There is no third face of agreement

The main claim of Alsina & Arsenijević 2012a, hence forth A&A, can be stated as in 1.

1. 1. There is no evidence for the idea that there are three potentially distinct sets of features needed for accounting for agreement facts—two syntactic sets and one semantic set.

The kind of argument that would show this claim to be false or incorrect is one that presents a fact, or a set of facts, that can only be explained by positing two syntactic sets of agreement features, in addition to the corresponding semantic information, that is, that requires the 2SAH (the dual syntactic agreement hypothesis). None of the arguments in Wechlser & Zlatić 2012, henceforth W&Z'12, is of this kind. The arguments in W&Z'12 are mostly about possible problems that the analysis in A& A might face when applied to constructions not discussed in A&A.

A&A focuses on hybrid nouns in Serbo-Croatian (S-C), since such nouns have been the main source of argumentation for the 2SAH in Wechsler & Zlatić 2000 (W&Z) and related work, and states generalizations within the 1SAH that account for the agreement facts involving such nouns. The argument in W&Z'12 based on the data in their examples 2-3 and 5 simply indicates that syntactic gender has a default association with semantic gender, in which masculine corresponds to male and feminine to female.¹ The argument illustrated by example 4 in W&Z'12 shows that syntactic number, like semantic gender, has a default semantic correlate, by which plural signals more than one entity, and that this semantic value emerges in certain constructions, such as the appositive construction in the example.² If singular agreement is shown to be the default choice for finite verb agreement, as argued in W&Z'12 on the basis of subjectless sentences and clausal subjects, all we need to do is to rephrase the statement about agreement in A&A's 4 (p. 373) so that it reads: 'A plural verb form is required if the subject is plural, either semantically or syntactically; otherwise, a singular form is used'. When the syntactic head triggering agreement is of a category that does not carry the features involved in agreement, such as a clause, an infinitive, or an adverb (see W&Z'12 exx. 6 and 8), singular agreement is predicted.³ As can easily be seen, all of these observations are not arguments for the 2SAH and are perfectly compatible with a theory that assumes the 1SAH.

The one instance in which the 2SAH seems to allow for a simpler statement than the 1SAH concerns personal pronouns. In W&Z, personal pronouns are said to agree with their antecedents either in index features or in semantic features. In the proposal in A&A, pronouns agree with their antecedents either syntactically or semantically, with syntactic agreement restricted to nominative case forms. The statement in W&Z avoids this restriction, but it hinges on the homophony of all neuter plural forms of pronouns with some other form: the nominative with the nominative feminine singular, and the remaining forms with the masculine/feminine plural form of the same case value. The price for dispensing with that restriction is (a) the massive complication that the 2SAH represents and (b) the stipulation that pronouns agree in index features, and not in concord features, with their triggers. And no empirical gain follows from this. In addition, restrictions relating case form to the choice of semantic or syntactic agreement are needed anyway, because relative pronouns allow semantic agreement only in certain case forms, as shown in W&Z (p. 819) (see Alsina & Arsenijević 2012b for an account).

W&Z'12 defends the neuter plural analysis of certain elements such as the pronoun ona agreeing with deca-type nouns by referring to Corbett (1983), rather than presenting the particular arguments. But A&A already explicitly takes care of these arguments, showing that they either are fully compatible with the feminine singular analysis of those elements or are based on incomplete empirical generalizations (as in the case of...