Beyond Words and Machines: Cognitive Pragmatics, Discourse Memory, and Contextual Reasoning in Odia Language Processing—Towards a Human-Centric Computational Framework

ABSTRACT Natural Language Processing (NLP) has undergone a paradigm shift from statistical modelling to deep neural architectures and, more recently, to foundation models capable of multilingual reasoning and generation. Despite these advances, computational systems continue to struggle with one of the most fundamentally human dimensions of language—pragmatic meaning. While machines increasingly recognise lexical patterns, syntactic structures, and semantic relations, they frequently fail to infer speaker intention, contextual implication, discourse continuity, irony, politeness, cultural presupposition, and implicit meaning. This limitation becomes especially significant for morphologically rich and culturally embedded low-resource languages such as Odia, where communication relies extensively on shared social knowledge, contextual inference, honorific distinctions, ellipsis, and discourse memory. Existing research in Odia Language Processing (OLP) has concentrated predominantly on orthographic normalization, machine translation, optical character recognition, speech technologies, morphological analysis, and sentiment classification. Comparatively little attention has been devoted to computational pragmatics, discourse cognition, or context-sensitive reasoning.   This paper proposes a novel Human-Centric Computational Framework (HCCF) for Odia Language Processing that integrates cognitive pragmatics, discourse memory, contextual reasoning, and explainable artificial intelligence. Drawing upon insights from pragmatics, cognitive linguistics, discourse analysis, computational linguistics, and contemporary transformer-based language models, the study argues that genuine language intelligence cannot emerge from lexical processing alone but requires computational modelling of communicative intention and evolving discourse context. The proposed framework conceptualizes language understanding as a dynamic interaction among linguistic form, cognitive memory, socio-cultural knowledge, and contextual reasoning. Rather than treating context as an auxiliary input, the model positions discourse memory as the central organizing principle of computational interpretation. Methodologically, the study adopts a qualitative and interdisciplinary approach based on theoretical synthesis and comparative analysis of recent developments in artificial intelligence and linguistic theory. It identifies the principal computational challenges posed by Odia pragmatics, including deixis, anaphora, conversational implicature, code-switching, ellipsis, honorificity, and culturally embedded expressions. The paper further outlines an architecture for integrating discourse-aware reasoning into future Odia NLP systems and discusses its implications for conversational AI, digital assistants, educational technologies, machine translation, legal informatics, and digital humanities. By introducing cognitive pragmatics into the core of Odia Language Processing, this article advances a new theoretical direction that transcends conventional lexical and syntactic approaches. The proposed framework contributes to the emerging discourse on explainable and culturally informed artificial intelligence while offering a research agenda capable of strengthening computational resources for one of India's oldest classical languages. The study thus positions Odia not merely as a low-resource language requiring technological support but as a linguistically sophisticated system capable of enriching contemporary theories of multilingual artificial intelligence.   Keywords: Odia Language Processing, Cognitive Pragmatics, Computational Pragmatics, Discourse Memory, Contextual Reasoning, Explainable Artificial Intelligence, Low-Resource Languages, Conversational AI, Computational Linguistics, Human-Centric NLP