A corpus-based approach to manual simultaneity

The most conspicuous modality difference between signed and spoken languages is the fact that in signing the two hands are used as primary articulators. The resulting potential for more simultaneous expression of information has intrigued researchers since the early days of sign linguistics, even though little empirical research has been done until recently (see Vermeerbergen, Leeson, and Crasborn 2007 for an overview). Examples of manual simultaneity have been described for several sign languages and at various levels of linguistic structure (prosody, syntax, discourse). While previous studies have described and analyzed a number of structures that exploit manual simultaneity, large-scale corpus-based analyses of the actual use of these phenomena are yet to be conducted. Studies aiming to provide grammatical description and analysis may rely on prototypical examples of the structures in question, but the corpus linguist deals with language use “in the wild” and has to identify and understand all instances of the phenomenon under investigation. The annotation process is crucial in any corpus-based approach (Aarts 2002) and clear annotation guidelines are indispensable for reliable and comprehensive data selection in the study of a given structure. The purpose of this paper is to describe a corpus-based approach to the study of manual simultaneity (in particular manual holds) and our first attempt at implementing it for the Sign Language of the Netherlands (Nederlandse Gebarentaal, NGT). After discussing previous research on the subject and the new insights we expect from a corpus-based analysis, we present our annotation guidelines and evaluate them based on a pilot study on the Corpus NGT.