Iconicity and transparency in Dutch Sign Language.

Iconicity in sign languages, visually motivated relationships between form and meaning, is closely tied to transparency, or the guessability of meaning based on its form. Yet, how to best measure this relationship remains contentious. This study compares iconicity ratings and transparency scores for 1412 lexical signs in Dutch Sign Language (Nederlandse Gebarentaal, NGT) across deaf NGT signers and Dutch and German non-signers. In this paper, we show that iconicity is perceived differently by the different groups of raters and replicate past findings that signers perceive signs from their own sign language as more iconic than non-signers do. No such differences are found between the two groups of non-signers. Our results show that Dutch non-signers' performance on a transparency task is best predicted by iconicity ratings provided by participants from the same population and by non-signers more broadly. This has important implications for psycholinguistic research, as we show that there is no single "correct" population providing the best iconicity ratings. Instead, rating participants should be matched to the target populations on relevant dimensions to meaningfully predict outcomes in psycholinguistic studies. With this paper, we provide an extensive database of NGT signs with iconicity ratings and transparency scores, available as a reference database for Dutch Sign Language research. The database is aligned with glosses and item-IDs used in the NGT Signbank and Corpus projects, allowing for easy combination of our rating data with phonological and semantic information provided by those databases.