Abstract / Summary
Background: Depression and anxiety affect university students’ academic and emotional well-being. Traditional self-report assessments may be limited by stigma and disclosure challenges. Language-based assessment offers a promising approach by identifying psychological patterns in communication, although validated tools remain limited. Methods: A quantitative validation study was conducted with 827 university students in Ghana. The Language-Based Depression and Anxiety Screening Instrument (LBDASI) was developed and evaluated using Natural Language Processing and psychometric analyses, including factor analysis, reliability testing, validity assessment, diagnostic accuracy, and measurement invariance testing. Results: The findings showed that students with depression and anxiety symptoms demonstrated distinct linguistic patterns, including higher negative emotional expressions, self-focused language, cognitive distortions, uncertainty expressions, and social withdrawal language, alongside reduced positive and achievement-related expressions. Language-based indicators significantly predicted psychological distress, achieving strong classification performance with 86.7% accuracy and an area under the curve of 0.92. The LBDASI demonstrated strong psychometric properties, including high internal consistency (α = 0.86–0.93), a stable three-factor structure, strong construct validity, and measurement equivalence across gender, academic level, and institutional contexts. The instrument also showed strong associations with established depression and anxiety measures. Conclusion: The study demonstrates that language can serve as a valid psychological measurement domain for identifying depression and anxiety symptoms among university students.