The present study investigates the semantics involved in the L2 English article use by Korean EFL learners. The examined usage data were extracted from the researcher compiled learner corpus, the Kyungpook National University Student English Learner Corpus-Writing (KSELC-W). The importance of current research is in the fact that the investigated data was not collected through formal elicitation tasks or grammar focused activities. Therefore, the learner corpus data represent Korean EFL learners' unconscious knowledge of the English article system and their performance on timed writing sessions which did not focus on grammar. The findings reveal that Korean EFL learners do not make random misuse or overuse errors in production of English articles the and a(n) and that the patterns are in accordance with the Fluctuation Hypothesis. Additionally, results show that the semantic feature specificity effects the Korean EFL learners in the current study and provide supporting data for direct UG access underlying article choice. These results support previously reported findings by Ionin et. al. (2004, 2009) and Ko et al. (2009).

1. Introduction
 2. Theroretical Background
  2.1. Article Semantics: Definiteness and Specificity
  2.2. L2 English article acquisition
 3. Method
  3.1. Learner Corpus and Participants
  3.2. Software
  3.3. Semantic context classification
 4. Results and Discussion
 5. Limitations and Conclusion
 References

• Daria Soon-Young Seog(Kyungpook National University)