Introduction: Iconicity, the non-arbitrary mapping between a sign’s form and its meaning, is a pervasive feature of sign languages (Perniss, Thompson, & Vigliocco, 2010, Front. Psychol.). Despite the fact that an average of one third of the lexical signs of any sign language are assumed to be iconically motivated, previous work has shown that the meaning of even highly iconic signs is not transparent to sign-naïve individuals (Bellugi & Klima, 1976, Ann. NY Acad. Sci.; Trettenbrein et al., 2021, Behav. Res. Methods). Here, we aimed to understand whether the different morphophonological properties (e.g., reduplication, use of sign space; see Finkbeiner et al., 2026, Proc. Sinn und Bed.) of noun-verb pairs for concrete and abstract concepts in German Sign Language (DGS) provide sufficient iconic motivation for the correct assignment of grammatical category (i.e., noun or verb) by sign-naïve participants in a forced-choice paradigm. Methods: 82 individuals viewed 20 videos of a native deaf signer of DGS producing abstract and concrete verbs and nouns (5 of each category, randomized) in a sentential context. Participants categorized each stimulus as a noun or verb via keypress. After completing the experiment, participants were shown the stimuli again, accompanied by the German translation and correct grammatical category. Data from 67 primarily hearing, sign-naïve individuals (61 hearing, 6 Hard of Hearing; 56 non-signers, 11 beginner signers) ages 18-80 (mean: 35.6) was analysed; 15 participants were excluded due to missing demographic information or inability to consent to data processing. Individuals were recruited through a community-based approach, as the experiment was carried out during a scientific outreach event focused on sharing sign language and neurolinguistics research with the public. Statistical analyses were carried out in R (v.2026.04). Results: Pairwise comparisons via Wilcoxon signed rank tests revealed an effect of concreteness dependent on grammatical category. Concrete nouns and verbs were identified with greater accuracy than their abstract counterparts, but concrete nouns were identified with lower accuracy than concrete verbs. Signal detection analysis revealed that participants could not discriminate nouns from verbs above chance (d’ = 0.08, p = .30), regardless of category (concrete or abstract; p = .59). However, participants had a strong, systematic response bias towards verbs (c = −0.20, p < .001) driven by concrete stimuli (c = −0.45; compared to abstract stimuli = +0.04). Basic knowledge of a signed language did not appear to consistently bolster accuracy; participants with some sign language knowledge were significantly more correct than non-signers in identifying concrete nouns, but performed at or below chance, like non-signers, in all other categories. Age or hearing status did not influence accuracy. Discussion: The present findings are consistent with previous work on American and British Sign Language (Perlman et al., 2018, Front. Psychol.; Fitch et al., 2021, Cogn Sci.) which observed verbs to be the most transparent lexical class. Going beyond previous insights, our analysis suggests that sign-naïve participants do employ an iconic strategy for lexical category discrimination but over-generalize its application by incorrectly labelling both the nominal and verbal variants of concrete signs as verbs.