variationbut

Variationbut is a methodological framework used in comparative linguistics and digital humanities to analyze systematic lexical or phonological changes while constraining the scope of variation examined. The term was coined in 2018 by Michael Varian in his study of Old English dialectal shifts, where he sought a way to catalog variations without conflating them with unrelated linguistic phenomena. Variationbut differs from traditional variationist approaches by explicitly limiting the analyst to a set of hypothesized sources of change—such as phonetic environment, morphological context, or contact influence—before allowing cross‑dialect comparison.

In practice, a variationbut analysis begins with a corpus of closely related texts. Researchers isolate a target feature (for example, the vowel shift in the North–South dialect boundary) and then generate a series of controlled test cases that exclude potential confounds such as orthographic conventions or scribal errors. The resulting data set is then mapped onto a model that predicts the frequency of variation under each constraint. If the observed frequencies significantly diverge from the model, the variation is considered to have additional explanatory variables, prompting further refinement of the hypotheses.

The approach has been applied far beyond Old English, including in studies of Arabic morphological change, Spanish vowel shifts in colonial texts, and even in computational phenotyping of genetic data where gene expression variation is constrained by known regulatory networks. Proponents argue that variationbut provides a reproducible, hypothesis‑driven method that reduces the risk of over‑interpretation, while critics point out the potential for oversimplification if too many constraints are imposed. The balance between restricting variation and capturing genuine linguistic change remains a key discussion point in contemporary variationist research.