Linguists generally assume that all languages have some words for parts of the human body such as ‘head’, ‘hands’, ‘mouth’ and ‘legs’, but it is not so widely agreed that speakers of all languages can speak—or even consciously think—of the designata of such words as ‘parts of the body’. In particular, it has been claimed that the Australian language Warlpiri lacks any suitable lexical equivalent of ‘part(s)’. Using data from the Warlpiri English Encyclopedic Dictionary, this study contests this claim, arguing that the relevant sense of ‘part’ exists in Warlpiri as one sense of the polysemous closed-class item yangka (whose main meaning can be stated, roughly, as ‘that one, you know the one’). Our argument is framed within the Natural Semantic Metalanguage (NSM) approach to meaning analysis, according to which meanings are stated as substitutable, cross-translatable paraphrases. NSM researchers have long maintained that PART(S) is a universal semantic prime, i.e. an indefinable meaning expressible by words or phrases in all human languages. The study also considers broader issues to do with semantic theory, polysemy and translation.