Intersemiotic Translation as a Factor in Subtitling Production through Digital Image Processing and Filmic Narrative

Abstract

This paper aims to apply digital image processing and information technology to analyze facial expressions of movie characters as an influential factor for the selection of verbal language in subtitling. The corpus used in the study is comprised of scenes selected from the films Magnolia (1999) and Bicho de Sete Cabeças (2001) aiming at: 1) analyzing character construction in filmic narratives through the application of digital image processing techniques, and 2) exploring information technology and digital image processing resources associated with Intersemiotic Translation to analyze images and the construction or omission of subtitles, considering non-verbal language in specific moments of the filmic narrative. To this end, the selected scenes were analyzed building on: triangulation of data obtained through digital image processing via software for facial recognition and emotional quantification (Affectiva, EMOVIE); the analysis of characters inserted in their filmic narratives (GAUDREAULT; JOST, 2009); and the translation choices for subtitles (DÍAZ CINTAS; REMAEL, 2014), considering Intersemiotic Translation as analogic signifying through distinct sign systems (PLAZA, 2003). Results showed that the textual reduction imposed to subtitles does not jeopardize spectatorial comprehension due to their insertion in audiovisual works through the concept of semiotic cohesion (DÍAZ CINTAS; REMAEL, 2014). They also showed that: 1) the human face and facial expressions are relevant in the filmic narrative as complex sign systems (EKMAN; FRIESEN, 2003), and 2) digital image processing and emotional quantification, albeit limited for dealing with filmic context, reveals new possibilities of technological use of software in favor of subtitling and the spectator.