Jaume Solà i Pujols - Departament de Filologia Catalana ...

Suppose that in both cases no DP is available for AGR to coindex with:
- in the impersonal passive case, because the Agent Argument is not projected as a DP;
- in the other case, because these verbs do not subcategorize for any DP not being
inherently Case marked. 50
The present account, based on parameter 0, gives a unitary explanation for the existence
of impersonal passives and the class of verbs in 0 in the same language: languages lacking
impersonal passives do not have verbs of this kind. Icelandic would be another instance of such a
correlation. 51 Actually, it is not clear that this correlation is a genuine one. To my knowledge,
Russian is a language having verbs similar to the ones in 0 without having impersonal passives.
Perhaps we could simply assume that the parametric option set for German is a necessary but not
sufficient condition for the existence of impersonal passives, whose licensing would depend on
other factors or, perhaps, would simply be a matter of idiosyncrasy.
As far as I know, languages allowing impersonal constructions like 0 (German, Russian,
Icelandic) are all languages overt Case marking on DPs. If this generalization is genuine, it
should be captured by the theory. I do not have any interesting proposal in this connection.
50 Some of these verbs subcategorize for an Accusative:
(i) dass (es) mich dürstet
that (it) me-ACC is-thirsty
We have to assume that this is an inherent Accusative, and
is already Case-marked by D-structure.
51 Italian has two verbs of that germanic type: importare
('concern') and dispiacere ('dislike') (thanks to L. Rizzi for
this remark):
(i) A me dispiace di questo
To me dislikes of this
(ii) A me importa di questo
To me concerns of this
Perhaps they are a residue of an earlier period where
Italian had impersonal passives (and so the German value of the
parameter).
French shows another case of a verb that allows AGR not to
coindex with any Argument:
(iii) Il faut ces livres
It need these books ('These books are needed')
These cases are, I think, not representative of the general
patterns of the languages in question and could be assumed to be
residual and not belonging to the core grammar.
1