Two main types of èto copular constructions exist in Russian. In the first one, the clause consists of a nominal phrase, followed by èto (identical in its form to the proximal neuter singular demonstrative ‘this’), a copula (in future and past tenses), and another nominal phrase, as shown in (1). We label these constructions ‘NOM_pre èto (be) NOM_post’, where the subscripts ‘pre’ and ‘post’ indicate whether a nominal phrase precedes or follows èto.

In the second type of èto copular constructions, èto is in the clause-initial position, followed by (a copula and) a single nominal phrase, as shown in (2). Accordingly, we label these constructions ‘Èto (be) NOM_post’.

In the existing literature, èto copular constructions have received various treatments. A major dividing line can be drawn between approaches that treat èto as a functional head on the clausal spine and analyses that consider it to be the predicate itself. The former typically treat èto as a topic head; see e.g., Junghanns (1997), Geist & Błaszczak (2000), Markman (2008).2 These accounts struggle to capture all the syntactic and semantic properties of the èto clauses, including the fact that èto only appears in identity clauses (as discussed in detail in section 2), the agreement pattern in èto copular constructions (as shown in (4) below), and the unavailability of multiple instances of èto in a single clause (especially problematic under the assumption that èto is merely an exponent of Top⁰, since a single clause may include multiple topics).

Geist (2008) puts forward an alternative approach, whereby èto spells out the predicate. Because it better captures the peculiar behavior of èto copular clauses, we take it as a starting point of our discussion. While the analysis that we propose in this paper differs from Geist’s in some important respects and does not build directly upon it, the two share the core idea that in copular clauses the èto constituent is a syntactic predicate at the clausal level. Thus, the present work may be considered a continuation of this line of research.

The novelty of our proposal lies primarily in the argument that there are two predication relations established in èto constructions: one on the clausal level (much as in Geist’s 2008 approach), and another one within the constituent headed by èto — which, we argue, constitutes a ‘big DP’ with a complement and a specifier, and a predication relation established between them with the help of the predicativizer èto. This approach will be shown to be restrictive enough to rule out the ungrammatical examples (for instance, examples where NOM_pre attempts to control agreement), but at the same time it leaves sufficient room to account for the seemingly exceptional examples of identity/identification constructions that have been overlooked by the existing literature.

The paper is structured as follows. Section 2 summarizes the typology of copular sentences and the main proposals regarding the syntax of èto-constructions, and identifies the contradictory aspects of the existing work that the current account seeks to address. In section 3, we turn to the ‘NOM_pre èto (be) NOM_post’ constructions and show that our proposal mobilizing a double predication structure (one within the ‘big DP’ and the other involving the ‘big DP’ as the predicate at the level of the clause) can account for them (3.1), and also spell out the workings of predication within the ‘big DP’ (3.2); the analysis is further extended to ‘èto (be) NOM_post’ constructions (3.3). Section 4 provides additional syntactic and prosodic evidence that is compatible with our analysis but would be hard to account for if NOM_pre was taken to be a (hanging) topic. In section 5, we discuss the wider implications of our approach. Section 6 concludes the paper.

Higgins (1973; 1979) has famously put forward a classification of copular clauses, distinguishing the following four types of constructions. Predicational copular clauses, such as John is tall, feature a predicational expression (syntactically, the predicate) assigning some property to a referential expression (syntactically, the subject). Next, specificational copular clauses ‘merely say what one is talking about: the subject in some way delimits a domain and the specificational predicate identifies a particular member of that domain’ (Higgins 1979: 213). As Mikkelsen (2005) clarified later, the pre-copular element of a specificational clause behaves as (containing) a variable and the post-copular element provides a value for that variable. Canonical examples of specification (not analyzable as any other type) are The number of planets is nine and What I like is apples. In these, one of the terms is superscriptional (the header of a list) and the other is specificational (specifying a value for the variable in the superscriptional term).3 The next type, identificational copular clauses, according to Higgins, are typically used to teach or state the names of people or things, as in That (man) is John. They usually contain two nominal phrases: one of them is referential, and the other one (most naturally occurring in the post-copular position) is described by Higgins as identificational (but see below on dependents of this sort also being plausibly characterized as referential). Finally, identity copular clauses (called ‘equative’ in later work) constitute the fourth type. These establish identity between two referential expressions, as in The morning star is the evening star.

Higgins considers the possibility of grouping identity clauses together with identificational clauses. Specifically, he notes that referentiality may be understood as a property of the expression itself, and if so, it becomes possible to consider an identificational clause such as That man over there is Joe Bloggs an example of identity, too. Although Higgins himself does not fully commit to this idea, we will adopt it in this paper. (See also Mikkelsen 2005 on grouping together identity clauses and identificational clauses with two referential entity-type dependents.) As will be shown in section 2.3, identity clauses in Russian form a natural class with identificational clauses in terms of the distribution of èto; in what follows, we will use the term ‘identity’ to refer to both subtypes of constructions.

Higgins also emphasizes that there is no one-to-one correspondence between (in)definiteness and referentiality/predicativity (Higgins 1973: 246), in the context of a discussion of the kinds of grammatical expressions that can be used in various types of copular clauses – proper names, (in)definite NPs, APs, etc.4 While typically referential expressions include deictic elements, proper names, and definite NPs, and predicational expressions include indefinite NPs, APs, or PPs, at least some indefinite NPs can be referential and at least some definite NPs can be predicative.

To appreciate this, consider first the sentence [One of the boys] was tall. The indefinite NP here (bracketed) has a specific referent and serves as the semantic and syntactic subject. On the other hand, (3a) shows that a definite NP (bracketed) can denote a property and serve as a predicate: here, John is attributed the property of being the president. Note that (3a) is ambiguous and can also receive an identity reading, if the president refers to a specific entity/individual; in this it contrasts with the predicative-only example in (3b). (Thus, only (3a) but not (3b) can be used in response to the question Which one is the president?, which requires identification; see also Higgins 1973, Doron 1988 for further discussion of minimal pairs of this sort).5

Importantly for our purposes, this means that the semantic type of a predicate, but not necessarily its grammatical definiteness, determines the type of the copular clause. This will be important for the discussion of èto as only appearing in identificational contexts in Russian.

One of the most syntactically well-developed analyses of èto copular clauses in Russian is Geist (2008). Because our approach shares some insights with this analysis, we summarize it below.

Geist (2008) considers clauses of the type ‘NOM_pre èto (be) NOM_post’ and argues that NOM_pre is a dislocated/external topic, while èto is an internal topic linked to it. The clause [èto (be) NOM_post] (excluding the external topic) is treated as predicational: èto serves as a predicate and takes NOM_post as its argument.

A beneficial implication of Geist’s proposal (though not one discussed by Geist herself: she only considers èto constructions in which the two nominal phrases have matching features) is that NOM_pre never controls agreement with the copula, and the copula obligatorily displays the phi-features of NOM_post, as shown by our examples in (4). This agreement pattern follows from the assumption that NOM_post is underlyingly the subject of predication while NOM_pre is not in a direct relation with the matrix T⁰ head (because, e.g., in Geist’s analysis, it is an external topic).

Geist treats èto as a demonstrative pronoun. To make it possible for it to be a predicate, she proposes that the èto-constituent has a special type of D⁰ head that takes èto as its complement and turns it into a predicate of the type <e,t>.6 The èto-predicate establishes an identity between two entity-type items: NOM_post (via predication) and NOM_pre (via co-indexation). This is schematized in (5).

The key strength of Geist’s (2008) proposal is that it allows for bringing together three semantic types of copular constructions that contain two nominal dependents: (i) predicational ones (cf. in English John is a teacher (by trade)), (ii) equative/identity ones (Mark Twain is Samuel Clemens), and (iii) specificational ones (The murderer is John); note that (ii) is the only type that contains and requires èto in Russian. The predicational constructions are taken to be the baseline, while in equative and specificational clauses, different types of type-shifting are applied, in order to combine the two non-property items and derive the corresponding semantics of the clause.

Despite its insightful nature, Geist’s analysis leaves aside some important data, and some of the assumptions that it rests upon are not tenable if the full range of data is to be accounted for. First, it is unlikely that NOM_pre always serves as a dislocated/hanging topic, in view of the fact (corroborated in section 4) that NOM_pre has a broader distribution than what is predicted for hanging topics.7

Second, as far as èto constructions are concerned, Geist (2008) only considers equative/identity clauses that contain two proper names, of the type Mark Twain is Samuel Clemens. However, èto can also appear in copular clauses with a proper name and a common noun, or with two common nouns, as in That girl on the left is Maša/my sister, and it is not immediately apparent how to treat such examples in her analysis. Geist discusses some similar examples (e.g., Ubijtsa staruxi – Raskol’nikov ‘The murderer of the old lady is Raskol’nikov’) but claims that they cannot contain èto and analyzes them unambiguously as specificational clauses. As we demonstrate in section 2.3, the data are more complex, and èto is not ruled out completely in such cases. This prompts a careful re-examination of the distribution of èto, which gives rise to the following questions: Is èto genuinely restricted to equative/identity constructions? Can it appear in specificational clauses too? Can it ever co-occur with a predicative, i.e. property-denoting (as opposed to referential) item?

To answer these questions, we expand the dataset by adopting the traditional classification of copular clauses developed by Higgins (1973, 1979), introduced in section 2.1, which is broader than that adopted in Geist (2008) and includes identity clauses with all kinds of referential dependents (e.g., That man is John) in addition to the equative clauses with two proper names. As mentioned in section 2.1, identity constructions include those with two proper names but are not limited to them, and may involve referential expressions of various types: pronouns, definite nominal phrases (those with a demonstrative pronoun or a possessor), and even indefinite nominal phrases (e.g., those modified by kakoj-nibud’, kakoj-to ‘some’, or odin iz ‘one of’). This becomes crucial: as we demonstrate in section 2.3, èto is only allowed in the case of identity in a broad sense. We further show that specificational clauses with èto are indeed prohibited; however, if presented out of the blue, the same expression can be ambiguous between a referential reading and a property reading (see the president examples in (3) above), which obscures the overall picture. To distinguish the possible identity, specificational, and predicational readings, we accompany every key example with a specific context. To bolster our argument that èto is only allowed in the case of identity, we also include examples (not previously discussed in the literature on copular clauses in Russian, to our knowledge) that are unambiguously specificational, based on Higgins (1973).

Finally, we bring to light another type of a ‘NOM_pre èto (be) NOM_post’ construction: one in which a predicative adjective is used instead of NOM_post, as illustrated in (6).

This construction falls outside the scope of Geist’s analysis because, in her analysis, NOM_post must be a referential nominal expression, given that it is an argument of the equating predicate [èto]. Because (6) is a copular èto-clause that is otherwise very similar to ‘NOM_pre èto (be) NOM_post’ constructions, it would be advantageous to accommodate examples of this kind as well. As we discuss in section 5.1, our analysis is well equipped to do so.

Despite the crucial differences, our analysis is similar to Geist’s in two respects. First, on our approach as on Geist’s, all ‘NOM_pre èto (be) NOM_post’ constructions have the same syntax, with the constituent headed by [èto] being a syntactic predicate at the clausal level. Second, our intuition also is that building an èto copular clause involves predicativization – a special operation that turns a referential expression of type <e> into a predicate of the type <e,t>. In contrast with Geist, however, we argue that the [èto] constituent is a ‘big DP’, in which èto is an exponent of the D⁰ head. The complement of D⁰ is a silent pro whose content is identified by NOM_post, and the specifier position is occupied by NOM_pre ‒ crucially, not an external topic. In the èto-clause, a syntactic predication relation is established on two levels: within the ‘big DP’ – between NOM_pre and pro, and on the clausal level – between the ‘big DP’ and NOM_post. Such a double-predication approach captures all the relevant properties of the expanded set of èto constructions under consideration in this paper and makes important predictions about their distribution.

The next subsection is dedicated to clarifying the data and determining the exact range of copular constructions that can include èto. Specifically, we show that, in Russian, the distribution of èto is restricted to identity constructions and does not cover specificational and truly predicational clauses.

Having introduced Higgins’ classification of copular clauses and shown that it has not yet been comprehensively discussed in the context of Russian data, we provide a detailed discussion of the different types of copular clauses in Russian below. One of our main conclusions is that the distribution of èto is restricted to identity contexts (including both identificational and strictly equative clauses).

Descriptively, ‘NOM_pre èto (be) NOM_post’ clauses consist of the invariant element èto, which is preceded by one nominative nominal phrase and followed by another one.10 The copula byt’ is inserted between èto and NOM_post in the past and future tenses and is absent in the present tense. This is illustrated in (20).

In line with Geist (2008), we argue that in constructions like (20), NOM_post is the underlying subject of predication on the clausal level, while the constituent that includes èto is the predicate. The two are combined within a small clause.11 The constituent that contains èto then undergoes movement to Spec, TP — an instance of predicate inversion, described in detail by Moro (1997) and Den Dikken (2006). This is illustrated in (21) with a bracketed example; we will address the status of NOM_pre below. The copula (overt or covert) is found in T⁰.

Predicate inversion is obligatory: èto can never follow the copula, which would correspond to the structure in (21) but without predicate movement. This is illustrated in (22), with (22a) presenting an example where both the subject and the predicate remain in situ, and (22b) showing one where NOM_post is promoted to Spec,TP instead.

The obligatoriness of predicate inversion is inextricably linked to the information-structural properties of èto constructions. The [_DP èto] constituent represents given information; this is in accordance with the obligatory identity nature of ‘NOM_pre èto (be) NOM_post’ clauses, in which NOM_pre is identified with NOM_post: to identify/equate NOM_pre with NOM_post, one needs to be familiar with NOM_pre. In Russian, constituents providing given information, as represented by [+presupposed] and [+referential] features, have been shown to precede others, e.g., in scrambling (Titov 2017). Given that, at the same time, Spec,TP is the natural host for presupposed/referential material in Russian (Bailyn 2004; Titov 2018), we suggest that a similar kind of movement, correlated with the given IS-status of the [_DP èto] constituent, obligatorily applies to this constituent in ‘NOM_pre èto (be) NOM_post’ clauses.12

Treating NOM_post as the notional subject of the clause allows us to derive the central property of identificational clauses that we have been discussing so far (instead of just postulating it): the fact that NOM_post must be referential, as discussed in detail in section 2.3 in the context of identity copular clauses. As Russian is an article-less language, the distinction between the two kinds of nominal phrases – referential vs. property-denoting/non-referential – is not expressed morphosyntactically, but the non-referential reading can be enforced by generalizing expressions corresponding to English ‘by trade’ and ‘by nature’. Examples of this sort are ungrammatical in the presence of èto, as shown in (23) (also demonstrating, once again, that non-identity copular clauses do not allow for èto).

In (24), another test for the referential nature of NOM_post is provided. The base sentence is the identity example ‘Petja was your boss’, (24a). A corresponding sentence with the topic marker čto kasaetsja ‘as for’, as in (24b), can be constructed and is accepted by native speakers. As argued by Reinhart (1981), only referential nominal expressions can be interpreted as topics; therefore, the acceptability of (24b) suggests that ‘your boss’ cannot be a property, must be interpreted referentially, and, accordingly, should not be analyzed as the predicate. Note that Petja in (24b) represents given material, since it picks out one referent from a group of known referents (several people that we have just met); this, according to our proposal, is correlated with Petja undergoing inversion and acting as NOM_pre.

The fact that NOM_post in ‘NOM_pre èto (be) NOM_post’ constructions is interpreted referentially and not predicatively can also be brought to light (following Heycock & Kroch’s (1999) lead) by adding a non-restrictive relative clause to modify NOM_post. In (25a), moj drug ‘my friend’ is potentially ambiguous between being a referential/definite expression (‘this friend of mine’) or denoting a property (‘a friend of mine’). The non-restrictive relative is grammatical only in the identity context (i.e., when èto is present), which confirms the referential reading of NOM_post.

To recapitulate, using the three diagnostics above – property-denoting phrases like po professii ‘by trade’ or po xarakteru ‘by temper’ in (23), a topic marker čto kasaetsja ‘as for’ in (24b), and a non-restrictive relative clause in (25b) – we can establish that NOM_post must have a referential and not a property-denoting/predicative reading in ‘NOM_pre èto (be) NOM_post’ constructions. The syntactic subjecthood of NOM_post fits well with this.

So far, we have argued that in copular clauses of the type ‘NOM_pre èto (be) NOM_post’, NOM_post is the subject, while the constituent that includes èto is the syntactic predicate.13 Let us now turn our attention to NOM_pre.

In section 2.3, we discussed that, in identity clauses, both nominal expressions are referential. For NOM_pre, this can be verified with the help of the čto kasajetsja ‘as for’ test, already mobilized in section 3.1:

But despite being referential, NOM_pre is not a viable candidate for subjecthood. NOM_pre never controls agreement with the copula in T⁰, as we showed in section 2.2. Two examples showing this are reproduced here in (27) and (28). In both examples, NOM_pre is plural and NOM_post is singular, and the copula obligatorily displays the phi-features of NOM_post.

Though NOM_pre is not itself the structural subject of the clause, we propose that it combines with èto in a single constituent of the following shape: [_DP [_DP NOM_pre] [_D’ èto [pro]]]. To refer to these complex DPs, we use the term ʻbig DP’, originally used in Romance linguistics to refer to the constituent that encompasses a nominal expression and a coreferential clitic (Uriagereka 1995; 2005; Kayne 2005, i.a.).14 Accordingly, what precedes the copula in the examples in (20) has the structure shown in (29).

Within the ‘big DP’, the specifier position is occupied by NOM_pre and the complement position is occupied by pro,15 with a predication relation established between the two. As both NOM_pre and pro are inherently referential, the symmetry must be broken, and we propose that this is done by èto in D⁰, which we analyze as a predicate-forming operator, a ‘predicativizer’. Its role is to turn pro in its complement position into a predicate and to mediate the structural relation between this predicate and its subject (occupying Spec,DP).16

Outside the ‘big DP’, the content of pro is contextually determined by NOM_post (‘Petja was it/this, viz., my boss’). We assume that the fact that pro is co-indexed with NOM_post allows it to be formally licensed as a silent element and its content identified.17 The fact that pro does not have its content identified within the confines of the ‘big DP’ (causing this ‘big DP’ to contain a free variable) allows, and indeed forces, this ‘big DP’ to be used as a predicate at the clausal level. In this respect, our ‘big DP’ is very much on a par with tough-movement constructions: there, too, the occurrence (inside an otherwise saturated constituent) of a variable whose range is left unrestricted inside the infinitival clause is what turns the infinitival clause into a predicate at the level of the copular clause.

Our proposed structure of ‘NOM_pre èto (be) NOM_post’ is shown in (30), with the predicativized pro boxed.

The èto-headed DP in ‘NOM_pre èto (be) NOM_post’ constructions is first merged as the predicate of a small clause, whose subject is NOM_post. The èto-headed DP then undergoes movement to Spec,TP, which results in the obligatory word order NOM_pre-èto(-copula)-NOM_post. As was shown in (22), the alternatives, without the èto-headed DP raising, are ungrammatical.

The hypothesis that [_DP NOM_pre èto] is a constituent provides an explanation for the behavior of ‘NOM_pre èto (be) NOM_post’ constructions, e.g., in wh-questions, which would otherwise be hard to explain. The relevant empirical observation is that ‘NOM_pre èto (be) NOM_post’ constructions are ungrammatical as root wh-questions, (31). This is expected on our analysis, since NOM_pre (kto ‘who’) can neither be merged as a hanging topic – for information-structural reasons: wh-words do not make good hanging topics – nor be raised to Spec, CP via sub-extraction from the ‘big DP’ [_DP NOM_pre èto] in Spec,TP, because of criterial freezing: the ‘big DP’ has been moved into a criterial position (Spec,TP), and is frozen as a result.18

We conclude that in the ‘NOM_pre èto (be) NOM_post’ constructions under consideration, NOM_pre and èto form an immediate constituent, a ‘big DP’, as is evident from the behavior of the constructions in wh-questions.19 Furthermore, the agreement facts show that NOM_post acts as the subject of the clause probed by T⁰.

So far in this paper, our attention has mostly been confied to ‘NOM_pre èto (be) NOM_post’ constructions, in which èto is preceded by a nominative noun phrase. At this point, we would like to extend the scope of the analysis to ‘èto (be) NOM_post’ constructions, as illustrated in (32), where èto is again invariant and NOM_post is a nominative noun phrase, but no additional nominative noun phrase shows up to the left of èto.

At the level of the copular clause, the structure of these constructions is identical to that of the other copular clauses with èto, as presented in the previous sections: NOM_post is the underlying subject of predication, while the constituent that includes èto is the predicate. That the èto-headed DP is once again the predicate of the sentence is supported by the ungrammaticality of predicative copular constructions or specificational constructions of the shape ‘It/this èto NOM_post’. Èto cannot appear in these constructions, as illustrated in (33). This fits well with our analysis of the èto-headed DP as the predicate, since there can only be one main predicate per clause. For analyses that treat èto as a copula spelling out the T⁰ head or a similar functional head on the clausal spine, examples like (33) posit a problem: under those approaches, the restriction on what can surface as NOM_pre would be unmotivated.

The main difference between ‘èto (be) NOM_post’ and ‘NOM_pre èto (be) NOM_post’ is that the specifier position in the ‘big DP’ is occupied by an overt NOM_pre in the latter but not in the former. But èto is still a functional head D⁰ within the ‘big DP’, and its role, in constructions with and without NOM_pre alike, is to predicativize the silent pronoun (pro) in its complement. The difference between ‘èto (be) NOM_post’ and ‘NOM_pre èto (be) NOM_post’ comes down to the silence/overtness of the subject of predication within the ‘big DP’: an overt nominative in the latter; a silent element (another pro) in the former.

Considering the reference of the pros in (34), we propose that pro_m is essentially NOM_pre being pro-dropped, eligible for omission because it is discourse-anaphoric, similarly to other topical/discourse-anaphoric subjects in Russian (McShane 2009; Tsedryk 2015, i.a.) The discourse-anaphoricity of pro_m becomes evident when (34) is used in a context, for instance, in response to the questions in (35), where another person or a statue acts as an antecedent.20

As for pro_k in the complement of ‘big DP’ in (34), we argue that, just as in other ‘NOM_pre èto (be) NOM_post’ constructions, its value is identified by NOM_post. Èto predicativizes pro_k, and a predication relation is established within the ‘big DP’, with NOM_pre in its specifier acting as the referential subject of predication.

To summarize, we have argued that in copular clauses of the type ‘èto (be) NOM_post’ a predication relation is established between the postcopular nominative DP (NOM_post, the subject of predication) and the èto-headed DP (the predicate). We have further argued that èto, when taking pro as its complement, turns this null proform into a predicate. The null proform’s content is identified via co-indexation with NOM_post, as in identity copular clauses with two overt nominal dependents. Thus, our account uniformly accommodates all kinds of copular clauses where èto appears, which is a favorable outcome from a theoretical perspective.

In the literature, it has often been suggested that NOM_pre is base-generated outside of the copular clause as a dislocated/hanging topic; see for instance Geist (2008), Markman (2008), and most recently Seres & Espinal (2019). Furthermore, Geist (2008) uses the hypothesis that NOM_pre is a hanging topic to explain the agreement pattern and the fact that the copula never spells out the features of NOM_pre: as a clause-external topic, NOM_pre is not in a local relationship with T⁰.

In this section we provide evidence against a hanging topic analysis being applicable to (all) examples of the ‘NOM_pre èto (be) NOM_post’ type, by looking more closely at their syntactic behavior in two contexts – when these constructions are embedded under non-bridge predicates and in conditional clauses – as well as their prosodic properties.

To start with embedded contexts with non-bridge predicates, consider the examples in (36). They show that udivitelʹno ‘surprisingly’ in Russian behaves as a non-bridge predicate, in that it does not allow sub-extraction out of an embedded complement clause without resumption, thus forming a weak factive island (Polinsky & Potsdam 2014).

As is typical for non-bridge predicates cross-linguistically, udivitelʹno ‘surprisingly’ cannot embed a finite clause with a hanging topic, as shown in (37a). However, udivitelʹno ‘surprisingly’ can take a clausal dependent of the type ‘NOM_pre èto NOM_post’, as demonstrated by (37b). That (37b) is grammatical falls out directly from a treatment of [_DP NOM_pre èto] as a ‘big DP’ constituent occupying Spec,TP. It would be incompatible with NOM_pre as a hanging topic, accompanied by an independent functional/pronominal element, èto.21

Further evidence against necessarily treating NOM_pre as a hanging topic comes from conditional dependent clauses. In English, as Haegeman (2007) shows, conditional if-clauses resist DP-topicalization within the embedded clause:

Russian does not allow for hanging topics to be embedded in conditional clauses either. This is shown in (39b), in contrast to an acceptable non-embedded clause with a hanging topic in (39a).

In contrast, a ‘NOM_pre èto NOM_post’ can felicitously be embedded in a conditional clause, which shows that it does not contain a hanging topic. This is illustrated in (40).

The grammaticality of (40) in contrast to (39b) tells us reliably that the noun phrase referring to Sean Connery in (40) is not a hanging topic (as the bulk of the existing proposals for èto copular clauses would treat it). As a non-hanging topic, Šon Konneri will have to originate somewhere inside the clause. With the èto-phrase and Džejms Bond already occupying the only two phrasal positions within the copular clause, this forces Šon Konneri into originating inside either of these two nominals – and since proper names have no DP-internal dependents (which takes the nominal of Džejms Bond off the table), this leaves the DP of èto as the only home for Šon Konneri. Furthermore, the fact that our analysis maps Šon Konneri and èto in (40) onto a constituent guarantees that the two of them together precede the copula byl; breaking up this constituent by subextracting Šon Konneri from it and topicalizing it separately would create syntactic difficulties no matter whether this were done after raising the complex èto-DP into Spec,TP (‘criterial freezing’; fn. 19) or prior to this (cyclicity), so the unity of [Šon Konneri èto] as a single syntactic constituent must arguably be preserved.

The constituent status of NOM_pre and èto is also compatible (perhaps surprisingly, at first blush) with the prosody of ‘NOM_pre èto (be) NOM_post’ constructions. We show this in the next section.

It is commonly assumed that there is an obligatory intonational break or pause between NOM_pre and èto, which might be regarded as undermining the ‘big DP’ analysis of NOM_pre and èto. We find, however, that there is considerable variability in the prosody of ‘NOM_pre èto (be) NOM_post’ constructions, and that, while a prosodic break between NOM_pre and èto is possible, it is certainly not required. Our analysis of a random sample of ca.150 tokens of ‘NOM_pre èto (be) NOM_post’ constructions from the Multimedia sub-corpus of the Russian National Corpus (https://ruscorpora.ru/new/search-murco.html) shows that NOM_pre and èto are often produced without a prosodic break.

More specifically, we find that in some ‘NOM_pre èto (be) NOM_post’ examples, the left edge of èto is indeed aligned with the left edge of a prosodic constituent, which may be realized as a pause and/or a strong glottal onset on the initial vowel in èto. It should be noted, though, that a prosodic break before èto is most commonly found in contrastive or strongly emphatic contexts, especially those involving ellipsis, as well as very formal contexts, such as speeches. This is illustrated in (41) and Figure 1, an example extracted from an interview with Alexei Navalny, which is pronounced with strong emphasis on Mosgorduma ‘Moscow City Duma’. As Figure 1 shows, there is a perceptible pause between Mosgorduma and èto (labelled <SIL>).

Nevertheless, examples without a pause or any evidence for a prosodic boundary before èto are also commonly attested in the corpus. Two subtypes of this pattern can be distinguished. In the first one, which is especially common in informal and/or rapid speech, no audible pauses or other prosodic boundary cues can be identified. This pattern is illustrated in (42) and Figure 2, extracted from an interview with Pyotr Miloserdov. The pronunciation of this example is entirely natural and, in contrast with the example in Figure 1, does not convey extra emphasis on any of the constituents.

The second subtype of examples that do not contain a prosodic boundary before èto in fact provide evidence for a prosodic boundary to the right of èto, which may be manifested by (i) a pause, (ii) a strong glottal onset on the initial vowel of the word following èto, and/or (iii) final lengthening on èto. These prosodic strategies were found in a variety of pragmatic contexts. Cues (i) and (ii) are illustrated in (43) and Figure 3, where èto is followed by a pause, and the first vowel in odnoèlektronnye ‘single-electron’ is realized with a strong glottal onset, labelled with ‘ʔ’. The example is extracted from a mini-lecture on popular science by Professor Artem Oganov.

Finally, cue (iii), final lengthening on èto, is illustrated with an example from an interview with Sergey Glazyev, shown in (44) and Figure 4; the extra duration, received by the final vowel in èto, is especially apparent in the waveform.

We hypothesize that the presence of a prosodic break between NOM_pre and èto in ‘NOM_pre èto (be) NOM_post’ constructions may be attributable to the expression of contrast or emphasis, or be a pragmatic marker of a formal context. It certainly does not represent an intrinsic prosodic property of the construction itself. ‘NOM_pre èto (be) NOM_post’ constructions have a range of prosodic realizations, including ones in which NOM_pre and èto as a unit are prosodically separated from the rest of the clause and likely form a prosodic constituent.

In section 2.3 we argued that the ‘NOM_pre èto (be) NOM_post’ clauses can only be of the identity type; in section 3 we have further shown that NOM_post typically is a referential nominal phrase. However, examples of ‘NOM_pre èto (be) NOM_post’ sentences can be found in which the postcopular constituent is represented by unambiguously predicational expressions – an adjectival phrase or a prepositional phrase. Èto is still present and still fails to exhibit number/gender concord with NOM_pre, always remaining neuter singular. Two such examples, the first featuring a postcopular AP (modno ‘fashionable’) and the second a postcopular PP (na zavtrak/obed ‘for breakfast/lunch’), are given in (45).23

These examples contrast with ordinary predicational copular clauses: here, èto is prohibited, concord/agreement with NOM_pre is required, and the interpretation changes – the property is now attributed directly to NOM_pre.

Existing analyses of èto constructions are ill-equipped to capture the data in (45). For instance, under Markman’s (2008) approach, NOM_post is always an ordinary predicate, and the prediction is that an AP in a context like (45a) should exhibit concord with the subject, and the overt past tense copula should spell out the subject’s φ-features. Under Geist’s (2008) approach, whereby NOM_post is always the main subject of predication and NOM_pre is a hanging topic, we expect no NOM_post to ever be a property-denoting expression. Our double-predication analysis, whereby a predication relation is established both within the clause and inside a ‘big DP’, leaves just enough room to accommodate the seemingly exceptional examples.

We have previously proposed that NOM_pre and èto form a ‘big DP’, whose complement position is occupied by pro. The content of pro is determined by NOM_post, and èto turns it into a predicate (by turning it into an expression of the type <e,t>). Both operations – determining the content of pro and semantic predicativization ‒ happen at LF, and we would like to suggest that they may in principle occur in two different relative orders, as follows.

In the ‘NOM_pre èto (be) NOM_post’ examples with two referential nominal phrases considered in the previous sections, the interpretation of the content of pro within the ‘big DP’ always took place first and predicativization by èto followed. As a result, the referential pro could only be co-indexed with a NOM_post of the type <e>. But now imagine that predicativization takes place first. In this scenario, èto turns pro into a predicate of type <e,t> and after that pro must get linked to an expression of the same type outside the ‘big DP’; naturally, such an expression is an adjectival/prepositional phrase. Note that, in an example like (45), èto cannot be omitted – because it is still required to predicativize pro.

Thus, in (45), modno ‘fashionable’ and na zavtrak ‘for breakfast’ represent abstract properties and determine the content of the predicativized pro within the ‘big DP’. The resulting interpretation is close to that of sentences of the type What Shakira was was fashionable, where fashionable is not predicated directly of Shakira but rather specifies the property associated with her as a phenomenon (as opposed to a person); hence, the perceivable difference in the readings of the examples in (45) and (46).

In syntax, the copula receives the default neuter gender specification. This is unsurprising, as NOM_pre does not show concord with the pro within the ‘big DP’ and, not itself being in a direct relationship with T⁰ (due to be embedded in the constituent in Spec,TP), cannot control agreement at the level of the clause. This is schematized in (47). Accordingly, no concord is possible between Shakira (f) and modno ‘fashionable’ (n) in examples like (45a).

To recap, we proposed that the two LF operations that involve the ‘big DP’ – namely, determining the content of pro and the predicativization of pro ‒ can come in two relative orders, which gives èto the potential to co-occur with a property-denoting expression (an AP or PP) in post-copular position in identificational copular clauses. We leave a detailed semantic analysis of these examples for future work.

Finally, we would like to draw a tentative parallel between the ‘NOM_pre èto (be) NOM_post’ and specificational pseudoclefts of the question‒answer pair type, illustrated in (48).

Based on previous work (Markman 2008; Burukina & den Dikken 2020), we take èto in examples of this kind to spell out the Top⁰ head. Its function is to connect a question CP, located in Spec,TopP, with the answer TP, which contains the focused XP and is located in the complement of TopP. This is schematized in (49); the parts of the sentence that are structurally present but elided are marked by a strike-through line.

Although èto in ‘NOM_pre èto (be) NOM_post’ copular clauses is a D⁰ head, not a Top⁰ head, there is an important parallel between the two kinds of clauses. In both cases, èto is a functional X⁰ that mediates a relationship between two terms that are in a semantic co-construal relationship. These are NOM_pre and pro in ‘NOM_pre èto (be) NOM_post’ constructions and the topic and the comment in èto-focus constructions. This brings closer the two sets of data with èto that have generally been viewed as irreconcilable.