L1 Interference in Students’ Translations: A Corpus-Based Analysis of Collocation Errors and Pedagogical Implications

Abstract

Collocation is challenging for translators as the words differ across languages. This study examines collocation errors in translations produced by English Department students who took the Indonesian-English Translations subject. The text discussed the Minangkabau tradition written in Indonesian. Employing a corpus-based approach, the research analyzes students' translations to identify recurring collocational mismatches, classify error types, and first-language (L1) interference in English as an L2, which is studied in terms of the insufficient mastery of English phraseology. The translations made by English Department students were compiled from student submissions and compared against reference corpora (e.g., COCA, BNC) as natural collocations. Tools such as AntConc and Kortara were used to quantify deviations and categorize errors into four primary types: (1) verb-noun, (2) adjective-noun mismatches, (3) unnatural noun-noun phrases, (4) adverb-adjective, (5) verb-preposition, and (6) clause base. Findings reveal that 58.94% of errors arise from noun-noun combinations, where students applied Indonesian syntactic or lexical patterns to English, resulting in unnatural collocations. The study highlights the pedagogical need for explicit collocation instruction in translation training, especially for language pairs with significant structural and cultural differences. It advocates for incorporating corpus tools into classrooms to enhance students’ awareness of natural collocations and reduce L1 interference.