This investigation responds to the need for longitudinal data-driven research on additional-language (AL) acquisition by examining grammatical-gender marking among AL learners of French during a 21-month period, which included an academic year abroad (LANGSNAP corpus). The analysis of oral production consists of a generalized linear mixed model that examines a range of linguistic and extralinguistic factors shown to be important for gender marking in previous research, as well as a random effect for participant. Results show evidence of both change across time and consistency in the interlanguage. Drawing on variationism and usage-based approaches, we argue that longitudinal investigations that are focused on how learners use their additional language have much to offer our understanding of AL acquisition processes.