AI-BASED ANALYSIS OF CODE-SWITCHING PATTERNS IN MULTILINGUAL DIGITAL COMMUNICATION AMONG PAKISTANI YOUTH

Abstract

This study investigated multilingual code-switching in digital communication among Pakistani youth using an artificial intelligence (AI)-based natural language processing (NLP) framework. A corpus of digital communication was systematically collected, cleaned, and annotated to identify language patterns, switching points, and major forms of code-switching, including inter-sentential, intra-sentential, and tag switching. AI-based language identification and classification techniques were employed to detect and quantify multilingual switching patterns, while model performance was evaluated using accuracy, precision, recall, and F1-score. The analysis focused particularly on Urdu–English and other multilingual combinations occurring in digitally mediated communication. The findings indicated that code-switching represented a systematic and context-sensitive linguistic practice, with Urdu–English interaction constituting a prominent pattern and intra-sentential switching demonstrating substantial prevalence. Fine-tuned AI models provided more effective identification of code-switched text than general multilingual approaches. The study contributes to computational sociolinguistics by integrating AI-based language detection with established sociolinguistic perspectives and provides a foundation for developing linguistically responsive NLP technologies for Pakistan's multilingual digital environment.