TOCFL學習者書面語料庫中關於日籍學習者對華語能願動詞「會」的「遺漏」現象探究

An Exploration of the“Omission”Phenomenon in Japanese Learners' Use of the Chinese Modal Verb Huìin the TOCFL Learner Written Corpus

本研究探討初中級日籍華語學習者在書面語中對能願動詞「會」之「遺漏」（omission）現象，以補足現有文獻中對「非能力」語意（如，未來、推測、習慣用法）遺漏類型之細部分析不足。雖然過去曾有論文指出，日籍學習者學習「會」時，非能力用法的遺漏偏誤最為顯著，然而系統性分類與統計證據尚不充足。本研究採用語料庫方法，以TOCFL學習者書面語料庫中日籍學習者的作文為研究對象，分析「會」的語例773筆。研究將「會」依語意功能分為四類：「未來用法」（對應日語中無標記形式）、「推測用法」（對應日語darou標記但常被省略）、「習慣用法」（對應日語中習慣性標記如monoda類亦易省略）、「能力用法」（對應日語可能形，其標記通常不可省略）。分析指標包括正確率、偏誤率與誤正比，並透過卡方檢定比較特定語境（如「包孕句」）中推測用法之遺漏與正確使用率。研究結果顯示：第一點，能力用法學習最為穩定，正確率最高；推測用法與習慣用法次之；未來用法為最弱項。第二點，在包孕句語境中，推測用法的正確使用顯著下降。本文建議教學者和日籍學習者應該特別注意這些母語遷移造成的學習困難點（如，「日語與華語情態系統之差異」、「包孕句中的推測用法」）。

This study investigates the omission of the modal verb huìin the writing of lower-intermediate Japanese learners of Mandarin Chinese. Existing literature has noted that non-ability meanings of huì(e.g., future, epistemic inference, and habitual readings) are frequently omitted by Japanese learners; however, deta iled classification and statistical evidence remain insufficient. Using a corpus-based approach, this study analyzes 773 instances of huìdrawn from compositions by Japanese learners in the TOCFL Learner Corpus. The analysis divides huìinto four semantic functions: future (corresponding to the unmarked form in Japanese), epistemic inference (corresponding to darou, though often omitted in actual use), habitual (corresponding to Japanese habitual markers such as monoda, likewise frequently omitted), and ability (corresponding to the Japanese potential form, whose marking is typically obligatory). Quantitative indices include accuracy rates, error rates, and over-marking ratios. Chi-square tests are further employed to compare omission patterns in specific environments, such as epistemic huìwithin embedded clauses. The results show two major tendencies. First, the ability use exhibits the most stable acquisition, with the highest accuracy rate; epistemic and habitual uses show intermediate performance; and the future use is acquired least accurately. Second, in embedded clause contexts, accurate use of epistemic huìdrops significantly. The findings highlight learning difficulties arising from L1–L2 mismatches in markedness and modal systems, particularly regarding the“unmarked future”and the weakened markedness of epistemic expressions inside embedded clauses in Japanese. Pedagogical implications for teachers and learners are also proposed.