Online news headlines must identify an event, establish its domain, and remain interpretable within severe spatial limits. This article examines how terminological units perform this work in English and Uzbek digital journalism. The material is a purposive micro-corpus of 21 headlines (10 from BBC News and 11 from Kun.uz) containing 28 salient terminological units, evenly divided between the two outlets. Each unit was assigned a dominant role within a five-part analytical model: domain identification, denotative specification, quantitative-parametric precision, audience accommodation, and interdisciplinary integration. A contrastive qualitative analysis was supplemented with descriptive frequencies. Domain identification was the most frequent dominant role (7/28; 25.0%), followed by denotative specification (6/28; 21.4%); each of the other three roles accounted for five units (17.9%). The BBC News sample showed a stronger concentration of medical, scientific, and technological terminology, whereas Kun.uz foregrounded public-administrative, statistical, diplomatic, and economic vocabulary. English headlines tended to package specialized meaning through compact noun clusters and premodification, while Uzbek headlines more often relied on relational morphology, extended attributive structures, and explanatory co-text. The findings indicate that terms in headlines do more than name specialist concepts: they delimit reference, encode measurable parameters, activate domain knowledge, and manage the distance between expert and general audiences. The study proposes a replicable functional model for comparative headline analysis while recognizing the limits of a small purposive corpus.