Why don't linguists publish their field recordings to the public?

[Sirabun]

Last century eye-witnessed a great development in all fields of linguistics including phonetics, phonology, morphology, syntax…but there is a problem now still need to be solved: how to get the language data , especially for those rare languages that are being used by a small number of people or(/and) in an isolated area which is hard to reach. Some of these languages have been distinct or changed since being recorded last time.So it'll be very convenient for other scholars and students to study them if linguists are open to publish there recordings(including video and audio) to others.
Only transcription with IPA is far from enough!
Apart from the rarity , publishing raw material is the need for falsifiability (better with intermediate statistic materials) because other scholar can do the same experiment and analysis again. There are even something wrong in the intermediate statistic procedure so others can point them out.
It is not only the problem of copyright or technology (online technology has developed so much and there are already some amateurs doing so)  but  nearly every linguist keep their own materials in the pocket and at most share them with several colleagues.
Comparing to archaeology, nearly every excavation follows a systemic publishing in the form of book even online video where you can see the original materials so why don't linguists learn from archaeologists?

[MalFet]

Welcome to the forums!

The answer to your question, unfortunately, is very simple: privacy. Ever since Tuskegee, academic linguists have had to pass their protocols through institutional review boards, which generally require extremely strict consent procedures for the publication of anything identifiable (especially from socioeconomically marginalized people). That makes audio hard and video even harder.

[Daniel]

Hmm... not necessarily. All that is required, at least at my university, is consent from the participants, and additional paperwork involved in sharing audio/video. If all field linguists were to get that permission before recording, then they could share later, as long as consent was involved.

In fact, I think this is the direction the discipline is headed.

Another problem has been technology, where audio recording was expensive or impractical for years, so it was not done all the time, and now even video recording is easy. You can fit 1000 hours easily on a single hard drive, without needing tapes or other storage.


But then the issue is again privacy: do these speakers want their recordings shared? It's very important that we respect what they want. Archaeology doesn't have that problem: no 3,000 year old mummy is going to complain if we post its picture on Wikipedia.


Regardless, some material is in fact available. There are various archives that host it, although sometimes it's in limited amounts.

Another question is how valuable sharing everything would be. Maybe allowing access to everything would help some, but usually it's just other researchers who want that much access. A sample of recordings is usually enough for the general public. And researchers often do share with other researchers-- so if you're working on a project, you could ask someone about their data.

[MalFet]

Hmm... not necessarily. All that is required, at least at my university, is consent from the participants, and additional paperwork involved in sharing audio/video. If all field linguists were to get that permission before recording, then they could share later, as long as consent was involved.

Have you ever actually gone through this process for real, primary fieldwork? In cross-cultural contexts, with people who are often non-literate, obtaining informed consent is often very complicated (as it should be). There are many field linguists who make it happen, but many more decide that it's either not possible or not worth the effort.

[panini]

I'll tell you a second reason why: data-mining. Things written in a linguist's notebook are not all "facts", not even the transcriptions. They have to be interpreted in the context of the eliciter's knowledge of the speaker's behavior, e.g. what the real meaning of "I suppose so" is in response to a "can you say?" question. Speakers often go through a process of learning how to respond to questions, and may need to re-familiarize themselves with facts of that language as opposed to a related dominant language in the area. Mindless data-mining is an accepted research strategy, and there is no way to protect against opportunistically harvesting an unrepresentative token from raw notes. By controlling access to the product of elicitation, the linguist can protect against such misuses.

There are other ways to approach data-gathering, and this leads to the distinction between "language documentation" versus "language description". The practice of "language documentation" focuses on gathering and making published a finite corpus of material on a language, along with elaborate meta-data (which calls for complete videotaping or at least session audio). Analysis is an optional aspect of documentation -- the primary goal is creating a document. "Language description" on the other hand puts the primary emphasis on discovering higher-level generalizations about the language, based on some gathered data. Since the number of minutes that you can spend on a project is finite, you have to make choices about what your primary goal is.

I also hold that even with signed consent forms, raw recordings should not be make public unless the participants very carefully review the material, since they are likely to have forgotten themselves and say something like "Do you know what my idiot brother did today?", forgetting that they were being recorded.

[Daniel]

I agree with all of those points, but I think data-mining should also be weighed against just having the information. Is it better to not record/share/retain/have information just because some of it may be wrong?

This is why I'd think that more complete releases of notes and recordings would be better: the idea would be that you could get the context. It wouldn't be meant for general public consumption or transmission on the radio. But a sufficiently motivated linguist could access them and do something useful, beyond what those finite-time constraints you mention limit the original research from doing.

But the bottom line is that once it's lost, it's lost. And I find it a hard argument to think that it's easier to redo the whole project than to properly interpret the notes and recordings. It should be done with thought of course! I've come across some unpolished descriptions of languages that explicitly ask the reader to be skeptical and note that there may be mistakes. I don't think there's anything terribly wrong with that, especially until someone else can actually do a better job. With the recordings, that would be possible, without needing to find speakers, or even after there are no more speakers.

they are likely to have forgotten themselves and say something like "Do you know what my idiot brother did today?", forgetting that they were being recorded.

True. So maybe there should be a delay on releasing the recordings? That seems backwards for science, but it would at least preserve the information.