Machine Translation Quality Assessment : A Quantitative-Qualitative Analysis of English- into- Arabic Translation in The Climate Change Discourse

The present research demonstrates a specialized quality-assessment of English -into-Arabic in Machine translation based on three speeches that were delivered in the UN .The translation is generated by two AI systems which are ChatGPT 3.5 and Gemini systems.The research includes a mixed-method that combines qualitative analysis with quantitative error-frequency evaluation. The two main frameworks guiding the assessment of this research are TAUS Error Typology, which classifies errors that occur in translation into four main classes, and the Eco-translation theory, which expresses the ethical, ecological, and cultural responsibility in translation.The results show the errors according to four main categories which are accuracy;terminology,verity,and fluency. The results show that ChatGPT 3.5 and Gemini proceed with a similar number of errors in translating metaphors and cultural reference, accuracy, also they have equal terminology errors rate .Although both systems maintain sentence fluency, they struggle with cultural and ecological adaptation. In light of these results,the research highlights the urgent need for an extensive integration of eco-conscious strategies in specific domains of AI models for enhancing its performance as well as for reliable and effective climate change communication.