Testing the Relationship of Multidimensional Linguistic Complexity Features to Textbook Volume in Chinese College English

Readability is a key concept for evaluating the difficulty and applicability of foreign language textbooks, although the present study does not measure reader-based readability directly. Traditional readability formulas, which primarily rely on surface-level features such as word and sentence length, are limited in capturing deeper linguistic features. This study analyzed 96 texts from Volumes 1–4 of Contemporary College English (3rd ed.; Yang, 2024), using textbook volume as an ordinal proxy for intended pedagogical level. Twenty-nine linguistic indices across lexical, syntactic, and discourse dimensions were extracted. One-way ANOVA and Kruskal–Wallis tests identified 17 nominal between-volume differences. An approximate Benjamini–Hochberg correction applied to the 29 rounded omnibus p values retained nine differences: four lexical and five syntactic, with none among the three discourse indices. F15 number of word types showed a substantial unadjusted association with volume (?² = .283), but this raw count is confounded by text length and is not a length-adjusted lexical-diversity measure. F17 attested constructions (?² = .169) and F24 mean dependency distance (?² = .140) also varied across volumes. The three-predictor stepwise linear model explained 43.3% of the variance in the numeric volume code; because the model is data-driven, the outcome is ordinal, and the LOOCV implementation does not establish nested feature selection, the model is interpreted as an exploratory in-sample association rather than validated out-of-sample performance. The findings indicate selected linguistic differences across textbook volumes, not independently validated readability or uniform difficulty progression. Length and genre effects and reader-based readability remain outside the scope of the present analysis.