• Keine Ergebnisse gefunden

A seminal work for the comprehension of how Wh-questions are derived is the one in which Rizzi (1996) argues for the existence of a Wh-criterion, responsible for the specific

115

configuration of questions. Rizzi proposes that a [+wh]39 feature marks all questions and constrains the way the sentence must be derived. The proposed Wh-criterion (Rizzi, 1996:64) consists of two mutual constraints:

a. A wh-operator must be in a Spec-head configuration with an X°+wh

b. An X°+wh must be in a Spec-head configuration with a wh-operator.

The presence of a [+wh] feature on the complementizer layer determines the nature of the sentence and, at the same time, it establishes that the sentence is well-formed only if the operator and a head endowed with the same interrogative feature enter into a Spec-head configuration.

Based on the observation that in English the only element which is allowed to appear directly to the left of a Wh-element is the verb (and no other constituent is allowed to intervene between the Wh-element and the verb), Rizzi claims that the verb is endowed with the [+wh] feature and it must enter the required Spec-Head configuration. At this point, once assumed the presence of the Wh-phrase in SpecCP, a second assumption must be done, namely the movement of the inflected verb to the CP. In its first formulation, the Wh-criterion predicts the presence of the Wh-element in SpecCP and of the Verb in C°

in order to satisfy the required Spec-Head configuration. In these terms, the proposal also accounts for the Subject-Auxiliary Inversion (SAI) that characterizes root questions in English. Rizzi’s analysis (1996) starts from the observation of English root Yes/No questions. Sentences entail a configuration that the author defines as a form of residual V2, because it consists in the movement of the auxiliary to the C° position (3):

(3) a. (*you) did you arrive this morning?

b. When (*you) did you arrive?

Both in the Yes/No question and in the adjunct question in (3), the subject can neither precede the auxiliary nor intervene between the Wh-element and the auxiliary. It is therefore assumed that the subject stays in a lower position within the IP, while the

39 Authors alternatively refer to the feature for Wh-questions as [+wh] or [+Q]. I will adopt [+Q] later in the study (see Chapter 6). However, Rizzi’s proposal is here represented in its original version with [+wh].

116

auxiliary moves from I°, the position actually endowed with the [+wh] feature, to C°, the position where the Wh-criterion is to be satisfied.

Building on the fact that Italian shows the same behaviour with respect to subject position in interrogatives, namely the fact that subjects follow the finite verb rather than preceding it, Rizzi (1996) extends his analysis to Romance languages too. However, the proposal has not received unanimous consensus. Rather, it has opened a debate on Romance interrogatives, which covers three different aspect of the derivation of Wh-questions: i) are Romance questions instantiations of CPs or of IPs? ii) what is the landing site of V in interrogatives?, and iii) what is the position of the subject in non-subject interrogatives?

For instance, Barbosa (2001) rejects the idea that Romance questions are CPs and claims that a structure starting with the IP node is sufficient in order to account for the derivation of Romance Wh-questions. Her claim builds on the observation of the positions that subjects can occupy in Romance interrogatives, in comparison to the positions they occupy in Romance main declarative clauses. Barbosa points out that in Italian interrogatives, the subject can intervene neither between the Wh-operator and the verb40, nor between the Auxiliary and the Past Participle. The latter holds true for declarative sentences too. For instance, the subject of Italian interrogatives can follow the past-participle, can precede the interrogative as a hanging topic or can be marginalized to the left of the interrogative (Antinucci & Cinque, 1977).

The parallelism between interrogative and declarative sentences with respect to the position of the subject is confirmed by other Romance varieties: in Spanish, for instance, the subject is tolerated in between the Auxiliary and the Past Participle only with the auxiliary estar and with the imperfect past form of haber (have), namely había. Whenever the auxiliary is in a different tense form (e.g., the present ha or the future habrá), the subject is banned from that position and it is allowed to appear only either before the Auxiliary or after the Past Participle. The reasons for this are not relevant here; what actually matters is the absence of asymmetries between declarative and interrogative clauses with respect to subject positions. Building on this observation, Barbosa (2001) rejects the idea of using subjects as a diagnostic tool for I° to C° movement.

40 In Italian, the subject can intervene between the Wh-operator and the verb only in the case of embedded interrogatives with a verb at the subjunctive. In this case, the [+wh] feature is represented in C° and not in I°, such that verb movement from I° to C° is not implemented (Rizzi, 1996).

117

In her view, it is not possible to assume verb movement to C° since there is no overt sign of the presence of the verb in the CP. Along this line of reasoning, Barbosa does not see any reasons for assuming that the Wh-criterion is satisfied at the CP level; rather, she assumes that the relevant Spec-head configuration between the Wh-phrase and the verb takes place in IP. Her proposal posits the presence of the Wh-phrase in SpecIP in order for this to enter in a relation with the element carrying a [+wh] feature in I°. Given that Rizzi’s (1996) proposal does not set any strict constraint on the exact position where the Spec-Head agreement is to be realized41, Barbosa actually leaves unchanged the core mechanism of the Wh-criterion. Several consequences arise at the syntactic level though.

The main issue concerns SpecIP and the nature of this position: in order to account for the presence of a Wh-phrase in SpecIP, it is necessary to assume that this is a position of the A-bar kind, able to host operators for scope-discourse features. This observation unveils a curious fact: Barbosa (2001) rejects verb movement to C° because of the lack of overt signs of movement, while she assumes Wh-movement to SpecIP, despite clear evidence against it. The main reason why Romance interrogative cannot be assumed to be root IPs is that SpecIP plays a relevant role in subject-verb agreement and therefore it cannot be exploited for other purposes. In particular, SpecIP is the landing position for pro in Italian. An example will help illustrate the point at issue:

(4) Cosa hai portato?

What Aux_2ps bring_PTCP 'What did you bring?'

If we analyse an object question with a null subject of the kind in (4) and claim that the Wh-phrase is in SpecIP, we automatically have to imply that pro occupies a different position, namely a post-verbal one. From this assumption, a clear clash arises, due to the fact that pro has been known for a long time to align with preverbal subjects rather than with post-verbal ones (Cardinaletti, 2004; Guasti, 1996a). A variety of observations supports the claim.

41 Strictly speaking, Rizzi (1996) does not explicitly claim that the Wh-criterion must be satisfied in the left-periphery; however, syntactic evidence points to this conclusion.

118

Guasti (1996a) in particular remarks the relevance of the following observations:

first, floating quantifiers can appear in between the auxiliary and the past participle only in the case of preverbal subjects and of pro, but not with post-verbal subjects (Rizzi, 2000). Second, the subject of a main clause can control the subject PRO of a temporal adjunct only if it is null or if it overtly appears in the preverbal position; in contrast, post-verbal subjects cannot control PRO in temporal adjuncts.

The detailed distribution of Italian subjects is certainly beyond the purpose of the present structure. However, for the sake of the present discussion it will be sufficient to follow Guasti (1996a) and assume that a null subject occupies the upper-portion of the IP and that this position cannot correspond to the landing site of Wh-operators. Through a careful analysis of a variety of phenomena, Guasti concludes that wh-operators must appear within the CP layer in order to satisfy the relevant [+wh] feature and they are not allowed to appear in IP. With respect to the position of the Wh-operator, her analysis is therefore identical to the one originally formulated by Rizzi (1996) for English interrogatives. However, the author also admits that no overt sign of verb movement to the CP layer is to be detected in Italian, an observation that very much resembles the main argument brought up by Barbosa (2001).

A relevant contribution to the issue of V movement to C in Romance languages comes from the analysis conducted by Poletto (2000) on Italian varieties. If standard Italian fails to provide the desired information, some indirect proof can be derived from the analysis of varieties of Northern Italian dialects. In particular, the author presents two clear cases in which the verb has moved to the CP layer of interrogative clauses.

The first piece of evidence comes from Fassano dialect, a Rhetoromance variety, characterized by Subject clitic inversion in root questions. The variety also disposes of a focus particle pa, which signals the interrogative character of the clause and presumably marks the lower boundary of the CP layer. The interesting piece of information for us is that the verb and the clitic subject can appear (inverted) in a position higher than pa. The example in (5) clearly signals that the verb has moved across the focus marker and has landed in the CP layer (Poletto, 2000:46):

119 (5) Co l fas-to pa?

How it do-you interr. Marker 'How do you do it?'

Second, Poletto (2000) finds instances of do-support phenomena in Eastern Lombardian varieties. In the dialect of Monno (Eastern Lombard), interrogatives imply the insertion of an auxiliary in the higher functional portion of the sentence, while the lexical verb is lower in the structure (Poletto, 2000: 49). As can be observed in (6), the choice of the auxiliary fare (“do”) and its position in the sentence clearly resemble the do-support phenomenon typical of English, a language for which V to C movement is widely assumed (Rizzi, 1996).

(6) Come fa-l comportas?

How does-he behave-himself 'How is he behaving? '

The two pieces of evidence for V to C movement in interrogatives in Northern Italian Dialects are only an indirect proof for the variety at stake here, which does not allow for certain conclusion with respect to standard Italian. Nonetheless, along with the observations concerning the impossibility of inserting wh-operators as low as the SpecIP position, their presence clearly points towards an analysis that assumes V to C movement.

Summing up, with respect to the first two questions concerning the analysis of Romance questions (see above), the present work assumes that standard Italian root interrogatives are CPs nodes, in which the higher functional field (the CP) hosts both the wh-operator and the verb. In particular, in the case of root argument questions the Wh-element will target the Focus position in CP42 (Rizzi, 2001).

As for the position of the subject in non-subject interrogatives, it will be assumed that this, when lexically realized, will stay in SpecvP and check its features against the verb from there (Belletti and Guasti, 2015); while a pro is generated in correspondence of SpecIP.

42 A complete taxonomy of the different positions that Italian Wh-operators can target in CP depending on the kind of interrogative to be realized (root/embedded, Yes/No questions and argument questions; Why questions) can be found in Rizzi (2001).

120 5.3.1 Bare Wh-elements vs WhichNP

In order to complete the overview on the syntactic characteristics of Wh-questions, it is necessary to consider one further factor, namely the kinds of Wh-element in use. For instance, Italian has two kinds of Wh-elements for argument questions: either bare Chi ('who') and Cosa ('what'), or Quale NP ('which NP'), namely a wh-element endowed with a lexical restriction. The two classes of elements are not equivalent though, as they realize different functions.

Pesetsky (1987) first observed that the English counterparts of the two kinds of elements behave differently. First of all, the author pointed out that in case of English multiple wh-questions, WhichNP phrases do not obey the Superiority Condition, while Who and What strictly do:

(8) Superiority Condition

In a multiple interrogation, where a wh-phrase is in Comp and another is in situ, the S-structure trace of the phrase in Comp must c-command the S-structure position in the Wh-in-situ. (Pesetsky, 1987: 104).

The claim derives from the following examples, in which movement of the interrogative subject phrase to CP is always allowed (in 9a and 10a). By contrast, movement of the interrogative object phrase across an interrogative subject phrase left in-situ is allowed only in case the wh-element in use is of the WhichNP type (10b), otherwise the structure is very marginal (9b):

(9) a. Who did you persuade to read what?

b. ?? What did you persuade who(m) to read?

(10) a. Which man did you persuade to read which book?

b. Which book did you persuade which man to read?

Pesetsky (1987) ascribes the asymmetry to specific properties of the two elements in use.

He suggests to distinguish the elements on the base of their relationship with the discourse set. In particular, WhichNP phrases usually refer to a set of elements already present in

121

the discourse; they are D(iscourse)-linked because they anchor the interrogative to the information shared by participants in the discourse and bias the answer, by suggesting that this should be retrieved among the elements already mentioned. On the other hand, bare Wh-elements of the Who/What type do not operate the same kind of presupposition and leave the answer open.

Rizzi (1990) also points out the relevance of referentiality in the formation of Wh-questions, although his observations substantially diverge from the one presented in Pesetsky (1987). Rizzi specifies that the notion of referentiality should be anchored to thematic roles. He distinguishes between theta-roles, which identify referents taking part in the event, and elements that further characterize the event (e.g. manner, measure). The former are proper arguments, while the latter only qualify as quasi-arguments. The distinction generates the dichotomy between arguments and adjuncts. In Rizzi’s view, arguments always receive a referential index through theta-role assignment, which allows for long-distance movement of the phrase. Quasi-argument do not receive theta-roles able to license a referential index. It follows that they obey different movement constraints.

Only arguments that receive a referential index are properly bound by the predicate, while quasi-arguments are not and must undergo a more local kind of movement.

Cinque (1990) builds both on Pesetsky (1987) and on Rizzi (1990), and suggests a sharp distinction between WhichNP and Who interrogatives in syntactic terms. The author limits the use of referential indexes to elements present in the discourse; it follows that only WhichNP phrases receive an index, because they refer to members of a set, known both to the speaker and to the hearer. In this sense, quantifiers of the Who/What kind do not bear an index. This distinction in indexes has crucial consequences on the kind of syntactic chains that the elements can enter. WhichNP phrases enter binding relations, while Who/What quantifiers enter in chains of the antecedent-government kind.

The syntactic formalization is rather outdated in its terms, but a number of experimental studies has nonetheless resort to it in order to account for speakers’ asymmetric performance on questions of the WhichNP and of the Who type. We will therefore go back to this issue in the following sections; for the moment, it should only be considered that the two kinds of Wh-elements can generate different presuppositions, which have to be integrated in the processing of interrogatives. In case of a lexical restriction, the number of represented features on the phrase is higher and the set of alternative answers

122

is limited to the elements known to the participants in the conversation. A lower number of features is represented on Who/What quantifiers and these do not set requirements on the expected answers (except for the distinction between human and non-human referents).

The distinction between bare and lexically-restricted Wh-elements represents one of the main factors in the empirical studies on Wh-questions; the topic will therefore be further discussed later in the present chapter.