Advances in Role and Reference Grammar

460 LINDA SCHWARTZ

syntactic accounts of split intransitivity provide no motivation for the con­

sistent behavior of attributives and identificationals as a class in some lan­

guages and not in others. Rather, without a semantic alignment hypothesis,

the expectation would be that attributives and identificationals would not

behave as a class; the fact that they do for all predicate-based parameters is

not explainable in a purely syntactic analysis. On the other hand, the

account of these constructions presented here entails their behavior as a

class with respect to a given criterion (such as undergoerhood or typical

undergoerhood) but allows for cross-linguistic and language-internal cross-

construction variability based on typicality conditions, defaults, and the

notion of argument-based vs. verb-based parameters. Moreover, this

analysis is consistent with the Motivated Alignment Hypothesis and thus

weakens the case for a multistratal syntactic account of the subjective-

objective distinction.

Notes

* I am grateful to Marshall Lewis and Gerald Sanders for discussing various ideas

on this topic with me and to Carol Rosen and Robert Van Valin Jr., whose com­

ments on earlier drafts have led to substantial improvements in the paper. Steve

Franks, Manuela Gieri, Françoise Mennechet, Paul Newman, Caroline Pour-

query de Boisserin, and Sammani Sani have provided data and information on the

languages cited here.

1. These classes are standardly referred to as "unergative" and "unaccusative"
   respectively in Relational Grammar, "intransitive" and "ergative" in Govern­
   ment-Binding.


2. As noted in Van Valin (1990), the discussion of objectivity vs. subjectivity does
   not always distinguish those patterns which are verb-encoded, such as auxiliary
   selection in Italian, from those which are noun-encoded, such as case-marking in
   Choctaw. This is a crucial distinction, since the factors which induce verb-based
   ergativity are claimed to be parameters of verbal semantics, while various other
   factors affect nominal case-marking patterns. In this paper, the distinction is rele­
   vant to the patterning of attributive and identificational constructions in Dakota.


3. This is not a universal pattern, however; see the Dakota data presented in this sec­
   tion.


4. Russian is somewhat more variable in the unaccusative-unergative than French,
   Dakota and Italian. Pesetsky (1982) suggests that this may be due to the lack of a
   clear and systematic morphological distinction such as auxiliary selection which
   makes the distinction more salient.