Author:Ike, Nneka Francisca
Date: 05/07/2026
The increasing use of Artificial Intelligence (AI) in translation has raised important questions about its ability to preserve meaning in culturally rich literary texts. This study examined the extent to which Grok AI and Google's Gemini AI preserve semantic meaning, pragmatic force and cultural fidelity across twenty selected excerpts from Nke M Ji Ka by G. I. Nwozuzu (2013). Adopting a qualitative descriptive design, the Igbo excerpts were translated into English by the researcher and subsequently back-translated into Igbo using both AI systems. The generated texts were compared with the original Igbo versions using Baker's (1992) Theory of Pragmatic Equivalence. The findings reveal that while both systems successfully preserved basic semantic meaning in direct literal contexts, they encountered considerable difficulties in maintaining pragmatic force and cultural fidelity in dialogues rich in cultural nuances. Figurative devices and indigenous cosmological frameworks proved particularly challenging, resulting in literal calques, phrasal verb intrusions, and anatomical impositions. Comparatively, Grok AI demonstrated stronger structural fidelity and lexical correspondence across the dataset, whereas Gemini AI exhibited significant lexical hallucinations and grammatical instability. The study concludes that although AI translation systems show significant potential for baseline Igbo translation, they remain structurally limited in preserving the deeper cultural, syntactic, and pragmatic dimensions of literary discourse. Expert human review therefore remains essential for ensuring the accurate translation and contextual preservation of culturally embedded texts.
Keywords: Artificial Intelligence in Translation, Back-translation, Pragmatic Equivalence, Cultural Fidelity.
Download PDF