Introduction: Does grammatical gender color everyday categorization outside explicit language use? We tested whether grammatical gender, semantic category, animacy, and task framing (personify vs. generic instructions) guide adjective choices in an all-English task among native Greek (three-gender system) and native English speakers.
Method: In a properties-judgment task, 24 black-and-white photos (balanced by grammatical gender in Greek; natural vs. artifact; animate vs. inanimate) were presented. Participants (Greek n=40; English n=33) chose three adjectives per photo from five "feminine" and five "masculine" terms; between-subjects factors were language group and instruction.
Results: There was no Language by Grammatical Gender interaction. We found a clear main effect of grammatical gender. Items that are grammatically masculine elicited more feminine adjectives than items that are grammatically feminine, with this pattern most evident under generic instructions. Natural images drew more feminine descriptors than artifacts, and this effect was stronger for Greek speakers. Inanimate images drew more feminine descriptors than animate images, and this effect was stronger for English speakers.
Conclusions: In an unlabeled task conducted entirely in English, the usual differences between language groups that are attributed to grammatical gender did not appear. Instead, responses aligned with semantic category, animacy, and small changes in the instructions, which fit the idea of "thinking for speaking" and broader gender stereotype schemas that shape choices. For applied cognitive psychology and assessment across cultures, these findings caution against assuming language group effects in tasks without explicit linguistic labels, and they show that modest wording changes can meaningfully shift responses. Practical uses include test design across cultures, HCI prompt writing that is sensitive to bias, and communication training.