This article addresses different ways scholars conceptualize the multilingual repertoire. Formal and functional approaches investigate crosslinguistic influence across clearly demarcated languages, while dynamic systems theory and translanguaging approaches question countability and boundedness of languages. Problems with both perspectives are discussed, and a unified approach that relies on the idea of languages as natural categories is presented as a form of conceptual reengineering. It embraces the possibility and the meaningfulness of the identification of languages in the repertoire while also accounting for their continuous, graded nature and the lack of clear boundaries of language categories. The proposal accounts for seemingly inconsistent findings on the effects of bi- and multilingualism on cognition and on language learning. Multilingualism as a field of inquiry is the investigation of an ordinary phenomenon; what is special about multilingualism is that multilingual speakers may have an extraordinary amount of language and linguistic variability at their disposal.