When discussing language histories, historical grammars traditionally tend to assume presence of sound change when it surfaces in texts, sometimes without questioning (a) systematicity within the sound system and intersection with morphosyntax, (b) systematicity of distribution across the language area in time and space and in relation to register, and (c) systematicity relative to the writing system. This paper takes up the much discussed but never resolved history of the *g > *γ (> h) lenition in Czech, reconsiders critically the proposed arguments, proposes a new way to approach the sparse data set, and draws concrete and principled methodological conclusions applicable to other instances of reconstruction. Spatial analysis of the attestations and different kinds of distributional phenomena are performed digitally and discussed relative to their time depth and in the context of linguistic and extralinguistic evidence from the same periods. In the context of Czech historical phonology, this paper offers methodological considerations for evaluating inner-linguistic and extra-linguistic distribution in reconstruction.
Jadranka Gvozdanović (Wed,) studied this question.