156 Małgorzata Krzek
would enter an Agree relation with the pronoun but the reflex of that agreement would
not be overtly represented in the form of inflectional marking as the suffixal functional
element has its morphological features established in the lexicon. I am not going to
commit myself to any of the accounts as both of them lead to the conclusion that the
pronoun in question does not trigger an overt agreement marking on the verb it enters
an Agree relation with.
Holmberg (2010b) argues, following Vainikka (1989), that in Finnish it is pos-
sible to determine with near-certainty whether a pronoun triggers agreement or
whether it is established by default. This is because there is a correlation between
case and agreement in Finnish; that is, the object is marked accusative when there is
a subject agreement marking present on the verb. This is illustrated by the example
in (37).
(37) Me ostetaan auto/*auton.^31
we.nom buy.pass car.nom/car.acc
‘We’re buying a car.’
(Finnish, Holmberg 2010b: 207)
The example in (37) demonstrates that it is the presence of subject agreement that is
a crucial factor triggering assignment of accusative on the object, not the presence
of an overt nominative subject. Therefore, when the object is marked nominative,^32
it is taken to indicate that the verb does not agree with the subject, and as a result
agreement is assigned by default. Taking this observation as a diagnostic, Holmberg
(2010b) shows that in the impersonal construction the object appears in the accusa-
tive, which indicates that the 3sg form is not assigned by default but is assigned by
the null generic pronoun. This, in turn, suggests that that the impersonal pronoun
in Finnish has 3sg value. However, a similar type of diagnostic cannot be applied for
Polish. The reason being that in Polish there is no correlation between the presence of
subject agreement and accusative case on objects. This is illustrated by the sentences
in (38).- This sentence is interesting for a number of reasons. Some of them are given here. First,
although the subject is nominative and located in SpecTP (Anders Holmberg p.c.), it still
does not trigger agreement. It appears then that only some nominative subjects can trigger
agreement but others cannot. The question is what makes a nominative subject trigger subject
agreement marking on the verb. Secondly, it is not clear why, although the morphology on
the verb is passive, the subject has not been demoted. This seems to run against all major hy-
potheses as to how passive constructions are generated. The discussion of these problems is,
however, beyond the scope of this paper. - Holmberg (2010b: 206) notes that nominative object case is different from nominative
subject case in that it does not trigger agreement.