As part of the First INNET conference at the University of Cologne, Sebastian Drude of The Language Archive initiated a panel discussion on interoperability between endangered languages archives. The panel consisted of Sebastian Drude, Gary Holton (Alaska Native Language Archive, Mary Linn (Sam Noble Museum, University of Oklahoma), Nick Thieberger (Paradisec), and David Nathan and Kakia Chatsiou (Endangered Languages Archive). Sebastian introduced the session by describing a scenario for archive interoperability in support of users’ resource discovery and access. In the scenario, archive users would be able, through a “single sign in” to a federated archive portal, search for and access not only all the member archives’ holdings for a particular language, but also take such search to the level of individual resources. For example, a user could search for all occurrences of relative clauses in the combined holdings.
A vigorous discussion ensued, with several points raised by the panel and audience for implementing this vision. The broad picture is that while TLA is involved in multinational European projects such as CLARIN involving nationally-recognised archives where interoperability is a key requirement, the other archives have few comparable aspirations for high levels of interoperability. Gary Holton wondered whether our archives should wait for other fields with more resources to develop such interoperability infrastructure, postulating they would do so in the next 5 years, and then we could adopt their solutions, just as much of the archives’ existing infrastructure is based on broader developments. Nick Thieberger described Paradisec’s focus on its data collection and preservation work, and this, combined with the heterogeneity of different archives’ approaches to metadata, access and presentation, means that interoperability should be at the collection level, using simple standard metadata that can be aggregated by the OLAC portal. There, users can find out about the location of resources and need to go to the holding archive and use its methods to gain access to materials. Most panelists agreed that interoperability would find the most common ground by applying to resource discovery using metadata at the collection/deposit level. Other points that were raised include:
- what are users’ actual needs? Have archive users been requesting high levels of interoperability?
- interoperability is one component of the broader concept of the accessibility of materials and content (which also includes usability of sites etc). Thus, appropriate interface mechanisms e.g. for visualisation of search results, presentation of interfaces in different languages, would also be important
- interoperability is also one aspect of the “distribution of labour”, i.e. where it is not implemented, users who would benefit from it must provide the effort themselves, for example by searching multiple times in different archives. Providing interoperability shifts efforts to archives to build and maintain the enabling services
- even under an advanced level of interoperability, it would be difficult to serve diverse users, for example if ornithologists or ethnobiologists were interested in bird calls, these are unlikely to have been annotated in searchable data. And for heritage speakers and broader audiences, access via terminologies and precise spelling of search terms can present difficulty
- at what level should interoperability apply? Paradisec and DoBeS store all metadata in the same directories as the relevant resources, so that even if the access application layer is changed or disrupted, the integrity of the collections can be recovered by a user or by building another access application layer.
- re access restrictions and access rights: how necessary are restrictions? While all the represented archives acknowledged that respect of access restrictions is inherent to working with endangered languages documentation, each archive has different policies and implementations. Some archives have a binary open/closed system, while others have systems offering graded access (depending on the materials and the status of the particular user). ELAR has a sunset clause in its deposit agreement which can be used by discretion to ask depositors to re-motivate restrictions after 3 years. DoBeS is currently revising its access policy.
There was also some discussion of sign-on/authentication systems, how to include community members and wider audiences within access mechanisms, and the feasibility of federated access using common methods such as Shibboleth, Facebook, or Twitter. An audience member drew attention to an open source solution, OpenID.
Finally, the panelists noted that interoperability is also applicable to other aspects of our work, for example in considering how well our websites operate on the full range of digital platforms (include mobile, tablets, iOS/Android). Also valuable is a basic interoperability for preservation, for example so that an archive can share its data for replication or even for transfer to another archive if an archive is closed and its data orphaned.
The panel thanked Sebastian for initiating such a stimulating discussion, and also Dagmar Jung, the organiser of the INNET conference.