By "generative grammar", I take the widest interpretation and do not mean "Chomsky's theory of syntax today", thus HPSG and LFG would be instances of GG(broad). Phonology has a concept "feature", and there are two competing theories of features. One is that a feature is a value-attribut pair, for example "-coronal" where the attribute "coronal" has the value "-". There are various theories of what the values are (+,-,integers,u,m....) and what the attributes are: I don't care about that. The competing theory is the privative theory, that features are only attributes.
Earlier versions of syntactic theory e.g. Aspects employed a value-attribute model so that there are things that are [+N,-V]. My question is where in any domain other than phonology, has it been posited that features are privative, that is, are they just attributes and never values. My recollection of HPSG is hazy enough that I don't know if that is an example. It would be useful for me to know about mixed theories (where some features might be binary and come privative). I don't know anything about the theory of features in minimalist syntax.