Designing Elicitation Stimuli and Tasks for the DGS Corpus Project

In recent years quite a few large-scale sign language data collections have been compiled. However, up until now there has been little discussion on the methods of data collection for sign language corpora, i.e. which kinds of stimuli have been used to reach which goals and why. This paper illustrates and exemplifies how stimulus material and tasks for eliciting data for a general sign language corpus can be systematically developed, tested and adapted. The first part of the paper provides an overview on the materials and elicitation tasks used for the data collection in the DGS (German Sign Language) Corpus Project. The second part describes motivation, advantages, difficulties and the effectiveness of different elicitation techniques and tasks. Finally, results of the pilot tests and a preliminary evaluation of the ongoing data collection are discussed.

The DGS Corpus Project, which has started in January 2009, is a long-term project with two major aims: (i) to establish an extensive corpus of DGS and (ii) to develop a comprehensive dictionary DGS–German based on the analysis of the corpus data. During the first three years the main focus is on data collection. Data from approx. 300 signers from various sociolinguistic backgrounds (e.g. gender, regional background, age, age of first exposure to sign language) enable us to build a well-balanced corpus. Signers are always filmed in pairs. The session for each pair lasts for approximately six hours. 

The data collection is designed to obtain language data that reflects natural or near-natural language use of DGS and covers a great range of lexical items which are used in everyday conversations and, at the same time, serves as a cultural resource of the German Deaf community. To this aim, a variety of different stimuli and tasks are adopted to encourage natural sign flow and signed interactions. The stimuli and tasks consider a range of possible modes of discourse including free conversation, discussion (argumentation), narration, re-narration, planning and description. At the same time, all discourse modes touch on as many different subject domains as possible. Some of the elicitation stimuli and tasks were either adapted from other projects (see e.g. Hong et al. 2009) or developed specifically for the DGS Corpus Project. Stimuli include pictures, picture stories, non-verbal film clips (e.g. cartoons and realistic film clips), signed movies and also some written German. Topics for conversations, discussions or description tasks are generally presented in signed form. Where necessary, slides with (written language) keywords and/or pictures serve as an aide-memoire afterwards.

All of the stimuli and tasks were tested during a pilot phase in order to assess their feasibility and effectiveness. Some of the tasks tested needed to go through several rounds of modifications while others did not work out and thus were not included for the actual data collection. In this regard, we discuss the criteria for designing different stimuli and tasks and evaluate the results and insights gained during the pilot phase. Finally, preliminary results of the initial phase of the data collection will be illustrated, especially on how well results from the pilot scale to full data collection sessions in the field.