Linguistic Complexity and Lexical Cohesion of Indonesian-Language State Textbooks Across Two School Transitions: A Fixed-Corpus Computational Audit

Abstract

Background. Textbooks set much of the reading demand students meet at the primary–junior and junior–senior secondary transitions, yet Indonesian evidence rests on readability graphs of short samples.

Objective. To compare the linguistic complexity and lexical cohesion of reading passages in national Bahasa Indonesia textbooks across the Grade VI–VII and IX–X transitions.

Methods. A fixed-corpus computational audit of one revised student book per grade (SIBI catalog). Twenty-four passages (six per book; 12,077 words; 1,015 sentences) were frozen before analysis and parsed with UDPipe (Universal Dependencies 2.5). Seventeen indices were aggregated with equal genre weights and compared with exact stratified permutation tests, genre-pooled t intervals, mixed models, and permutational multivariate analysis of variance (PERMANOVA).

Results. No primary index differed at either transition: mean sentence length changed by +0.92 words from Grade VI to VII (95% CI −3.50 to 5.33; p = 0.576) and by −1.01 words from IX to X (95% CI −5.42 to 3.41; p = 0.662), and the moving-average type–token ratio (MATTR-50) by +0.0180 (p = 0.410) and −0.0289 (p = 0.176). Grade order predicted no index (all p ≥ 0.097). Books explained 1.0%–4.4% of sentence-level variance, passages within book and genre 5.2%–11.6%. Exploratory: every Grade VII passage had wider sentence-length dispersion than every Grade VI passage (exact p = 0.005; Holm-adjusted p = 0.124).

Conclusion. Within these books, the selected passages do not form a grade-ordered complexity ladder; variation lies within books and passages. The evidence supports passage-level description of language demand; effects on students' comprehension require reader studies.