자동화된 어휘 분석 프로그램을 활용하여 어휘의 정교화와 쓰기 능력간의 관계 탐색

Abstract

This study explores the relationship between newly developed indices of lexical sophistication and essay scores written by Asian college students. The study employs the Tool for the Automatic Analysis of Lexical Sophistication(TAALES) program. A total of 440 essays written by EFL students from Korea, China, Japan, Thailand, Taiwan, and Indonesia were selected from the ICNALE(International Corpus Network of Asian Learners of English) corpus. The following two research questions were addressed. First, what is the best lexical sophistication predictor of essay scores? Second, what are differentiating lexical sophistication indices among high-, mid-, and low-level EFL students? Two statistical analyses such as the stepwise regression analysis and the discriminant function analysis were performed using the SPSS version 26.0. Results showed that ‘age of acquisition content words’ turned out the best predictor of essay score, indicating that essays with content words that on average are perceived to be learned later received higher scores. Also, two multiword units such as ‘COCA magazine trigram strength of directional association’ and ‘COCA magazine bigram mutual information’ contributed to distinguishing among EFL students with different English proficiency levels. Therefore, EFL students who produced two-word and three-word phrases with stronger associations were classified as the high proficiency group, hence judged to be more proficient in writing.