On the surface, Basque looks like an ergative language: Case morphology groups direct objects and (most) intransitive subjects together as unmarked absolutive, separate from transitive subjects, which have marked ergative case. However, its case system has so far defied satisfactory explanation, and the two most promising current theories of ergative cannot account for the distribution of this case in Basque subjects. In the dependent case approach (i.a. Marantz 1991, Baker 2015), ergative is restricted to transitive subjects, as it is assigned to a nominal that c-commands another nominal in a sentence. However, many Basque intranisitive verbs govern ergative subjects, and attempts to reanalyze them as underlyingly transitive predicates have been shown to be wrong (i.a. Preminger 2012). An alternative approach to ergative case claims that ergative is inherent, assigned specifically to external arguments (i.a. Woolford 1997, Legate 2008). This makes correct predictions for transitive subjects, which are typical external arguments, and also for many agent-like ergative intransitive subjects in Basque. However, this entails that ergative cannot be assigned to derived subjects (since inherent case is by definition restricted to underived positions), which has been refuted in a number of recent works (most prominently, Rezac, Albizu, and Etxepare 2014).
In this talk, I propose that so-called ergative case in Basque is in fact not ergative case. Rather, the case that surfaces on transitive subjects is nominative of a special kind, labelled "marked nominative" in the typological literature (referring to the fact that, unlike most instances of nominative, the exponence of subject case in some languages is marked). Specifically, I adopt Baker's (2015) analysis of marked nominative, according to which it is assigned in the clausal domain (TP) to a nominal that is not c-commanded by another nominal. In addition, I also adopt Baker's proposal that a clause has two case domains. The higher case domain is the clause itself (TP), and the lower one is VP. Thus, a subject is assigned marked nominative if and only if it is the highest VP-external nominal in a clause. This predicts, correctly, that external arguments of transitive verbs have marked nominative, since they are generated VP-externally. While many intransitive subjects are VP-internal, others are VP-external (either by base-generation or movement), which explains why some intransitive subjects have marked nominative.
As evidence for this approach to ergative as marked nominative, I argue that it overcomes the difficulties for standard theories of ergative mentioned above. In a nutshell: Assignment of marked nominative is not dependent on the presence of a lower nominal, and it need not be assigned to a nominal in its base position. Furthermore, together with the claim that VP is a lower case domain in Basque, the analysis also makes correct predictions for the distribution of dative case in sentences with ditransitives, causatives, psych predicates, and differential object marking (widely attested in substandard Basque, i.a. Rodriguez 2016).