M. Yu. Knjazev
UiL OTS, Utrecht
NULL DEMONSTRATIVES IN NOMINALIZED CLAUSES AND THE ECONOMY PRINCIPLE
Let’s start with a contrast first noted in [Svedova 1980] that that-clauses in Russian can be “doubled” by a demonstrative pronoun if this pronoun carries oblique case but (generally) can not be doubled if that pronoun carries accusative case. The clause can (and, in fact, must) be doubled only if it receives a contrastive focus stress. The whole paradigm is illustrated in (1):
(1) a. Ja pomnju (o tom), cto Petja ljubit Masu.
I remember (about it.LOC) that Peter loves Masha
‘I remember that Peter loves Mary’.
b. Ja znaju (*to), cto Vanjapobil Petju.
I know it.ACC that John beat Peter
‘I know that John has beaten Peter’.
c. Ja znaju tol’ko *(TO), cto Vanja pobil Petju.
I know only it.ACC that John beat Peter
‘I only know that John has beaten Peter’.
The question then is what accounts for the contrast between (1a) and (1b). Let’s first observe the intuition that the “doubled” version of (1b), is not radically bad. In the Minimalist framework of [Chomsky 1995] this intuition might be interpreted as economy violation as opposed to convergence violation (under the assumption that the latter tend to be more severe than the former). The “doubled” version then should be less economical than the “bare” version of it.
If this approach is on the right track, we have two theoretical options, namely, either the “bare” version of (1b) lacks demonstrative element altogether (so that having it is uneconomical) or it contains an unpronounced version of the demonstrative element (so that having a pronounced version of it is uneconomical).
I am going to pursue the second option since it makes possible to state the relevant economy condition at work straightforwardly, namely, that, as suggested by Chomsky in the 1995 book (to account for absence of a pronounced complementizer in English root clauses), the operation Merge (on a par with Move) might be subject to Procrastinate, so that Merging a silent version of an element is more economical than the pronounced one thereof.
Parsing the demonstrative pronoun as a D head (having CP as its complement in [Kayne 1994] fashion), the proposal then would be that (1b) has underlying structure as in (2).
(2) Ja znaju, [DP D [CP cto Vanja pobil Petju]].
I know that John beat Peter
‘I know that John has beaten Peter’.
One of the advantages of this structure is that it explains how clauses can check (accusative) case features of a verb and at the same time appear in case-less positions. This would be conditioned by absence or presence of a D head. (Cf. [Boskovic 1995] idea that it is conditioned by the absence/presence of a case feature on the complementizer.) A similar structure for English clauses has been a. o. proposed in [Takahashi 2010] (and references therein) for independent reasons although there it was assumed for dislocated clauses only.
The immediate question this structure raises is the following. If we have a silent D head preceding the clause, then the ungrammatically of a clause being a complement of a preposition (at least in Russian and English) becomes a mystery since there is no obvious reason why a preposition wouldn’t have a DP as its sister.
(3) *Ja pomnju o [DPD, cto Petja ljubit Masu].
I remember about that Peter loves Masha
‘I remember that Peter loves Mary’.
One of the accounts for this ungrammaticality is due to [Pesetsky 2010], who claims that prepositions require that their complements bear phi-features, whereas clauses don’t bear phi features (as claimed a. o. in [Iatridou, Embick 1997]). Under the structure in (1b) this explanation is apparently lost since determiners are standardly
assumed to bear phi-features. What I am going to do, however, is to
claim that silent D (at least the version of it used in (2)) in Russian
lacks phi-features. Thus we can maintain Pesetsky’s account for (3) as well as for the D-less version of it.
Since (3) is ungrammatical (presumably, non-convergent), no question arises with respect to compliance to economy of the “doubled” version of (1a). The same is true for (1c), the silent version of which is ungrammatical presumably because silent element can’t bear focus stress.
A further question is why bare indicative CPs can’t function as subjects in Russian, as in (4), whereas they can be topicalized, as in (4).
(4) *(To), cto Vanja pobil Petju, menja udivilo.
it.NOM that John beat Peter me surprised
‘That John beat Peter surprised me’.
(5) Cto Vanja pobil Petju, ja znaju.
that John beat Peter I know
‘That John beat Peter I know’.
The question is what accounts for the contrast in grammatically between the “bare” versions of (1b) and (4). Let’s assume that since in Russian there is no object agreement, (the head of) the object may bear phi-features optionally, which, I propose, is the case in (1b), whereas subject agreement must be triggered by an element bearing phi-features, therefore both silent D and lack of D altogether is precluded, and again no issue of economy arises.
Another empirical problem is why overt demonstratives bearing oblique case assigned by the verb are acceptable, as (6) shows.
(6) Ja udivilsja (tomu), cto Petja ljubit Masu.
I was.surprised (it.DAT) that Peter loves Masha ‘I was surprised that Peter loves Mary’.
Under the advocated approach, (6) should not contain a silent D head, otherwise the “doubled” version of (6) would be ruled out by economy. The question then is what forces it. One possibility is that silent D may not check oblique case features. Note that if that is true then this rules out (3) as well (so that it has a double source of ungrammatically). Another possibility is that in (6) oblique case is assigned by a silent prepositional head so that the non-occurrence of a silent D in (6) has the same source as in (3).
Independent support for the claim that the silent D might have features different from those of the overt one comes from free relatives.
Russian free relatives involving a nominative counterpart of what may optionally have a D head in accusative environments, as in (7 a)
(7) Petja vzjal ?(to), cto lezalona stole.
Peter took it.ACC what lay on table
‘Peter took what was on the table’.
However, if the matrix verb subcategorizes for an animate (neuter) object, the D becomes obligatory, as in (8).
(8) Petja udaril *(to), cto sidelo na ego krovati.
Peter beat it.ACC what sat on his bed
‘Peter beat what was sitting on his bed’.
Although (8) is pragmatically odd, still the contrast between “doubled” and “bare” versions of it holds.
The contrast between (7) and (8) might be accounted for if silent D lacks [Person] feature required for the convergence of (8).
So, the claim that embedded clauses are headed by a silent D lacking phi-features seems compatible with the data (although not strictly speaking entailed by them) and gives an account of the contrast between (1a) and (1b), which is Minimalist in spirit. A broader question this discussion might bear on is whether there are differences in feature content between pronounced and unpronounced functional elements.
References
Boskovic Z. Case properties of clauses and the Greed Principle // Studia Linguistica 49, 1995.
Chomsky N. The minimalist program. Cambridge, MA, 1995. latridou S., Embick D. Apropos pro // Language 73, 1997.
Kayne R. The antisymmetry of syntax. Cambridge, MA, 1994.
Khomitsevich O. Dependencies across Phases: From Sequence of Tense to Restrictions on Movement. PhD Thesis, Utrecht Institute of Linguistics, 2007.
Pesetsky D. LOT course handout, 2010.
Svedova N. Yu. (ed.). Russkaja grammatika [Russian Grammar]. Moscow, 1980.
Takahashi S. The hidden side of clausal complements // Natural Language & Linguistic Theory 28, 2010.