An annotation system for signed language interpreting corpora
Hermeneus. Revista de Traducción e Interpretación
Year of publication
Because of the complexity in transcribing and annotating a visual language, there exists a paucity of corpus-based/driven sign language interpreting research. In this paper, a transcription and annotation system for signed language corpora is outlined. The system was designed to analyse a corpus of interpretations into South African Sign Language, specifically to analyse corpora of signed interpretations using text-based concordance software, such as Wordsmith Tools or Antconc. Transcription of the manual signs is undertaken using lemmatized ID glosses, allowing for semantic comparison between the source message and its interpretation. The annotation systems is designed to analyze five main groups of factors, namely linguistic features of the interpreter's sign language produced, characteristics pertaining to the quality of production of a face-to-face message, characteristics of the type of language used by the interpreter (i.e. parole) and features of the interpreting process, both in terms of describing differences between the source and target messages and in terms of identifying phonological, syntactic and discourse errors made by the interpreter.