Background: Most oral language assessments measure competency in just one language, which risks describing multilingual children by their deficits as opposed to capturing their knowledge across languages (assets). Methods: The Multi-Language Assessment (MLA) was developed to be a brief, direct tool to capture young children’s expressive language skills across three languages. This study explored its use in a longitudinal design that followed kindergartners (N = 200, with exposure to three languages (English, Kiswahili, Kikamba) over 7 months in government schools in rural Kenya. Results: The MLA loaded on a single factor (0.21–0.60), was valid for the construct, stable, reliably administered (97% consistency), and informative as a supplement to measurements of reading achievement. There was sufficient power to describe individual changes in language use as well as predictive power. It explained 26% of the variance in letter sounds and 21% of the variance in spelling. Conclusions: It was most sensitive to showing language changes over time for children with middle and lower abilities. Children who could demonstrate complete or partial knowledge of concepts, improved their mastery and expressed some preference for mother tongue. The MLA fills a gap to capture the expressive language skills of young children in multilingual environments.