How can discrete grammatical rules produce gradual changes in the way a community speaks? This research article approaches that question by treating linguistic knowledge as a weighted distribution of competing grammars. Children encounter variable expressions, test grammatical possibilities against them, and gradually favor the grammars that successfully analyze their input. Internal constraints define the available possibilities, while external linguistic evidence shapes their relative weights. The discussion challenges models in which learners abruptly switch grammatical parameters. Evidence concerning children’s omitted subjects instead motivates a learning process that rewards successful analyses and penalizes unsuccessful ones. Drawing an analogy with population genetics, the article explains how stable grammatical alternatives can acquire changing frequencies without the alternatives themselves being transformed. A mathematical account then follows competing grammars across successive generations. Under its stated assumptions, differences in their compatibility with linguistic evidence determine which grammar spreads, producing the familiar S-shaped trajectory of historical change. The analysis also distinguishes changes in community input from isolated individual learning errors. The supplied historical discussion focuses on the loss of verb-second word order in French. Corpus statistics show how omitted subjects weaken the observable advantage of verb-second grammar, while constructions placing the verb later become more frequent. An application to Old English is announced. Together, these arguments connect child language development, grammatical variation, and historical textual evidence within a quantitative explanatory framework.