Two groups of 12-year-old ethnic minority (EM) users of alphasyllabary (66 Tibetan and 45 Yi) were compared with 42 Han Chinese students in comprehending Chinese narrative and expository texts, each with inferential questions requiring short open-ended written answers. Three constructs (verbal working memory, orthographic and sentential processing), each with two indicators, were hypothesized to predict text comprehension differentially in the three groups. A 43-item Students' Approaches to Learning (SAL) scale showed the EM students might not have developed effective strategies in learning Chinese. A task x group MANCOVA with SAL as covariate showed substantial differences in the students' performance. Multiple comparisons found group differences changed across tasks. Structural equation modeling, multiple regression analyses point to the significant and differential contribution of the constructs and the tasks to the two genres of Chinese text comprehension in the different groups. Educational implications include strengthening teaching the structure and function of Chinese characters, words, syntax and the need for adaptive curriculum materials.