YouTA: A context-aware corpus annotation tool for Critical Discourse Analysis of YouTube comments

YouTA (YouTube Text Annotator) is presented as a context-aware corpus annotation tool that bridges Critical Discourse Analysis (CDA) theory and computational analysis by addressing limitations in existing annotation environments for Systemic Functional Linguistics (SFL)-informed research. The tool integrates four innovations: (i) context-aware annotation linking text to YouTube metadata; (ii) SFL-compatible hierarchical multi-label annotation; (iii) a human-augmented Large Language Model (LLM)-assisted workflow with provenance tracking; and (iv) an ISO-standardized interface design minimizing cognitive load. This architecture preserves situational and interactional context while supporting complex annotation schemes and enabling qualitatively informed quantification. To demonstrate its methodological contribution, a case study is included in which mental processes are annotated across a subset of YouTube comments using the human–LLM workflow. Results show that LLMs provide effective first-pass annotations (F1 = 0.578; partial-overlap recall = 0.693), with approximately 41% of annotations directly accepted, around 11% requiring boundary refinement, and the remainder either rejected or missed, reflecting a workflow in which the majority of relevant instances are either correctly identified or partially captured by the model. YouTA thus facilitates the systematic integration of linguistic theory, annotation practice, and scalable computational analysis, while maintaining human interpretive agency.