Comparison of human and machine translation – especially in terms of the preservation of contextual meaning – has already attracted the attention of many researchers. Accordingly, the present study intended to compare human (Daryabandari) and machine (Google Translate) translations of The Remains of the Day into Persian to describe translation error types in conveying contextual meaning. In all, 120 errors were extracted using MQM model from each translation. The study identified four main categories of errors: lexical (35%), semantic (30%), syntactic (23%), and pragmatic (12%). These findings highlighted the limitations of current MT systems in handling the nuances of language, cultural references, and literary style present in sophisticated narratives. The presence of syntactic errors (23%) further demonstrated the difficulty machines face in correctly interpreting and reconstructing grammatical structures across languages. Pragmatic errors, while less frequent (12%), underscored the machine’s inability to grasp and appropriately translate cultural nuances and context-dependent meanings. Further, the Chi-square test result (χ² = 81.4, df= 3, P= .000) revealed a statistically significant difference between the translation type and error categories. This implied that human and machine translations exhibited different error patterns. The findings of this study suggested that translators and translation students should incorporate both traditional human translation techniques and machine translation technologies into their curricula.