The soundness of documentation: an epistemology for audio in documentary linguistics

Documentary linguistics for endangered languages emphasizes primary data - recordings and representation of linguistic events - and audio and video recordings are a priority for both fieldwork and for archiving because those events are unlikely to occur into the future. Audio is the focal component of our primary data. While video offers much promise, it is by definition less close to the principal concerns of documenting (spoken) languages, and introduces many costs. Recent debates about the motivations for and value of video have raised valuable questions such as: What/who is it for? What makes a good recording? How can we figure the documentation value? What aspects of events are captured/not captured? But these questions, in turn (and some of the arguments for video), when applied to audio, reveal that we have tended to take audio for granted; at worst (and frequently enough), audio is simply an inconvenience on the way to creating symbolic representations such as transcriptions and analyses. Up till now, there have been some developments, such as increasing the accountability of analyses by providing direct links to "primary audio data", and tools built to support this such as Thieberger's Audiamus. But these on their own are not enough; they do not challenge audio's emasculated status as evidence rather than performance. Finnegan (2008), for example, has recently pointed out the almost unbounded richness of audio phenomena in linguistic performance, for which we don't have a vocabulary nor even recognition that without such a framework we are effectively hiding these phenomena from research. And we have been reminded (by Dietrich Schueller) that linguistics is one of the least scientific of disciplines when it comes to audio data collection. This paper considers audio from several perspectives, including documentary linguistics, experiences in training young documenters, information theory, and acoustics, to propose an epistemology for audio within documentation that surpasses descriptors such as formats and resolutions, and restores criteria such as informativeness, replicability, and representational validity for audio in language documentation. References Finnegan, R. 2008. Data – but data from what? In Peter Austin (ed) Language Documentation and Description. London: SOAS Schueller, D. 2006. Audio recording, digitisation and archiving. Workshop held at ELAR, SOAS, 13 February 2006