Purpose: Classifying children into two language ability groups, with and without language impairment, may underestimate the number of groups with distinct language ability patterns, or, alternatively, there may be only a single group characterized by a continuum of language performance. The purpose of the current study was to identify the number and characteristics of latent (unobservable) language ability groups in an unclassified sample of predominantly Spanish­speaking children. Method: An unclassified sample of 431 predominantly Spanish-­speaking 5-­ to 7­-year-­olds learning English participated in the study. The groups were identified on the basis of (a) language sample analyses (semantic, grammatical, and sentence-­length measures); (b) language processing tasks (phonological working memory and processing speed measures); and (c) nonverbal cognitive abilities assessed using a standardized measure. All tasks were administered in Spanish. Latent profile analysis was used to examine the number and nature of distinct language ability groups in the unclassified sample. Results: Results indicated that a three-group model best represented the data, characterized by low grammaticality in one group, low phonological working memory in another group, and average skills in a third group. Conclusion: Classifying children into two groups, those with and without language impairment, may lead to misidentification of language impairment.