1 Introduction
In this article I focus on the prosodic, syntactic and semantic properties of a particular type of predicative clause attested in Romance (and other languages) and exemplified in (1a).Footnote [2]
It is a kind of reduced clause where the missing verb is interpreted as a silent copula; the predicate linearly precedes the subject and is separated from it by a very neat change in the intonational curve, rendered graphically by means of a comma. From an interpretive point of view, the structure in (1a) corresponds roughly to a canonical predicative clause like (1b), although the linear order of the elements entering the predication relation is inverted with respect to the canonical one.Footnote [3]
In structures like (1a) the preposed predicate is primarily, though not exclusively, represented by a specific class of adjectives and nouns which have a strong affective connotation; the subject must be a referential expression and has topic-like properties. Prosodically, the preposed predicate in (1a) bears stress, while the subject is destressed.
The article is organized as follows: Section 2 presents the relevant crosslinguistic properties of verbless predicative structures; Section 3 discusses two possible analyses of verbless predicative clauses and Section 4 presents the technical details of the syntactic derivation I propose for the relevant structure; in Section 5 I discuss the derivation of the structure in which the predicate is fronted along with an overt copula; in Section 6 I argue for a distinction between the verbless predicative structure and an apparently related construction in which the preposed predicate is preceded by a wh-item; finally, Section 7 is devoted to some concluding remarks.
2 Crosslinguistic properties of the verbless predicative structure
In this section I present some relevant properties of the predicate in verbless predicative structures. The crosslinguistic syntactic and interpretive hallmarks of this kind of structure will be described mainly on the basis of data from Italian, Spanish and French. Throughout, I will not discuss the properties of the subject, as they appear to be constant across languages: in speech the subject is prosodically destressed, separated from the predicate by a slight intonation break and characterized by a falling intonational contour; moreover, the subject is generally represented by a referential expression.Footnote [4]
2.1 General remarks
As pointed out by Benincà (Reference Benincà, Renzi, Salvi and Cardinaletti1995), the preposed predicate in Italian can be a bare adjective or a noun modified by an adjective, as in (2), or a noun introduced by the indefinite article, as in (3).
These sentences tend to be temporally bound to the time of the utterance, as in (4), but they can also be used to express a generic present in statements of general validity, as in (5).Footnote [5]
An infinitival clause can appear as subject, as in (5) above, or as predicate, as in (6):
In this last case, the infinitive is preceded by the preposition da ‘from’ and has a deontic interpretation.
Also in French, as pointed out by Vinet (Reference Vinet1991: 81–84), the predicate can be represented by an adjective, as in (7), by a noun preceded by the indefinite article, as in (8), or even by a deontic infinitival clause used with an adjectival function, as in (9):
The structure under analysis has been described for Spanish by Hernanz & Suñer (Reference Hernanz, Suñer, Bosque and Demonte1999), who stress that these sentences have a default deictic temporal interpretation – referring to the speaker and the utterance time – as they are generally uttered as a reaction to a given situation. The preposed predicate can be adjectival or nominal. Adjectival predicates can appear alone, as in (10), or be modified by the emphatic article un ‘a’, as in (11), or, as in (12), by lexical elements favouring an evaluative reading:
When the preposed predicate is a noun, it is generally accompanied by a quantificational element, like the indefinite article un:
2.2 On the evalutative nature of the predicate
As anticipated in the previous section, in this particular structure adjectives and nouns of the evaluative class are often used in Italian:Footnote [6]
Vinet (Reference Vinet1991: 77) observes that also in French the predicate – adjectival or nominal – must express a personal evaluation of the speaker:Footnote [7]
Vinet (Reference Vinet1991: 82–83) also points out that nouns indicating a function or a profession are generally unacceptable in these constructions as they do not express a personal judgement unless they are accompanied by an adjective or a relative clause expressing a form of appreciation:
An ungrammatical structure can be made acceptable also through the addition of a degree modifier:Footnote [8]
According to Hernanz & Suñer (Reference Hernanz, Suñer, Bosque and Demonte1999), the pattern is similar in Spanish, where ungrammatical examples become acceptable through the addition of an evaluative modifier to a nominal predicate indicating a function or a profession:
On the other hand, predicates not expressing an evaluation – in particular relational adjectives and adjectives denoting objective properties – are excluded from this construction.Footnote [9]
2.3 The individual-level vs. stage-level predicate distinction
Vinet (Reference Vinet1991: 85) emphasizes that in French only predicates expressing permanent or inherent properties can occur in a verbless predicative clause, and not predicates referring to a transitory situation; the two types reflect the distinction between individual-level and stage-level predicates, first proposed by Kratzer (Reference Kratzer, Carlson and Pelletier1995). The predicate can acquire a character of permanence – thereby making the sentence acceptable – through the insertion of a lexical element which – functioning as an aspectual marker – triggers a generic interpretation:
Similarly, Hernanz & Suñer (Reference Hernanz, Suñer, Bosque and Demonte1999) point out for Spanish that the predicate must express properties not undergoing change or evolution, so that no stage-level predicates are admitted; still, the insertion of elements which can modify or limit the transitory character of the predicate can make the sequence grammatical:
The same restriction seems to hold for Italian, where non-evaluative predicates can appear provided that an adequate lexical item is added which turns the stage-level predicate into a permanent (i.e. individual-level) property:
Evaluative adjectives are generally interpreted as expressing inherent properties of the subject, in particular if there are lexical elements favouring that interpretation, as in (30b):
2.4 A restriction on degree
As expected in an evaluative context, the adjectival predicate can appear in the (absolute) superlative form:
Rather surprisingly though, if the adjective is in the relative superlative or in the comparative form, there is a considerable degradation in the acceptability:
According to Hernanz & Suñer (Reference Hernanz, Suñer, Bosque and Demonte1999), a consequence of the fact that verbless predicative clauses are interpreted on the basis of contextual factors is that the degree quantification of the predicate must be grammatically fixed (otherwise it would not be appropriately limited). The superlative is therefore grammatical because in this case it is possible to fix the right degree of the relevant property. This conclusion is supported by the fact that adjectives which are normally not admitted in this structure can appear in the absolute superlative form or be preceded by adverbs functioning as degree quantifiers which contribute an affective component:Footnote [10]
The ungrammaticality produced by the comparative is explained by the hypothesis that it does not permit to fix the exact degree of the property expressed by the predicate:
As for French, as we have seen above, the adjectives most readily used in this structure are the evaluative ones:Footnote [11]
However, the restriction holds also for French comparatives and relative superlatives:
The two interpretive restrictions on verbless predicative structures discussed here can be easily accounted for within the analysis that I put forth in the next section. In particular, I would like to propose that the ban on comparatives and relative superlatives can be reduced to a more general ban on stage-level predicates, in that the property expressed by predicates in the comparative and superlative form is always evaluated with respect to a reference set which is temporally limited, hence has an intrinsically transitory (i.e. stage-level) character.
3 Two possible derivations of verbless predicative clauses
In this section I address the question of the derivation and the internal structure of verbless predicative constructions in an attempt to account for the crosslingustic properties noted in the previous sections.
As indicated in connection with example (1) above, repeated here as (40), the interpretation of (40a) is roughly equivalent to that of (40b), where the predication relation between the subject DP and the adjective is expressed by an overtly realized copula, which is absent in (40a).
From the theoretical point of view, the grammaticality of the structure in (40a) raises two major questions: How is the linear order of predicate and subject derived? and How is the empty copula licensed? Following previous proposals on this topic, and consistently with Kayne (Reference Kayne1994), I will argue in what follows that the surface linear order is derived from the basic order – in which the subject precedes the predicate – by raising the predicate across the subject. As for the non-overt realization of the copular verb, I will be led to conclude that the copula spells out the agreement features of the AgrS° head only when the subject raises overtly to the corresponding specifier position, which is arguably not the case in the structure under discussion. However, first, in the present section, I focus on two previous analyses which deal with the above questions.
3.1 Against a bare small clause approach
A first possible approach to the internal structure of verbless predicative clauses is based on the traditional analysis according to which predication relations can be structurally represented as bare small clauses as in (41) (see the seminal work of Williams Reference Williams and Kimball1975 and Stowell Reference Stowell1981).
This kind of approach has been pursued by Moro (Reference Moro1997) in his analysis of copular constructions, where he proposes that the structure in (41) can be embedded under a copular predicate and that the surface linear order of canonical and inversed copular constructions is derived by raising either the subject NP or the nominal predicate respectively to a small clause external position:Footnote [12]
Moro (Reference Moro2000) develops further the analysis of copular constructions in terms of dynamic antisymmetry. Building on Kayne’s (Reference Kayne1994) antisymmetry approach to syntactic structure, Moro claims that the trigger for the raising of the subject or the predicate lies in the necessity to break the structural symmetry between the two constituents inside the small clause and to produce an asymmetric structure exactly by raising either noun phrase to the precopular position.
Assuming that subject and predicate of a verbless predicative structure can be base-generated inside a small clause not selected by a copular verb, as in (41), such a proposal might at first sight provide us with an immediate answer for the obligatory displacement of the predicate around the subject, as (43b) (= (40a) above) could be derived from (43a) by preposing the AP to a small clause external position in order to restore the asymmetrical character of the structural configuration. However, under this analysis, we would expect to move also the subject to a small clause external position, deriving the linear sequence in (43c), which is ungrammatical without an overt copula.
Moreover, this approach has nothing to say about the individual-level vs. stage-level predicate distinction discussed in Section 2.3 above. Finally, it would leave completely unexplained the possibility of leaving the copula phonetically unrealized in this kind of structure.
On the basis of these shortcomings, I therefore reject the analysis of verbless predicative structures in terms of bare small clauses and turn to a different analysis.
Instead, as a working hypothesis, I adopt an approach to predicative structures defended by various scholars over the last two decades, according to which every predication relation is mediated by an intervening functional element working as a connective between the subject and the predicate (see e.g. Bowers Reference Bowers1993, Adger & Ramchand Reference Adger and Ramchand2003, Baker Reference Baker2003, den Dikken Reference den Dikken2006). Within this approach, den Dikken (Reference den Dikken2006: 11) proposes that the basic template for predication relationships is syntactically represented by the following structural configuration, where R° is the small clause internal relator, which heads the maximal projection RP hosting the subject XP and the predicate YP in the specifier and in the complement position respectively:
Starting from the nuclear predicative structure in (44), a copular sentence like (45a) can be derived as seen in (45b), that is, by raising the subject XP to Spec,TP to satisfy the EPP feature:
In the minimalist framework, this is the most economical solution towards a converging derivation, as the subject XP is closer to T than the predicate YP.
Predicate inversion, that is, raising of the predicate YP to Spec,TP across the subject, as provisionally sketched in (46b), is forced, according to den Dikken (Reference den Dikken2006: 95), by the need to license a null pro-predicate located inside YP, a need which overrides least effort considerations:
A more detailed analysis of the derivation of predicate inversion in Romance will be provided in Section 4 below.
3.2 An instance of predicate inversion?
The structure under analysis has been viewed by Paul (Reference Paul2006) as an instance of predicate inversion, in the sense of den Dikken (Reference den Dikken2006), with the predicate YP in (44) raising out of RP to a higher structural position.Footnote [13]
According to Paul, the predicate moves to Spec,TP in order to license the null copula head; if the copula is overtly realized, then movement is blocked. The syntactic representation of (47a) would be as in (47b):
However, this analysis is confronted with three problems. First, as noted by den Dikken (Reference den Dikken2006: 82–83), predicate inversion is typically associated with the lack of focus on the predicate, whereas in (47a) the predicate is focused and receives the most prominent stress, while the subject is obligatorily destressed. Secondly, predicate inversion is possible even when the copula is overtly realized, as, for example, in (48), where the preposed predicate is an evaluative one:
In fact, predicate inversion leads precisely to the obligatory overt realization of the copula, as witnessed by (49b):
Thirdly, according to Paul, only predicative adjectives are allowed in the structure in (47b) above. This prediction is not borne out, since stage-level adjectives can be used predicatively but are not allowed in the relevant structure:
Gonzàlez-Rivera (Reference Gonzàlez-Rivera2011) discusses in detail the properties of verbless clauses in Spanish, among which the structure analyzed here. Following Paul (Reference Paul2006), he proposes to view it as an instance of predicate inversion, albeit caused by a different trigger (though the exact landing site of the predicate is not explicitly identified). Thus, the derivation in (51a) would be parallel to the one of the English example in (51b), modulo the overt realization of the copula in the second example:
In particular, Gonzàlez-Rivera proposes that the raising of the predicate is due to the necessity of discharging the strong semantic feature evaluativity [+E] in the course of the syntactic derivation; the relator head R° is not empty, as it accommodates the tense feature.
This analysis, despite its prima facie attractiveness, faces a number of shortcomings, which I quickly review in what follows. First, unlike in genuine predicate inversion structures like (51b), in (51a), the subject is not interpreted as a focus, but as a topic, as Gonzàlez-Rivera explicitly recognizes by noting that the topic nature of the subject imposes strong referential constraints on the subject itself, which must be specific. Secondly, it is not at all clear that the landing site of the predicate (presumably Spec,TP if the small clause is headed by a tense feature) is a suitable one for discharging the relevant semantic feature. Thirdly, this analysis does not account for the fact that, unlike (51b), the structure in (51a) cannot be embedded under a matrix predicate like considerar ‘consider’; only a DP marked for accusative case can appear there, as shown by (52b).
Finally, the analysis does not account for the fact that in pro-drop languages a copular verb can be optionally realized at the left of the raised predicate:
Given these shortcomings, I discard the possibility of identifying the structure under analysis as an instance of predicate inversion and put forth an alternative proposal.
4 An alternative analysis
4.1 Movement to Spec,FocP
In this section I suggest an alternative analysis of verbless predicative clauses, showing that it can account for the attested properties without the drawbacks of the analyses presented in Section 3.
I assume, following Benmamoun (Reference Benmamoun, Freidin, Otero and Zubizarreta2008) and Hazout (Reference Hazout2010), that verbless sentences are not reduced clauses, in the sense that they project at least a Tense node (although, according to Benmamoun, Tense does not necessarily cooccur with a verbal head). In fact, I would like to defend here an even more radical assumption, strictly related to the cartographic approach to syntactic structure fostered primarily by Cinque (Reference Cinque1999), namely that the whole functional endowment of the clause is always projected, including the functional skeleton usually associated with the left periphery of the clause.
In particular, I propose that the verbless predicative structure analyzed here results from the fronting of the predicate YP in (44) to a left-peripheral specifier position motivated by the need to check a discourse-related focus feature: I suggest that the final landing site of the predicate is the specifier of FocusP, the left-peripheral functional projection postulated by Rizzi (Reference Rizzi and Haegeman1997) among others.
This hypothesis provides us with a straightforward account of the prosodic pattern instantiated in this kind of construction, that is, for the strong prominence of the predicate (resembling the one associated with contrastively focalized constituents), and for the consequent destressing of the following constituent, the subject, which is therefore associated with an abrupt falling of the intonational curve.Footnote [14]
In Section 4.2 I will spell out the technical details of the syntactic derivation of verbless predicative structures, showing how they can account for the other properties displayed by this construction.
4.2 Phase extension and predicate inversion
On the assumption that small clauses such as the RP in (44) qualify as phases (see den Dikken Reference den Dikken2006), an obvious question is how the predicate YP in (44) can be accessible to an external probe, given the Phase Impenetrability Condition as defined in Chomsky (Reference Chomsky and Kenstowicz2001: 14):
My answer to this question will rely on den Dikken’s (Reference den Dikken2006: 115) treatment of predicate inversion in terms of A-movement of the predicate facilitated by ‘domain extending head-movement’, or phase extension, defined as follows (italics in the original):
On the basis of (55), he assumes that movement of the R° head up to an external functional head F° extends the RP phase up to FP. Merging a functional head outside the small clause RP in (44) provides a landing site both for the R° head of the small clause and for either phrasal constituent of the small clause (in the corresponding specifier position Spec,FP).
Given (55), movement of the predicate YP to Spec,FP will not illegitimately cross a phase boundary, as movement of R° up to F° shifts the phase up to FP, which dominates the first available landing site of the raised predicate.Footnote [15]
The functional head F to which the ‘relator’ R° raises is referred to as the ‘linker’, the head that provides the link between the raised predicate and the small clause it comes from. Raising the ‘relator’ to the ‘linker’ creates a minimal domain that includes Spec,FP and Spec,RP, thereby enabling the predicate to raise to Spec,FP in conformity with the minimalist locality theory.
After A-movement of the predicate YP to Spec,FP, the predicate, being at the edge of the (extended) phase FP, becomes visible to external probes, among which Foc°; it then A’-moves to Spec,FocP to check its discourse related feature:
Under this view, in verbless predicative structures, predicate inversion is only the first step of a more complex movement operation extracting the predicate constituent from the small clause.
Note that, as pointed out by den Dikken (Reference den Dikken2006: 117), raising of the predicate to Spec,FP has the effect of trapping the subject of the small clause inside the newly extended phase, as it ends up being embedded within a phase as a result of predicate inversion, and will not be visible to any outside probes, including T°.
At this point a natural question arises as to the EPP feature, which might be checked by the predicate YP moving through Spec,TP on its way to Spec,FocP. I will suggest that this is indeed the case.
A crucial assumption in this respect is that the projection TP can in fact be decomposed into different functional projections encoding different formal features, as proposed by Cardinaletti (Reference Cardinaletti and Rizzi2004: 116) among others. Cardinaletti identifies at least two preverbal subject positions, the specifiers of two functional projections, SubjP, associated with a ‘subject of predication’ feature (which can be easily identified with the EPP feature), and AgrSP, encoding phi-features and nominative case assignment (and, I will assume, the overt realization of the copula).
If the EPP is checked by the raising predicate on its way to Spec,FocP, we expect the predicate to be able to occupy Spec,SubjP, as shown by (57b), where the postcopular subject must be contrastively focalized, witness the ungrammaticality of (57a) with flat intonation:
Furthermore, Spec,TP can also be occupied by predicates such as relative superlatives or comparatives, which, as noted above (see Section 2.4 above), cannot normally appear in the corresponding verbless structure:
This could be due to the fact that the verbless predicative structure under discussion is typically uttered in ‘out of the blue’ contexts, while comparative and superlative predicates always require a background of reference against which the relevant property is evaluated. Hence, they are intrinsically discourse-bound which makes them incompatible with the Focus feature activated in verbless predicative clauses while making them perfectly compatible with the ‘subject of predication’ feature encoded in Spec,SubjP.
On the basis of this evidence, I admit that the predicate raising from Spec,FP to Spec,FocP has Spec,SubjP as its second, intermediate landing site. As represented in (59) below, I crucially assume that the raising predicate skips the specifier of the AgrSP projection, which is the canonical position of the overt realization of the copula, hence the absence of the copula in this particular structure:
I propose that Spec,AgrSP is inaccessible to the raising predicate because the predicate (which in most cases is adjectival) does not qualify as a suitable checker for the relevant formal features associated with AgrS° (that is, phi-features and nominative case).Footnote [16]
As an alternative, after phase extension has taken place, we may choose to move the subject XP to Spec,FP (leaving the predicate YP stuck inside the small clause), where it becomes accessible to the EPP feature of T°, and raises subsequently to Spec,TP (or, more precisely, to Spec,AgrSP and Spec,SubjP) for that reason:
Raising of the subject to Spec,SubjP through Spec,AgrSP will have as immediate consequence the overt realization of the copula as the spell out of the agreement features codified in AgrS°.
4.3 On the individual-level vs. stage-level predicate distinction
We have seen in Section 2.2 above that the prototypical predicates appearing in verbless predicative structures are the evaluative ones:
However, it has also been observed that even stage-level predicates can appear in the relevant structure, as long as there are lexical elements turning them into individual-level predicates:
Given that evaluative predicates are a subset of the individual-level predicates, we can generalize by saying that only individual-level predicates can appear in the relevant structure. How can we account for this descriptive generalization?
I think that a suitable answer to this question can be provided by the proposal put forth by Park (Reference Park1997). In order to account for the restrictive distribution of individual-level (IL) predicates within small clauses compared with the relatively free distribution of stage-level (SL) predicates, he suggests that IL predicates and SL predicates are grammatically distinguished by two different types of the selecting head Pred, which functions as a mediator between a predicate and its associated subject. Park argues that the Pred which selects an IL predicate raises to a higher Tense position to undergo temporal interpretation, whereas the Pred selecting an SL predicate does not have to raise, hence it does not raise.Footnote [17]
Capitalizing on this analysis, I suggest that the Pred discussed by Park (Reference Park1997) should be assimilated to the head R° in den Dikken (Reference den Dikken2006), as they both function as mediators between the predicate and its associated subject. This amounts to saying that whenever in (44) R° selects an IL predicate, R° is bound to raise to a higher functional head (probably Tense°), thereby extending the phase RP up to (at least) FP and providing a possible landing site for the predicate YP, which becomes then eligible by an external probe, as discussed in the previous section. Alternatively, if the subject raises to Spec,FP, we get the standard copular structure, as represented in (60) above.
If, on the other hand, R° selects a stage-level predicate, no raising of R° will obtain, hence no phase extension will take place, and no possible landing site for the predicate will be provided; the only possible derivation will be the one represented in (45b) above, with raising of the subject.Footnote [18]
5 On the overt realization of the copula
We still have to account for the possibility of realizing overtly the copula before the preposed predicate. This possibility is attested with both IL and SL predicates, as exemplified in (63a) and (63b), respectively:
My proposal is that in this case what raises to Spec,FocP is a larger chunk of structure, including both the copula and the predicate, and excluding the subject.
Adopting Cardinaletti’s (Reference Cardinaletti and Rizzi2004) view presented in Section 4.2 above, a canonical copular sentence like (45a) should be assigned the following structure:
Here the subject XP raises to Spec,AgrSP in order to check nominative case and phi-features, triggering agreement with the copula, and raises subsequently to Spec,SubjP to check the ‘subject of predication’ feature.
Let us consider first the case of SL predicates, where no raising of R° obtains; the subject XP, situated at the left edge of RP, raises to Spec,AgrSP and subsequently to Spec,SubjP to check the relevant features, as represented in (65a). Suppose, however, that the predicate is associated with a discourse-related focus feature, to be discharged in the appropriate Spec–head agreement configuration with the left-peripheral head Foc°: after raising of the small clause subject XP, the only way for the predicate YP to enter the relevant structural configuration will be to pied-pipe the AgrSP projection containing the copula to Spec,FocP, as in (65b), which is the structural representation of (63b):
In the case of IL predicates, we have R° raising to F°, with phase extension: if the subject XP raises to (Spec,FP, Spec,AgrSP and) Spec,SubjP we get (66a); if the predicate has a focus feature to check, being trapped inside the small clause as a consequence of subject raising, it must pied-pipe AgrSP to Spec,FocP, as in (66b), the structural representation of (63a):Footnote [19]
Differently from den Dikken (Reference den Dikken2006), I assume that the copula spells out the AgrS° head, and not the lower F° head, that is, overt realization of the copula, at least in Romance, should not be seen as an automatic consequence of raising a small clause constituent to Spec,FP, otherwise we would expect it to be overtly realized even when predicate inversion obtains (and the predicate YP raises alone to Spec,FocP), contrary to fact:Footnote [20]
Note furthermore that we must also exclude the possibility for the predicate YP to pied-pipe only RP (with SL predicates) or only FP (with IL predicates), stranding the copula in AgrS°:
In fact, the hypothesis that the copula occurs in AgrS° as the spell out of the agreement features with the clausal subject straightforwardly accounts for the ungrammaticality of (68): the extraction of the predicate (through Spec,FP) traps the subject inside the small clause preventing its raising to Spec,AgrSP and the overt realization of the copula; conversely, if the subject raises first (to Spec,FP and) to Spec,AgrSP, triggering the overt realization of the copula, this results in the impossibility for the predicate to escape from the small clause, so that (68) is doomed to crash in any case.Footnote [21]
6 A seemingly related structure
In this section I compare the verbless predicative construction discussed so far, which I have analyzed as a structure involving focus fronting of the predicate, with the corresponding exclamative structure in which the preposed predicate is preceded by a wh-item.
The relevant minimal pair is exemplified in (69) by Italian sentences; if the preposed predicate is an adjective, the overt realization of the introductory wh-element che ‘how’ is optional:
On the other hand, if the preposed predicate is a bare noun, the omission of che requires either the insertion of adjectives like gran (de) ‘big’ and bel (lo) ‘beautiful’ or the presence of the indefinite article:Footnote [22]
As for Spanish, Hernanz (Reference Hernanz2001: 94) correctly underlines that the structure under analysis reveals the existence of an – a priori unexpected – correlation between the semantic properties of evaluative adjectives and the syntactic behaviour of wh-elements, as both these types of elements can introduce a verbless predicative structure:
According to the author, the well-formedness of (72) is due to semantic factors (Hernanz Reference Hernanz2001: 101–102); more precisely, it is the evaluative character of the predicate that licenses its presence in initial position in these constructions, as witnessed more clearly by (73).
Hernanz (Reference Hernanz2001: 102–103) assumes that evaluative terms encode semantically what in other cases is expressed through explicit morphological markings; their syntactic behaviour in verbless predicative structures is therefore motivated by the same factors which underlie the preposing of constituents endowed with an exclamative feature codified by the wh-element qué ‘how’:Footnote [23]
As for French, unlike Italian and Spanish, it does not allow the preposed adjective to be preceded by any wh-modifier, as noted by Vinet (Reference Vinet1991: 77–78):Footnote [24]
Despite the apparent similarity between the two constructions, there are some sound empirical reasons to believe that when the wh-item is realized we have to do with a different kind of structure. Looking closer at the Italian minimal pair in (69) above, it is easy to recognize that the difference between (69a) and (69b) does not simply lie in the overt phonetic realization of the wh-item; on the contrary, there are prosodic, syntactic, semantic and pragmatic differences distinguishing the two structures.
A first crosslinguistic argument in favour of the hypothesis that the similarity between (69a) and (69b) is only superficial is the fact that French admits the structure with bare predicate preposing while disallowing the corresponding structure with overt realization of the wh-item, which would be completely unexpected if the two constructions were indeed minimal variants of each other.
A second argument is based on the prosodic properties of the two constructions, which are clearly different: the preposed bare predicate is associated with a rising prosodic curve followed by an abrupt falling on the subject, while the presence of a wh-item imposes a more straight intonation with a regular degradation towards the end of the clause.
A third argument comes from the fact that the two examples in (69) can be paraphrased – modulo the overt realization of the copula – as in (76a) and (76b), respectively:
Interestingly, in the two cases, the copular verb surfaces on different sides of the preposed predicate; that is, the copula can be preposed along with the predicate only when the predicate itself is not modified by a wh-item. This syntactic difference has interesting interpretive consequences, which I will turn to next.
A fourth argument for drawing a clear distinction between (69a) and (69b) is the fact that the structure with the wh-item is not limited to individual-level predicates, while, as we have seen above, stage-level predicates are systematically banned from canonical verbless predicative structures:
A final argument for a distinct analysis of (69a) and (69b) is the fact that the two sentences are by no means synonymous. Here a short digression on the interpretation of exclamative clauses is in order. Obenauer (Reference Obenauer1994) identifies a basic semantic property of exclamative wh-phrases: differently from what happens in wh-interrogatives, in wh-exclamatives the value of the variable is determined, and situated outside the pertinent domain. This view has been further developed by Zanuttini & Portner (Reference Zanuttini and Portner2003), who identify, among the defining properties of exclamative clauses, the scalar implicature, i.e. the fact that exclamatives indicate that some entity has a property which falls at the extreme end of some contextually given scale of properties. This extreme quality may give rise to the feeling that the fact expressed is surprising or noteworthy in some way. Given this notion, in (69b) the scalar implicature effect is much stronger than in (69a); in this sense, (69b) is closer to (78) than (69a):Footnote [25]
In (78), the extreme nature of the property expressed by the predicate is conveyed by the prototypical morphological template of high degree, the superlative, thereby triggering the scalar implicature effect induced in (69b) by the wh-item. This effect is minimally present in (69a), where it is due to the mere presence of an evaluative predicate, which is compatible with the scalar component of the semantics of exclamatives.
Indeed, what characterizes (69b) – as opposed to (69a) – is the fact that the speaker shares with the addressee a previous common evalutation of the relevant property, that is, they share a generic appreciation and agree on the fact that the book is interesting to some extent; uttering (69b) the speaker intends to express an emphatic confirmation of this judgement. In contrast, in (69a), by using the bare adjective, the speaker introduces a property of the subject, involving the addressee in a first evalutation, that the addressee may not share.
That the pragmatic function of the structure with the bare adjective is to submit the relation between subject and predicate to the attention of the addressee, encouraging his approval, is confirmed by the fact that only in this case a question tag like vero ‘true’ can be added:Footnote [26]
Summarizing, in a verbless predicative structure in which the predicate is not modified by a wh-item, the speaker is simply interested in stating a predication relation between the entity functioning as subject and an evaluative predicate. This entails that unlike in canonical exclamative clauses, in verbless predicative structures there is no real commitment on the part of the speaker as to the degree to which the relevant property is present, and that such property is situated at a point just above the standard one in a contextually given scale of default expectations.
7 Summary
In this paper I have provided a formal account for the prosodic, syntactic and interpretive properties of a predicative structure attested throughout Romance (as well as in other languages), in which the predicate precedes the subject and the predicative link between the two constituents is determined by a phonetically unrealized (but interpretively relevant) copular verb. The analysis developed here crucially relies on the assumption that the functional skeleton of the clause is always fully projected and potentially active, even in what appears to be a structurally reduced syntactic object.
The syntactic derivation that I have proposed for this structure is based on den Dikken’s (Reference den Dikken2006) analysis of predicative constructions, and in particular on the process of predicate inversion, which in turn exploits the notion of phase extension. On the basis of several empirical arguments, I have rejected the hypothesis that verbless predicative structures are the result of the application of predicate inversion, arguing that in this kind of structure predicate inversion is in fact only the first step of a more complex derivational process by which the predicate raises – across the subject – from its basic, small clause internal, position to a left-peripheral functional specifier in order to check a discourse-related feature. Predicate raising from the predicative small clause, an instance of A-movement, is made possible by prior raising of the small clause relator head R° to a higher functional head F°, thereby extending the phase domain and providing a possible landing site for the predicate in the corresponding specifier position. From this intermediate landing site, the predicate is attracted by the external probe Subj° to satisfy the EPP feature of the clause, and subsequently by the left-peripheral head Foc°, and A’-moves finally to the specifier of FocusP in order to check a Focus feature. I have proposed that, crucially, raising of R° to F°, hence phase extension, is only possible in small clauses associated with individual-level predicates, which straightforwardly accounts for the fact that verbless predicative structures are crosslinguistically limited to this kind of predicates. The exclusion of comparatives and relative superlatives from the relevant structure is traced back to the same restriction, on the assumption that in this case the property expressed by the predicate is evaluated with respect to a transitory situation, hence has a stage-level character.
Finally, several arguments have been presented which suggest that the structure under analysis should be clearly distinguished from the minimally different construction in which the preposed predicate is preceded by a wh-modifier, despite their superficial similarity.
