Transient Languages & Cultures

All over Australia now people are writing reports on the progress of their grants - to attach to their begging-letters for more grants. Reading the reports gives you the sense that Australia is a garden of projects, each a mass of bright blossoms fragrant with success. (So why haven't we solved world poverty or climate change yet?) That's why it was really really good to go along to the ARC E-Research post-funding workshop (14-15 February), where participants were encouraged to report on the problems they encountered in their projects...

E-research is about using new technologies to create new kinds of data and to visualise data in new ways. The workshop brought together a range of humanities and science projects, based on their common use of large data stores, high speed grids, and similar computing issues. The format was speed-posters with four minute presentations. It was held in the Members' Dining Room in Old Parliament House, an elegant but wireless-less room. So participants had to listen or make notes on their nice new notepads (branded with the name of the caterer, source of rather good small eclairs), rather than sink into their e-mail. Project spruikers couldn't use their allocated four minutes to show us their stuff - a pain for projects whose goal was to increase connectivity by web access, and for anyone who wanted to click on the URLs of the projects we were hearing about.

Two projects of relevance to linguists involved annotating language data, the Ethno-ER project (partly sponsored by PARADISEC) which has produced EOPAS, a tool for online representation of interlinear text linked to audio and video, which Nick Thieberger talked about, and DADA-HCS, which Steve Cassidy and Roland Goecke presented, and which might provide a back-end for EOPAS.

It was striking how many concerns were shared by the projects (science, medicine, humanities), and how many of their concerns are familiar to people trying to manage linguistic fieldwork data.

Take the International data exchange for global gravitational wave astronomy. They want to share data internationally - just as digital archives of language material do. But..

Data sharing.. is all special cases, both bureaucratically and technologically. Just tracking software and configurations is a full-time job.

And take the MRI Data and computational facility people. They want to develop a work flow for acquiring data, providing meta-data and distributing MRI scans. Just like the kind of work-flow PARADISEC has been developing. But the presenter said that they'd realised that "Our data wasn't well enough organised to take advantage of" - so they hooked up with a commercial digital assets management company to help them. PARADISEC and Ethno-ER have managed to collaborate with researchers in universities and CSIRO to get some of the programming work done, but DADA-HCS noted the problem in competing with industry for recruiting good programmers.

Something which sent a shudder of guilty recognition through me came from a consortium of experimental protein folders. They want to exchange complex data on protein folding which requires accurate metadata, but

Users are too lazy to deposit their data. The simplest UI (User Interface?) is often too perplexing for the average biologist!

This was echoed by a project to develop web-surveying for epidemiological research: "If they can break it they will", and, more politely on their powerpoint: 'Participant proofing' is necessary for survey administration and getting clean data.

The Earthbyte people (makers of fine time-maps showing plate tectonics) emphasised how much longer it takes to develop software and infrastructures that conform to standards, rather than what they called "hero code" (which I think meant once-off stuff). Again, familiar to followers of PARADISEC.

A CI on another brain scan project that led to the development of the Australian Schizophrenia Research Bank was talking about how to cope with lots of little collections of brain scans and "aggressive host firewalls" to create a repository. Again, familiar to language data archives.

The complications of bureaucracy and collaboration were mentioned by the iDIG CI Howard Morphy (Indigenous collections and knowledge archives research network) - they had a prototype web portal for museums to make their data accessible to Indigenous communities. Another achievement was getting an MOU between the collaborators to allow the material to be accessed. It mightn't seem much, but in these days of brutal contracts, expensive insurance and paranoia over worst case scenarios, it really is an achievement. (Ethno-ER suffered from a fifteen month delay in getting its MOU signed, despite much prodding).

Many of the projects involved the creation of repositories. Thus the Australian Schizophrenia Research Bank hopes to be "the world's biggest online mental health research facility", and the other repositories of scientific data look pretty large too. One problem I bet they share with language data archives is the need for an eternity pill. That is, guaranteed long-term funding to maintain the repository.

There's not much hope of the eternity pill until governments and grant agencies and publishers require supporting data-sets to be lodged in public repositories. And even then, it will take a while for those public repositories to get the funding to maintain the data properly. In the meantime I fear that all too many important data-sets will disappear into black holes of old technology.

Back to begging-letters..