Eventive nominalizations (ENs) and gerunds in English and other languages are typically analyzed as verbal structures embedded under DP (see, e.g., Abney 1987, Grimshaw 1990). Since many Slavic languages lack articles, the availability of the DP projection in those languages has been hotly disputed (see Pereltsvaig 2013, 2015 for overviews of the debate). Curiously, Russian nominalizations have been used to mount arguments for both the pro-DP position (see Engelhardt and Trugman 1998a,b, Rappaport 1998, 2001, 2004) and the anti-DP position (see Bošković 2008, 2012, Bošković and Şener 2014). Obviously, both arguments cannot be correct: the same data cannot show both that Russian does not have the DP projection and that it does. As the debate is not yet settled, it is important to verify if either argument based on nominalizations is valid. To date, this has not been done.
In this squib, I reconsider the anti-DP and pro-DP arguments and show both nominalization-based arguments to be wrong. Instead, I argue that all of the nominalization-specific morphosyntax (including genitive case marking) happens lower than the putative DP projection; therefore, nominalizations can tell us nothing special (that we cannot learn from ordinary noun phrases) about the presence or absence of the DP in article-less languages.
Before I proceed, a clarification regarding the scope of the squib and the terminology is in order. This squib is concerned with Russian nominalizations denoting events, as in (1).
kollekcionirovanie redkix monet professorom
collecting [rare coins].GEN [professor
‘Professor Pupkin’s collecting of rare coins’
razrušenie armiej žilišč mestnogo
destruction army.INSTR [houses indigenous
‘army’s destruction of indigenous population’s houses’
First, note that the internal argument may precede the external argument, as in (1a), or may follow it, as in (1b). The order of the arguments appears to be subject to weight-related and information-structural considerations, but a detailed discussion of this issue goes beyond the scope of this squib. Second, as the examples illustrate, eventive-denoting nouns may be derived with the nominalizing suffix -ni (or its allomorph -ti).1 However, as Lyutikova (2014) first noted, eventive nominals without the -ni/-ti suffix (kritika ‘critiquing’, zapret ‘ban’, kontrol’ ‘control’, podderžka ‘support’, etc.) are indistinguishable from nominals containing the -ni/-ti suffix in terms of argument structure, word order possibilities, case marking, binding patterns, and aspectual properties. Here, I use both types of event-denoting nouns interchangeably. Finally, unlike their English counterparts in -ing or -ion, Russian ENs can also denote achievements, as in (2a), and even (more rarely) states, as in (2b). My use of the term eventive (rather than process; see Grimshaw 1990) is meant to highlight this difference.
otkrytie Ameriki Kolumbom
discovery America.GEN Columbus.INSTR
‘the discovery of America by Columbus’
neznanie det’mi geografii
not.knowing children.INSTR geography.GEN
‘children’s not knowing geography’
In the following section I reconsider the anti-DP argument based on nominalizations, and in section 3 I do the same for the pro-DP argument.
2 The Anti-DP Argument
Bošković (2008, 2012) and Bošković and Şener (2014) claim that the structure of noun phrases in languages without articles, including Russian, differs radically from that of languages with articles, such as English, German, and Italian: in the former type of language, noun phrases lack the DP projection (hence, their term “NP-languages,” in contrast to “DP-languages” such as English). One of their arguments for this hypothesis comes from ENs: citing Willim (2000), they note that double genitives in ENs are possible only in “DP-languages,” illustrating the observation with the following examples from German (a “DP-language”) and Russian (an “NP-language”). Furthermore, they maintain that one of the genitives in (3a) is checked by D0; the presumed absence of the DP projection in (3b) is supposed to explain its ungrammaticality.
Hannibals Eroberung Roms
Hannibal.GEN conquest Rome.GEN
‘Hannibal’s conquest of Rome’
*zaxvat Gannibala Rima
conquest Hannibal.GEN Rome.GEN
Intended: ‘Hannibal’s conquest of Rome’
However, this supposed contrast emerges from comparing “apples and oranges,” by likening the boldfaced prenominal genitive in the German example with the boldfaced postnominal genitive in the Russian example. The nonparallelism of the two examples is partially obscured by the fact that German has the same morphological form for (some) preand postnominal genitives.2 Notably, in parallel with (3b), the structure with two postnominal genitives in German is ungrammatical.
*die Entdeckung Amerikas Kolumbus’
the discovery America.GEN Columbus.GEN
Intended: ‘Columbus’s discovery of America’
The same is true of English, another supposed “DP-language”: the grammatical pattern includes one prenominal and one postnominal argument (e.g., Hannibal’s conquest of Rome), whereas two postnominal arguments are ungrammatical (e.g., *the conquest of Hannibal of Rome or *the conquest of Rome of Hannibal); see Alexiadou, Haegeman, and Stavrou 2007:543. The ungrammaticality of two postnominal genitive arguments has been noted for other languages with articles: Italian (Cinque 1980, Alexiadou 2001:83–84), Catalan (Picallo 1991, Alexiadou 2001:81–83), French and Spanish (Alexiadou 2001:83), and Greek (Alexiadou, Haegeman, and Stavrou 2007:543).
Crucially, the same generalization holds for article-less languages like Russian, where two postnominal genitive arguments are ungrammatical, as in (3b), but one of the arguments can be expressed by a prenominal “ov-possessive.”3
Gannibal-ov zaxvat Rima
Hannibal-POSS conquest Rome.GEN
‘Hannibal’s conquest of Rome’
(see Lyutikova 2014:132)
Following Babyonyshev (1997) and Trugman (2007), I assume that the prenominal ov-possessive appears in D0.4 This analysis is similar to Abney’s (1987) placement of Saxon genitives (SGs) in English in Spec,DP.5 The analysis that places both Russian ov-possessives and English SGs in the same projection accounts for empirical similarities between them. First, there can be only one ov-possessive or SG per noun phrase (e.g., *the enemy’s the city’s destruction; see Schoorlemmer 1998:56 for Russian examples). Second, ov-possessives and SGs are similar in that if an external argument is lacking, both denote the internal argument, but if a postnominal genitive is present, both can only denote the external argument.
moë uvol’nenie šofëra
my.POSS dismissal driver.GEN
‘my dismissal of the driver’ ≈ ‘I dismissed the driver’
(not: ‘the driver dismissed me’)
my dismissal of the driver
≈ ‘I dismissed the driver’ (not: ‘the driver dismissed me’)
So, if we compare like with like, the same generalization emerges for both types of languages, with or without articles: only one argument can be expressed as a postnominal genitive phrase, while the second argument can be expressed as a prenominal possessor (or as an oblique phrase), but not as another postnominal genitive. Thus, two postnominal genitives in ENs are ungrammatical, regardless of whether the language has articles. Thus, the alleged correlation between the availability of articles (and by hypothesis, of the DP projection) and the possibility of two (postnominal) genitives does not hold.6 Consequently, ENs in Russian do not provide evidence for the distinction between “DP-languages” and “NP-languages.” For a critical review of Bošković’s other arguments in support of this distinction, see Pereltsvaig 2013, 2015 and Lyutikova and Pereltsvaig 2015.
3 The Pro-DP Argument
In a number of articles, Engelhardt and Trugman (1998a,b) and Rappaport (1998, 2001, 2004) use Russian nominalizations to argue the opposite of Bošković’s proposal: namely, that article-less languages like Russian indeed have the DP projection. They concede that two postnominal genitives are not possible in ENs (as discussed in the previous section), but point out that two postnominal genitives are possible in another kind of nominalization, which they call result nominals. (Throughout this section, examples are from Russian.)
kollekcija monet professora Pupkina
collection coins.GEN [professor Pupkin].GEN
‘Professor Pupkin’s collection of rare coins’
In result nominals, they claim, the internal argument (e.g., monet ‘of coins’ in (7)) receives genitive from the N0, while the external argument (e.g., professora Pupkina ‘Professor Pupkin’s’, which they place in a rightward Spec,DP) receives genitive from the D0. (Their analyses differ in that Engelhardt and Trugman treat the latter genitive as a structural case and Rappaport considers it an inherent case.) As for ENs, recall that only one (postnominal) genitive is possible. Contrary to Bošković’s claims, for Engelhardt and Trugman this is because D0 is present but N0 is not; instead, in their analysis a verbal projection is embedded directly under D0. (The trees proposed by Rappaport contain an N0, but he offers no explicit explanation for why the derived noun in (7) licenses inherent genitive to its complement but the derived noun in (1a) does not.)
However, this analysis associating the genitive in ENs with D0 runs into two serious problems. The first problem, noted by Lyutikova (2014), concerns so-called small nominals (SNs; i.e., nominals in which the DP is not projected; see Pereltsvaig 2006). For example, as I have shown elsewhere (Pereltsvaig 2006), quantified subjects that trigger plural agreement on the predicate are DPs, but their counterparts that do not trigger agreement on the predicate (which thus appears in the default 3rd singular neuter form) lack DP. Thus, only agreeing subjects can contain demonstratives.7
V ètom fil’me igral-i [DP (èti) pjat’ izvestnyx
in this film played-PL these five famous
‘(These) Five famous actors played in this film.’
V ètom fil’me igral-o [SN<DP (*èti) pjat’
in this film played-NEUT.3SG these five
‘(These) Five famous actors played in this film.’
Assuming both my (2006) analysis and Engelhardt and Trugman’s (1998a,b) and Rappaport’s (1998, 2001, 2004) proposals, one expects that nonagreeing predicates (whose subjects are less than a DP, per my analysis) would be incompatible with a genitive internal argument of an EN that functions as a clausal subject (since the genitive is supposed to be licensed by D0, per Engelhardt and Trugman’s and Rappaport’s proposals). Yet this expectation is not met.
V tot god bylo zaregistrirovano [SN 25
in that year was.NEUT.3SG registered.NEUT.3SG 25
‘In that year, 25 hijackings of airplanes were registered.’
Note further that the possibility of quantified ENs, as in (9), is curious in and of itself. Following Bailyn (2004), among others, I assume that the numeral is merged in the NumP. Setting aside the possibility of approximative inversion (see Yadroff and Billings 1998 for a detailed discussion), the derived noun (zaxvatov ‘hijackings’ in (9)) follows the numeral, strongly suggesting that the derived noun does not raise past NumP to land in DP, as proposed by Engelhardt and Trugman and by Rappaport.
A second argument against Engelhardt and Trugman’s and Rappaport’s hypothesis that the genitive on the internal argument in ENs is licensed by D0 comes from interference by numerals. Even in nominals where the DP layer is projected, as in quantified subjects that trigger agreement on the predicate (per my (2006) analysis), the D0 cannot be checking the genitive on the internal argument because the numeral is an intervening case checker: it is a well-described fact of Russian that if a quantified noun phrase appears in a structural case context—that is, nominative or accusative case—the numeral assigns genitive case to its complement. (Babby (1987) calls this the “heterogeneous case pattern” since the numeral appears in the nominative/accusative, but the noun and any attributive adjectives below the numeral appear in the genitive.) Crucially, the case checked by a head outside the nominal (e.g., nominative in (10)) cannot penetrate the case domain of the numeral.
pjat’ izvestnyx kartin
five.NOM known.GEN paintings.GEN
‘five known paintings’
*pjat’ izvestnye kartiny
five.NOM known.NOM paintings.NOM
Intended: ‘five known paintings’
Going back to ENs, D0 (whether overt or null), which is outside the case domain of the numeral, cannot possibly check the genitive on the internal argument in an EN (samolëtov ‘airplanes.GEN’ in (9) and (11)), which is inside the case domain of the numeral, marked by boldfaced brackets subscripted with GEN.8
4 Russian Eventive Nominalizations: Take-Home Message
The discussion in the preceding section suggests that the derived noun in Russian ENs cannot be moving as high as D0, nor can the genitive case on the internal argument in an EN be licensed by D0. Thus, ENs provide no direct evidence for the presence of DP in Russian. Thus, the debate about the existence of DP in article-less languages will have to be settled on the basis of evidence from other empirical domains.
1 For a discussion of the two allomorphs, see Pazel’skaya and Tatevosov 2008:350–351.
2 As a reviewer notes, the pattern in (3a) holds only for “bare” genitival noun phrases. When a definite article is present, the genitival noun phrase can only occur in postnominal position.
3 These “ov-possessives” may be derived by the suffix -ov (hence the name) or its variant -in, depending on the morphophonological properties of the root (see Babyonyshev 1997 for details); pronominal possessives have the same morphosyntactic properties as well.
4 A similar analysis is proposed for the Bulgarian counterparts of the Russian ov-possessives in Harizanov 2018.
5 Note that the terms possessive (as in ov-possessive) and genitive (as in Saxon genitive) are used here as traditional labels for morphological forms, without implying a particular syntactic analysis. In fact, as discussed in the main text, although these two constructions have different labels, the analyses they receive are very similar; however, placing ov-possessives in D0 rather than in Spec,DP explains why they can be formed only from one word, not from a multiword phrase (see Babyonyshev 1997 for details).
6 As a reviewer points out, two genitives of the same morphological format are possible in some article-less languages: for example, in Japanese, a head-final article-less language (argued by Bošković to lack the DP layer), two genitive arguments—both marked by no—are possible in eventive nominalizations.
yuubokumin no toshi no hakai
nomad GEN city GEN destruction
‘the nomads’ destruction of the city’
Contrary to the reviewer’s suggestion, such data are not problematic for the argument advanced in this squib: since I reject Bošković’s claim that article-less languages lack the DP layer, nothing prevents postulating two genitive assigners for (i), D0 and n0. Thus, the only difference between Russian and Japanese (besides the obvious word order/headedness contrast) concerns the realization of the genitive assigned by D0: in Japanese (and German) it is realized in the same way as the genitive assigned by n0, whereas in Russian (and English) it is realized by a special form, hosted in D0 (ov-possessives in Russian and SGs in English). Crucially, the presence of articles in the language does not predict the realization of the external argument in eventive nominalizations.
7 Following Alexiadou, Haegeman, and Stavrou (2007:219–220), who write that “Russian has no articles, but demonstratives have a special status among modifiers that justifies treating them as occurring in a higher functional projection,” I assume that demonstratives appear in DP. As a reviewer points out, (8b) is grammatical if the demonstrative appears in the genitive form, ètix. Such a string, however, is a marked discontinuous variant of the partitive pjat’ ètix izvestnyx aktërov ‘five of these famous actors’ and is therefore irrelevant for the discussion at hand.
8 Moreover, the presence of the instrumental external argument (here, terroristami ‘by terrorists’) indicates that the case checked by the numeral does not apply to the arguments of the EN. Thus, the genitive on the internal argument (e.g., samolëtov ‘airplanes’ in (9) and (11)) cannot be licensed by the numeral itself. This conclusion is further supported by the preservation of the genitive on the internal argument in the “homogenous case pattern,” where the numeral does not assign genitive (see Babby 1987, Bailyn 2004 for details), as in (i), and under paucal numerals, where the numeral requires a special “paucal” form of the noun (see Pereltsvaig 2010 for details), as in (ii).
soobščenija o pjati zaxvatax samolëtov
reports about five.LOC hijackings.LOC airplanes.GEN
‘reports about five hijackings of airplanes by terrorists’
V tot god bylo zaregistrirovano [dva zaxvata
in that year was.NEUT registered.NEUT two hijacking.PAUC
‘In that year, two hijackings of airplanes were registered.’
This work has benefited greatly from discussions with (in alphabetical order) M. Teresa Espinal, Boris Harizanov, Ekaterina Lyutikova, Ora Matushansky, David Pesetsky, and Roberto Zamparelli. I am also grateful to the three anonymous reviewers, who helped improve this squib. All remaining errors are mine alone.