Comments on ISLE Metadata set

I had a go at applying the coding schema to the following language data resource:

Elicitation session in Nyamal/English (recorded near Port Hedland, Western Australia, 8 May 1993)

Are the categories clearly named and described?

For the most part, yes. However, the difference between “Session . Description” and “Content . Description” was not clear enough. “Informant . Language+” also not clearly defined.

Do they allow you to enter the right information?

I had some concerns about how best to manage different media recordings made at the one session. For example, an elicitation session includes an audio-taped record as well as a set of prepared questions (in some cases) and fieldnotes. It appeared to me that a decision needed to be made as to the primary record and other sets of information are then listed as associated resources. The logical decision to make is to list the audio-tape as the primary record with the transcript of this record, its keyboarded (electronic) version (prepared soon after the session), questions prepared before the session and notes made during the session, listed together as associated resources. It is possible that not all recorders would make these same decisions about the primary and affiliated resources.

This complex then raises the question of access rights. The coding scheme appeared to address access rights as applying mainly to the primary resource. However, the different media involved in the recording of a session and their relative levels of sensitivity as well as the extent to which they represent different kinds of intellectual property owned by different participants (speaker, interviewer, granting agency etc.) make this potentially quite complex. It seemed to me that the coding for access rights needed to be applied to each linked resource. It may be that this is assumed, but if so it wasn’t obvious to me.

Are these the categories you would find useful in retrieving other data or documentation?

Yes. Some of the categories used here are quite rich . For example, “Interviewer . Role” allows the coding of information which may be very useful in decoding some aspects of the speech events in the recording (eg. kin relationships between participants which determine choices in code) and which might also allow sophisticated searching.

However, one thing that became very clear to me in the attempt was how time consuming and resource intensive the preparation of metadata sets is likely to be. (This is noted in the introduction to the ISLE document.) I have not made detailed records of my own collected material that approaches the degree of delicacy envisaged by the coding scheme given here (mea maxima culpa) and I have some sense of the costs of retrospectively cataloguing this material. As the ISLE coding scheme suggests, single recordings might be sub-divided (at the depositor’s discretion) and the parts coded as separate “sessions”. Thus the elicitation session I chose involved conversation in (non-standard) English, elicitation in English/Nyamal, and conversation in Nyamal (when the session was joined by a third party) as well as some mixed (rather than code-switching) English-Nyamal. The audio-taped song ceremony includes songs in various Australian languages, interspersed by conversations, harangues, etc. in (non-standard) English, and Australian languages relating to the business of the ceremony as well as radio broadcasts of cyclone warnings in standard English which were also discussed in a variety of languages. In each of these cases, sub-division into a series of sessions/speech-events and the metadata coding of each of these would provide a very rich resource for future researchers. But it would come at a price.

All users of the database will want as much detail as possible (about everyone else’s resources, at least) but it is likely that any database will include metadata of widely varying degrees of detail with respect to the resources they represent. What problems/frustrations do such systems create for their users/managers and is there a way to code for the complexity of the metadata. Meta-metadata?

Alan Dench