366 LAURA Α. MICHAELIS
3.3 Jensen (1983)In contrast to the two analyses discussed previously, Jensen's analysis of
deviant case-marking in Latin takes as its primary focus the relationship
between impersonal passivization and deviant case, arguing for a lexical
formalism for passive that also incorporates the distinction, advanced
within Government and Binding theory, between lexically and structurally
assigned case (i.e., inherent case and case assigned under government).
Quirky case is said to be of the former type, normal case of the latter.
According to Jensen, all instances of deviant case-marking reflect a situa
tion in which "a particular case is required of a complement phrase of a lex
ical item by virtue of being stated in that item's (lexical) subcategorization
frame," thus restricting lexical insertion to "inserting an item from the lexi
con generated with that case" (p. 25). Jensen also assumes that case is
determined by government where no case is specified in the subcategoriza-
ton frame, hence yielding the standard nominative case for subjects (via
INFL government) and accusative case for objects (via verb government).
The distinction between inherent and structural case is relevant to the
lexical redundancy rule proposed by Jensen to link active verbs to their pas
sive forms. A lexical rule for passive which did not incorporate this distinc
tion would not, according to Jensen, enable one to explain the fact that
non-accusative objects never become passive subjects in Latin or, for that
matter, in any other language allowing such objects. Jensen's lexical rule
allows only a verb that is both morphologically non-passive and followed by
noun phrase with no lexically assigned case features to be related to another
verb phrase containing a morphologically passive verb followed by an
empty noun-phrase node (said by Jensen to bear a trace of some sort,
despite the fact that passive for him does not constitute a movement rule).
In keeping with GB theory, Jensen assumes that the subject of passive verb
is assigned a theta-role by dint of this object "trace". This rule excludes
from passivization both one-place predicates (which, of course lack the
requisite object) and two-place predicates having an object which bears a
case feature. Hence, both of these verb classes must form impersonal rather
than personal passives. With respect to the passives of such verbs as curro
("I run"), Jensen argues (p. 32) that "[s]ince passive verbs do not assign
theta-roles to their subjects, the subject of an intransitive passive cannot be
assigned a theta-role normally given to passive subjects through the object
trace. Such passives are therefore impersonal" and assume the unmarked