Corpus Analysis of Collocation Errors and Corpus Informed Instruction Outcomes in Indonesian Junior High EFL Learners
Year: 2026
Type: article
Study design: Clinical Trial
Abstract: Lexical collocation errors persist among Indonesian Junior High School EFL learners (A1–A2 CEFR), mainly due to L1 transfer and limited authentic input. These errors appear in unnatural verb–noun (e.g., do mistake), adjective–noun (e.g., strong rain), and verb–preposition pairings. This sequential mixed-methods study investigates error patterns and the effects of corpus-informed instruction. Twenty learners generated a 10,000-word descriptive corpus, analyzed via AntConc 4.0 (MI > 3.0, 5L:5R span, min. freq. = 3) against British National Corpus benchmarks to spot deviant collocations. A six-session (60-min each) intervention used data-driven activities like concordance lines, matching games, and guided tasks targeting verb–noun and adjective–noun pairs. Pre/post-tests (30 items; α = 0.87–0.89) and questionnaires assessed outcomes. Errors dropped 53.7% (95 to 44), with verb–noun errors down 55.6% (45 to 20). Scores rose from M = 52.3 (SD = 9.2) to 72.1 (SD = 8.5); t (19) = −8.42, p < 0.001, d = 1.62. ANCOVA showed strong intervention effect: F (1,18) = 12.45, p = 0.002, η² = 0.41. Learners reported better error awareness and confidence. This advances corpus pedagogy for beginner EFL contexts in resource-limited settings.
Authors: Nurul Aeni, Yudhi Arifani, Morve Roshan Khalil
Institutions: Universitas Muhammadiyah Gresik, Southwest University
Language: English
Subfield: Developmental and Educational Psychology
Field: Psychology
Domain: Social Sciences
SDG: Quality education
Open Access status: diamond
OpenAlex ID: W7161574446