Purpose: Identification of children with specific language impairment (SLI) can be difficult even though their language can lag that of age peers throughout childhood. A clinical grammar marker featuring tense marking in simple clauses is valid and reliable for young children but is limited by ceiling effects around the age of 8 years. This study evaluated a new, more grammatically challenging complex sentence task in children affected or unaffected with SLI in longitudinal data, ages 5-18 years. Method: Four hundred eighty-three children (213 unaffected, 270 affected) between 5 and 18 years of age participated, following a rolling recruitment longitudinal design encompassing a total of 4,148 observations. The new experimental grammaticality judgment task followed linguistic concepts of syntactic sites for finiteness and movement within complex clauses. Growth modeling methods evaluated group differences over time for four different outcomes; three were hypothesized to evaluate optional omissions of overt finiteness forms in authorized sentence sites, and one evaluated an overt error of tense marking. Results: As in earlier studies of younger children, growth models for the SLI group were consistently lower than the unaffected group, although the growth trajectories across groups did not differ. The results replicated across four item types defined by omissions with minor differences for an item with an overt error of tense marking. Covariates of child nonverbal IQ, mother's education, and child sex did not significantly moderate these effects. Conclusion: The outcomes support the task as having potential screening value for identification of children with SLI and are consistent with linguistic interpretations of task demands.