LINGUIST List 3.28

Mon 13 Jan 1992

Disc: Nominative in Non-finite Clauses

Editor for this issue: <>

Directory

1. , Nominative in nonfinite clause
2. Jouko Lindstedt, 3.6 Nominative with Infinitive

David Pesetsky suggests that the Lezgian construction with a nominative in a
non-finite clause in (1) could be analyzed as (the equivalent of) subject-to-
(nominative)object raising:

(1) Didedi-z gada-0 agaq'-na k'an-zawa. 'Mother wants the boy to arrive'
 mother-DAT boy-NOM arrive-NONFIN want-PRES

However, such an analysis is clearly impossible for Lezgian. As Ivan Derzhanski
observes, Daghestanian languages generally show gender agreement of nominative
arguments with their verbs, and no Daghestanian language shows agreement in
constructions like (1). As it happens, Lezgian is one of the few Daghestanian
languages that have lost gender agreement, so this cannot be illustrated from
Lezgian. But word order makes it absolutely clear that there is no raising
involved (in fact, Lezgian shows no raising or other grammatical function-
changing rules at all).
 The 'problem' of nominative arguments in non-finite clauses appears in many
parts of Lezgian grammar, and it is clear that it cannot be explained away.
The following three examples are illustrative (for more details, see my
forthcoming Grammar of Lezgian, Berlin:Mouton de Gruyter, Mouton Grammar
Library).

 Complement clause, non-finite marker -n-
(2) Dide-0 gada-0 hele xtu-n-al tazub xa-na.
 mother-NOM boy-NOM already return-NONFIN-about surprised be-PAST
 'Mother was surprised that the boy had returned already.'

 Complement clause, non-finite marker -di
(3) Am taxsirlu tusir-di askara ja.
 she:NOM guilty be:NEG-NONFIN clear be
 'It is clear that she is not guilty.'

 Non-finite relative clause (participial marker -j)
(4) Lezgi-jar-0 jasamis zezwa-j xur-er-0 Azerbajzand-a awa.
 Lezgian-PL-NOM living be-PARTCP village-PL-NOM Azerbaijan-in be.in
 'In Azerbaijan there are villages where Lezgians live.'

Thus, nominative arguments in non-finite clauses are quite normal in Lezgian.
Crucially, unlike the Portuguese personal infinitive (cited by Geoff Nathan)
and the Bulgarian da-construction (cited by Ivan Derzhanski), non-finite
verbs do not show person agreement with the nominative argument in Lezgian
(or Dravidian).
 In GB theory, nominative case is not assigned by the verb but by something
else (INFL, AGR) that is not present in non-finite clauses. In this way, case
assignment is more complicated in finite clauses, although finite clauses
would seem to be the unmarked member of the finite/non-finite opposition.
 If the native language of GB theorists were Lezgian or Dravidian, presumably
GB would show a different bias: Nominative case would be assigned by the verb,
just like the other cases, and a special blocking mechanism would take care
of a strange phenomenon in exotic languages like Indo-European, where nominative
arguments are barred from certain marked non-finite constructions. It seems to
me that such an approach could well be closer to the truth.

Martin Haspelmath, Free University of Berlin

Ivan A. Derzhanski presents a Bulgarian example of a nominative subject
in a "non-finite clause":

> Let me add the Bulgarian counterpart:
>
> Tja iska toj da pristigne. `She wants him to arrive.'
> she-NOM want-Pres3Sg he-NOM to arrive-Pres3Sg
>
> where _da_ is a de-finitising particle. (There is no infinitive.)
> I'm using pronouns, because nouns in Bulgarian don't decline.

I think that we have here a good example of how it is difficult in
syntax to distinguish between primary data and its theoretical
interpretation. First, how do we now that _da_ is a "de-finitizing
particle"? We do not say that English prepositions are (surface) case
markers. _pristigne_ is clearly a finite form, it does not become
non-finite simply because it has got a particle before it. Actually,
_da_pristigne_ can even form a complete main-clause utterance by
itself ('let him come!').

Second, how do we know that _toj_ is not simply the subject of an
embedded (finite) complement clause? Certainly _da_ can be a
complementizer though it does not begin the clause. I see two
finite clauses here:

 Tja iska [toj da pristigne]

 Jouko Lindstedt
 U of Helsinki
 <Jouko.LindstedtHelsinki.FI>