TIGA Corpus: German-Arabic Telephone-based Interpreting
Blog post by Rahaf Farag
The Center for Sustainable Research Data Management at the University of Hamburg hosts the first corpus of German-Arabic interpreter-mediated encounters via telephone (TIGA). It was released in August 2023. You can access the corpus with your internet browser and download files to work with them offline. Access (for teaching and research purposes only) can be given upon request:
Meyer, Bernd / Farag, Rahaf. (2023). Telephone Interpreting German-Arabic (TIGA) / Telefondolmetschen Arabisch-Deutsch (TeDo) (Version 1.0) [Data set]. [http://doi.org/10.25592/uhhfdm.13191].
Background and research project
We created the TIGA Corpus in our research project Turn-Taking and Ensuring Understanding in Arabic-German Telephone-based Interpreting (2019-2022) in the Department of Intercultural Communication at the Johannes Gutenberg University of Mainz, Germany, in the Faculty for Translation Studies, Linguistics, and Cultural Studies. The research was funded by the German Research Foundation (Deutsche Forschungsgemeinschaft; DFG). Drawing on conversation analysis and interaction-oriented interpreting studies as an analytical framework, its aim was to investigate the dynamics of turn-taking and the negotiation of understanding during authentic counselling sessions in social settings mediated by telephone interpreters.
Information and communication technologies, such as the telephone and App-based solutions, provide ad hoc access to interpreting services irrespective of the interlocutors’ physical locations (i.e. remote interpreting). Given the frequent lack of alternatives, telephone-based remote interpreting (TI) has become an increasingly common practice in care and counselling settings to facilitate communication with refugees and migrants. The presence of a third party (i.e. interpreter) in interpreted encounters usually presents additional communicative challenges. The conventional coordinative actions of communicative interaction are unlikely to be sufficient when facing language barriers and knowledge asymmetries between the interlocutors. This is especially true for turn-taking and turn allocation as well as initiatives to ensure (mutual) understanding like queries and reformulations. In view of the particularities of the TI setting a question that arises is which linguistic-communicative strategies are used by the telephone interpreters to compensate for the absent co-presence of primary interlocutors and the constraints on the auditory and visual access.
Impairments of a technical or situational nature (e.g. poor audibility, volume fluctuations, background noises), which were sometimes unpredictable and difficult to rectify, posed challenges for the interpreters. Our quantitative and qualitative analyses, however, show that such challenges are not as frequent as one might expect. There are no indications that coordinative and communicative moves (smooth turn-taking, problem-solving actions, clarification of misunderstandings, repair of non-understanding, etc.) were particularly compromised in remote settings (Farag/Meyer 2022, 2023).
Corpus Processing and Annotation
The corpus provides transcripts of audio and video recordings of interpreter-mediated counselling sessions about general asylum-related topics. It documents twelve encounters (with a length of approximately 30 to 60 minutes each) between German migration and integration counsellors, Arabic-speaking clients (CL), and telephone interpreters. The clients and the counsellors were co-located and physically co-present. They communicated with the interpreters, who had to be called in from afar, via a speakerphone. The interpreting was performed consecutively and in both directions. To arrange the recording set-up, all encounters took place by prior appointment. The counsellors called an interpreter at the appointed time. Yet, the setting was not controlled in other respects. Counsellors and interpreters were not acquainted. The course of interaction was not arranged; topics were not discussed in advance.
The clients were refugees, mostly from Syria, who did not have a (sufficient) command of conversational German. They were in serious need of advice on family reunification, language courses, employment opportunities, legal matters, and further urgent issues. All interpreters held a university degree in translation and language studies. Most of them were sworn interpreters. They had several years of professional experience in working with German authorities and communal institutions, but they lacked (solid) experience in working under remote conditions.
Each transcript attempts to reconstruct one counselling session. Except for two four-party sessions with two clients (relatives), the interactions were triadic. The interpreters and most of the counsellors participated in two sessions, at least. The clients are the only participants who appear once in the corpus. We used the EXMARaLDA Partitur-Editor to transcribe and annotate the recordings in a musical score notation, largely following the conventions of the semi-interpretative working transcription (Ger. Halbinterpretative Arbeitstranskription; see Rehbein et al. 2004). Due to technical constraints resulting primarily from the opposing and incompatible directionalities in multi-script notations, I developed a phonological-orthographical system to represent the Arabic sequences with Latin characters (Farag 2019). We provided relatively idiomatic, yet denaturalised (i.e. unpolished) German translations for the Arabic utterances. Using the EXMARaLDA Corpus-Manager (Coma) we were able to link the recordings and their transcriptions with metadata. Sensitive personal data that could be traced back to the participants or the counselling offices were de-identified before publication.
The transcriptions are largely based on the audio recordings of both interaction rooms ‒ the interpreters’ and the counsellors’ workspaces. We added descriptions of non- and paraverbal actions only when relevant. The video recordings have proven to be essential for obtaining more clarity and precise information when identifying or designating actions that were difficult to allocate solely by auditory means. Besides, the videos gave us access to various actions performed on both sides, allowing interesting findings, for instance regarding their audibility. Accordingly, we annotated acoustic perception precisely and systematically (e.g. completely drowned, partly drowned, barely comprehensible, inaudible, only visible, etc.).
Transcripts and metadata are available as PDF. Raw data can be downloaded as EXB-files and can therefore be processed with other transcription tools. They hold enormous untapped potential for further analyses and methodological research (linguistics, translation and interpreting studies).
Further References
Farag, Rahaf / Meyer, Bernd (2022): Telefondolmetschen Arabisch-Deutsch: Gesprächstranskription im Spannungsfeld von Mehrsprachigkeit, schriftlichem Standard und Varietätenvielfalt. In: Grawunder, Sven / Schwarze, Cordula (Hg.): Transkription und Annotation gesprochener Sprache und multimodaler Interaktion: Konzepte, Probleme, Lösungen. Tübingen: Narr. S. 213-237.
Farag, Rahaf / Meyer, Bernd (2023): Coordination in telephone-based remote interpreting. In: Interpreting 26.1, 80-113. [https://doi.org/10.1075/intp.00097.far]
Farag, Rahaf (2023): Computergestützte Transkription arabisch-deutscher Gesprächsdaten: ein methodischer Beitrag zur Untersuchung gedolmetschter Gespräche. FTSK: Publikationen des Fachbereichs Translations-, Sprach- und Kulturwissenschaft der Johannes Gutenberg-Universität Mainz in Germersheim 75. Lausanne: Peter Lang.