<![CDATA[In contemporary digital media, news headlines function not only as informational summaries but also as strategic discourse units that shape readers’ interpretation and engagement. One prominent feature of such discourse is sarcasm, which enables implicit evaluation and critique through indirect linguistic expression. However, the role of recurrent phraseological patterns in constructing sarcastic meaning remains underexplored. Drawing on corpus linguistics, particularly the lexical bundle framework proposed by Douglas Biber, this study examines how recurring multi-word sequences contribute to sarcasm in digital news headlines. This research adopts a corpus-assisted discourse analysis approach, combining quantitative and qualitative methods. The dataset consists of 500 sarcastic headlines (250 from The Onion and 250 from HuffPost) obtained via Kaggle. The corpus was analyzed using AntConc to extract lexical bundles of three to five words based on frequency thresholds. The findings reveal that stance expressions constitute the dominant category (48%), followed by discourse organizers (30%) and referential expressions (22%), indicating that sarcasm is primarily constructed through evaluative language and implicit judgment. Comparative results show that The Onion employs more explicit and stylized sarcasm, while HuffPost demonstrates a more subtle and context-dependent use. This study concludes that lexical bundles function as key linguistic resources for encoding sarcasm and evaluative stance in digital discourse.]]>
Copyrights © 2026
