This paper reconsiders the relationship between the study of dialect variation and the explanation of language change. After an introductory section on the relevance of dialectology and the use of dialect data in diachronic linguistics( I), I move on to discuss the alternative between internal and external explanation for language change ( II). It is suggested that much current work in formal linguistics uses dialect data, if at all, without exploiting fully their potential import. This is largely the product of a well-entrenched a priori preference for internal vs. external explanation, which atomizes dialects reducing them to mere collections of parametric options. To show this, the rest of the paper discusses a case study drawn from variation over time and space in the pronominal clitic systems of some Italo-Romance varieties( III). A recent account of the changes involved is discussed( IV), which puts forward a purely internal explanation based on a formal theory of syncretism. In V, I argue for an alternative explanation, which crucially relies on factors external to morphology-phonetic, on the one hand, and geographical and sociolinguistic, on the other hand-and proves superior for both empirical and theoretical reasons.