Subtitling for the deaf and hard-of-hearing (SDH) and eyetracking. Results of an experimental research

This study is part of Accessible Audiovisual Translation (TAVa in Portuguese) studies, more specifically, Subtitling for the Deaf and Hard-of-Hearing (SDH) studies. It analyzes, based on information provided by the use of an eye-tracker, the reading/processing of two groups of deaf and hearing Brazilian participants as they view the subtitling of Brazilian language in two audiovisual programs: a political program and a documentary. It will focus on two subtitling technical parameters: linguistic segmentation (division of speech into semantic blocks based on semantic and syntactic units) and subtitle rate. The hypothesis of this study originated from exploratory research accomplished with 34 deaf participants from four regions of Brazil. The assumption was that an appropriate linguistic segmentation between the lines of a subtitle, respecting the highest syntactic nodes possible, would facilitate the processing of information by deaf people. Therefore, we observed the participants’ reception of the subtitles which were presented in four experimental conditions: slow well-segmented (SWS), slow illsegmented (SIS), fast well-segmented (FWS), fast ill-segmented (FIS). The hypotheses are: 1) Slow well-segmented subtitles can facilitate the reception of deaf and hearing participants; 2) Slow ill-segmented subtitles can hamper the reception of deaf and hearing participants; 3) Fast wellsegmented subtitles can facilitate the reception of deaf and hearing participants; 4) Fast ill-segmented subtitles can hamper the reception of deaf and hearing participants. Both studies had the same methodological procedures. In order to test them, we accomplished an experiment with 8 deaf and 8 hearing participants from the city of Fortaleza, in each study. In the first study, we used scenes from a documentary. In the second, we used videos from local political campaigns on tv. The participants watched four different parts of the documentary and four different video campaigns, with the SDH manipulated into the four different experimental conditions involving rate and segmentation: SWS, SIS, FWS and FIS. The results of both studies revealed that ill-segmented subtitles caused discomfort and a higher processing cost for both groups in the subtitles reading process. Also, for both groups, speed was not an obstacle in understanding the videos, but they processed the quick subtitles more easily.