Open AccessQuality assessment in interlingual live subtitling: The NTR Model
Author(s) -
Pablo Romero-Fresco,
Franz Pöchhacker
Publication year - 2018
Publication title -
linguistica antverpiensia new series - themes in translation studies
Language(s) - English
Resource type - Journals
ISSN - 2295-5739
DOI - 10.52034/lanstts.v16i0.438
Subject(s) - computer science , grading (engineering) , natural language processing , quality assessment , quality (philosophy) , linguistics , grading scale , set (abstract data type) , artificial intelligence , speech recognition , evaluation methods , engineering , medicine , philosophy , civil engineering , surgery , epistemology , reliability engineering , programming language
This article presents a model for assessing the quality of interlingual live subtitling, as carried out by way of respeaking and automatic speech recognition or other means. The first part reviews the requirements and formulas for quality assessment in intralingual live subtitling and presents the NER model, developed by the first author, which serves as a foundation for the new model. Drawing on insights from the literature on both prerecorded interlingual subtitling and simultaneous interpreting, the authors then introduce the NTR model, an NER-based formula for calculating the accuracy rate of (interlingual) live subtitles. The model features a set of categories for scoring the accuracy of content and form as well as a three-level grading scale for translation and recognition errors. The application of the NTR model is exemplified, highlighting the role of the accuracy rate as part of an overall qualitative assessment.