Introduction 5
functional head is devoid of a referential index, and the projection it heads is con-
strued as non-specific. The two constructions differ semantically in an important way:
thus, the subject of the się construction is construed as in inclusive generic (i.e. one
which includes the speaker and hearer), while the subject of the -no/to is construed as
an exclusive generic (also called ‘arbitrary’).
Gréte Dalmi focuses on generic inclusive lexical and null DPs in Hungarian. The
chapter is couched against the background of the typology of Null Subject Languages
outlined by Roberts and Holmberg (2010), which recognizes four major language
types, in particular: Type 1, which allows only expletive null subjects; Type 2, in which
generic subjects are always null, and third person singular null referential subjects
are allowed in finite clauses, but only if they are anteceded by an overt expression in a
higher clause; Type 3, which freely allows null referential subjects in finite clauses, but
disallows null generic subjects; Type 4, which allows any argument of a verb to be null,
with either unique reference or generic inclusive import. Generic exclusive DPs are,
however, always overt.
As is often the case, not all languages illustrate the ‘pure’ types of a typology, some
of them evincing a ‘mixed’ type. Dalmi demonstrates that Hungarian is such a lan-
guage. Specifically, she provides evidence that Hungarian allows null DPs in all argu-
ment positions of a finite clause, just like type 4, but differs from the latter in excluding
alternation between unique reference and generic inclusive reference null DPs. Dalmi
shows that in Hungarian, the free occurrence of generic inclusive DP must be overt,
while the bound variable occurrence must be null, both in the scope of GN; this is dif-
ferent from what is found in the four major types of Null Subject Languages indicated
above, as well as from British English, where generic inclusive one may be used with
either free or bound import, as in One never admits that one may be wrong.
Dalmi builds on the semantic account of the generic inclusive/generic exclusive
DPs given by Moltmann (2006, 2010 , 2012 ) and assumes a GN operator in SAPP, the
leftmost functional projection of the C-domain within the cartographic model (Rizzi
1997 , 2004 , 2006 , 2013 ; Cinque 2002). She proposes to derive the generic reference/
unique reference alternation found with Italian si ‘one/people/they’ and Polish się
‘one/people/they’ from their feature composition: the [+GN] and [+SAP] features are
checked in SAPP via Agree. In the absence of the [+SAP] feature the generic exclusive
‘people’ reading emerges. In the absence of the [+GN] and [+SAP] features the unique
reference ‘they’ reading emerges.
The article extends the discussion to PRO (the null subject of infinitival clauses).
Based on the syntactic properties of Hungarian lexical and null generic inclusive DPs,
it claims, contra Moltmann (2006, 2010 , 2012 ), that neither unique reference PRO, nor
arbitrary reference PROarb qualifies in Control infinitival constructions where the ante-
cedent is a generic inclusive DP. Therefore a third type, PROGEN, must be introduced.