The structure of null subject DPs and agreement in Polish impersonal constructions 141
its referent has already been mentioned in the discourse. In (23c) and (23d), an
overt 3rd person plural pronoun receives specific interpretation; that is, the speaker
knows which group of people the pronoun ‘they’ refers to. When it is null, however,
the meaning is equivalent to that of an indefinite plural ‘somebody’. If it is bound,
then the referent is a specific group of people, just as in the case of the overt 3rd
person plural pronoun. In other words, only null 3rd person pronouns, and not
the 1st/2nd person pronouns, can be variables whose interpretation will change
depending on the specification of an A-topic that these null 3rd person pronouns
are bound by.^11 The 1st/2nd person pronouns, on the other hand, receive their speci-
fication from the speech features that provide values for Φ-features. However, when
it comes to null subject pronouns in impersonal constructions, things are not as
straightforward. First of all, the range of interpretations that these null impersonal
pronouns can have exceeds the ones of 1st/2nd/3rd pronouns. Impersonal null
pronouns can be interpreted as generic inclusive (referring to all people with the
speaker included) and exclusive (with the speaker excluded), arbitrary (referring
to some group of people), and specific (referring either to the speaker or a group
including the speaker; may be regarded as an equivalent to ‘we’). Because of these
various interpretations, null impersonal pronouns cannot be said to behave exactly
like either 1st/2nd person pronouns or 3rd person pronouns; that is, their inter-
pretation cannot be simply due to them being bound by either the speech features
or the referential A-Topic. Rather, the interpretation of null impersonal pronouns
appears to be the result of them being bound by speech features (among other ele-
ments), but not by an A-topic, as the very existence of non-specific (understood
here as unfamiliar/non-identifiable) topics is questionable. Instead, in order to
acquire generic/arbitrary readings, they will have to be bound by operators (e.g. the
generic and existential operator), Q-adverbs and speech features, all of which are
located in the CP. This, however, is only possible if it is hypothesized that pronouns
are internally complex bundles of unbound but interpretable features that can be
- A classic example illustrating that fact is the observation that only null pronouns can be
bound by a quantifier. Compare the sentences below:
(i) Każdy studenti sądzi że proi jest inteligentny.
every student thinks that (he) is intelligent
‘Every studenti thinks that hei is intelligent.’
(ii) Każdy studenti sądzi że on*i/j jest inteligentny.
every student thinks that he is intelligent
‘Every studenti thinks that hej is intelligent.’