Traditionally in sign language research, the issue of whether a lexical sign is articulated with one hand or two has been treated as a strictly phonological matter. We argue that accounting for two-handed signs also requires considering meaning as a motivating factor. We report results from a Swadesh list comparison, an analysis of semantic patterns among two-handed signs, and a picture-naming task. Comparing four unrelated languages, we demonstrate that the two hands are recruited to encode various relationship types in sign language lexicons. We develop the general principle that inherently “plural” concepts are straightforwardly mapped onto our paired human hands, resulting in systematic use of the two hands across sign languages. In our analysis, “plurality” subsumes four primary relationship types —interaction,location,dimension, andcomposition— and we predict that signs with meanings that encompass these relationships — such as ‘meet’, ‘empty’, ‘large’, or ‘machine’ — will preferentially be two-handed in any sign language.