In two studies, we tested the hypothesis that not having a potentially group-defining attribute (e.g., in-group language) can affect social identification and support for group goals (e.g., national autonomy). Focusing on the Welsh minority in the UK, Study 1 provided evidence that Welsh language fluency predicted Welsh identification and support for national autonomy, and that identification accounted for the language–autonomy association. Study 2 extended this by (1) examining British and English as well as Welsh identification; and (2) quasi-manipulating the surrounding context (Welsh speaking vs. non-Welsh speaking). As predicted, low Welsh language fluency predicted stronger British and English identification, but only where language was criterial (Welsh-speaking regions). British identification, in turn, predicted lower support for national autonomy. Implications and prospects for future research are discussed.