Applying Library Science Models to the Arrangement of Language Collections

Abstract

Language archives support language scholars and communities by stewarding language resource collections. Discovery services are paramount for maintaining connections with audiences. Digital language archives have deployed a variety of methods for arranging the contents of their collections (Paterson III 2021b, 2021a). Popular across language archives is the ‘bundle’ (Johnson 2002; Burke and Zavalina 2019) wherein several resources might be housed under a single metadata record. Bundles can be internally complex and have external relationships to other resources. For example, multiple interview recordings on the same audio tape, or several audio and video recordings of the same interview. Another popular arrangement is the inclusion of transcriptions with their associated audio or video resources. Huber (2023) points out that bundles present challenges for rights management, while Burke and Zavalina point out that they present challenges for API access to individual resources.

Libraries around the world have centered their record keeping around the IFLA-LRM model (Riva, Le Bœuf, and Žumer 2017) which has as its core the Work-Expression-Manifestation-Item (WEMI) ontology. This model is not attested to be in use at any language archive in the DELAMAN community. While the full IFLA-LRM and its closely associated Resource Description and Access model allow for a number of relationships, Coyle (2022) points out that a very minimal ontology of WEMI entities allows for accurate record keeping and the flexible building of discovery services across a variety of industries.

In this paper we present some examples of how WEMI would be applied to records for two types of resources: (1) a legacy audio recording which has been digitized and (2) an audio/video recording with an accompanying transcript. We then contrast this with the kinds of records we currently see in digital language archives. We address critical questions about the independent nature of transcriptions and annotations as works versus expressions under WEMI models. We then describe how record independence or conflation impact citation and referencing of resources in bibliographic contexts. Our conclusions point out that the adoption of WEMI within language archives would provide more consistent discovery services within and across archives, increase compatibility of metadata with library systems which often host language archives, improve citation and referencing capabilities by providing a framework for the assignment of persistent identifiers, and make the management of metadata easier by aligning records with unique units.

Tags: Training Curriculum Development Language Archives Stewardship WEMI 1-to-1
Categories: Peer Reviewed Conference
Content Mediums: MovingImage Text Event