A New Way to Distinguish Raising from Fronting in Russian1

David Perlmutter
UC San Diego

1 The Problem

The hypothesis that led to the development of relational grammar in the 1970s was that the class of movement rules in early transformational grammar consisted of two distinct sets of phenomena: those that affect nominal-clausal grammatical relations such as subject, direct object, and indirect object, and those that do not. The former set includes raising; the raisee is a subject or object of the matrix clause and typically behaves like one both syntactically and morphologically. The second set includes various kinds of fronting, including WH-movement and topicalization. This division has been interpreted in different ways in subsequent theories, e.g. the division between lexicon and syntax in LFG and that between A-movement and A'-movement in GB and its successors, among others. While different theories have characterized the two classes of phenomena differently, there is widespread agreement that the two classes of phenomena are distinct and have different properties.

It is sometimes not clear, however, which phenomena in a given language belong to which class. This squib addresses the question of whether Russian sentences like (1) exemplify raising or fronting.

(1) Borisu [DAT] nachalo byt' stydno / skuchno / grustno.
 'Boris began to be ashamed / bored / sad.'

Here Borisu is in the dative case and the predicates nachalo and stydno/skuchno/grustno are neuter singular. The complement of (1) would be realized in isolation as (2), whose morphology parallels that of (1).

(2) Borisu [DAT] stydno / skuchno / grustno.  'Boris is ashamed / bored / sad.'

Under both analyses of (1), Borisu has a thematic role in the complement. Under both, its dative case results from a requirement imposed by the complement predicate.2 Under the raising analysis, Borisu is raised to matrix subject. Under the fronting analysis, Borisu occupies sentence-initial position but is not the matrix subject. The term "fronting" here implies no more than that Borisu occupies sentence-initial position.

Nachalo is one of the small class of raising predicates in Russian:3

 prekrashchat'/prekratit''stop, cease'

Raising is exemplified by the sentences in (4), where the raising predicates nachal, prodolzhal, and dolzhen are masculine singular in agreement with their nominative-case subject, the raisee Boris.

(4)a. Boris [NOM] nachal/prodolzhal rabotat'. 'Boris began/continued to work.'
 b. Boris [NOM] dolzhen rabotat'. 'Boris ought to work.'

In uncontroversial cases of raising such as (4), where the raisee determines agreement on the matrix predicate, the raisee is always the complement subject.

The morphology of (1) contrasts with that of (4) in two respects: the putative raisee Borisu is in the dative case and the predicates nachalo and stydno/skuchno/grustno do not agree with it. This provides evidence that (1) exemplifies fronting, not raising. The fronting analysis accounts for the following:

(5)a.Case: Borisu is not nominative because it is not the matrix subject. It is dative in (1) because it is in (2); it is merely fronted in (1).
 b.Agreement: Borisu does not determine agreement on nachalo because it is not the matrix subject.

Consistent with the fronting analysis is the claim that Borisu is not the subject of (2) or of the complement of (1).4 Consequently, it is not eligible for raising.

The raising analysis of (1) includes these elements:

(6)a.Case: Borisu is a dative-marked subject in (2) and in both the matrix and complement clauses in (1).
 b.Agreement: Borisu does not determine agreement because only nominative subjects determine agreement.

Raising is posited because Borisu is in initial position. It is claimed to have dative case as the subject of a complement predicate that idiosyncratically marks its subject with dative case, as in (2). Its failure to determine agreement is attributed to its dative case.

A major difference between the fronting and raising analyses concerns the status of the putative raisee. Under the fronting analysis it is not the subject of (2) or of the complement of (1). Hence it is ineligible for raising and is merely fronted in (1). The raising analysis, on the other hand, claims that it is the subject of (2) and the subject of both the matrix and complement clauses of (1). The status of the putative raisee in (1) is thus closely bound up with the general issue of what is a subject in Russian. Of particular relevance is the issue of "dative subjects" in Russian and how they behave.

2 Dative Subjects, I-Nominals, and Predicate Agreement

Two distinct phenomena have been analyzed as "dative subjects" in Russian. The first consists of dative nominals in finite clauses, e.g. Borisu in (2). The second consists of subjects of infinitival clauses, e.g. (7).

(7)Borisu [DAT] ne rabotat' tam.
 'It's not in the cards for Boris to work there.'

"In the cards" is used to indicate the special meaning of root infinitival clauses. Subjects of several types of subordinate infinitival clauses also appear in the dative case.

The dative nominals in finite clauses such as (2) ("I-nominals") are not surface subjects; they contrast systematically with dative subjects of infinitival clauses, which are true subjects that do determine predicate agreement (Moore and Perlmutter 2000):

(8)Takim fil'mam [DAT] ne byt' dostupnymi publike.
 'It's not in the cards for such films to be accessible to the public.'

Here dostupnymi is plural in agreement with its dative subject takim fil'mam.

The raising analysis of (1) relies on (6b) to account for the lack of agreement on the predicates nachalo and stydno/skuchno/grustno. This entails the claim:

(9)Only a nominative subject determines predicate agreement.

Sentences such as (8) show that this claim cannot be sustained.5 While I-nominals, which are not surface subjects, do not determine predicate agreement, true dative subjects do. This undermines a key element of the raising analysis of (1).

3 The Case of Complement Predicates in Raising Constructions

The argument that (1) involves fronting rather than raising is based on the generalization in (10), illustrated by (11), in which vezhlivym and lenivym have the instrumental suffix -ym.

(10)When the non-expletive subject of an infinitival clause is raised, its adjectival predicate can appear in the instrumental case.6
(11)Boris [NOM] nachal byt' vezhlivym / lenivym.  'Boris began to be polite / lazy.'

If the complement of a raising predicate does not have a non-expletive subject, its adjectival predicate appears in what is called the "short form" in (12a), not in the instrumental case in (12b):

(12)a.  Nachalo byt' zharko / temno / syro. 'It began to be hot / dark / damp.'
 b.* Nachalo byt' zharkim / temnym / syrym.7

The complements of (12a) do not have a non-expletive subject:

(13)Zharko. / Temno. / Syro.  'It's hot. / It's dark. / It's damp.'

The possibility of instrumental case on adjectival predicates in the complements of raising constructions is thus a diagnostic of whether a non-expletive NP is raised.

What is the structure of (1)? Since the raising analysis claims that its complement has a non-expletive subject (Borisu) that is raised to subject of the matrix clause, (10) predicts that the complement adjectives stydno/skuchno/grustno can be in the instrumental case. This prediction is wrong:

(14)* Borisu [DAT] nachalo byt' stydnym / skuchnym / grustnym.

Only the short forms stydno/skuchno/grustno, like the short forms in (12–13), are possible. This shows that Borisu is not the raisee in (1). The raising analysis of (1) must therefore be rejected in favor of the fronting analysis.

The ungrammaticality of (14) also shows that Borisu is not the subject in (2). If it were, it would be able to raise, yielding (14).

4 Independent Evidence from True Dative Subjects

The argument against the raising analysis of (1) is even stronger. True dative subjects of infinitival clauses do raise, resulting in complement adjectival predicates in the instrumental case:

(15)Borisu [DAT] ne nachat' byt' vezhlivym / lenivym.
 'It's not in the cards for Boris to begin to be polite / lazy.'

The matrix predicate nachat' is an infinitive, the raisee Borisu is its dative subject, and the complement predicates vezhlivym and lenivym are in the instrumental case, exactly as (10) predicts. With respect to (10), sentences such as (15) with true dative subjects behave like those with nominative subjects such as (11); the case of the raisee is irrelevant.

The claim that both nominative and dative subjects raise is strengthened by evidence that the positions from which they raise are nominative and dative, respectively. The evidence comes from the semipredicatives sam 'oneself' and odin 'alone', whose agreement is accounted for by (18):8

(16)Boris [NOM] rabotaet odin [NOM].'Boris works alone.'
(17)Borisu [DAT] ne rabotat' odnomu [DAT].'It's not for Boris to work alone.'
(18)A semipredicative agrees with its subject in gender, number, and case.

Paralleling (11) and (15) are:

(19)Boris [NOM] nachal [PAST] rabotat' odin [NOM].
 'Boris began to work alone.'
(20)Borisu [DAT] ne nachat' [INFIN] rabotat' odnomu [DAT].
 'It's not in the cards for Boris to begin to work alone.'

(18) automatically accounts for the contrasting case of odin and odnomu in (19–20) if the complement subject positions from which the raisee raises in (19–20), like those in (11) and (15), are nominative and dative, respectively. This strengthens the evidence that both nominative and dative subjects raise.

The raising analysis cannot restrict raising to nominative subjects. If it did so, it would fail to account for (15) and (20). It would also be abandoning its basic claim that Borisu is the raisee in (1). Similarly, the raising analysis cannot claim that complement predicate adjectives can be instrumental only when a nominative subject is raised, for complement predicate adjectives are instrumental in (15), where the raisee is dative.

5 Independent Evidence for Fronting

The fronting of Borisu in (1) is not found with I-nominals alone, but also with other non-subjects in clauses with no overt subject. For example, the experiencer of toshnit' 'nauseate' appears as an accusative object (Borisa in (21)).

(21)a.Borisa [ACC] toshnilo. 'Boris was nauseous.'
 b.Borisa [ACC] nachalo toshnit'.'Boris began to be nauseous.'

(21a–b) exemplify a neutral or unmarked word order. Since (21a–b) do not have an overt subject, the object is fronted to shield the verb from initial position. In unmarked word order other non-subjects are similarly fronted when there is no overt subject. Fronting is the same in sentences with and without a raising predicate.

(22)a.V dokladax [LOC] ne pojavilos' oshibok [GEN].
   'There appeared no mistakes in the reports.'
 b.V dokladax [LOC] ne perestalo pojavljat'sja oshibok [GEN].
   'Mistakes did not stop cropping up in the reports.'
(23)a.U Borisa [GEN] ne xvataet deneg.
   'Boris doesn't have enough money.'
 b.U Borisa [GEN] nachalo ne xvatat' deneg.
   'Boris began not to have enough money.'
(24)a.V Moskve [LOC] morozit.
   'It is freezing in Moscow.'
 b.V Moskve [LOC] nachalo morozit'.
   'It began to be freezing in Moscow.'
(25)a.Borisu [DAT] zhal' vsju sem'ju.
   'Boris feels sorry for the whole family.'
 b.Borisu [DAT] nachalo byt' zhal' vsju sem'ju.
   'Boris began to feel sorry for the whole family.'

The verb is usually shielded from initial position by an overt subject. Where there is no overt subject, fronting has this function.

6 Conclusions

Based on the generalization in (10), it has been argued here that sentences like (1) exemplify fronting rather than raising. (10) thus provides a novel diagnostic, within Russian, of the contrast between raising and fronting. We have also seen another contrast, like those in Moore and Perlmutter (2000), between true dative subjects and I-nominals. Raising of dative subjects, as in (15), results in the appearance of instrumental case on predicate adjectivals, while fronting of I-nominals, as in (1), does not.


Comrie, Bernard (1974) "The Second Dative: A Transformational Approach," in Richard Brecht and Catherine Chvany (eds.) Slavic Transformational Syntax, Michigan Slavic Materials, No. 10, University of Michigan, Ann Arbor, pp. 123–150.

Franks, Steven (1995) Parameters of Slavic Morphosyntax, Oxford University Press, Oxford.

Moore, John and David M. Perlmutter (1999) "Case, Agreement, and Temporal Particles in Russian Infinitival Clauses," Journal of Slavic Linguistics, Vol. 7, No. 2.

Moore, John and David M. Perlmutter (2000) "What does it take to be a Dative Subject?" Natural Language and Linguistic Theory 18, 373–416.


1 Thanks to Andrew Kehler, James McCloskey, John Moore, and Maria Polinsky for discussing aspects of this work with me, and to Maria Polinsky for checking the Russian examples. Errors are my own. [Back]

2 The nature of this requirement is different under different analyses of sentences like (2). [Back]

3 These raising predicates occur with infinitival complements and are listed here as imperfective/perfective pairs in infinitival form. Other predicates occur with raising from small-clause complements. [Back]

4 Detailed arguments for this are presented in Moore and Perlmutter (2000). If Borisu is a surface indirect object in these examples, its dative case is automatically accounted for. [Back]

5 Argumentation against (9) can be found in Moore and Perlmutter (1999, 2000). [Back]

6 The use of the instrumental case on predicate adjectivals is more general than is indicated here. Although consideration of additional data would result in a different formulation, the generalization in (10) will suffice for this squib. [Back]

7 Whether such clauses have a silent expletive subject that raises in sentences like (12a) or are subjectless is irrelevant here. [Back]

8 These semipredicatives are discussed by Comrie (1974) and Franks (1995), among others. [Back]