New Methods for
Creating, Exploring and Disseminating
Linguistic Field Data

The new NSF TalkBank Project is sponsoring a workshop on computational support for linguistic fieldwork, to be coordinated by Steven Bird (University of Pennsylvania).

The workshop will bring together linguists and computational linguists committed to empirical research on large datasets, through the combination of traditional field methods and new technologies for exploring and visualizing complex datasets. The languages under study may range from the undescribed to the well-studied, and the fieldworker may operate in a village or a laboratory. The focus is the exploratory mode of research, where elicitation, analysis and hypothesis-testing form a tight loop. The workshop will contribute to the evaluation and evolution of methodologies that integrate traditional practices with new technologies, leading to increased accessibility, accountability, and stability of empirical linguistic research.

The workshop will address a selection of the following issues:

1. Representation - what are good data models for interlinked, heterogeneous, multimodal linguistic field data, including lexicons, (interlinear) texts, field notes, (annotated) recordings, paradigms, grammar sketches, maps, photographs, folios, course notes and problem sets?
2. Tools - what are the existing and new tools for manipulating linguistic field data, and what are their strengths and weaknesses vis-a-vis creating, browsing, searching, querying and transforming this data? How well do the tools accomodate the fieldworker's continuously evolving conception of the data? What statistical corpus-analysis methods are suitable for datasets whose items number in the hundreds rather than the hundreds of thousands?
3. Collaborative knowledge discovery - how can a geographically distributed network of linguists and native speakers cooperate on the construction, validation and enrichment of multimodal field data? How do we bridge the gap between the field and the laboratory?
4. Online repositories - how can a collection of online multimodal field data covering many languages be archived and curated? What are the corpora that people are currently willing to share? What are the confidentiality issues, and what mechanisms exist to protect privacy?
5. Dissemination and citation - how are datasets to be accessed by researchers, native speakers, language learners, field-methods students, and so on? How can we facilitate durable citations to shared linguistic resources, and track the provenance of a data item from a published transcription, through any intermediate databases, right back to a digitized speech recording?

A sample of ongoing work which is closely relevant to the topic of this workshop is available via the Linguistic Exploration page.

To register for the Linguistic Exploration Workshop, and to receive future updates about the workshop and related activities, please contact Steven Bird: sb@ldc.upenn.edu.

PROGRAM