It is impossible to get information with examples on the intended usage of pVar.
One would think that section 9.4 at http://www.tei-c.org/release/doc/tei-p5-doc/en/html/DI.html#DIHW would explain the intended usage of this element, but it only mentions it as if it were illustrated in the examples. (And the "as if" is why I classify this as a bug report).
In particular, the "'duhv(z)" part feels like a candidate here, but in the example, it is handled by means of <pron> with the @extent attribute.
The fact is that pVar, pRef, oVar and oRef are making a bad system and the extensive revamping activity we are doing with Jack Bowers on the dictionary chapter makes us think of a much simpler system (basically just two elements for o- and p- with integrated Var/Ref, the difference being superfluous). You'll get a feature request soon.
Indeed, this system looks pretty bad to me on the whole. I wonder whether it's used and to what extent (like, whether it's used as a complete set of elements). I recall using just oRef for some nifty presentational effects once, but that was long ago, in the time of Firefox 2.x, which was still a relatively good visualiser for XML.
Another issue is the discrepancy between @type attributes for oRef and oVar (but, as is often the case with @type, this issue can be weaseled out of, I imagine).
Providing a clear guidance on the value for @type should be indeed part of the revamping activity. More in 12 hours, when the plane will have landed.
Laurent has opened [feature-requests:#545], which should will address this, so closing this bug in favor of that.
Related
Feature Requests: #545