In linguistics, a principled definition of what constitutes a ‘language’ in opposition to a ‘dialect’ has been notoriously elusive. The intelligibility criterion, possibly the only criterion that could form the basis of such definition, has often been considered inadequate, leading to the widespread conclusion that languages may not be linguistically definable objects at all (e.g. Chambers and Trudgill, 1998). This paper reconsiders some of the objections typically raised against the intelligibility criterion and argues that one of these objections — namely that intelligibility is a scale to which no meaningfully discernible segmentation may be applied— can be formulated as a testable empirical claim. Three experiments are then presented with the explicit aim to test this claim. Results indicate that, contrary to what has been frequently claimed, the intelligibility scale does allow for potentially meaningful segmentation, providing empirical evidence in favour of adopting intelligibility as an empirically sound criterion of demarcation for the identification of languages and dialects.