On Beyond TM: When the Translator Leads the Design of a Translation Support Framework

Commercial off-the-shelf machine translation engines and translation support tools, such as translation memory (TM), have been developed primarily for translating grammatically well-formed, edited text. The real-world, foreign language (FL) document collections that our translators work with consist instead of noisy and complex image files. We are currently conducting experiments that involve building and evaluating the effectiveness of different multi-component workflows for the automated processing and translation of these FL images into English text. To support the project’s ongoing needs for translations, we are developing a software framework designed with and for our translator that (i) streamlines users’ access to and capability to add-in and modify existing online tools and data resources (ex. MT, TM, dictionaries, morphological analyzers, LM), (ii) builds persistent data objects for later re-use, and (iii) provides users with a configuration screen page to select the tools and data resources for their sessions, to set their tools’ options and display options for their translation screen page. This paper introduces our extreme programming approach to the software engineering of this new, hands-on translator’s framework called TREAT (Toolkit and Resource Environment to Assist Translation), where the translator---as subject matter expert and experienced software user---participates fully in the software design, evaluation, and iterative modification processes.