Diamesic Translation: A Qualitative and Quantitative Assessment of Intralingual Live Subtitles

At the crossroads between Simultaneous Interpreting (SI) and subtitling, respeaking is a technique that enables live events to be accessible to a wide audience, as in the case of the d/Deaf and hard-of-hearing. To do so, respeakers use an Automatic Speech Recognition (ASR) software to repeat or paraphrase the original speech, thus producing Live Subtitles (LS). This paper focuses on furthering the studies of this discipline by analysing an LSed conference called Les Assises Régionales de l’Accessibilité (‘The Regional Accessibility Conference’). The service was provided by Le Messageur cooperative, whose work aims at facilitating communication for the d/Deaf and hard-of-hearing community. Both the audio and video transcripts will be analysed from the point of view of quality and accuracy thanks to a new taxonomy for Diamesic Translation (DT). Then, the various strategies will be classified and analysed. In particular, the research will focus on the following: 1) the quality of the subtitling, with particular reference to DT strategies: litteratim, verbatim, sensatim, signatim; 2) the overall quality of the subtitles assessed against renditions and semantic and non-semantic non-renditions; 3) the number of words per minute (wpm) 4) the delay between the speech and the subtitles appearing on screen; 5) the reduction of the subtitles compared to the source text (ST).