2.1 Connectivity effects

The most frequently mentioned connectivity effects concern binding, quantifier scope and the bound interpretation of anaphora. Thus, in (6)–(10), the element in the post-copular part of the specificational copular sentence behaves as if it occupied the position of the gap in the pre-copular relative sentence (indicated by __).

1. (6) Binding: Principle A

   1. (a) What he_i is __ is angry with himself_i.

   2. (b) He_i is angry with himself_i.

      (den Dikken et al. Reference Dikken, Meinunger and Wilder2000: 42)

   1. (a′) *What she_i is __ is angry with himself_j.

   2. (b′) *She_i is angry with himself_j.

1. (7) Binding: Principle C

   1. (a) *What he_i claimed __ was that Ian_i was innocent.

   2. (b) *He_i claimed that Ian_i was innocent.

      (Heycock & Kroch Reference Heycock, Kroch, Mikkelsen and Potts2002: 103)

1. (8) Binding: Principle B

   1. (a) *What John_i is __ is proud of him_i.

   2. (b) *John_i is proud of him_i.

1. (9) Bound anaphora interpretation Footnote ⁵

   1. (a) What every linguist_i loves __ is her_i first syntax class.

   2. (b) Every linguist_i loves her_i first syntax class.

      (Cecchetto Reference Cecchetto, Cecchetto, Chierchia and Teresa Guasti2001: 96)

1. (10) Negative Polarity Footnote ⁶

   1. (a) What I don't have __ is any bread.

   2. (b) I don't have any bread.

      (Heycock & Kroch Reference Heycock, Kroch, Mikkelsen and Potts2002: 103)

The contrast between (6a) and (6a′) and between (6b) and (6b′) shows that an anaphor in the post-copular part of the specificational sentence behaves as if it were a clause-mate with an NP in the pre-copular part of the sentence. The examples in (7)–(8) demonstrate that, in both specificational copular sentences and canonical sentences, NPs and pronouns cannot be co-referential with a preceding antecedent in the same sentence. The examples in (9) show that a variable (her) in the post-copular part of a specificational copular sentence can be bound by a quantifier (every in every linguist) in the pre-copular part (9a), similar to what happens in the simple sentence (9b). In (10a) the presence of a negative element in the pre-copular part of the specificational sentence licenses the appearance of a negative polarity item (any) in the post-copular part, just as the negation in (10b) licenses the appearance of a negative polarity item in the object position.

In generative grammar, binding, quantifier scope, variable interpretation and the licensing of negative polarity items are considered to be a reflex of the structural relations between the elements of a sentence, which are expressed in terms of c-command (cf. Haegeman Reference Haegeman1994). Hence, the reflexive in (6b), the noun phrase in (7b), the pronoun in (8b), the possessive in (9b) and the negative polarity item in (10b) are argued to be c-commanded by the preceding elements on which they depend for their interpretation. Reflexives must be c-commanded by a co-referential expression (Principle A of Binding Theory), pronouns cannot be c-commanded by a co-referential expression in the same sentence (Principle B), and NPs can never be c-commanded by a co-referential expression (Principle C). Moreover, it is commonly assumed that a variable can only have a bound interpretation if it is in the c-command domain of the relevant quantifier (cf. Reinhart Reference Reinhart1983: 122, Heim & Kratzer Reference Heim and Kratzer1998) and that a negative polarity item is licensed when it is c-commanded by a negative or near-negative element.

In contrast with canonical non-copular sentences, specificational copular sentences are highly problematic for the just-mentioned c-command constraints. The post-copular reflexive in (6a), the noun phrase in (7a), the pronoun in (8a), the possessive in (9a) and the negative polarity item in (10a) behave just as if they were c-commanded by the element in the pre-copular part on which they depend for their interpretation. This is, however, impossible, since an element inside the pre-copular phrase cannot c-command outside this phrase.

2.2 Anti-connectivity effectsFootnote ⁷

In recent linguistic literature, it has been observed that specificational sentences are also characterized by anti-connectivity effects, which exhibit the opposite pattern from the connectivity effects in (6)–(10) (cf. Sharvit Reference Sharvit1999; Cecchetto Reference Cecchetto, Hirotani, Coetzee, Hall and Kim2000, Reference Cecchetto, Cecchetto, Chierchia and Teresa Guasti2001; den Dikken et al. Reference Dikken, Meinunger and Wilder2000; Schlenker Reference Schlenker2003). Most anti-connectivity effects concern binding, quantifier scope and the interpretation of bound variables.

For instance, the following examples show that an NP in the post-copular part of a specificational copular sentence cannot be co-referential with a pronoun in the pre-copular part of the specificational copular sentence, whereas the corresponding non-copular sentences are grammatical, both in Italian (11) and in French (12):Footnote ⁸

1. (11)

   1. (a) *Chi __ lo_i vide è la sorella di Gianni_i.who him saw is the sister of Gianni ‘The sister of Gianni saw him.’

   2. (b) La sorella di Gianni_i lo_i vide.the sister of Gianni him saw ‘The sister of Gianni saw him.’

      (Cecchetto Reference Cecchetto, Cecchetto, Chierchia and Teresa Guasti2001: 108)

1. (12)

   1. (a) ??Ce qui lui_i a coÛté cher,that what to-him has cost expensive,c'est la Toyota de Jean_i.it is the Toyota of Jean ‘The Toyota of Jean cost him a lot of money.’

   2. (b) La Toyota de Jean_i lui_i a coÛté cher.the Toyota of Jean to-him has cost expensive ‘The Toyota of Jean cost him a lot of money.’

      (Schlenker Reference Schlenker2003: 204f.)

Also, the quantifier in the pre-copular part of the Italian (13a) and French (14a) specificational sentence can bind the variable in the post-copular part, but the corresponding simple sentences (13b) and (14b) are marginal:Footnote ⁹

1. (13)

   1. (a) Ciò che _sfilava dietro a ogni generale_i era il suo_i battaglione. this that was-marching behind every general was his battalion ‘His battalion was marching behind every general.’

   2. (b) ??Il suo_i battaglione sfilava dietro a ogni generale_i.his battalion was-marching behind every general ‘His battalion was marching behind every general.’

      (Cecchetto Reference Cecchetto, Cecchetto, Chierchia and Teresa Guasti2001: 106)

1. (14)

   1. (a) Ce qui marchait derrière [chaque général]_i, c'était son_i bataillon.this that was-marching behind every general it was his battalion ‘His battalion was marching behind every general.’

   2. (b) ?Son_i bataillon marchait derrière [chaque général]_i.His battalion was-marching behind every general ‘His battalion was marching behind every general.’

      (Schlenker Reference Schlenker2003: 205f.)

Moreover, Cecchetto shows that the scope interaction of the universal quantifier and the existential quantifier in Italian is not the same in specificational copular sentences as in the corresponding simple sentences, regardless of whether the existential quantifier is in the post-copular (15a) or the pre-copular (17a) part of the specificational copular sentence. The author argues that the distributive reading (in which the universal quantifier in ogni studente ‘each student’ has scope over the existential quantifier un ‘a’ in professore severo ‘a strict professor’) is missing in the specificational sentence (15a), whereas it is present in the canonical sentence (15b):

1. (15)

   1. (a) * distributive reading

      Chi __ ha controllato ogni studente è un professore severo.who has checked every student is a professor strict ‘The person who checked every student was a strict professor.’

   2. (b) ok with distributive reading

      Un professore severo ha controllato ogni studente.a professor strict has checked each student ‘A strict professor checked every student.’

      (Cecchetto Reference Cecchetto, Cecchetto, Chierchia and Teresa Guasti2001: 115)

The same pattern is observed in French:

1. (16)

   1. (a) * distributive reading

      Celui qui a contrôlé chaque étudiant, c'est un professeur sévère.who that has checked each student it is a professor strict ‘The person who checked each student was a strict professor.’

   2. (b) ok with distributive reading

      Un professeur sévère a contrôlé chaque étudiant.a professor strict has checked each student ‘A strict professor checked each student.’

Consider also the following example, where the existential quantifier (in un bravo avvocato ‘a good lawyer’) is in the pre-copular part of the specificational sentence (17a). Here again, the distributive reading does not occur in the copular sentence (17a), whereas it does in the corresponding canonical sentence (17b):

1. (17)

   1. (a) * distributive reading

      Ciò che un bravo avvocato ha difeso __ è stata ogni multinazionale del tabacco.this that a good lawyer has defended was been every company of-the tobacco ‘What a good lawyer defended was every tobacco company.’

   2. (b) ok with distributive reading

      Un bravo avvocato ha difeso ogni multinazionale del tabacco.a good lawyer has defended every company of-the tobacco ‘A good lawyer defended every tobacco company.’

      (Cecchetto Reference Cecchetto, Cecchetto, Chierchia and Teresa Guasti2001: 117)

Schlenker (Reference Schlenker2003: 207) mentions a similar example in English, attributing it to Eisner (Reference Eisner1995) and Uli Sauerland (p.c.):

1. (18)

   1. (a) * distributive interpretation

      What some student admires __ is every teacher.

   2. (b) ok with distributive interpretation

      Some student admires every teacher.

2.3 Consequences

The pairs of sentences in (6)–(10), which exemplify connectivity effects, contrast with the anti-connectivity cases in (11)–(18) in that, with respect to binding, quantifier scope and the bound interpretation of anaphora, the specificational copular sentence in the former pairs yields the same grammaticality judgment as its corresponding canonical sentence, whereas this is not the case in the latter pairs of sentences.

The existence of both connectivity and anti-connectivity effects in specificational sentences has led to two fundamentally different conceptions of specificational sentences, and, by consequence, of phenomena such as binding, scope and anaphor interpretation.

According to some authors, connectivity and anti-connectivity effects show that phenomena such as binding and scope cannot be analyzed on the basis of the structural c-command constraints, and have to be accounted for in a purely semantic way (Jacobson Reference Jacobson, Harvey and Santelmann1994, Sharvit Reference Sharvit1999, Cecchetto Reference Cecchetto, Hirotani, Coetzee, Hall and Kim2000, Reference Cecchetto, Cecchetto, Chierchia and Teresa Guasti2001; cf. Sharvit Reference Sharvit1999 and Schlenker Reference Schlenker2003 for a detailed survey of these semantic theories). The main disadvantage of these accounts is of course that they have to reanalyze each c-command test to show that it is not a c-command test but is actually due to other (semantic) properties.

Besides these semantic approaches, there has also been a syntactic answer from linguists suggesting that, in copular specificational sentences, the corresponding simple sentence is in fact present at some level of syntactic representation. In the approaches advocated by Akmajian (Reference Akmajian1970), Grosu (Reference Grosu1973), Culicover (Reference Culicover1977) and Bošković (Reference Bošković1997), it is argued that the corresponding simple sentence is reconstructed at LF by movement of the post-copular focused constituent, which is interpreted in its original position. In Heycock & Kroch's (Reference Heycock and Kroch1999) proposal, which postulates an additional level of syntactic representation, it is argued that the derivation of the corresponding simple sentence is a post-LF process, triggered on semantic grounds. The proponents of the Question-in-Disguise Theory (den Dikken et al. Reference Dikken, Meinunger and Wilder2000 and Schlenker Reference Schlenker2003, drawing on original insights from Ross 1972) analyze pseudo-clefts as a question-answer pair, wherein part of the answer is elided – [what John likes __ ] is [John likes himself] (Schlenker Reference Schlenker2003: 159).Footnote ¹⁰

The existence of anti-connectivity effects has been used by proponents of the semantic approaches as an argument against any syntactic theory. It is clear that, if the simple sentence counterpart of a specificational sentence is present at some level of representation, then both sentences should display the same behavior with respect to binding and scope, which is obviously not the case. Hence, whereas connectivity effects could be accounted for in a syntactic analysis, anti-connectivity effects constitute a problem for any syntactic analysis of specificational sentences and phenomena such as scope, binding and anaphor interpretation.

In the next section, I show that the distinction between the connectivity and anti-connectivity effects mentioned above is not independent. I demonstrate that anti-connectivity effects with respect to binding disappear when the influence of information structure is neutralized, and that anti-connectivity effects with respect to scope result from independent semantic properties of specificational sentences.

3.2 The distinction between connectivity and anti-connectivity effects

Information structure thus has an influence on phenomena like binding, variable interpretation and quantifier scope. However, information structure is not taken into account in the comparison between the specificational copular sentences and the corresponding simple sentences in the pairs of sentences in (6)–(18). This comparison is consequently less than perfect.

It is well known that specificational copular sentences have a fixed information structure wherein the pre-copular constituent (the variable part) is interpreted as the topic, the ground, the discourse-old or presupposed part of the sentence, and the post-copular constituent (the value part) is discourse-new, the focus of the sentence (cf. Prince Reference Prince1978; Higgins Reference Higgins1979: 234–236; Declerck Reference Declerck1988: 12; Van Peteghem Reference Van Peteghem1991; Heycock & Kroch Reference Heycock and Kroch1999: 394, 2002: 148f.; Partee Reference Partee, Kusumoto and Villalta2000: 199f.; Mikkelsen Reference Mikkelsen2002: section 4).

As for canonical sentences it is commonly assumed that, in their neutral, unmarked reading (with the main stress of the sentence on the most deeply embedded constituent), the subject is interpreted as the topic or theme, while the elements in the verb phrase are interpreted as the focus or rheme (cf. Lambrecht Reference Lambrecht1994, Zubizarreta 1998, among many others).Footnote ¹¹

In all the connectivity cases in (6)–(10), the two constituents involved in the connectivity effect have the same information status in the specificational copular sentence as in the corresponding simple sentence (with the neutral, unmarked stress pattern). Thus, he is part of the topic and himself is part of the focus in both the specificational copular sentence in (6a) and the corresponding canonical sentence in (6b). In (7), the pronoun he is part of the topic and the NP Ian is part of the focus in both the specificational copular sentence and the corresponding simple sentence. The NP John is also part of the topic and the pronoun him part of the focus in both the copular sentence and the canonical sentence in (8). Similarly, in both the specificational copular sentence in (9a) and the corresponding canonical sentence in (9b), the universal quantifier in every linguist is part of the topical part of the sentence, whereas the bound variable (her in her first syntax class) is part of the focal part of the sentence. In the same way, the negative polarity item is part of the focus in both the specificational copular sentence and the canonical sentence in (10).

In the anti-connectivity examples in (11)–(16), by contrast, the information status of the two relevant constituents in the specificational copular sentence and in the canonical sentence is not the same. The NPs Gianni (in la sorella di Gianni ‘John's sister’) and Jean (in la Toyota de Jean ‘John's car’) are in the focus of the pseudo-clefts in (11a) and (12a), whereas they appear in the topical part of the sentence (i.e. the subject) in the corresponding canonical sentences (11b) and (12b). Furthermore, the universal quantifier (in ogni generale and in chaque général ‘each general’) is part of the topic in the specificational copular sentences in (13a) and (14a), whereas it is part of the focus in the corresponding canonical sentences (13b) and (14b). In (15)–(16), the universal quantifier (ogni in ogni studente ‘each student’ and chaque in chaque étudiant ‘each student’) is part of the topic in the copular sentences (15a) and (16a), but part of the focus in the corresponding canonical sentences (15b) and (16b).

Hence, in contrast with the connectivity examples in (6)–(10), the anti-connectivity examples in (11)–(16) exhibit a different information structure for the relevant constituents involved in binding, quantifier scope or variable interpretation.

Note, however, that this is not the case in the anti-connectivity examples (17)–(18). In (17), ogni multinazionale del tabacco ‘each tobacco company’ is in focus position and un bravo avvocato ‘a good lawyer’ is part of the topic in both the copular sentence (17a) and the canonical sentence (17b). The same holds for (18), where some student has and have the same information status in both sentences, as does every teacher. The impossibility of the distributive reading in (17a) and (18a) could, however, be due to the fact that the universal quantifier is in post-copular position. Williams (Reference Williams1994) argues that a universal quantifier in the post-copular position of a pseudo-cleft is unable to have wide scope over an existential quantifier in the free relative in the pre-copular part. Moro (Reference Moro1990) observes the same for non-cleft specificational copular sentences. It thus seems to be a general constraint on specificational copular sentences that a universal quantifier in post-copular position cannot have scope over an existential quantifier in pre-copular position. According to Heycock & Kroch (Reference Heycock and Kroch1999: 386), this is due to the specific semantics of specificational sentences, which the authors analyze as equatives:

Cecchetto (Reference Cecchetto, Cecchetto, Chierchia and Teresa Guasti2001: 116), however – in contrast with Moro (Reference Moro1990), Heycock & Kroch (Reference Heycock and Kroch1999) and Williams (Reference Williams1994)– argues that the claim according to which a universally quantified pivot can never take wide scope over a quantifier inside the pre-copular part of the specificational sentence is too strong, since the following example does have a distributive reading:

1. (22) ok with distributive reading

   Chi ha scelto un bravo avvocato è stato ogni politico corrotto.who has chosen a good lawyer has been every politician corrupt ‘Who chose a good lawyer was every corrupt politician.’

   (Cecchetto Reference Cecchetto, Cecchetto, Chierchia and Teresa Guasti2001: 116)

The author argues that an (apparent) universal quantifier in post-copular position is not interpreted as a quantifier, but has a group reading, just like the corresponding plural DP:

1. (23) ok with distributive reading

   Coloro che hanno scelto un bravo avvocato sono stati i politici corrotti.those who have chosen a good lawyer have been the politicians corrupt ‘Those who chose a good lawyer were the corrupt politicians.’

   (Cecchetto Reference Cecchetto, Cecchetto, Chierchia and Teresa Guasti2001: 117)

The author's claim is supported by the fact that, whenever a distributive reading is possible for the (apparent) universal quantifier in post-copular position, it is also possible for the plural DP in post-copular position, and vice versa. Hence, the distributive reading is possible in both (22) and (23), whereas it is impossible in both (17a) and the corresponding example (24), in which the universally quantified NP has been replaced by a plural DP:Footnote ¹²

1. (24) * distributive reading

   Ciò che un bravo avvocato ha difeso __ sono state le multinazionale del tabacco.this that a good lawyer has defended were been the companies of-the tobacco ‘What a good lawyer defended were the tobacco companies.’

In consequence, the fact that the distributive reading is impossible in the specificational copular sentences (17a) and (18a), whereas it is available in the corresponding canonical sentences, cannot be considered as an independent anti-connectivity effect, since it is due to the semantic properties of specificational sentences, regardless of the specific analysis adopted.

The conclusion from the preceding is that in the anti-connectivity cases in (11)–(16), the two constituents involved in the phenomena under consideration (co-reference, quantifier scope and bound anaphora interpretation) do not have the same information status in the specificational copular sentence as in the corresponding canonical sentence, in contrast with the connectivity cases in (6)–(10). As far as I know, this has not been observed previously in the linguistic literature on connectivity and anti-connectivity effects.

The immediate consequence of this is that no conclusions whatsoever can be drawn on the basis of the different acceptability judgments of specificational copular sentences and the corresponding simple sentences in these examples, since such differences can potentially be due to the different information structure of the constituents. Hence, in order to neutralize the influence of information structure, specificational copular sentences have to be compared with simple sentences with the same information structure. This is the only way to determine whether or not the examples in (11)–(16) involve real anti-connectivity effects, which are not the consequence of a different information structure status of the involved constituents.

In the following section I show that the anti-connectivity effects presented above disappear when the influence of information structure is neutralized and when the semantics of the whole construction is taken into consideration.

3.3 Neutralizing the influence of information structure in anti-connectivity

First consider the pairs of sentences in (11)–(12). In these examples of anti-connectivity, la sorella di Gianni and la Toyota de Jean are part of the focus of the copular sentence, whereas they function as the topic in the unmarked reading of the corresponding simple sentence. Cecchetto (Reference Cecchetto, Cecchetto, Chierchia and Teresa Guasti2001) and Schlenker (Reference Schlenker2003) agree that the specificational copular sentences (11a) and (12a) are degraded in acceptability, in contrast with the fully acceptable canonical sentences (11b) and (12b). These judgments, however, do not take into consideration information structure. In order to annul the influence of information structure, both sentences have to be put in a context where the corresponding constituents have the same information structure status. Since the post-copular constituent of a specificational copular sentence is by default interpreted as the focus (see above), the obvious way to achieve this is to put the corresponding canonical sentence in a context where this constituent is narrowly focused, i.e. in the context of the answer to a question about the referent of this constituent.

Note, however, that when the simple sentences in (11b) and (12b) appear in a context where the whole subject NP is focused, i.e. in the answer (A) to a question (Q) about the subject, different grammaticality judgments are obtained in French and Italian: French (12b) becomes ungrammatical, as (26b) illustrates, in contrast with Italian (11b), as (25b) shows.

1. (25)

   • Q:Chi vide Gianni?who saw John ‘Who saw John?’

   • A:

     1. (a) *Chi __ lo_i vide è [_F la sorella di Gianni_i]who him saw is the sister of John ‘The person who saw him is John's sister.’

     2. (b) [_F La sorella di Gianni_i] lo_i vide. the sister of John him saw ‘The sister of John saw him.’

1. (26)

   • Q:Qu'est-ce qui a coÛté cher à Jean?what has cost expensive to John ‘What cost John a lot?’

   • A:

     1. (a) ???Ce qui lui_i a coÛté cher, c'est [_F la Toyota de Jean_i].that what to-him has cost expensive it is the Toyota of Jean ‘What cost him a lot, is John's Toyota.’

     2. (b) ???[_F La Toyota de Jean_i] lui_i a coÛté cher. the Toyota of John to-him has cost expensive ‘The Toyota cost John a lot.’

This would lead to the surprising conclusion that the neutralization of information structure annuls the anti-connectivity effect in French but not in Italian. However, the NPs la sorella di Gianni and la Toyota de Jean in these examples are problematic to the extent that they consist of two NPs, and that, when focused, they can be interpreted either as having the last N focused or the whole NP.Footnote ¹³ In Italian, when a complex NP containing two NPs is focused, as in (25b), the intonation pattern is ambiguous between a reading where only the NP Gianni is focused, or the full NP la sorella di Gianni. In French, however, where preverbal focused subjects do not occur very frequently, the interpretation which is most likely to be associated with the answer in (26b) is one where only Jean is focused. As it may be hard for native speakers of Italian to distinguish between the two readings of (25b), the comparison is less than perfect.

In order to avoid this ambiguity and to have real minimal pairs, I contrast sentences where only the first N of the NP is contrastively focused. When information structure is neutralized in this way, the marginal sentences (11a) and (12a) both turn out to be fully acceptable, as (11′a) and (12′a) show:

1. (11′)

   • Q:Chi vide Gianni?who saw John ‘Who saw John?’

   • A:

     1. (a) Chi __ lo_i vide è [_F la sorella] di Gianni_i … non la sua maestra.who him saw is the sister of John not the his teacher ‘The person who saw him is John's sister, not his teacher.’

     2. (b) [_F La sorella] di Gianni_i lo_i vide … non la sua maestra. the sister of John him saw not the his teacher ‘The sister of John saw him, not his teacher.’

1. (12′)

   1. Q:Qu'est-ce qui a coÛté cher à Jean?what has cost expensive to John ‘What cost John a lot?’

   2. A:

      1. (a) Ce qui lui_i a coÛté cher, c'est [_F la Toyota] de Jean_i … et pas sa Peugeot.that what to-him has cost expensive it is the Toyota of John and not his Peugeot ‘What cost him a lot is John's Toyota, not his Peugeot.’

      2. (b) [_F La Toyota] de Jean_i lui_i a coÛté cher … et pas sa Peugeot. the Toyota of John to-him has cost expensive and not his Peugeot ‘The Toyota of John cost him a lot of money, not his Peugeot.’

These examples show that, when the influence of information structure is neutralized in an unambiguous way, the specificational copular sentences have the same grammaticality judgment as the corresponding canonical sentences. In other words, (11′)–(12′) show that the anti-connectivity effects in (11)–(12) disappear when the influence of information structure is neutralized.

Consider now the anti-connectivity examples in (13)–(14). The marginal canonical sentences (13b) and (14b), where the subject is interpreted as the topic of the sentence in an unmarked reading, are fully grammatical when they appear in a context where the subject is narrowly focused, both in Italian (13′) and in French (14′):

1. (13′)

   • Q:Che cosa sfilava dietro a ogni generale?what thing was-marching behind every general ‘What was marching behind every general?’

   • A:[_F Il suo_i battaglione] sfilava dietro a ogni generale_i.his battalion was-marching behind every general ‘His battalion was marching behind every general.’

1. (14′)

   1. Q:Qu'est-ce qui marchait derrière chaque général?what was-marching behind every general ‘What was marching behind every general?’

   2. A:[_F Son_i bataillon] marchait derrière chaque général_i. his battalion was-marching behind every general ‘His battalion was marching behind every general.’

The contrast between (13b) and the answer in (13′), and the contrast between (14b) and the answer in (14′) demonstrate that the anti-connectivity effect disappears when the influence of information structure is neutralized.

Exactly the same observations can be made with respect to the following example in French, which is analogous to (14), with the difference that the universal quantifier is not in a PP (derrière chaque général ‘behind every general’) but in an object NP (chaque livre ‘every book’):

1. (27) [Context: At a book exhibition]

   1. (a) Celui qui signera chaque livre_i, c'est son_i auteur.the one who will-sign every book it is its author ‘The one who will sign every book is its author.’

   2. (b) *Son_i auteur signera chaque livre_i.its author will-sign every book ‘Its author will sign every book.’

This example, which at first sight would seem to be a good example of anti-connectivity, is also influenced by information structure, as is shown by the contrast between (27b), which has a neutral reading, and the answer in (27′) where the subject is focused, which is the real counterpart to (27a):

1. (27′)

   1. Q:Qui signera chaque livre?who will-sign every book ‘Who will sign every book?’

   2. A:[_F Son_i auteur] signera chaque livre_i.its author will-sign every book ‘Its author will sign every book.’

Hence, all the apparent anti-connectivity effects considered above disappear when the influence of information structure is neutralized.

This is, however, not the case for the anti-connectivity examples (15)–(16), repeated here for convenience:

1. (15)

   1. (a) * distributive reading

      Chi __ ha controllato ogni studente è un professore severowho has checked every student is a professor strict ‘The person who checked every student was a strict professor.’

   2. (b) ok with distributive reading

      Un professore severo ha controllato ogni studente.a professor strict has checked each student ‘A strict professor checked every student.’

      (Cecchetto Reference Cecchetto, Cecchetto, Chierchia and Teresa Guasti2001: 115)

1. (16)

   1. (a) * distributive reading

      Celui qui __ a contrôlé chaque étudiant, c'est un professeur sévère.who that has checked each student it is a professor strict ‘The person who checked each student was a strict professor.’

   2. (b) ok with distributive reading

      Un professeur sévère a contrôlé chaque étudiant.a professor strict has checked each student ‘A strict professor checked each student.’

When the canonical sentences (15b) and (16b) are put in a context where the adjective is contrastively focused, however, as in (15′) and (16′), the distributive reading remains possible:

1. (15′) ok with distributive reading

2. Q:Chi ha controllato ogni studente?who has checked every student ‘Who checked every student?’

3. A:Un professore [_F severo] ha controllato ogni studente. (non un professore intelligente).a professor strict has checked every student not a professor intelligent ‘A strict professor checked every student (not an intelligent professor).’

1. (16′) ok with distributive reading

2. Q:Qui a contrôlé chaque étudiant?who has checked every student ‘Who checked every student?’

3. A:Un professeur [_F sévère] a contrôlé chaque étudiant (au lieu d' un professeur intelligent).a professor strict has controlled every student instead of a professor intelligent ‘It was a strict professor who checked every student (instead of an intelligent professor).’

The acceptability of these examples is thus not affected by information structure. However, they involve only apparent anti-connectivity effects, to the extent that they are due to a semantic constraint. It is well known that the post-copular phrase in specificational sentences is highly referential (cf. among others Declerck Reference Declerck1988 and Van Peteghem 1991), as its function is to single out the element that satisfies the description given in the pre-copular phrase. Hence, an indefinite in post-copular position is preferably interpreted as a specific indefinite, or, as Cecchetto (Reference Cecchetto, Cecchetto, Chierchia and Teresa Guasti2001: 116) argues, the quantificational interpretation of the indefinite pivot is blocked. Note, however, that cases in which an indefinite pivot seems to be scopally dependent on a universal quantifier in the wh-clause are not excluded:

1. (28) ok with distributive reading

   Chi ogni studente prima o poi ha insultato è un professore severo.who every student sooner or later has insulted is a professor strict ‘(The one) who every student sooner or later insulted was a strict professor.’

   (Cecchetto Reference Cecchetto, Cecchetto, Chierchia and Teresa Guasti2001: 114)

Cecchetto (Reference Cecchetto, Cecchetto, Chierchia and Teresa Guasti2001) argues that, since indefinite pivots are not quantificational, the distributive reading in cases like (28), which he calls a pseudo narrow scope reading, cannot be the result of the indefinite quantifier being in the scope of the universal quantifier. Instead, the author convincingly shows that the distribution of this reading with indefinite pivots in pseudo-clefts is explained by their choice function interpretation (cf. Kratzer Reference Kratzer and Rothstein1998).Footnote ¹⁴ This semantic property holds not only for indefinite pivots, but for indefinites in general. Hence, the fact that the distributive reading is impossible in the pseudo-clefts (15a) and (16a), whereas it is possible in the unclefted counterparts, as the answers in (15′) and (16′) show, cannot be considered as an anti-connectivity effect, because it is due to semantic properties which are independent of the specific analysis adopted for specificational sentences.

3.4 Conclusion and consequences

The data presented in this section lead to two main descriptive conclusions.

On the one hand, I have shown that the anti-connectivity effects with respect to binding (examples (11)–(14) and (27)) disappear when the influence of information structure is neutralized, i.e. when the corresponding constituents in the specificational copular sentence and the parallel canonical sentence have the same information structure status. As far as I can see, this has not been noted as such before.

On the other hand, I have shown that the anti-connectivity effects concerning quantifier scope (examples (15)–(18)) do not disappear when information structure is controlled for, but that they are due to specific semantic constraints, and, hence, do not constitute independent anti-connectivity effects.

The overall descriptive conclusion of this section is thus that the anti-connectivity effects mentioned in the linguistic literature disappear when the specific semantics of specificational sentences is taken into account and when the influence of information structure is neutralized. Two more theoretical consequences immediately follow from this.

First, the discussion above has shown that, since binding anti-connectivity effects disappear if information structure is taken into account, they can no longer be used as an argument against a syntax-based approach to specificational sentences and, more generally, binding. Moreover, the discussion about the scope anti-connectivity effects, which survive when information structure is neutralized but are dependent on specific semantic properties, has shown that the syntactic analysis of specificational sentences alone cannot account for the full set of data. Hence, the new data I adduced with respect to the (anti-)connectivity effects mentioned in the literature show that the analysis of specificational sentences must contain both a syntactic and a semantic component.Footnote ¹⁵

A second consequence of the data presented in this paper is that, since binding connectivity is only observed in pairs of sentences where the relevant constituents have the same information structure, information structure must be taken into account in the syntactic analysis of specificational sentences. As far as I can see, the only existing syntactic approach to specificational sentences that is based on their information structure is the one offered by Heycock & Kroch (Reference Heycock, Kroch, Mikkelsen and Potts2002). In the following section, I present their analysis (section 4.1) and provide additional empirical evidence in favor of it (sections 4.2 and 4.3).

4.1 Heycock & Kroch (Reference Heycock, Kroch, Mikkelsen and Potts2002)

Following Steedman (Reference Steedman2000a, b), Heycock & Kroch (Reference Heycock, Kroch, Mikkelsen and Potts2002: 109) argue that ‘the focus/ground (rheme/theme) division of a sentence is not established at some late point in its interpretation but rather in the initial syntactic parse’. Hence, according to the authors, a sentence like (29a) – if used in the answer to a question like What did Mary read? – enters into the syntactic derivation as the pair of constituents in (29b).

1. (29)

   1. (a) Mary read the newspaper.

   2. (b) [_ground Mary read NP], [_focus the newspaper].

      (Heycock & Kroch Reference Heycock, Kroch, Mikkelsen and Potts2002: 109f.)

The authors argue that connectivity effects arise because the initial phrase, rather than being a question (recall section 2.3 above), is the ground of the assertion made by the sentence. The special fact about specificational sentences is that they allow the ground to be constructed from the initial phrase, rather than from the entire sentence. Given this possibility, the connectivity effects then arise from the normal mechanisms for putting ground and focus together. More particularly, Heycock & Kroch argue that LF explains connectivity, and that the LF representation (30c) accounts for the sentences in (30a) and (30b), which exhibit connectivity effects:

1. (30)

   1. (a) What he claimed was [_focus that John had fallen down].

   2. (b) He claimed [_focus that John had fallen down].

      (Heycock & Kroch Reference Heycock, Kroch, Mikkelsen and Potts2002: 111)

   3. (c) [_FocP [_CP that John had fallen down]_i [_Foc° [_ground he / his claim / what he claimed t_i ]]]Footnote ¹⁶

This is further demonstrated on the basis of evidence showing that the connectivity effects in canonical pseudo-clefts (31a) are different than in what are called reverse pseudo-clefts, which are illustrated in (31b):

1. (31)

   1. (a) Canonical pseudo-cleft

      *What he_i has always claimed is that Cain_i is innocent.

   2. (b) Reverse pseudo-cleft

      That Cain_i is innocent is what he_i has always claimed.

      (Heycock & Kroch Reference Heycock, Kroch, Mikkelsen and Potts2002: 104)

   3. (c) Canonical sentence

      *He_i always claimed that Cain_i is innocent.

The fact that reverse pseudo-clefts, in contrast with canonical pseudo-clefts, do not exhibit connectivity effects is straightforward, since the information structure of the sentences has also been altered: that Cain is innocent is part of the focus in the canonical pseudo-cleft (31a) and the corresponding simple sentence (31c), but not in the reverse pseudo-cleft (31b), where it is typically interpreted as part of the ground.Footnote ¹⁷

The basic tenet of Heycock & Kroch's (Reference Heycock, Kroch, Mikkelsen and Potts2002) analysis, which I adopt here, is thus that connectivity effects arise in sentences which are marked for the ground-focus partition. This immediately predicts that not only specificational copular sentences should display connectivity effects, but also other (non-copular) constructions with an explicit ground-focus partition. One example of such a construction is the verb–object–subject (henceforth: VOS) word order in French, illustrated in (32), and Spanish, illustrated in (33):

1. (32)

   • Q:Qui a mangé des/les gâteaux?who has eaten some/the cakes ‘Who ate some/the cakes?’

   • A:Ont mangé des/les gâteaux: Marie, Pierre et Stéphanie.have eaten some/the cakes Marie Pierre and Stéphanie ‘Marie, Pierre and Stephanie are those who ate the cakes.’

     (Kesik Reference Kesik1985: 60)

1. (33)

   • Q:¿Quién ganó la lotería ayer?who won the lottery yesterday ‘Who won the lottery yesterday?’

   • A:Ayer ganó la lotería Juan.yesterday won the lottery Juan ‘Yesterday, it was Juan who won the lottery.’

     (Ordóñez Reference Ordóñez2000: 29)

On the basis of attested examples retrieved from several corpora, Lahousse (Reference Lahousse2006, Reference Lahousse2007) shows that in French and Spanish VOS, the verb phrase is necessarily interpreted as the ground of the sentence, whereas the postverbal subject necessarily has a focal interpretation.Footnote ¹⁸ Hence, VOS in French and Spanish has exactly the same information structure partition as specificational copular sentences. According to Heycock & Kroch's (Reference Heycock, Kroch, Mikkelsen and Potts2002) proposal presented above, VOS structures would then enter into the syntactic derivation as in (34a), and have the LF representation (34b):

1. (34)

   1. (a) [_ground NP verb object], [_focus subject]

   2. (b) [_FocP [_NP subject]_i [_Foc° [_ground t_i verb object ]]]

This immediately predicts that VOS word order in French and Spanish will also exhibit connectivity effects. In other words, given my claim that connectivity effects in specificational sentences are dependent on information structure, and Heycock & Kroch's (Reference Heycock, Kroch, Mikkelsen and Potts2002) analysis of connectivity effects in copular specificational sentences as a consequence of their explicit ground-focus structure, it is predicted that VOS should also exhibit connectivity effects. In the following sections I show that this is indeed the case, both in French (section 4.2) and Spanish (section 4.3).

4.2 Binding connectivity in French VOS

In the following example, the possessives in son auteur et son éditeur ‘its author and its editor’ are bound by the quantified NP chaque livre ‘every book’ in the VOS example in (35a), the pseudo-cleft in (35b) and the corresponding canonical sentence with a focused subject (35c).

1. (35) [Context: At a book exhibition]

   1. (a) Signeront chaque livre_i [_F son_i auteur et son_i éditeur].will-sign each book its author and its editor ‘Those who will autograph each book are its author and its editor.’

   2. (b) Ceux qui signeront chaque livre_i, ce sont [_F son_i auteur et son_i éditeur].those who will-sign each book they are its author and its editor ‘Those who will autograph each book are its author and its editor.’

   3. (c)

      • Q: ‘Who will sign every book?’

      • A:[_F Son_i auteur et son_i éditeur] signeront chaque livre_i. its author and its editor will-sign each book ‘Its author and its editor will autograph each book.’

In (36) too, the possessive is bound by the quantifier in the VOS example (36a) and the pseudo-cleft (36b), as well as in the corresponding canonical sentence with a focused subject (36c):

1. (36)

   1. (a) Devront accompagner leur_i enfant [_F tous les parents_i qui se sont inscrits la semaine passée].will-have-to accompany their child all the parents who themselves have inscribed the week last ‘Those who will have to accompany their child are all the parents who registered last week.’

   2. (b) Ceux qui devront accompagner leur_i enfant, ce sont [_F tous les parents_i qui se sont inscrits la semaine passée].those who will-have-to accompany their child they are all the parents who themselves have inscribed the week last ‘Those who will have to accompany their child are all the parents who registered last week.’

   3. (c)

      • Q:Which parents will have to accompany their child?

      • A:[_F Tous les parents_i qui se sont inscrits la semaine passée] devront accompagner leur_i enfant. all the parents who themselves have inscribed the week last will-have-to accompany their child ‘All the parents who registered last week will have to accompany their child.’

Examples (35)–(36) show that, with respect to binding, VOS in French behaves exactly like the corresponding pseudo-cleft and the corresponding canonical sentence with a focused subject. Hence, these are clear cases of connectivity.

4.3 Binding connectivity effects in Spanish VOS

VOS in Spanish also exhibits connectivity effects, as is illustrated in the following examples involving, respectively, anaphora resolution and co-reference between reflexives and NPs:

1. (37)

   1. (a) Aquí besó a su_i amiga [_F cada niña_i].here kissed Prep her friend each girl

      (Ordóñez Reference Ordóñez2000: 39)
   2. (b)

      • Q:Who kissed her friend here?

      • A:Aquí [_F cada niña_i] besó a su_i amiga.here each girl kissed prep her friend ‘Here, it was each girl who kissed her friend.’

1. (38)

   1. (a) Hablaron de sí mismas_i [_F tus hermanas_i]talked about themselves your sisters ‘It was your sisters who talked about themselves.’

      (Ordóñez Reference Ordóñez2000: 40)
   2. (b)

      • Q: Who talked about themselves?

      • A:[_F Tus_i hermanas] hablaron de sí mismas_i. your sisters talked about themselves ‘It was your sisters who talked about themselves.’

The following examples show that the possessive (su) in the subject NP refers to the quantified object in both the VOS (39a) and the SVO (39b) patterns:

1. (39)

   1. (a) El primer día de escuela deberá acompañar a cada niño_i [_F su_i MADRE] (y no su padre).the first day of school will-have-to accompany prep each child its mother and not its father ‘The first day of school, it is the mother (and not the father) that will have to accompany each child.’

   2. (b) El primer día de escuela, [_F su_i MADRE] deberá acompañar a cada niño_i.the first day of school its mother will-have-to accompany prep each child ‘The first day of school, it is the mother (and not the father) that will have to accompany each child.’

      (Zubizarreta Reference Zubizarreta1998: 143f.)

Consider also (40a) from the ARTHUS Corpus, where the distributive anaphoric determiner sendos ‘each’ appears between the verb and the post-verbal subject:

1. (40)

   1. (a) Por la noche patrullan el campus en sendos_i vehículos [_F dos_i parejas]: un guarda de seguridad conductor y un vigilante.at the night patrol the campus in as-many vehicles two pairs a guard of security driving and a policeman ‘At night two pairs patrol the campus in as many vehicles: a driving security guard and a policeman.’

   2. (b)

      • Q: Who patrols the campus at night?

      • A:Por la noche [_F dos_i parejas] patrullan el campus en sendos_i vehículos.at the night two pairs patrol the campus in as-many vehicles ‘At night two pairs patrol the campus in as many vehicles.’

According to Bosque (Reference Bosque, Cartagena and Schmitt1992) and Aranovich (Reference Aranovich, Beyssade, Bonami, Hofherr and Corblin2003), sendos ‘each’ behaves like reflexives in that it must be preceded by a co-referring expression in the same sentence, as the grammaticality contrast between (41) and (42) shows:

1. (41) Dos_i niños montaban en sendas_i bicicletas.two children rode on as-many bicycles ‘Each child was riding on a bicycle.’

   (Aranovich Reference Aranovich, Beyssade, Bonami, Hofherr and Corblin2003)

1. (42) *[Sendos niños]_i montaban en [dos bicicletas]_i.as-many children rode on two bicycles ‘Each of two children was riding on a bicycle.’

   (Aranovich Reference Aranovich, Beyssade, Bonami, Hofherr and Corblin2003)

Hence the acceptability of both the VOS example (40a) and the corresponding SVO example (40b) with a focused subject is a case of binding connectivity.

4.4 Conclusion

In this section, I have presented Heycock & Kroch's (Reference Heycock, Kroch, Mikkelsen and Potts2002) approach to specificational sentences, according to which specificational sentences are encoded as ground-focus structures and connectivity effects arise because the ground and the focus are put together for semantic interpretation. I have argued that this immediately predicts that any kind of syntactic configuration with a ground–focus information structure should exhibit connectivity effects, and I have shown that this is indeed the case in French and Spanish VOS word order.