Frequency, Dispersion and Abstractness in the Lexical Sophistication Analysis of A Learner-Based Word Bank: Dimensionality Reduction and Identification

The current study incorporated a number of lexical sophistication indices including frequency, dispersion and abstractness of words. A learner-based word bank (inclusive of a Chinese middle-school vocabulary list, a Chinese high-school vocabulary list and a Chinese college-English-test vocabulary list) was manually coded based on two existing corpora: Corpus of Contemporary American English (COCA) and British National Corpus (BNC). Indices of frequency, dispersion and abstractness of the word bank were analysed to shed light on the predetermined categorization of lexical sophistication among second language learners. Based on the principal component analysis, the results demonstrated that dispersion was a unique factor loaded on all entered eight variables while word frequency and abstractness were extracted by the same factor in the learner-based word bank. Moreover, a follow-up MANOVA analysis with post hoc comparisons showed that lexical sophistication indices in general produced pronounced differences among the three levels of word lists. More critically, dispersion was found to be the only significant indicator to differentiate the three levels of word lists. Discussion centred on the uniqueness of dispersion in lexical sophistication and the shared algorithm in frequency and abstractness.