Transplanted languages are widely held to differ from their sources, but synchronic differences are silent on the existence, time and locus of change. When the discrepancies involve variants that are salient and stigmatized, the transplanted variety is usually implicated as the innovator, especially if it has been in sustained contact with another language. Such is the case of Canadian French, whose myriad differences from European French are widely attributed to influence from the majority language, English. Determination of whether the features in question are in fact innovations, or rather retentions, and if the former, whether contact-induced or independent developments, requires a real-time benchmark predating the current situation. Written texts, the standard fare of historical linguists, turn out to be of limited use in this context, since the vernacular features of interest are either grossly underrepresented or not attested at all. The most appropriate benchmark is an oral precursor, but such precursors necessarily have a time depth too shallow to chart the full course of change.
In this paper I describe a series of diachronic corpora, each of which represents at least one aspect of the speech of earlier times, and the novel uses we make of them in our efforts to trace the locus and trajectory of change. Each has complementary sources of error, but interpreted in conjunction with synchronic age distributions, they converge in showing, for a series of morphosyntactic variables, that change tends to be very slow, infrequent, highly circumscribed and independent of language contact. Rather, they suggest that many of the synchronic discrepancies may result from post-migration changes in the reputedly conservative European French source, while vernacular Canadian French has remained remarkably close to its roots.