This is very much my own perspective (at least on the work side). We use the TEI because is has a standard vocabulary that allows us to capture distinctions that are commonly made by scholars when they publish the kinds of texts we work on, and because the format supports presentation and search in the ways we need it to. So I agree with Doug, even though I wish he'd stop making fun of obscure languages.
Of course, nothing stops you from using that vocabulary for literary criticism either.
I think we're off in the weeds here. I'm not sure there's anybody who can say, with any authority, "the primary purpose of TEI is to do X." Different projects use it for different (if likely related) purposes because they find it useful.
Something Martin said makes me think though:
>> I have no idea how many people actually use
>> TEI, or how we could know, but it certainly isn't 50.
Why don't we know this, at least in a general way? Has anyone made an effort to find out? I keep coming back to this, but I suspect that the TEI "community" bears little resemblance to the TEI membership, and I think that's both a testimony to the success of the project and a serious problem for it going forward.