A Comparative Analysis of Translation and Back-Translation Vs. Machine Translation With Post-Editing

Abstract

Background

The translation and back-translation (TBT) method is widely regarded as the gold standard for translating health status measures into different languages, ensuring accuracy and cultural relevance. While machine translation (MT) can provide immediate translations, it often overlooks context and cultural nuance. A new method, machine translation with post-editing (MTPE), addresses these shortcomings by involving human translators to refine machine translations.

Methods

The study evaluated the TBT and MTPE methods in translating 15 self-administered (SA) scales of the Standard for Clinicians’ Interview in Psychiatry (SCIP) from English into Arabic.

Results

The findings demonstrated that the MTPE method was highly efficient. Through the TBT method, 4 translators required 120 hours to complete the translation of 15 SA SCIP scales. In comparison, the MTPE method required only 2 translators, who used ChatGPT for the initial translation and completed the editing in just 24 hours. The final translations produced through the MTPE method were both contextually accurate and culturally appropriate.

Conclusions

The integration of machine translation with human post-editing provides the combined advantages of speed, accuracy, and cultural sensitivity. The authors recommend the development of formal guidelines to support the standardized use of the MTPE method in translating health measures.