MAPPING SEMANTIC SHIFTS: A COMPARATIVE STUDY OF JAWAB-I-SHIKWA IN HUMAN AND AI-GENERATED TRANSLATIONS

Abstract

The manner of translating classical Urdu poetry is drastically extra problematic than simply replacing one word with another; it requires deliberate engagement with profound theological, cultural, and emotional factors. This study investigates the semantic shifts that occur when translating Allama Iqbal’s *Jawab-i-Shikwa* (*The Answer to the Complaint*), a seminal text of twentieth-century Islamic philosophy. By employing a Comparative Semantic Mapping (CSM) framework, this research contrasts the nuanced, culturally mediated human translation of M.A.K. Khalil with the probabilistic, algorithm-driven output of the Large Language Model GPT-4. Grounded in Mona Baker’s Translation Universals and classical semantic shift theories, the study maps how each medium navigates the complex Persian-Arabic diction, Sufi-philosophical terminology (such as *khudi*, *ishq*, and *tawhid*), and ecstatic rhetorical structures of Iqbal’s poetry. The analysis reveals that while GPT-4 exhibits high fluency and structural coherence, it is constrained by algorithmic literalism and statistical centralization, resulting in "semantic flattening." This flattening diminishes the text's rhetorical fervor, philosophical depth, and emotional resonance. Conversely, Khalil’s human translation demonstrates deliberate interpretive judgments—utilizing explicitation and functional equivalence—to preserve the spiritual "spirit" and cultural manifestation of the original, even at the expense of formal structural equivalence. The findings highlight that AI-generated translations, while efficient, currently lack the hermeneutic capacity required for high-register, culturally anchored philosophical poetry. This research contributes to the growing discourse on the intersection of digital humanities and translation studies, advocating for hybrid translation models where algorithmic efficiency is coupled with humanistic, cultural mediation.