A query...

[Daniel L. Everett]

>
> We're talking about two different things - data accountability, and  
> widespread data access. Full accountability of the researcher and  
> replicability of results is one thing, and I'd argue that this is  
> one of the things which makes any particular piece of research  
> "scientific" versus "humanistic". How data are disseminated is  
> another question. In the article that started this discussion, for  
> example, the issues wasn't that the guy didn't put his data on the  
> web, it's that he fabricated the results and then lied about it. If  
> he'd put the fabricated data on the web, no one would have been  
> better off.

To fabricate linguistic sound files would be hard to pull off well,  
though. We wouldn't be putting up statistics. Our data is more  
concrete in that sense.

Of course, someone can fail to include information that is contrary  
to them. So we need to do as much as we can to ensure that the bonds  
of trust in the field stay as strong as possible.

How can you hope to replicate results in a grammar (which is never  
fully replicable, of course) unless the data are available in a  
neutral space or you make your own field trip.

Ultimately, I think we need both. All the data available to all and  
more fieldwork on every grammar.

Dan