Morphological awareness in arithmetic word-problem solving among Chinese early adolescents

Arithmetic word-problem solving requires translating and integrating textual information into mathematical representations and executing solutions, differentiating it from pure calculation due to its reliance on text comprehension. Prior work shows reading subskills correlate with word-problem performance in younger students. This study focuses on morphological awareness—a structural language skill involving analysis and manipulation of morphemic structures—as a potential contributor to word-problem solving among Chinese early adolescents (Grades 5–6). The goals were to (a) integrate morphological awareness into models of arithmetic word-problem solving to test linguistic demands and (b) assess whether relations between reading subskills and word-problem solving persist into late elementary years. The authors hypothesized that Chinese morphological awareness would uniquely predict arithmetic word-problem-solving abilities in early adolescents.

Chinese relies heavily on compounding morphology, with 75–80% of words composed of multiple morphemes; thus, morphological awareness supports decomposition, morphemic structure analysis, and meaning inference. Prior studies show morphological awareness facilitates Chinese children’s vocabulary, character reading, and reading comprehension. Interface arguments posit parallels between Chinese compounding and the base-10 compounding number-naming system (e.g., 十一 ‘ten-one’ for 11), suggesting morphological awareness may support mathematical vocabulary acquisition and numerical processing. Longitudinal evidence with preschoolers and kindergartners (e.g., Zhang & Lin, 2015; Liu et al., 2016; Liu et al., 2020) shows unique contributions of morphological awareness to counting, arithmetic skills, and word-problem solving, even after accounting for age, IQ, visual-spatial skills, and phonological awareness. In elementary students, morphological awareness predicts word-problem solving directly and indirectly via inference-making and comprehension monitoring (Ng et al., 2021). Collectively, literature indicates morphological awareness bolsters both reading comprehension and the translation of text to equations, motivating investigation in older children transitioning to ‘reading to learn.’

Participants: 668 typically developing Chinese students (Grades 5: n=330; Grade 6: n=338; 379 boys, 289 girls; mean age = 11.3 years) from two public schools in coastal southeast China, instructed in Mandarin. Ethical approval obtained; informed consent collected.

Measures:

• Receptive vocabulary (control): A checklist (80 two-character items; 64 real, 16 nonwords) adapted from Anderson & Freebody (1983). Students selected known words; performance computed via signal detection–based index using real hits and false alarms. Scores converted to percentages (0–100).

Morphological awareness (three subtests; 20 items each):

• Morpheme recognition (Ku & Anderson, 2003): Judge whether a component character/word relates semantically to a compound (e.g., 小心 vs 小). Reliability α = 0.725.
• Morpheme discrimination (Ku & Anderson, 2003): Identify the odd compound with a different morphemic meaning among three sharing a morpheme (e.g., 白米, 小米, 厘米). Reliability α = 0.734.
• Compound structure awareness (Liu & McBride-Chang, 2010): Choose legitimate novel compounds based on prompts across subject–predicate, verb–object, subordinate, and coordinate structures (e.g., 蛙跳 vs 跳蛙). Reliability α = 0.749.

Arithmetic word-problem solving: 12 problems spanning addition, subtraction, multiplication, division, and fractions; controlled for linguistic and numerical complexity. Text length: 37–57 characters (mean 44). Linguistic complexity indices (number of characters/words, difficult words, sentences, content words, log mean content word frequency) computed using CRIE; Grade 5 pool: 473 chars, 337 words, 82 difficult words, 38 sentences, 306 content words, log mean 1.82; Grade 6 pool: 419 chars, 302 words, 85 difficult words, 27 sentences, 260 content words, log mean 1.87. Numerical complexity controlled by multi-step and operation difficulty; all two-digit operations. Two subscales:

• Equation formation: Translate text to equations/operations. Scoring: 2 points per correct equation; no partial credit. Total 24; α = 0.820.
• Arithmetic computation: Compute correct answers. Scoring: 2 points per correct; no partial credit. Total 24; α = 0.822.

Procedure: Paper-and-pencil tasks administered in classrooms; tasks counterbalanced across rooms; separate administrations to avoid carry-over; 60 minutes total for all five tasks; data collection over 8 weeks.

Analytic approach: Descriptive statistics, correlations, hierarchical multiple regressions (controlling age and receptive vocabulary), and covariance-based path analysis. Assumptions checked: Normality (Shapiro–Wilk; morpheme discrimination square-root transformed), homoscedasticity, multicollinearity (VIF 1.08–1.19).

• Descriptives indicated adequate variability across measures; vocabulary showed wide dispersion due to percentage scaling.
• Equation formation and computation were extremely highly correlated (r = 0.987, p < 0.001), suggesting accurate equation formation virtually guaranteed correct solutions.
• Receptive vocabulary correlated with both equation formation and computation (r ≈ 0.367 and 0.366, p < 0.001). Morphological awareness facets correlated with word-problem indicators (r range 0.164–0.320, all p < 0.001).
• Hierarchical regressions: After controlling for age and receptive vocabulary, morphological awareness facets explained additional variance in word-problem solving: ΔR² = 0.047 (equation) and 0.048 (computation); overall R² ≈ 0.18 for both models; F ≈ 12.15–12.44, p < 0.001. Unique predictors: morpheme recognition (β ≈ 0.18, t ≈ 4.56–4.71, p < 0.001) and morpheme discrimination (β ≈ 0.10, t ≈ 2.54–2.58, p < 0.05). Compound structure awareness was not significant (β ≈ 0.04, ns). Age was not significant (β ≈ 0.04, ns). Receptive vocabulary was a strong predictor (β ≈ 0.36–0.37, p < 0.001).
• Path analysis (standardized estimates): Receptive vocabulary predicted equation formation (β = 0.279, p < 0.001) and computation (β = 0.277, p < 0.001). Morpheme recognition predicted equation (β = 0.192, p < 0.001) and computation (β = 0.196, p < 0.001). Morpheme discrimination predicted equation (β = 0.080, p = 0.033) and computation (β = 0.080, p = 0.033). Compound structure awareness did not predict either outcome (equation β = 0.042, p = 0.248; computation β = 0.040, p = 0.272). Age was non-significant in both paths.

Findings support the hypothesis that Chinese morphological awareness uniquely contributes to arithmetic word-problem solving in early adolescents, extending prior evidence from preschool and early elementary samples. Results align with text comprehension frameworks (dual representation and construction–integration models), indicating that structural linguistic skills facilitate constructing a textbase and translating word problems into mathematical representations. The near-identical performance across equation formation and computation suggests that, at this age, successful equation translation is the principal bottleneck rather than calculation per se. Among morphological facets, recognition and discrimination of compounds were uniquely predictive, consistent with the importance of extracting and mapping sublexical morphemic meanings in multimorphemic words. In contrast, compound structure awareness did not show direct effects on word-problem outcomes, implying that sensitivity to compounding structures may relate more to early counting and basic number naming than to forming propositions and equations in complex problem texts for adolescents.

The study demonstrates that morphological awareness—specifically morpheme recognition and discrimination—contributes uniquely to arithmetic word-problem solving among Chinese early adolescents beyond age and receptive vocabulary. This highlights the language-specific role of Chinese morphological processing in mathematical text comprehension and problem translation. Educationally, integrating morphological awareness training with mathematics instruction may enhance students’ ability to interpret problem texts and form correct equations. Future research should use longitudinal designs to track developmental trajectories and include broader cognitive and linguistic covariates to further test theoretical models of word-problem solving.

• Cross-sectional/concurrent design limits causal inferences; longitudinal data are needed to assess developmental pathways and stability of effects.
• Limited variable set; future work should include orthographic knowledge, syntactic knowledge, working memory, and general text comprehension to better isolate linguistic and cognitive contributions.
• Although linguistic and numerical complexity were controlled, external validity across diverse curricula and problem types warrants further examination.