This is perhaps a more theoretical question (better for the mailing list?), but how do DELPH-INers and HPSGers generally treat alienability distinctions? Is it typically a lexically required SPEC, or something else?
There may be something useful in Elizabeth Nielsen’s thesis. I don’t think she took on alienable v. inalienable directly, but she does have a way of marking the two parts of doubly marked possessive constructions, so that a possessum marked as such needs to co-occur with a possessor.
In general, is it the case that alienable/inalienable always lines up with optional/obligatorily co-occurring with a possessor though?
- Elizabeth Nielsen 2018 Modeling Adnominal Possession in the LinGO Grammar Matrix
Empirically, no, alienability and obligatory possession don’t always line up. But “alienability” means a lot of things to different people. Alexiadou 2003 says that underlyingly you have a difference anyway in terms of the argument structure of the noun, even if on the surface they look the same. I’m not saying I buy this, but it has been said.
But yes I meant more strictly how obligatorily possessed nouns were modeled.
It seems that there is some connection here to the ERG’s treatment of possessed idioms as in “She wracked her brains” or “We waited our turn”. The analysis is described in this 2015 paper:
The implementation in the ERG closely follows this analysis, with the semantics of such idiomatic verbs adding an identity relation between the index of the possessor and that of the relevant argument of the verb. See specifically the definition of the type np_trans_refl_verb in erg/lextypes.tdl, which inherits from the semantic linking type poss_idiom_lt defined in the same file. It might be a better design to eventually have this type instead add an ICONS identity rather than an additional id_rel EP.