Spatial language in signed language is assumed to be shaped by affordances of the visual-spatial modality –where the use of the hands and space allow the mapping of spatial relationships in an iconic, analogue way– and thus to be similar across sign languages. In this study, we test assumptions regarding the modality-driven similarity of spatial language by comparing locative expressions (e.g., cup is on the table) in two unrelated sign languages, TİD (Türk İsaret Dili, Turkish Sign Language) and DGS (Deutsche Gebärdensprache, German Sign Language) in a communicative, discourse context. Our results show that each sign language conventionalizes the structure of locative expressions in different ways, going beyond iconic and analogue representations, suggesting that the use of space to represent space does not uniformly and predictably drive spatial language in the visual-spatial modality. These results are important for our understanding of how language modality shapes the structure of language.