Gender-biased communication is pervasive in digital environments. since it can reproduce and amplify socially established beliefs about women and men. This chapter presents an exploratory analysis of gender-associated English vocabulary by combining lexical data from the online Cambridge Thesaurus with computational measures of sentiment classification and online prominence. A set of 47 terms was obtained from the thesaurus entries for manly, masculine, womanlike, and feminine: 29 terms in the masculine-associated set and 18 in the feminine-associated set. Each isolated term was classified with DistilBERT, fine-tuned on the Stanford Sentiment Treebank 2, while approximate Google result totals were collected through SerpApi as a proxy for online prominence. The masculine-associated set contained a larger proportion of negative classifications (31.0%) than the feminine-associated set (22.2%), although the negatively classified feminine-associated terms received higher and less dispersed model-confidence values. Search-result estimates were highly skewed and strongly affected by the literal nouns male and female. Contrary to the initial expectation that polar or negative items would necessarily be more prominent, signed classifier output correlated positively with search-result rank (pooled Spearman’s rho =.51; rho =.63 after excluding male and female). The results therefore reveal asymmetries in lexical coverage and model behaviour but do not establish the sentiment of online discourse or a causal effect of search algorithms. The chapter argues for a cautious, audit-oriented interpretation of decontextualised sentiment classification and outlines a contextualised design for future research on gendered language online.