Longitudinal Language Growth Stagnation as a Predictor of Academic Achievement Among Middle School Multilingual Learners: Evidence From a Seven-Year Pilot Study

Early identification of English Learner (EL) students at risk of academic underperformance remains a critical unresolved challenge in U.S. public education. Current early warning systems are predominantly retroactive, relying on year-end standardized assessment results that confirm academic failure rather than enabling preventive response. This paper presents findings from a seven-year longitudinal pilot study conducted at a Virginia public middle school (N = 118 Research_ID students; 47.5% ESOL, 14.4% SPED) that developed and validated a predictive analytics framework — the Longitudinal Growth Stagnation Framework (LGSF) — for early academic risk identification among multilingual learners. The framework operationalizes annual WIDA ACCESS proficiency growth trajectories, a novel Stagnation Load Index, a five-signal Composite Risk Score, and empirical k-means trajectory clustering into a single, district-deployable early warning system. Primary results demonstrate that language growth stagnation — defined as annual WIDA composite growth below 0.3 levels — is significantly associated with lower Virginia SOL Reading performance (M = 338.8 vs. 370.3; Welch's t = 4.619, p = 6.68 × 10⁻⁶, d = 0.619). Assessment data available in August predicted quarterly academic outcomes throughout the academic year (r = 0.48–0.70, all p < .001), establishing an eight-month prospective identification window. ESOL students demonstrated a 16–22 percentage point gap relative to non-ESOL peers across all quarterly assessments (d = 1.09–1.70). The framework requires only data routinely collected by all 44 WIDA-member states and is directly replicable in districts serving the 5.3 million English Learner students enrolled in U.S. public schools.