Ileana Comorovski and Klaus von Heusinger, Existence: Semantics and Syntax, xx­xx. Springer LJUDMILA GEIST PREDICATION AND EQUATION IN COPULAR SENTENCES: RUSSIAN VS. ENGLISH* It is a disgrace to the human race that it has chosen to employ the same word "is" for these two entirely different ideas (predication and identity) ­ a disgrace which a symbolic logic language of course remedies. (Russell 1919:172) 1. INTRODUCTION Copular sentences are a source of embarrassment for syntacticians and semanticists alike. The long standing question concerns their ambiguity: copular sentences may express either identity or predication. In a predicational sentence like (1), the property expressed by the predicate noun phrase a teacher (XP2) is predicated of Mary (XP1). The expression by profession is added to indicate that sentence (1) in the table below is interpreted as a predicational sentence. By contrast, equative sentences like (2) assert that the referent of the expression Mark Twain and the referent of the expression Samuel Clemens are identical. Semantically, both NPs are arguments of type e. Table 1. Three types of copular sentence types examples XP1 XP2 (1) predicational John is a teacher (by profession). e (2) equative Mark Twain is Samuel Clemens. e e (3) specificational The murderer is John. ? e Key: the labels XP1, XP2 refer to the actual linear order of the noun phrases at issue independently of their syntactic category (NP vs. DP), their possible syntactic function (subject vs. predicate nominal), and the semantic type shown in the last two columns. It has been assumed that the copula in a predicational sentence takes a predicate of type and applies it to its argument of type e, see (1). Given that in an equative sentence, see (2), both noun phrases are arguments of type e, they do not match the argument structure of the predicational copula. In addition, equatives express an identity relation that is not available in predicationals. The problem which arises is how the identity relation enters the semantic interpretation in an equative sentence. In addition to these basic types of copular sentence, some authors assume a further type which is called 'specificational' (see (3)) because it specifies the "value" of the description given in the pre-copular phrase XP1. In our example, the XP the L. GEIST murderer restricts the variable for which the noun phrase John specifies a value. In (3), XP2 is clearly referential; the denotational status of XP1, however, is controversial. These three types of copular sentence raise the following questions: - What is the basis on which the distinction between predication, identity, and specification rests ? - What does the copula be contribute to each of these interpretations ? - Are these types of copular sentence interrelated in a way that would suggest that one type, say the predicational, can be considered basic and the other types are to be derived from it ? - If so, how are equative and specificational sentences to be derived from their predicational source in a compositional way ? English is a language in which, at least superficially, equative, predicational and specificational copular sentences resemble one another in surface morpho-syntax. Insights into the above questions concerning the ambiguity of English copular sentences can be gained by comparing them with a language which appears to use radically different morpho-syntactic means for expressing predication and equation. We will consider Russian, a language with a rich morphology, but without articles. With respect to copular sentences, Russian displays the following peculiarities: a) defective paradigm: the copula verb byt' in modern Russian has no form for present tense, but only a past and a future form2 b) overt equation marker: equative sentences require the addition of a morphological element, a pronoun, and have a complex syntactic structure c) case alternation: in predicational sentences, post-copular noun phrases (XP2) can occur in the Nominative or in the Instrumental case, as can the initial noun phrase (XP1) in specificational sentences In what follows, we will examine the Russian equivalents of the three types of copular sentence shown above in Table 1. The main issues we want to tackle are these: - Is there a common semantic basis from which predication, equation and specification possibly emerge ? - If so, how could we handle this in a compositional manner in both languages ? The paper argues for a single underlying semantics of the copula in predicational, equative und specificational sentences. The different types of copular constructions can be accounted for if we assume that non-overt type-shift operations suggested by Partee (1986, 1987) and Chierchia (1984) are at work in semantics. We will argue in favour of the view that ­ despite being morphologically non-overt ­ such typeshifting operations can be justified on a structural basis. The way to do this is to examine the Russian counterparts of the three types of copular sentence in Table 1 and exploit the morpho-syntactic distinctions they show to re-analyze the less PREDICATION AND EQUATION IN COPULAR SENTENCES transparent English data. We will attempt to find out how these differences can be derived from language specific parameters. 2. PREDICATIONAL SENTENCES The typical examples of predicational sentences are those shown in (4), i.e. where the post-copular expression describes a property of the subject referent. For example, in (4a) the property of being a teacher by profession is assigned to John. The indicator by profession serves as a marker for the predicative status of XP2. Besides predicative noun phrases, APs and PPs can occur in the post-copular position. (4b) presents being tall as a characteristic of Nina. In (4c), the subject referent is assigned the property of being located in the garden. (4) a. John is a teacher by profession. b. Nina is tall. c. She is in the garden. In a predicational sentence, the copula serves as a link between the subject term and the predicative. Following Partee (1987), we will assume the following (still incomplete)3 lexical entry for the predicational copula: (5) bePred : P x [P(x)] According to Partee (1987), the function of the copula is to indicate that the property denoted by its complement P holds of its external argument x. The be of predication selects for a predicative complement of type but doesn't impose any restrictions on the syntactic nature of it. Different expressions like NPs, APs and PPs, but under some conditions also DPs, can occur in the complement position of the copula. In this paper we will concentrate on predicational sentences with noun phrases (NPs and DPs). The predicative use of noun phrases can clearly be distinguished from the referential use by considering the selection of wh-pronouns in questions. On standard assumptions, the use of the pronoun what instead of who, even if ranging over human beings, is a diagnostic for an NP in predicative use. In clear-cut predicational questions like (6), what cannot be replaced by who, cf. Williams (1983:426). (6) What/*Who is she by profession ? What/*Who is she like ? (7) What did John become ? A doctor. ? What did John talk to ? A doctor. Higgins (1973) suggests that even if related to human individuals, what asks for a property, hence the answer a doctor in (7) is a non-referential noun phrase. L. GEIST Partee (1987:119) mentions an additional diagnostic for the predicative use of noun phrases, namely occurrence in the complement position of consider. The verb consider takes -type complements that are APs or indefinite NPs. (8) Mary considers John intelligent / a genius. As can be seen in (9), definite noun phrases can also occur in the predicative position and hence serve as predicates. (9) I consider Chris the best dancer. Since definites primarily occur in argument positions, the predicative interpretation of definites should be assumed to be derived. The details of such a derivation will be discussed later in this section. Now, consider predicational sentences in Russian as in (10a/b). Russian does not make the who/what distinction with respect to referential vs. predicate noun phrase: the wh-pronoun for properties related to humans is always kto 'who', see (10b). (10) a. Ivan vysokij / v sadu / lučšij tancor. Ivan tallSg.Masc.Nom /in garden/ best dancerSg.Masc.Nom 'Ivan is tall/ in the garden / the best dancer.' b. Kto Ivan po professii ? Učitel'. whoNom Ivan by profession ? A teacherSg.Masc.Nom 'Who is Ivan by profession ? A teacher.' Predicate noun phrases can occur in the complement position of sčitat' 'consider'. (11) Maša sčitaet Ivana intelligentnym/ geniem / lučšim tancorom. Mary considers Ivan intelligentIns / geniusIns / best dancerIns. 'Mary considers John intelligent / a genius / the best dancer.' Russian predicational sentences display a peculiarity which distinguishes them from the other types of copular sentence. In Russian, the case of the predicate XP2 can alternate, provided the copula is non-zero, that is, in the past or in the future form. While in copular sentences with a zero copula the predicate NP always occurs in the Nominative, see (10a/b), in those with an overt copula, the predicate NP can occur in the Nominative or in the Instrumental, cf. (12): (12) Ivan byl xorošim učitelem / xorošij učitel'. Ivan was [good teacher]Ins / [good teacher]Nom 'Ivan was a good teacher.' This case variation encodes a semantic difference, which can be briefly described in approximately the following way: the predicate occurs in the Instrumental if the situation described is temporally bounded, while the Nominative occurs otherwise. PREDICATION AND EQUATION IN COPULAR SENTENCES In this paper we will not go into the details of case variation, for various formal proposals see Geist (1999), Matušansky (2000), Pereltsvaig (2002), and classic literature on this topic by Nichols (1981) and Wierzbicka (1980). In both English and Russian predicational sentences, XP1 has a referential meaning, whereas XP2 is clearly non-referential, i.e. an -type. Although we will analyze properties intentionally, we will (mis)represent them as type for ease of exposition. At this point I would like to briefly introduce some assumptions about the syntax and semantics of noun phrases, which will be important for the analysis I advocate. I follow Partee (1987) in assuming that noun phrases can have intepretations of different types. In copular sentences we will consider at least two interpretations of noun phrases: an e-type and an -type interpretation.4 The -type is the semantic type of predicate noun phrases. A predicate NP as in (13a) is semantically represented as a one-place predicate with an unsaturated referential argument x as in (13b). (13) a. a teacher / učitel' b. [NP a teacher / učitel']: x [TEACHER(x)] An NP with such a denotation can occur in the predicate position in a copular sentence. While the referential argument x of a predicate noun phrase remains unbound, the referential argument of a referentially-used noun phrase like the president of the club in The president of the club came in is bound. For the binding of the referential argument we will make use of the iota-operator as suggested in Partee (1987:117). I follow Zamparelli (2000) in assuming that this operator is the semantic instantiation of some functional head. Following Abney (1987) I will call this functional head D and the maximal projection DP. (14) a. the president of the club b. [DP [D the [NP president of the club ]]]e : x [PRESIDENT-OF-CLUB(x)] In English, the functional head D can be identified by a strong determiner, cf. Langobardi (1994) and Zamparelli (2000) among others. According to Zamparelli, determiners in predicate noun phrases like a in He is a teacher are weak, that is they cannot bind the referential argument of the NP, but rather behave like adjectives, and are thus not located in D. In general, determiners can be assumed to be ambiguous since they can be interpreted in D or within the NP. In a nutshell, I assume that only referential NPs are DPs. The functional head D contains a binding operator for the referential argument of the embedded NP. The referential binding causes the referential interpretation of the embedded NP. Proper names and pronouns, which are inherently referential, are entities of the DP-level.5 (15) a. Mary b. [DP mary]e: mary L. GEIST Again, we assume not only a semantic but also a syntactic difference between referentially and predicatively used noun phrases. The semantic interpretation of a noun phrase is reflected in its syntactic structure. Returning to Russian, Russian is a language without articles and hence there is no syntactic need to assume the DP layer. But because of the strong correspondence between syntactic and semantic categories assumed in this paper, the DP layer as a layer for referentially used noun phrases can be assumed in Russian for semantic reasons (cf. Steube & Spaeth 1998, but see Zlatic 1997 for a different analysis for noun phrases in another articleless language, Serbian). Since Russian is a language lacking articles, the operator in D has no overt counterpart, i.e. D remains syntactically empty. (16) [DP [D [NP prezident kluba ]]]e: x [PRESIDENT-OF-CLUB(x)] As we have seen above, definite descriptions can occur in the complement position of consider/sčitat' and hence can be interpreted as denoting a property. Additionally, the question test in English in (17a/b) shows that the definite description the president of the club can correspond both to the interrogative pronoun who and to the interrogative pronoun what in questions. In the answer to a question with who, the definite DP is interpreted as denoting an individual, while in the answer to a question with what, it is interpreted as denoting a property. (17) a. Who is John ? John is the president of the club. b. What is John ? John is the president of the club. Depending on the interpretation of the DP the president of the club, the copular sentence has a predicational or an equative reading. To account for the predicative use of definite DPs in predicational sentences, we follow Partee (1987) in using the operator ident defined in (18). (18) ident: y x [y = x] or equivalently6 y x [x = y] This operator maps any element y onto the property of being identical to y. The application of ident to the definite description prezident kluba / the president of the club yields the representation in (19). (19) [predDP prezident kluba / the president of the club]: x [y [PRESIDENT-OF-CLUB(y)] = x] The shifted DP denotes the property of being identical to the president of the club. Theoretically, predicates can be created by the ident-operator from all e-type expressions, even from deictically used personal pronouns. However, such pronouns cannot occur in the predicative complement position of consider either in English or in Russian. PREDICATION AND EQUATION IN COPULAR SENTENCES (20) a. *I consider Mary her / she. b. *Ja sčitala Mašu ej. I considered MaryAKK sheIns. They are also excluded in the predicative position of predicational sentences. To force the predicational interpretation of a copular sentence, we use a predicational question with what to ask for a property. As can be seen in (21), the copular sentence with a deictically used personal pronoun in the predicative position is not an appropriate answer to such a question: (21) What is Mary ? #Mary is her / she. Personal pronouns like he or she specify their referents immediately by pointing at them. As Mikkelsen (2003:166ff.) points out, pronouns lack descriptive content. This characteristic of pronouns seems to prevent them from occuring in predicative positions. In contrast, definite descriptions, which can be used as denoting properties like the president of the club, introduce their referent via their descriptive content, which restricts their denotation. Having presented our main assumptions concerning the internal structure of noun phrases, we return to the structure of predicational sentences. To illustrate how the derivation works, we consider a predicational sentence, (22a) with the underspecified syntactic structure in (22b) and the semantics of syntactic constituents in (22c). (22) a. b. John is a teacher. [S [DPJohn] [is [NP a teacher]]] c. (john) P x [P(x)] y [TEACHER(y)] Combining the predicate NP a teacher with the copula, we get the expression in (23a), which denotes the property of being a teacher. (23b) is derived by replacing the variable x with the subject argument john via lambda conversion. (23) a. [ is a teacher]: [P x [ P(x)]] (y [TEACHER(y)] ) x [TEACHER(x)] b. [S John is a teacher]: [x [TEACHER(x)]] (john) [TEACHER(john)] The result corresponds to the paraphrase: "John has the property of being a teacher". We now turn to an examination of equative sentences. L. GEIST 3. EQUATIVE SENTENCES 3.1. Equatives in English Equative sentences as in (24a) resemble predicational sentences as in (24b) in consisting of two noun phrases and the copula verb. But the similarity between predicational and equative sentences is merely superficial.7 (24) a. Cicero is Tully. b. Cicero is an orator and philosopher. The interpretation of the equative sentence differs radically from that of the predicational sentence. While the predicational sentence in (24b) ascribes the property to the referent of XP1, the equative sentence (24a) asserts that XP1 and XP2 have the same referent. There is much controversy in the syntactic and semantic literature about how to account for the fact that copular sentences with the same verb be can render both the predication and the identity relation. The main question is how the identity relation enters the semantic interpretation in equative sentences. At least two ways of dealing with this problem have been discussed in the literature. One possibility is to trace back the distinction predication vs. identity to the copula verb itself, as Higgins (1973), Sharvit (1999), and Schlenker (1998, 2001) do, following the philosophical tradition since Frege and Russell. They posit two verbs be, a "be of identity" and a distinct "be of predication". While the be of predication, selects a predicative complement and an argumental e, the be of identity takes two arguments of type e, and, hence, has a different argument structure, see (25a). After instantiating the variable x by Tully in (25b) and y by Cicero in (25c), we get the resulting Semantic or Logical Form shown in (25d). (25) a. beIdent : x y [y = x]8 b. [is Tully]: [x y [y = x]] (tully) y [y = tully] c. [S Cicero is Tully]: [y [y = tully]] (cicero) d. [cicero = tully] Another possibility is to locate the source of the ambiguity in the arguments. Williams (1983) and Partee (1986, 1998) offer such an alternative account by assuming a single be of predication plus some type-shifting operations on arguments. To account for equative sentences with two e-type noun phrases, Partee uses the operator ident, which shifts the type of the post-copular referential DP to the corresponding predicative reading, see (26a) in our notation. Applied to Tully in (26b), ident converts it into the property of being (identical to) Tully. Note that the ident operation locates the identity relation in the type-shifted meaning of XP2, not in the copula; this is what allows us to dispense with a separate be of identity. PREDICATION AND EQUATION IN COPULAR SENTENCES (26) a. ident: x y [y = x] b. ident (tully): y [y = tully] The predicational copula in (27a) can take the predicate (26b) as its complement. The derivation of the sentence S is given in (27b/c). (27) a bePred: P x [P(x)] b. [is Tully]: [P x [P(x)]] (y [y = tully]) x [tully = x] c. [S Cicero is Tully]: [y [y = tully]] (cicero) cicero = tully The ident type-shift allows identity sentences to be semantically interpreted as predicational. This analysis has the desirable result of avoiding an ambiguity with respect to the copula verb. However, as Partee herself notes, one would like to have further evidence for such an analysis. In English equative sentences, there is no explicit counterpart of the ident-operator and hence no independent evidence showing that XP2 is a predicate and not an argument. In other words: we cannot confirm the assumption that the identity relation enters the semantic interpretation of the sentence Cicero is Tully via the nominal complement and not via the copula verb. The analysis in (27) suggests that XP2 is used as a predicate. If this is correct, we expect the shifted noun phrase (ident (tully)) to be allowed in predicational small clauses governed by the verb consider. However, Tully is in fact barred from this position, whereas a true predicative NP like a talented politician is allowed (cf. Rothstein 2001:245 for a similar observation): (28) a. *They considered Cicero Tully. b. They considered Cicero a talented politician. Obviously, the ungrammaticality of (28a) is due to the fact that XP2 ­ Tully ­ by its very nature as a proper name cannot serve as a predicate. However, given the analysis in (27), it should be convertible to a predicate but ­ as (28a) shows ­ it is not, at least in certain cases. Rothstein (2001:237) notes that the insertion of the copula in the small clause in sentences like (28a) improves their acceptability. (29) They considered Cicero to be Tully. If the ident-operator were to apply to the referential XP2 and convert it into a predicate, a predicational small clause would emerge. In predicational small clauses the copula makes no semantic contribution to the sentence and it can be omitted as was shown above. Given that, it is not clear why the copula cannot be omitted in (29). From this we may conclude that the analysis suggested in (27) is in need of some sort of adjustment. In order to overcome the flaws of this analysis while maintaining its advantage of assuming only one copula, we suggest having the ident L. GEIST operator apply not to XP2 but to the copula verb. The idea is this: we combine the formula for ident in (30a) and bePred in (30b) via Functional Composition, yielding (30c). (30) a. ident: u y [y = u ] b. bePred: P x [P(x)] c. beIdent: [P x [P(x)]] [u y [y = u]] u [P x [P(x)] (y [y = u])] u [x [ y [y = u] (x)]] u x [x = u] The shifted copula is the copula of identity. The semantic derivation of the whole sentence proceeds in the same way as the derivation shown in (25). The advantage of the type-shift analysis suggested in (30) over the "two copula account" is that we still have only one predicational copula in the lexicon from which we can derive the "be of identity", which is a desirable result. 3.2. Equatives in Russian In Russian, equative sentences differ from predicational sentences syntactically. They require the occurrence of a constant form of the demonstrative pronoun to 'thisSg.Neut'. In predicational sentences, however, to is excluded, cf. (31) Mark Twain ­ *(to) Samuel Clemens. (equative) Mark T.Nom this Samuel C.Nom 'Mark Twain is Samuel Clemens.' (32) Mark Twain (­ *to) pisatel' po professii. (predicational) Mark T. this writerNom by profession 'Mark Twain is a writer by profession.' The demonstrative pronoun seems to be needed to express that the two noun phrases have the same referent. Another difference between predicationals and equatives in Russian is the following: in equative sentences, there is no case alternation, XP2 occurs in the Nominative only. In order to warrant that XP2 in the Instrumental is excluded from to-sentences (and not because of lacking an overt copula), we have to use an overt form of the byt' copula. (33) a. Ciceron ­ to byl Tullij. CiceroNom thisNeut wasMasc TullyNom b. *Ciceron ­ to byl Tulliem. CiceroNom thisNeut wasMasc TullyIns There is strong evidence that in (33), XP2 is the underlying subject: the copula agrees with XP2 and not with to, which remains Singular Neuter Nominative. PREDICATION AND EQUATION IN COPULAR SENTENCES Now, what does to contribute to the syntax and semantics of the equative sentence ? Why is it obligatory ? Błaszczak & Geist (2000a/b) have shown that to-sentences are not simplex sentences like predicationals but, rather, they exhibit a cleft-like structure. They consist of two parts, see (34) below, the to-clause and the dislocated XP1. XP1 is separated from to-clause by an intonation break, indicated by the dash in (34). (34) [S [DP XP1] ­ [IP to cop XP2]] We adopt the view suggested by Junghanns (1997) that demontrative to has the status of a base-generated internal topic. The pronoun to connects XP1, which serves as an external topic, with the clause. The semantic relation holding between XP1 and to will be discussed below. Syntactically, to-sentences are similar to left dislocations. However, in contrast to typical left dislocation constructions, the dislocated noun phrase in Russian equatives is resumed not by a personal pronoun but by an uninflected demonstrative. The demonstrative pronoun to 'this' in equative sentences has a constant morphological form independent of Gender and Number of the dislocated XP1, i.e. to remains uninflected. As can be shown in (35) the uninflected to cannot be used to pick out a human referent of the dislocated DP, only the inflected form of to (ti 'thisPl') can, but it sounds archaic. Personal pronouns are more common as resumptives in left dislocation constructions in Russian. (35) Devočki, oni / *to / ti obyčno lučše učatsja čem mal'čiki. girls theyPers.3.Pl/ this Dem.Sg.Neut /thisDem.Pl usually better learn than boys. 'Girls are usually better in school than boys.' To account for the fact that the inflected to can be used as an anaphor for referential DPs we assume that the inflected to is itself a DP and can be semantically represented as an e-type variable xi restricted to range over human and non-human individuals, cf. (36a).9 The anaphoric relation of to to some antecedent-DP is indicated via indexation. (36) [DPtoi]: xi (type e) However, in equative sentences, another type of to, the uninflected to, is used. Since the uninflected to cannot trigger the agreement of the copula and does not agree with the dislocated DP in Gender and Number, we assume that it is predicative. We can predicativize to by means of the operator ident as is illustrated in (37a). The result is a predicate expression of type , cf. (37b): (37) a. ident (xi) = y [xi = y] b. [predDPtoi ]: y [xi = y] (type ) L. GEIST As is shown in (37b) the predicate to is not just semantically but also syntactically more complex than the argument to. Syntactically, the argument to corresponds to a DP, whereas the predicate to to a predDP. The head of predDP contains the identoperator. We now have all the ingredients to derive an equative sentence. We begin with the IP in (38a). The semantics of the constituents is given in (38b). (38) a. b. to Tullij. [IP [predDPtoi] [ cop [DP Tullij ]]] y [xi = y] yP [P(y)] (tully) We follow Williams (1983) and Partee (1986) in assuming that the external argument and the complement of the copula can occur in either order: the argument can precede or follow the predicate. In to-sentences, the predicate to precedes the argument-DP. The predicational copula applies the pronominal predicate to to the individual tully denoted by Tullij. The compositional result for the to-clause is the representation in (39): (39) [IP to Tullij]: [xi = tully] (type t) The interpretation of (39) can be paraphrased as "some individual x is identical to Tully". In order to interpret the sentence, the variable x has to be identified via the co-indexing-relation with its antecedent, the DP Cicero, cf. (40). (40) Ciceron ­ to Tullij ciceroi xi = tully Due to co-indexing, the individual variable x takes the constant cicero as its value. If we instantiate the predicate variable x by cicero we get [cicero = tully] which is logically equivalent to the representation we derived in (27c) for the corresponding equative sentence in English. To summarize: our analysis suggests that equative sentences in Russian are syntactically but also semantically more complex than hitherto assumed. Equation in Russian is mediated by inverted predication, employing the pronominal predicate to. Comparing equatives in Russian and English, we notice that the identity relation enters the semantic interpretation at different places: in English, it is the copula which is shifted via ident; in Russian equatives, the operator ident is applied to the demonstrative pronoun to. The following question arises: what is the reason for such a different syntaxsemantics mapping for equation in English and Russian. Why does Russian employ an additional pronominal element to convey equative semantics ? Why does English differ from Russian in this respect ? The situation we face is this: in Russian, the copula verb byt' lacks an overt present form which ­ in a way to be explored in a separate study ­ somehow PREDICATION AND EQUATION IN COPULAR SENTENCES prevents type-shifting operations from applying. But a type-shift is needed to form a predication structure from two unrelated referential DPs, here: Cicero and Tully. This need for the type-shift seems to be the reason for the use of the additional pronominal item in Russian equatives. The pronoun forms a predicate that relates the two DPs. In English, an overt copula is available in all tenses, so there is no obstacle preventing application of the type-shifting operator.10 The use of pronouns in equative sentences seems to be a widespread phenomenon among the languages of the world. A situation similar to that in Russian obtains in Polish where the demonstrative pronoun to is used. But languages which are genetically not related to Russian, such as Scottish Gaelic (Adger & Ramchand 2001), Standard Arabic and Modern Hebrew also employ a pronoun in equative sentences. Unlike Russian, these languages employ a personal and not a demonstrative pronoun, cf. an example from Modern Hebrew: (41) dani *(hu) mar yosef. (Rothstein 2001:207) dani PronMasc.Sg mr yosef Dani is Mr Yosef. As argued by Rothstein (2001), the personal pronoun hu is necessary to trigger the type-shift of a DP to allow equative sentences to be formed. Interestingly, in Modern Hebrew, the copula stem lacks a present form, as in Russian. However, the pronominal elements in Russian and Hebrew differ in distribution and categorial status and hence cannot be treated in exactly the same way. In the next section, we will examine specificational copular sentences, for short: specificationals. 4. SPECIFICATIONALS In this section we will focus on sentences like (42): (42) a. The president of the club is John. b. The murderer was Raskolnikov. c. My teacher is Mary. Such copular sentences were called specificational by Higgins (1973) because, intuitively, XP2 specifies the "value" of the description given by XP1. In our example (42a), the president of the club (=XP1) restricts the variable for which XP2 specifies the referent of John as a value. One of the peculiar aspects of specificational sentences that has been pointed out by many people is their fixed focus-background structure (cf. Heycock & Kroch 2001:148). Specificational sentences invariably come with a focused post-copular DP (=XP2) and an XP1 which provides the background, and hence cannot be stressed.11 In specificational sentences like those in (42), XP2 is clearly referential. However, the denotational status of XP1 is controversial. In some accounts, XP1 is L. GEIST analyzed as a predicative NP and the sentence is considered an inverse predicational sentence (e.g. Heggie 1988). If XP1 is analyzed as referential, then the sentence can be classed as equative, as assumed by Heycock and Kroch (1999). Partee (1998) observes that cross-linguistically, both possibilities are available. In Italian, as noted by Moro (1997:28), specificationals display a different agreement pattern from English. This can be seen when XP1 and XP2 do not match in Number. In English, the copula invariably agrees with XP1 in both predicational and specificational sentences, see (43). In Italian, however, in predicationals the copula agrees with XP1, while in specificationals, it agrees with XP2, cf. (44). Syntactically, XP2 can thus be assumed to be the subject, and XP1 to be the predicate. (43) English a. The pictures of the wall *was / were the cause of the riot. b. The cause of the riot was / *were the pictures of the wall (44) Italian a. Le foto muro *fu / furono la causa della rivolta the pictures of the wall was / were the cause of the riot b. La causa della rivolta *fu / furono le foto del muro the cuase of the riot was / were the pictures of the wall The inverted agreement pattern can be regarded as strong evidence for a predicate fronting analysis in Italian. In English, instead, XP1 is the syntactic subject of specificational sentences and hence is not a predicate. Thus, Heycock & Kroch (1998, 1999) argue for subsuming English specificationals under equatives. If the analyses of specificationals as (inverted) predicational in Italian and as equative in English are correct, the problem that arises is that the meaning of "specification" is expressed in different languages by different semantic structures: in one language "specification" gives rise to the semantics of (inverse) predication, and in another language to the semantics of equation. This is an unexpected and undesirable consequence, which we want to overcome, if possible. In what follows, we will derive the invariable semantics of specificational sentences based on Russian examples and then compare the syntax-semantics mapping in Russian and English. We will show how language specific differences in the syntax-semantics mapping can be traced back to independently attested differences between the two languages. 4.1. Specificational sentences in Russian Consider the following specificational sentences: (45) a. Ubijca staruxi ­ (*to) Raskol'nikov. murdererNom of-old-lady this Raskolnikov 'The murderer of the old lady is Raskolnikov.' PREDICATION AND EQUATION IN COPULAR SENTENCES b. Pričinoj avarii *byla /byli neispravnye tormoza. reasonSg. Fem.Ins of-accident wasSg.Fem/werePl broken brakesPl 'The reason for the accident was broken brakes.' c. Edinstvennyj, kto stal na našu storonu, *byl /byla Varvara only-personMasc.Nom who came to our side wasFem/wasMasc BarbaraFem 'The only person who defended us was Barbara.' (Padučeva & Uspenskij 1997:178) To characterize these sentences, we will compare them with equative and predicational sentences, which we introduced in the previous sections. 1) Comparing the specificational sentence in (45a) with equatives, we observe that the predicate proform to is excluded in the specificational sentence. This fact suggests very strongly that, at least syntactically, Russian specificationals do not belong to the equative type, and therefore XP1 and XP2 cannot both be referential. 2) As (45) shows, XP1 can occur in the Nominative or in the Instrumental. Since the case alternation Nominative vs. Instrumental is only possible with predicative noun phrases, the case alternation on XP1 can be seen as a crucial argument in favour of assigning it predicate status (cf. Padučeva & Uspenskij 1997, Partee 1998). In contrast to predicationals, the predicate noun phrase is in the initial position in specificationals. 3) Concerning the agreement marking of the copula, Russian, as shown in (45b/c), patterns with Italian; that is, it displays an inverted agreement pattern, hence XP2 serves as the syntactic subject. 4) Personal pronouns cannot occur in the initial position of specificational sentences or, more precisely, if such a pronoun is placed in the initial position of a copular sentence with a proper name occurring in the post-copular position, the sentence is interpreted as an equative and, in Russian, the pronoun to has to be inserted. (46) My ­ *(to) Marija i Ivan. we this Maria and Ivan 'We are Mary and Ivan.' The inability of personal pronouns to occur as XP1 in specificational sentences can be considered to be a direct consequence of their inability to occur in the predicative position (cf. section 2). From these observations we can conclude that specificationals are related to predicationals and can be syntactically analyzed as predicate inversions. However, a frequent objection to the analysis of specificationals as inverted predicationals in other languages is that not all predicate expressions occurring in the predicative position in predicational sentences can also occur in the initial position of specifi- L. GEIST cational sentences (Heycock & Kroch 1999:379). Although predicate expressions like APs and NPs can occur in predicational sentences in Russian, they are excluded from the initial position of specificationals. Only descriptions like prezident kluba 'the president of the club' and ubijca 'the murderer' in (43c/d) are felicitous in both sentence types: (47) a. predicational specificational Ivan ­ dobryj. #Dobryj Ivan.12 Ivan good-natured good-natured Ivan 'Good-natured is Ivan.' 'Ivan is good-natured.' b. Ivan byl učitelem po professii. #Učitel' po professii ­ Ivan. Ivan was teacher by profession. teacher by profession Ivan 'Ivan was a teacher by profession.' 'A teacher by profession is Ivan.' c. Ivan byl prezidentom kluba. Prezidentom kluba byl Ivan. Ivan was presidentIns clubGen PresidentIns clubGen was Ivan 'Ivan was the president of the club.' 'The president of the club was I.' d. Raskol'nikov byl ubijcej. Ubijcej byl Raskol'nikov. Raskolnikov was the-murderer the-murderer was Raskolnikov 'Raskolnikov is the murderer.' 'The murderer is Raskolnikov.' I will call predicates that can only occur in predicational sentences 'core predicates', since predicate interpretation is the most natural interpretation for such expressions. If specificational sentences are inverted predicational sentences, it is not clear why core predicates are excluded in specificationals. I think that the prohibition against the occurrence of core predicates in the initial position in specificational sentences is due to information structure, more precisely, due to one dimension of it ­ the so called topic-comment-structure.13 The topic is defined as the entity which the predication is pragmatically about (Reinhart 1982, Molnár 1991:41-43, 1993, among others). At least in Russian and English, the topic is normally associated with the sentence initial position. The part of the sentence without the topic is the comment. Now, consider the topic-comment-structure in predicational sentences in comparison to specificational sentences. The predicational sentence (48a) is about Raskol'nikov, hence this expression is the topic. The specificational sentence (48b) is about the murderer or more correctly, about somebody who is the murderer, hence ubica 'the murderer' serves as topic. (48) a. Raskol'nikov ­ ubijca. 'Raskolnikov is the murderer' b. Ubijca ­ Raskol'nikov. 'The murderer is Raskolnikov.' PREDICATION AND EQUATION IN COPULAR SENTENCES To verify our intuitive hypothesis concerning the topics in the above sentences, we use a topic-marking construction introduced by čto kasaetsja X 'as for X' (cf. Reinhart 1982:10, among others): (49) a. Čto kasaetsja Raskol'nikova, ja dumaju, čto on ubijca. what concerns Raskolnikov I think that he murderer 'As for Raskolnikov, I think that he is the murderer.' b. Čto kasaetsja ubijcy, ja dumaju,čto to Raskol'nikov. what concerns murdererAkk I think that it Raskolnikov 'As for the murderer, I think that it is Raskolnikov.' Native speakers agree that (49a) is a paraphrase of (48a) and not of (48b), whereas (49b) is a paraphrase of (48b) and not of (48a). According to this test, in the predicational sentence (48a) the topic is Raskolnikov, in the specificational sentence (48b) the topic is ubijca. Interestingly, core predicates like učitel' po professii 'teacher by profession' cannot occur at all in this topic-marking construction, as is shown in (50). (50) #Čto kasaetsja učitelja po professii, ja dumajučto to Ivan. what concerns teacherAkk by profession, I think that it Ivan 'As for a teacher by profession, I think that it is Ivan.' This suggests that the choice of a noun phrase as the topic expression of a given sentence is sensitive to the semantic properties of this noun phrase. The crucial factor for the infelicity of (50) seems to be that the NP following čto kasaetsja cannot be interpreted as being referential or as presupposing a referent. The observation that topics are sensitive to existential presupposition is not new, it goes back to Strawson (1964) who states that topic noun phrase expressions carry existential presuppositions. Reinhart (1982:11) makes similar assumptions. She proposes that a noun phrase can be interpreted as topic only if it is referential. Existential presupposition or referentiality can be considered a topichood condition for noun phrases. Now we have to check if the topics in the predicational and specificational sentences in (48) satisfy this topichood condition. In the predicational sentence (48a) the topic Raskolnikov is referential and hence satisfies the topichood requirement proposed by Reinhart. In the specificational sentence (48b), we analyze the murderer as an expression of type e shifted to a predicate via the ident-operator. In the predicational sentence (48a) the topic Raskolnikov is referential and hence satisfies the topichood requirement proposed by Reinhart. In the specificational sentence (48b), we analyze the murderer as an expression of type e shifted to a predicate via the ident-operator. The logical form of such a predicate DP can be represented analogously to the predicate DP prezident kluba / the president of the club, which was analyzed in section 2, as shown here in (51): L. GEIST (51) [predDP ubijca]: x [y [MURDERER(y)] = x] This is the property-type denotation for a noun phrase based on its e-type denotation. The predDP presupposes the existence of an individual fitting the descriptive content of the DP. Such a predDP fulfils the topic requirement assumed by Strawson. Core predicate expressions like učitel' po professii 'a teacher by profession', and dobryj 'good-natured', are predicates without (existentially presupposed) referents, they do not fulfil the topic requirement and as a consequence they cannot occur in specificational sentences. However, there is no topic requirement on the predicative position of a predicational copular sentences, that is why core predicate expressions are felicitous in this position. What is important in my analysis is that it is not the copula itself that restricts the realization of the predicative complement in specificational sentences. The copula takes a predicate of type as complement in both predicational and specificational sentences, and does not impose any restrictions on the syntactic or semantic nature of this complement. The restriction on the realization of the predicate complement in specificational sentences comes from the pragmatics. In a nutshell, assuming that XP1 in Russian specificationals is a topic, we can explain why core predicate expressions like APs and NPs as in (47a/b) are barred from initial position in specificationals. We are now in a position to analyze the specificational sentence (52a). Since in this paper we concentrate on the semantic structure of copular sentences, for the sake of simplicity we use underspecified syntactic structures, leaving out indicators of movement and/or traces. According to our discussion of examples (45), in specificational sentences in Russian, XP2 is the syntactic subject and XP1 is a predicate of a special sort. The semantics of syntactic constituents is given in (52c). (52) a. Ubijca ­ Raskol'nikov b. [S [predDP ubijca] [cop [DP Raskol'nikov]e ]] c. x [y [MURDERER(y)] = x] x P [P(x)] (rask) The copula simply "instructs" us to predicate its argument of its e-type argument. In this case, the combination of the subject DP with the copula yields (53a). Having instantiated the predicate variable P we get (53b). (53) a. [cop Raskol'nikov]: P [P(rask)] "the property that holds of Raskolnikov" b. [S ubijca cop Raskol'nikov]: y [MURDERER(y)] = rask The result shown in (53b) can be paraphrased as "the property of being the murderer holds of Raskolnikov". This result suggests that specificationals in a way combine features of both equative and predicational sentences. PREDICATION AND EQUATION IN COPULAR SENTENCES 4.2. Some speculations on specificational sentences in English Heycock & Kroch (1999) argue that English specificationals have to be treated as equatives. Under such an analysis, the mapping between syntax and semantics could be schematically represented as in (54): both NPs are referential, and the copula maintains the identity relation. (54) a. The murderer was Raskolnikov. b. x [MURDERER(x)] = rask Heycock & Kroch (1999) show evidence in favour of XP1 being referential. Both noun phrases in specificational sentences may be modified by non-restrictive relative clauses as shown in (55). (55) The duty nurse, who is very efficient, is Rina, who I am very fond of. (Heycock & Kroch 1999:374) Such data point to the referential status of XP1 in specificational sentences and provide an argument in favour of the analysis of specificational sentences as equative. However, the analysis of specificational sentences as equatives faces a problem. As Mikkelsen (2002b, 2003) notes, the interpretation of XP1 in specificationals differs from the interpretation of XP1 in equatives. The difference becomes obvious in a pronominalization test using tags suggested by Mikkelsen. It is well-known that the pronoun in the tag is always anaphorically related to the subject of the clause. In the equative sentence (56a), the personal pronoun he can refer back to the subject Samuel Clemens. But in the tag of the specificational sentence in (56b), the Singular Neuter it has to be used instead. (56) a. Samuel Clemens is Mark Twain, isn't he/*it ? b. The murderer of the old lady is Raskolnikov, isn't *he/it ? If specificationals were equatives, the failure of the personal pronoun in (56b) to pick out the referent of XP1 would be unexpected. Note that it in subject position of the tag governs agreement of the verb, hence it cannot be a predicate of type . Since the semantic type of an anaphor must match that of its antecedent, the antecedent of it, here the murderer, cannot be an -predicate either. In addition, the agreement facts in specificationals in (43) where the copula invariably agrees with the XP1 also suggest that XP1 is the syntactic subject and hence has argumental status. The referential status of it and its antecedent the murderer in (56b) can be accounted for straightforwardly by the approach suggested by Chierchia (1984). It can be anaphorically related to entities that are analyzed as "nominalized properties" in the sense of Chierchia. Compare the following sentences: L. GEIST (57) a. Blue is a nice colour, isn't it ? b. To be home is nice, isn't it ? The antecedent of it in these examples is a nominalized AP and an infinitival VP respectively. They serve as subjects of predication and hence are arguments, just as their anaphor it in the tag is. Chierchia (1984) proposes to analyze property expressions with argumental status as entities of a special type; that is, as "nominalized properties". Although Chierchia suggests the symbol for representing the semantic type of primitive properties, we will (mis)represent "nominalized properties" here as entities of type e for simplicity. Chierchia assumes two operators: nom for nominalization and pred for predicativization, respectively, see (58). The operator nom maps -type properties onto their entity-correlates of type e. This is the operation which, in Chierchia's analysis, is involved in nominalization, e.g. for conversion of the property denoted by the adjective blue in The coat is blue into the "nominalized property" as denoted by the common noun blue in (57a). The operator pred applies to an entity which is the entity-correlate of a property, such as the infinitive in (57b), thus making it into a corresponding property as in John is home as suggested by Chierchia. The operators nom and pred are inverse to each other. (58) nom (P): P (type e) pred (x): x (type ) We will now make use of pred for the analysis of XP1 in English specificational sentences. We assume that the semantic representation of it in (57) is a designated variable xi of type e which is restricted to range over entity-correlates of properties. To derive a specificational sentence such as (56b) The murderer is Raskolnikov compositionally, we will start with the sentence It is Raskolnikov, where the subject XP1 is pronominalized. First, we have to specify the semantics of the copula. The copula in such a sentence has to combine two arguments of type e: a nominalized property and an individual. To adjust the predicational copula, which always takes an predicate and an e argument, we have to change its argument structure. The copula of specification must have the argument structure as represented in (59). (59) beSpec: z y [ z (y)] (where z ranges over entity-correlates of properties, and y over individuals) The specificational copula can be derived from the predicational one by additional operations. The predicational copula can be combined with a type-shifter, as in (60a), that transforms the property P in the argument structure of the copula into a nominalized property. The Functional Composition of the two functions in (60b) yields the representation for the copula. (60) a. type-shifter: z [ z] b. P y [P(y)] z [ z ] z y [ z(y)] PREDICATION AND EQUATION IN COPULAR SENTENCES The function of the specificational be is to apply the nominalized property z to an individual y. Assuming that the complement and the subject of the copula can occur in either order (cf. Partee 1986 and Williams 1983), we change the order of the variables. ­ I have rephrased the two paragraphs following formula (60). We now have all the ingredients to derive a specificational sentence in a compositional way: (61) a. It is Raskolnikov. b. [S [XPit] [ is [DP Raskolnikov]] ] xi y z [ z(y)] (rask) The semantics of the whole sentence is derived below. (62) a. [is Raskolnikov]: [y z [ z(y)]] (rask) z [ z(rask)] b. [S it is Raskolnikov]: [z [ z(rask)]] (xi) xi(rask) The result achieved in (62b) can be paraphrased thus: "the contextually specified property x holds of Raskolnikov". Let us return to the specificational sentence The murderer is Raskolnikov. As the pronoun it and the XP1 the murderer can be anaphorically related, as shown in (56b) above, we can assume that the murderer has the same referential status as the pronoun; that is, it denotes a nominalized property. But for the XP the murderer such an interpretation is not basic but is derived. For the semantic derivation, we would suggest the following device: a) we predicativize the individual the murderer via ident type shift, thus obtaining the property "being identical to the murderer", cf. (63a); b) this property, in turn, can be nominalized by the nominalization operator ; the result is the entity-correlate of the property "being the murderer" in (63b); c) now we can compose the shifted meaning of XP1 with the meaning of the constituent [copula + XP2] from (62a) repeated in (63c); the result is given in the last line of (63d). (63) a. ident (x [MURDERER(x)]) y [x [MURDERER(x)] = y] b. nom (y [x [MURDERER(x)] = y]) y [x [MURDERER(x)] = y] c. [ be Raskolnikov]: z [ z(rask)] d. [z [ z(rask)]] ( y [x [MURDERER(x)] = y]) [[ y [x [MURDERER(x)] = y]] (rask) [y [x [MURDERER(x)] = y]] (rask) x [MURDERER(x)] = rask L. GEIST Interestingly, the result we receive in (63d) amounts to the same as the one we obtained for Russian in (53b) above. What distinguishes the two languages is the semantics of the constituents. While in Russian, the XP1 position is filled by a predicate, which denotes a property, XP1 is of an argumental type in English; that is, it denotes a nominalized property. The differences between Russian and English regarding the syntax-semantics mapping in specificational sentences are attributable to language specific grammatical restrictions. In English, the property of being the topic and belonging to the background coincides with the property of (preferably) being the subject. English does not allow a predicate in subject position. As Partee (1998) notes, in Russian, because of the relatively free word-order it has at its disposal, the topic and syntactic subject need not be the same element. This has consequences for the syntaxsemantics mapping in specificational sentences. 5. CONCLUDING REMARKS In this paper we have explored the mapping between the syntax and semantics of copular sentences in Russian in comparison to English. We argued for a single underlying semantics of the copula in predicational, equative und specificational sentences in both languages. The difference in interpretation between types of copular sentence can be explained by the ident-type-shift, which enters the semantic composition at different places in the sentence. In equative sentences, the ident operator applies to the element that combines two DPs. In English, this element is the predicational copula. Since in Russian, the copula in the present tense is zero, the ident operator applies to the demonstrative pronoun to. As for specificational sentences in English and Russian, the ident operator applies to the first referential DP, turning it into a predicate. Since in English, unlike in Russian, XP1 has argumental status, some additional type-shift operators have to be assumed in the semantic composition of English specificational sentences. The additional operations are the following: the change of the argument structure of the predicational copula which, in specificational sentences, takes a nominalized property and applies it to an argument, and the nominalization of the XP1. The paper derives the invariant semantics for predicational, equative and specificational sentences and explores how this invariant semantics is mapped to the syntactic structure in the two languages. The differences in mapping can be traced back to independent differences in the morpho-syntax of the two languages. The paper contributes to the general understanding of conditions and domains for application of type-shift operators. Originally, the type-shift operators such as ident were assumed to account for different interpretations of noun phrases. In this paper, we extended the domain of application for the type-shifter ident to verbs. In English, we assume that in equative and specificational sentences, the predicational copula has to be shifted. Interestingly, the contrast between the type-shifted copulas and the predicational (i.e. basic) copula shows up in small clauses. The verb consider can take predicational, equative and specificational small clauses as its complement. While in the predicational small clause of consider in (64a), the copula of PREDICATION AND EQUATION IN COPULAR SENTENCES predication can be omitted, the shifted copula of identity and the shifted copula of specification in (64b/c) cannot be. (64) a. They considered Cicero (to be) a talented politician. predicational be b. They considered Cicero *(to be) Tully. equational be c. They considered the best politician *(to be) Cicero. specificational be These data, which are already known in the literature, suggest that type-shift operators can only apply to overt elements in the sentence. This can explain why in Russian, a language with a zero present form copula, an additional overt element, the pronoun to, is employed in equative sentences. The ident-type shift can apply to this pronoun to convey equative semantics. However, we have for the moment no explanation for why it is a demonstrative pronoun that appears in Russian instead of a verbal copula. In further research, other types of copular sentence will have to be explored and used as a test for the type-driven analysis suggested in this paper. The conditions and restrictions for the application of type-shifters also merit further explorations. Ljudmila Geist Institut fuer Linguistik / Germanistik Universitaet Stuttgart D-70174 Stuttgart Germany NOTES * This research was carried out within the DFG-sponsored project "Copula-predicative constructions" directed by Ewald Lang at the Centre for General Linguistics (ZAS) in Berlin. Special thanks go to Philippa Cook for improving the language of the paper and for very helpful comments on English data. All remaining errors are my own. I would like to thank Joanna Błaszczak, Klaus von Heusinger, Gerhard Jäger, Ewald Lang, Line Mikkelsen and Ilse Zimmermann for very intensive and helpful discussion. I am also grateful to the audience of the Nancy "Existence" workshop 2002, especially Vladimir Borshev, Agnes Bende-Farkas, Barbara Partee and Georges Rebuschi, for their helpful comments and criticisms. I owe special thanks to the organizer of the workshop Ileana Comorovski and two anonymous reviewers whose comments I have tried to do justice to in the final draft. 2 The inherited present tense form est', which was originally the 3 Person Singular form of byt', has lost other forms and can be used for all Persons and Number. This archaic form is sometimes used in definitions and if tense is focused, cf. (i) Astrologija est' nauka. astrology is science 'Astrology is a science.' (ii) On byl est' i budet tvoim otcom he was is and will-be your father 'He was and will always be your father.' As an anonymous reviewer points out, there is a further, frequently used, copular verb in modern Russian: javljat'sja. Javljat'sja has no precise counterpart in English, but can be approximately translated in English as 'be'. Unlike byt', javljats'ja has an overt form in present tense. However, contrary to byt', the L. GEIST copula javljat'sja has a very restricted distribution as it can only combine with predicative NPs which denote some quality, and never with NPs denoting just professions or nationality. 3 We neglect, throughout the paper, tense and aspect markings of the copula verb. 4 In addition to these two types, Partee (1987) proposes the quantificational type <,t>. I will disregard this third type and will analyze noun phrases in argument positions generally as e-type. 5 In what follows, we will use "NP" to refer to nominals that do not contain the functional D-layer, and "DP" to refer to nominals that do contain such a layer. The term noun phrase or XP will be used if we want to let the status of a noun phrase remain unspecified. 6 Since the identity relation is a symmetric relation, the variables xi and y in the formula can stay in either order. 7 As an anonymous reviewer points out, the second noun phrase in an equative sentence cannot be an indefinite, whereas it can be an indefinite in a predicational sentence. 8 The predicate "=" "identical with" means simply "has the same denotation as". 9 The pronoun to has many different interpretations. It can be related to concrete individuals but also to abstract individuals and propositions, cf. Junghanns (1997) among others. In this paper only the former use of to is considered. 10 An anonymous reviewer points out that in earlier stages of Russian, the archaic present form copula est' was used in equative sentences instead of to. He/she quotes a statement from Lenin: (i) Kommunizm est' sovetskaja vlast' pljus lektrifikacija vsej strany. communism is Soviet power plus electrification of-whole conutry 'Communism is Soviet power plus the electrification of the whole country.' The reviewer states that, later, people misquoted this quotation as Kommunizm to sovetstkaja vlast' pljus lektrifikacija vsej strany. This datum illustrates the diachronic change from verbal copula in present tense, which gradually got lost, to a pronominal copula with the same function of maintaining the identity relation between two referents. 11 Sentences like The TEACHER was JOHN with two focused phrases or The TEACHER was John with the main focus on XP1 are not interpreted specificationally. They occur in contexts different to those of specificationals and also differ structurally, (see Heycock & Kroch 2002 and Mikkelsen 2002a, 2003 for details). 12 Such sentences are acceptable only if the inverted predicate gets a focus accent and a falling intonation. But in this case, the construction is not a specificational sentence. In specificationals, the pre-copular phrase belongs to the background and hence is not focused. For the structural differences between predicate inversion and specificational sentences, see Mikkelsen (2002a, 2003). 13 I follow Molnár (1991, 1993) in assuming the topic-comment structure as one of the three levels of information structure besides the focus-background structure and theme-rheme structure. REFERENCES Adger, D. & Ramchand, G. (2001). Predication and Equation. In N. Zhang (Ed.), Syntax of Predication. ZAS papers in linguistics, 26 (pp.1-22). Berlin: ZAS. Baszczak, J & Geist L. (2000a). Kopulasätze mit den pronominalen Elementen to/to im Polnischen und Russischen. In E. Lang (Eds.), Copular and AUX ­ Constructions. ZAS Papers in Linguistics, 16 (pp. 115-139). Berlin: ZAS. Baszczak, J. & Geist L. (2000b). Zur Rolle des Pronomens to/to in spezifizierenden Kopulakonstruktionen im Polnischen und Russischen. In G. Zybatow at al. (Eds.). Current Issues in Formal Slavic Linguistics. Linguistik International, 5 (pp. 247-257). Peter Lang: Frankfurt/Main. Chierchia, G. (1984). Topics in the Syntax and Semantics of Infinitives and Gerunds, Ph.D. dissertation, Univ. of Mass/Amherst. Declerck, R. (1983). 'It is Mr. Y' or 'He is Mr. Y' ? Lingua, 59, 209-246. Geist, L. (1999). Kopula byt' (sein) eine funktionale und/oder eine lexikalische Kategorie ? In E. Lang & L. Geist (Eds.), Kopula-Prädikativ-Konstruktionen als Syntax/Semantik-Schnittstelle. ZAS Papers in Linguistics, 14 (pp.1-39). Berlin: ZAS. Heggie, L. A. (1988). The syntax of copular structures. Ph.D. dissertation, Univ. of Southern California. PREDICATION AND EQUATION IN COPULAR SENTENCES Heycock C. & Kroch A. (1998). Inversion and equation in copular sentences. ZAS Papers in linguistics, 10, 71-87. Heycock, C. & Kroch A. (1999). Pseudoclefts Connectedness: Implications for the LF Interface Level. Linguistic Inquiry, 30 (3). 365-397. Heycock C. & Kroch A. (2002). Topic, Focus, and Syntactic Representations. In L. Mikkelsen & C. Potts (Eds.), Proceedings of the 21st WCCFL (pp. 141-165). Somerville, MA: Cascadilla Press. Higgins, R. (1973). The pseudo-cleft construction in English, Ph.D. dissertation, Univ. of Mass/Amherst. Junghanns, U. (1997). On the so-called oeto-cleft construction. In M. Lindseth & S. Franks (Eds.), Annual Workshop on Formal Approaches to Slavic Linguistics. The Indiana Meeting (pp. 166-190). Ann Arbor, Michigan: Michigan Slavic Publications. Langobardi, G. (1994). Reference and Proper Names: A theory of N-Movement in Syntax and Logical Form. Linguistic Inquiry, 25/4, 609-665. Matushansky, O. (2000). The Instrument of Inversion. Instrumental Case and Verb Raising in the Russian Copula. In R. Billerey & L. Brook (Eds.), Proceedings of the 19th WCCFL (pp. 101-115). Somerville, MA: Cascadilla Press. Mikkelsen, L. (2002a). Specification is not inverted predication. In M. Hirotani (Ed.), Proceedings of NELS 32 (pp. 403-422). University of Massachusetts, Amherst: GLSA. Mikkelsen, L. (2002b): Two types of definite description subjects. In Proceedings from the Student Session at ESSLLI 14. Trento. ESSLI 14. Mikkelsen, L. (2003). Specifying Who: On the Structure, Meaning, and Use of Specificational Copular Clauses. PhD University of California Santa Cruz. [SLUG Pubs]. Molnár, V. (1991). Das TOPIK im Deutschen und Ungarischen. Lunder germanistische Forschungen 58. Stockholm: Almqvist and Wiksell. Molnár, V. (1993). Zur Pragmatik und Grammatik des Topik-Begriffs. In Marga R. (ed.), Wortstellung und Informationsstruktur (pp.155-202). Tübingen: Niemeyer. Moro, A. (1997). The raising of predicates. Cambridge: Cambridge University Press. Nichols, J. (1981). Predicate Nominals. A Partial Surface Syntax of Russian. (Linguistics 97). Berkeley, Los Angeles, London: University of California Press. Padučeva, E.V. (1985). Vyskazyvanie i ego sootnesennost' s dejstvitel'nost'ju. Moskow: Nauka. Padučeva, E.V. (1992). Značenie i sintaksičeskie funkcii slova to. In Poblemy strukturnoj lingvistiki 1980 (pp. 78-9). Moscow: Nauka. Padučeva, E.V. & Uspenskij V.A. (1997). Binominativnoe predloženie: problema soglasovanija svjazki. In L.P. Krysin (Ed.), Oblik slova. Sbornik statej pamjati D.I. Šmeleva (pp. 170-82). Moscow: Nauka. Partee, B. (1986). Ambiguous pseudoclefts with unambiguous Be. In S. Berman, J.-W. Choe & J. McDonough (Eds.), Proceedings of NELS 16 (pp. 354-366). Amherst, Mass.: University of Mass. Partee, B. (1987). Noun Phrase Interpretation and Type-Shifting Principles. In J. Groenendijk et al (Eds.), Studies in Discourse Representation Theory and the Theory of Generalized Quantifiers (pp. 115- 143). Dordrecht: Foris. Partee, B. (1998). Copula Inversion Puzzles in English and Russian. In K. Dziwirek at al. (Eds.), Annual Workshop on Formal Approaches to Slavic Linguistics: The Seattle Meeting (pp.361-395). Ann Arbor: Michigan Slavic Publications. Pereltsvaig, A. (2001). On the Nature of the Intra-Clausal Relations: A study of Copular Sentences in Russian and Italian. Dissertation McGill University Montréal. Reinhart, T. (1982). Pragmatics and Linguistics: An analysis of Sentence Topics. Bloomington. Reproduced by the Indiana University Linguistic Club. Rothstein, S. (2001). Predicates and Their Subjects. (Studies in Linguistics and Philosophy, vol 74). Dordrecht, Boston, London: Kluwer. Russell, B. (1919). Introduction to Mathematical Philosophy. London: Allen & Unwin. Schlenker, P. (1998). Pseudocleft Connectivity and the Structure of NP. Ms, MIT, Cambridge MA. Schlenker, P (2001). Clausal Equations (A note on the connectivity problem). NLLT, 21, (1), 157-214 Sharvit, Y. (1999). Connectivity in Specificational Sentences. NLS, 7, 299-339. Steube, A. & Spaeth, A. (1998). DP-Semantik und Informationsstrukturierung im Russischen (auf der Basis eines Vergleichs mit dem Deutschen). Sprache & Pragmatik, 46, 1-70. Strawson, P. F. (1964). Identifying reference and truth values. Theria 30. Reprinted in Steinberg, D. & Jakobovits, L. (eds.) (1971), Semantics. London: Cambridge: University Press. Wierzbicka, A. (1980). The Case for Surface Case. Ann Arbor, Michigan: Karoma. L. GEIST Williams, E. (1983). Semantic vs. Syntactic Categories. Linguistics & Philosophy, 6, 423-446. Zamparelli, R. (2000): Layers in the Determiner Phrase. New York: Garland. Zlatic L. (1997). The Structure of the Serbian Noun Phrase. PhD. Universtity of Texas Austin.