[1] Among the enduring controversies of historical linguistics is the status of the Neogrammarian hypothesis, first advanced in the 1870s, which holds that sound change proceeds without exception. [2] On this view, when a phonetic shift takes hold in a speech community, it applies with mechanical uniformity to every word in which the conditioning environment occurs, irrespective of a word's meaning, its grammatical role, or how often speakers utter it. [3] The apparent counterexamples that pepper any dictionary, the Neogrammarians insisted, are illusory: they arise not from a defective sound law but from the intrusion of borrowing or of analogy, the process by which speakers reshape irregular forms on the model of regular ones. [4] So austere a claim proved extraordinarily productive, for it licensed the reconstruction of ancestral languages from the residue of their regular correspondences.
[5] Yet the doctrine has never wanted for critics. [6] Beginning in the mid-twentieth century, several scholars documented changes that seemed to advance not across the whole lexicon at once but word by word, a pattern they termed lexical diffusion. [7] In such cases a new pronunciation establishes itself first in a handful of items and then, over generations, colonizes others, so that at any moment the vocabulary is divided between words that have shifted and words that have not. [8] Crucially, frequency appears to govern the order of march, though the direction of its influence is disputed: some studies find that the commonest words succumb earliest, others that they resist longest, entrenched by the very familiarity that shields them from reanalysis. [9] Neither result sits comfortably with the Neogrammarian premise, which grants frequency no role whatever.
[10] More recent commentary has tempered the opposition between the two accounts rather than declaring a victor. [11] A growing number of linguists suspect that regularity and diffusion describe successive phases of a single trajectory: a change may originate as a gradual, phonetically fine-grained adjustment that sweeps the lexicon in Neogrammarian fashion, only to be reinterpreted by later speakers as an abrupt substitution of one sound for another, at which stage it propagates word by word. [12] If this synthesis holds, the two camps have not been describing rival mechanisms so much as observing the same phenomenon at different points in its life. [13] The reconciliation is attractive, yet it remains provisional, for the historical record rarely preserves a change in the act of transition, and the linguist is left inferring the intermediate stages from endpoints alone.
