. Specifically we focus on facilitating learners’ fluent language production in two aspects: formulaic expressions and paraphrases. The lexical items of a formulaic sequence tend to form a fixed pattern to carry out language functions which contribute to language use. Paraphrases are formulaic sequences sharing similar meaning but various usages. Grouping and learning such sequences effectively and systematically facilitates fluent language use. To achieve this goal, we have utilized natural language processing techniques to develop two computer-assisted phrase learning systems: GRASP and PREFER. These were designed to promote formulaic sequence learning and paraphrase learning, aimed at providing immediate and comprehensive assistance to learners and helping them develop productive competence for speaking and writing, instead of passively recognizing text meaning only. With both tools, users can input multi-word querying to find their desired phrases. The systems also provide syntactic patterns, usage information and example sentences to illustrate real world language use. Most importantly, these systems summarize usage information, unlike existing concordancers which display many exeample sentences. In other words, learners do not have to duduce the usages of the formulaic expressions and paraphrases when using GRASP and PREFER. Automatic summarization from language data lends support to the idea of data-driven learning (DDL). In order to evaluate the effectiveness of these two systems, we designed pre-test and post-test assessment to compare participants’ performance on a sentence composition task and a paragraph paraphrasing task. The results indicate that students showed significant improvement using both GRASP and PREFER.

Metrics

1 Record Views

Details

Logo image