This study compares Spanish morphosyntax error types and magnitude in monolingual Spanish and Spanish–English bilingual children with typical language development (TD) and language impairment (LI). Performance across groups was compared using cloze tasks that targeted articles, clitics, subjunctives, and derivational morphemes in 57 children. Significant differences were observed between bilingual TD and LI groups on all tasks; however, no differences were observed between bilinguals with TD and monolinguals with LI except on a sum-score across all tasks. There were no observed differences between bilinguals and monolinguals with TD; however, 60% of bilinguals with TD were misclassified as LI when using a cut score derived from monolingual-only data. Results support evidence that Spanish morphosyntax is vulnerable to error in monolingual and bilingual Spanish–English children with LI. However, the grammatical deficit seems clinically relevant only when children are compared to the same language peer group (i.e., bilinguals compared to bilinguals).

This study examined object clitic pronouns (OCPs) and verb inflections in twenty-five school-age children with typical development (TD) and twenty children with bilingual language impairment (BLI). MANOVA and ANOVA were used to explore differences according to grade level and language status (TD vs. BLI). Although children with BLI produced higher rates of grammatical errors overall, accuracy on number and gender assignment for OCPs was better for both groups in the higher grades. Although the rate of verb inflection errors did not differ for children with TD and BLI in the lower grades, a significant interaction yielded higher error rates on subject–verb agreement for third person singular and plural inflections in the later grades for children with BLI. Greater accuracy on OCP use in later grades weakens claims that bilingualism exacerbates language impairment. For BLI, whether incomplete acquisition or delayed development is the determining factor for verb inflection errors remains undetermined.