Post-editing through Speech Recognition: A Feasibility Study with Post-editor Trainees

In the past couple of years automatic speech recognition (ASR) software has quietly created a niche for itself in many situations of our lives. Nowadays it can be found at the other end of customer-support hotlines, it is built into operating systems and it is offered as an alternative text-input method for smartphones. On another front, given the significant improvements in Machine Translation (MT) quality and the increasing
demand for translations, post-editing of MT is becoming a popular practice in the translation industry, since it has been shown to allow for larger volumes of translations to be produced saving time and costs. The translation industry is at a deeply transformative point in its evolution and the coming years herald an era of converge where speech technology could make a difference. As post-editing services are becoming a common practice among language service providers and speech recognition is gaining momentum, it seems reasonable to explore the interplay between
both fields in a feasibility study.
In the context of machine-aided human translation (MAHT), different scenarios have been investigated where human translators interact with a computer through a variety of input modalities (i.e. typing, handwriting and speaking) to improve the efficiency and accuracy of the translation process. However, further studies need to be conducted to build up new knowledge about the way in which state-of-the-art speech recognition
software can be applied to the post-editing process. As a continuation of the pioneering work done in the SEECAT project, our presentation will report on a feasibility study where post-editor trainees will be asked to post-edit raw MT using voice and keyboard as an input method. This feasibility study will explore the potential of combining one of the most popular computer-aided translation workbenches in the market (i.e. MemoQ) together with one of the most well-known ASR packages (i.e. Dragon Naturally Speaking from Nuance). Two data correction modes will be considered: a) keyboard vs. b) keyboard and speech combined. These two different ways of verifying and correcting raw MT output will be examined making comparisons in terms of: i) overall time to complete the task, ii) final quality of the target text, and iii) user satisfaction.