Change in the context of variation

It is clear that native-learner errors must survive in the speech of individuals until the end of the developmental window for acquisition in order to become language changes; that is the first bottleneck or filter through which they must pass, so to speak. But if they are to “succeed” as linguistic changes, learner errors must then become accepted variants in the system of linguistic variation in the adult speech community; that is the second external constraint on language change (see Weinreich et al. 1968: 99–102). Virtually nothing is known about that process, for the reasons discussed in the preceding chapter; we can only infer that for social reasons operating on the level of individual relationships some idiosyncrasies of speech are not only tolerated but imitated, and that for structural reasons some innovations are inherently more likely than others to succeed. In any case, a change that begins to spread has crossed an important threshold in its development, because it thereby acquires a significant identity in the speech of the community at large.

From that point forward linguistic change occurs in the context of variation unless and until an innovation becomes universal in a speech community, when it is said to have “gone to completion.” During that part of its trajectory one can study the change only by studying the variation in which it participates – a type of study pioneered and still led by William Labov.