ATTRIBUTIVES AND IDENTIFICATIONALS 459of incorporation, to my knowledge. In the examples in (74), the attribute is
categorially a noun marked with the preposition da, which is a very general
linker of coordinations, comitatives, instrumentais, means, etc. In (75), the
attribute is again categorially a noun, marked by the preposition a which in
other contructions marks location or goal.^19 These constructions at least
seem to be clear cases where there is no incorporation of the attribute or
identification-set. This is clearest in (74) and (75), where the NP corres
ponding to that property is marked like an oblique NP.
5. Conclusion
I have tried to show in the above sections that the varied cross-linguistic
patterning of attributive and identificational constructions with respect to
the subjective-objective opposition is consistent with the Motivated Align
ment Hypothesis. The status of the subject of attributives and identifica-
tionals as undergoer will sometimes determine their patterning, in which
case they will go with the objective class as in Italian auxiliary selection.
The status of their subjects as non-typical undergoers will sometimes pre
clude their patterning with objectives, in which case they will go with the
subjective class as a default, as in Italian ne-insertion, French //-inversion
and en-insertion, and Russian genitive of negation. In a case where an argu
ment-based parameter rather than a predicate-based parameter is relevant
to the pattern, e.g., animacy in Dakota, they will not distribute as a class if
the parameter cross-classifies them (as it does in this case). This account
will entail the specification in the lexicon of the members of the marked
class in those instances where the semantic conditions are only necessary
but not sufficient for membership; this would be the case for Dakota, since
in Dakota the subjective-objective opposition is fully lexicalized (i.e. where
each predicate can appear in only one pattern).
Burzio (1986) claims that the fact that objectivity is construction-
specific rather than verb-specific (e.g. passives, and middles patterning con
sistently as objectives, while active transitive clauses with the same verbs
pattern consistently as subjectives in Italian) is evidence that a syntactic
account of objectivity is necessary. What has been shown here and also in
Centineo (1986) and Van Valin (1987, 1990) is that it is the logical structure
of the syntactic construction which determines the subjective or objec
tive pattern. More specific to the problem addressed in this paper, strictly