Abstract

In this article, I discuss the interaction of locality phenomena with the left periphery in Italian as elaborated in Rizzi 1997, 2001a, 2004b. It turns out that long-distance crossing possibilities fully predict the local orderings entailed by Rizzi’s left-peripheral template. In fact, both descriptive gains (in terms of topic positions) and explanatory gains (regarding the position and behavior of topics and of Rizzi’s (2001a) Int) can be made if local ordering is reduced to locality. This suggests that the left-peripheral template should be derived from some appropriate theory of locality and should not be taken as a theoretical primitive.

Syntactic constraints have the effect of filtering out unwanted structures or derivations. When multiple constraints apply to a given structure or derivation, it must satisfy all of them to count as grammatical. Accurately describing the properties of these constraints constitutes one of the goals of linguistic theory. This article starts with a remark on the relation between two such constraints: locality theory and cartographic templates. Both have an impact on the shape of the clausal left periphery and both need to be taken into account when constructing theories of the clausal left periphery. Specifically, to determine the nature of locality constraints, we need to be able to observe them independently of templatic effects; to determine the nature of templatic constraints, we need to be able to observe them independently of locality effects. Locality is easy to disentangle from the template. Traditional investigations of locality are already free from templatic contamination, since multiple moving items land in the left peripheries of different clauses. The left-peripheral template is much harder to observe without contaminating locality effects because the left periphery is largely occupied by elements that move there. We have potential evidence for a genuine templatic effect only if a structure obeys independently observed locality effects and is still ungrammatical. In other words, evidence for the template must come from filtering effects not attributable to locality.

In sections 2 and 3 of this article, I study the ramifications of this single methodological observation. The empirical domain is the structure of the clausal left periphery in Italian as developed in Rizzi 1997, 2001a, 2004b. I will take more or less for granted that the set of syntactic categories employed by Rizzi is complete, that Rizzi’s description of the possible and impossible orders is correct, and that the description of the locality behavior of the various operations involved given in Cinque 1990 and Rizzi 1980, 2004b is essentially correct. On these assumptions, it turns out that Rizzi’s (1997, 2001a, 2004b) left-peripheral template does not, as a matter of empirical fact, have any filtering effect. The elements that reach the left periphery through movement behave exactly as though their behavior was dictated by locality alone without any templatic structure imposing further constraints. In other words, the left-peripheral template cannot be detected empirically because, if it exists, its effects are masked completely by locality effects. It seems like a natural move, then, to give up the template for the left periphery as a theoretical construct.

None of the empirical assumptions mentioned in the previous paragraph are innocent by any means. In particular, various scholars have argued that Rizzi’s notion of topic is too coarse (see Belletti 2004, Benincà and Poletto 2004, Bianchi and Frascarelli 2009, Frascarelli and Hinterhölzl 2007, Samek-Lodovici 2006, 2008 for relevant discussion). I set these objections aside and adopt Rizzi’s description of the categories involved and of the locality behavior of the various operations here for two reasons. The first is that this allows me to demonstrate the power of the methodological point made in section 1. The second reason is that the locality behavior of the subtypes of topics posited by the authors just mentioned has not been studied in sufficient detail to allow the methodological principle to be applied. Preliminary results suggest that the picture painted here is confirmed nicely if more fine-grained distinctions between topics are made. The same comments apply, mutatis mutandis, for the finer-grained typology of relative clauses discussed in Bianchi 2004.

Section 4 contains a discussion of what a theory of locality that yields the desired result might actually look like.

The starting point for the present investigation is Rizzi 2004b, which presents a considerable refinement and elaboration of two theoretical tools, locality and the templatic structure of the left periphery, on both of which Rizzi had previously published seminal work (Rizzi 1990, 1997). Rizzi (2004b) introduces modifications and refinements of the original proposals in both areas. In the domain of locality, Rizzi refines the classification of phrasal types underlying Rizzi 1990; thus, instead of the monolithic Ā-type of Rizzi 1990, he now posits three separate categories: modifier, quantificational, and topic. The class of modifiers contains all adverbs, possibly even all adverbials no matter what their syntactic category may be. The class of quantificational elements encompasses various syntactic operators (wh, focus, negation, and quantificational adverbs—the last two are simultaneously also members of the modifier class). Topic (in Italian) picks out phrases that have undergone clitic left-dislocation (CLLD). I discuss and revise this classification in section 4.

The templatic structure of the left periphery (i.e., the hierarchy of functional projections) is also refined in Rizzi 2004b. In sections 2 and 3, I explore Rizzi’s claim in this area. Rizzi’s (2004b) template is as follows:

(1) Force Top* Int Top* Foc Mod* Top* Fin IP

As in Rizzi 1997, the topmost head, Force, hosts the complementizer che ‘that’ in the head position and relative operators in the specifier position, while the lowest head, Fin(iteness), hosts the nonfinite complementizer di ‘of ’. Sandwiched between these are a position for unstressed fronted modifiers (Mod, a position introduced in Rizzi 2004b), a position for fronted foci and fronted wh-phrases (Foc), the position of the interrogative complementizer se ‘if ’ and a few exceptional wh-phrases (Int—on which see Rizzi 2001a), and a number of topic positions (Top). (The asterisk on Top and Mod indicates that these positions are recursive.)

Locality Proposes—The Template Disposes

Suppose that in a particular domain two kinds of elements can appear, squares (▓) and circles (●), and that in this domain squares never precede circles. That is, suppose that descriptively the following holds:

(2)

graphic

Assuming that ▓ and ● are both moved to their position, there are (at least) two factors that could give rise to the generalization in (2). On the one hand, there could be a templatic prohibition against having ▓ locally precede ●. Such a prohibition would standardly be implemented as a ban against ▓’s host F10 taking ●’s host FP2 as its complement. Alternatively, the ban might be attributable to a locality ban against ▓ crossing ●. Both of these options are schematized in (3).

(3)

  • a.

    Violation of the template:

    graphic

  • b.

    Violation of locality:

    graphic

In other words, in any situation where both ▓ and ● move to the same domain, locality factors and templatic factors are confounded. To unconfound them, ▓ and ● simply have to be placed in different templatic domains, that is, in different CPs. This unconfounding procedure is based on two assumptions: (a) templates impose local rather than global ordering constraints and (b) the clause (CP) is the local domain to which templates apply. Both assumptions are standard.

First, it is a common assumption in current syntactic research that templates regulate local structure building but do not constrain global structure (see Cinque 1999, Grimshaw 2000, 2005, among many others). That is, for (2) to count as a templatic effect, the ban against ▓ preceding ● holds if ▓ and ● appear in the same local domain, and it does not hold when ▓ and ● appear in different domains, as in (4).

(4) ✓▓ . . . ▓ . . . [domain boundary . . . ● . . . ● . . . ]

Factually, this state of affairs is the norm rather than the exception. Thus, while a focus cannot precede a relative pronoun within the left periphery of the same clause, a focus in a higher clause is compatible with a relative operator in a lower clause, and so on.

Second, it is an equally common assumption that the templatic domain is the extended projection of the relevant lexical head. The CP is therefore usually considered the verb-related templatic domain.

Assuming then that empirically we are faced with a local ordering effect (i.e., both (2) and (4) hold), templatic and locality factors can be unconfounded by the following simple experiment: If ▓ can cross ● (both originating in the same domain) without violating locality, then the configuration in (5) will be grammatical. On the other hand, if ▓ cannot cross● without violating locality, then the configuration in (5) will be ungrammatical.

(5)

graphic

This experiment unconfounds templatic factors and locality factors for the following reason. If (5) is acceptable, then the locality explanation in (3) is not available and the templatic explanation might be indicated. On the other hand, if (5) is unacceptable, then the locality explanation for (2) is viable, and hence the templatic explanation is not necessary.

Observe the asymmetry between the two approaches to the effect in (2). The locality approach is the stronger approach, as it generalizes from (2) to (5). The templatic explanation, on the other hand, is strictly local and makes no global predictions.1 Because of the asymmetry between the two types of explanation sketched in (3), a templatic explanation will need to be invoked only in cases where a locality explanation is unavailable. The slogan “Locality proposes and the template disposes” is to be understood in this way. The template is an additional strictly local filter, necessary only to capture constraints not already captured by locality. In the following two sections, I will go through Rizzi’s template for the left periphery and show that as far as the moving elements are concerned, it adds no further restrictiveness.

Before engaging with the empirical facts, let us sum up by considering the space of logical possibilities. (I will write A < B to indicate that A may precede B and *A < B to indicate that A may not precede B. Likewise, A > B indicates that A may follow B, and *A > B that A may not follow B.)

(6)

graphic

Possibilities III and IV in (6) are inherently uninformative regarding the template: the function of the template is to restrict local orders.2 Possibilities I and II instantiate restriction (2). Because of the locality/template confound, however, only the state of affairs in II provides an argument for a templatic explanation; the state of affairs in I is properly explained by locality alone.

Relative Operators, Foci, Modifiers, and Topics according to Rizzi

In this section, we will look at symmetries and asymmetries in the ordering of elements in the left periphery as described in Rizzi 1997, 2004b. The discussion of wh-related elements will be postponed for the most part until section 3.

The discussion is structured as follows. Each element with a unique position in the template (i.e., all but topics) is discussed in its own subsection, starting at the top with relative operators in Spec,Force and working down from there.

The Position of Relative Operators

Since relative operators occupy the highest specifier position in Rizzi’s template, they asymmetrically precede all other material in the left periphery.

In particular, relative operators may locally precede topics.3 This is shown in (7).

(7) Rel < Top

  • a.

    un uomo a cui, il premio Nobel, lo daranno senz’altro

    a man to whom the Nobel Prize they will give it undoubtedly

    (Rizzi 1997:289, (12a))

  • b.

    *un uomo, il premio Nobel, a cui lo daranno senz’altro

    a man the Nobel Prize to whom they will give it undoubtedly

    (Rizzi 1997:289, (12b))

The nonlocal interactions between relative operators and topics are discussed for example by Cinque (1990), who claims that relativization can cross topics, (8a). Topicalization, by contrast, is strong-island sensitive and therefore cannot escape from a relative clause, (8b).4 (Here and throughout, I use the terms strong and weak island as synonymous with absolute and selective island (see, e.g., Cinque 1990, Szabolcsi 2006, Szabolcsi and Den Dikken 2003 for discussion). In this usage, the strength of the island does not necessarily correlate with the perceived strength of the violation.)

(8) Rel < Top

  • a.

    Questo é l’uomo, a cui tu pensi

    this is the man to whom you think

    che, il premio Nobel, lo daranno senz’altro.

    that the Nobel Prize they will give it undoubtedly

    (Chiara D’Ippoliti, pers. comm.)

  • b.

    graphic

    to Gianni I will talk to you only about people who undoubtedly will give him

    il premio Nobel.

    the Nobel Prize

    (Chiara D’Ippoliti, pers. comm.)

Example (8) shows that the templatic explanation for the asymmetry in (7) is redundant with the locality explanation independently needed for (8).

The examples in (9) show that relative operators asymmetrically precede foci in the local condition.

(9) Rel < Foc

  • a.

    graphic

    here is a man to whom THE NOBEL PRIZE they should give

    (non il premio X).

    (not prize X)

    (Rizzi 1997:298, (44a))

  • b.

    graphic

    here is a man THE NOBEL PRIZE to whom they should give

    (non il premio X).

    (not prize X)

    (Rizzi 1997:298, (44b))

This asymmetry in the local condition is matched by an asymmetry in the long-distance condition, (10).

(10) Rel < Foc

  • a.

    graphic

    your brother to whom we believe that MARIA they have introduced

    (non Francesca)

    (not Francesca)

    (Vieri Samek-Lodovici, pers. comm.)

  • b.

    *MARIA abbiamo incontrato tuo fratello, a cui avevano presentato.

    graphic

    (Chiara D’Ippoliti, pers. comm.)

Example (10a) is as expected under Cinque’s (1990) characterization of relativization as sensitive to strong but not weak islands. Given this characterization, we correctly expect relative operators to move past fronted foci. Similarly, in his classic discussion of subjacency in Italian, Rizzi (1980) states that relativization can escape (single) wh-islands, (11). Since fronted foci tend to create the same type of weak islands produced by wh-operators, (10a) is again as expected. In contrast to relative operators, foci are sensitive to both strong and weak islands. Relative clauses create strong islands; hence, (10b) is sharply ungrammatical.

(11)

  • Tuo fratello, a cui mi domando che storie abbiano raccontato,

  • your brother to whom I wonder which stories they told

  • era molto preoccupato.

  • was very troubled

  • (Rizzi 1980:50, (6b))

We see again that a templatic explanation for the asymmetry in (9) is redundant with the locality explanation independently needed for the asymmetry in (10).

Finally, as illustrated in (12), relative operators precede nontopical, nonfocal, fronted modifiers according to the template in (1). I will refer to such elements as fronted modifiers. When this term appears without further qualifications, it is to be understood as ‘fronted, nontopical nonfocal modifiers’.

(12) Rel < Mod

  • a.

    graphic

    this is the book that yesterday they brought to Gianni

    (Rizzi 2004b:241, (56b))

  • b.

    graphic

    this is the book yesterday that they brought to Gianni

That relativization can cross fronted modifiers in the long-distance case as well is shown in (13). The reverse case, a structure where such a modifier is fronted across a relative operator, is ungrammatical. Indeed, Benincà and Poletto (2004) and Rizzi (2004b) claim that modifiers in this use never leave the clause whose predicate they modify: all types of CPs are islands for fronted modifiers. (See section 2.2 for further discussion.)

(13)

  • Rel < Mod

  • graphic

  • this is the man to whom you have said that tomorrow I should speak

  • (Vieri Samek-Lodovici, pers. comm.)

This section has shown that, once templatic effects are disentangled from locality effects, the positioning of the relative operator follows entirely from locality; there is nothing left for the template to do.

The Position of Focus

Here, the result reached above for relative operators will be replicated for foci: once locality effects are taken into account, all ungrammatical examples are accounted for and we are left with no evidence for the template as an additional filter.

We have already seen how the long-distance condition for relative operators and foci predicts their relative local ordering. We now turn to the ordering of foci with respect to topics and fronted modifiers.

The examples in (14) illustrate that topics may either precede or follow fronted foci.

(14) Top > < Foc

  • a.

    Credo che a Gianni QUESTO gli dovremmo dire.

    I believe that to Gianni THIS we should say to him

    (based on Rizzi 1997:295, (37a))

  • b.

    Credo che QUESTO, a Gianni, gli dovremmo dire.

    I believe that THIS to Gianni we should say to him

    (based on Rizzi 1997:298, (37b))5

It will not come as a surprise that focus fronting and topicalization do not interact for purposes of locality in the long-distance condition either. As already mentioned, Cinque (1990) claims that topicalization is sensitive to strong but not weak islands. Topics themselves do not create islands for foci. The judgments in (15) confirm this characterization.

(15)

  • a.

    A Gianni, credo che QUESTO gli dovremmo dire.

    to Gianni I believe that THIS we should say to him

    (Vieri Samek-Lodovici, pers. comm.)

  • b.

    QUESTO credo che, a Gianni, gli dovremmo dire.

    THIS I believe that to Gianni we should say to him

    (Vieri Samek-Lodovici, pers. comm.)

Again, the long-distance condition fully predicts local ordering without the need to invoke an additional template.

The interaction of modifier fronting and focalization gives the following results. In the local condition, fronted modifiers may follow foci but must not precede them. When a modifier precedes a focus, the modifier is interpreted and pronounced as a contrastive topic (Rizzi 2004b; see also the discussion of adverbs in Benincà and Poletto 2004:55 and 72n3).

(16) Foc < Mod

  • a.

    QUESTA PROPOSTA, rapidamente, tutti i deputati hanno accettato.

    THIS PROPOSAL rapidly all the representatives have accepted

    (Vieri Samek-Lodovici, pers. comm.)

  • b.

    Rapidamente, QUESTA PROPOSTA tutti i deputati hanno accettato.

    rapidly THIS PROPOSAL all the representatives have accepted

    ✓ with rapidamente a contrastive topic

    * with rapidamente an unstressed modifier

    (Vieri Samek-Lodovici, pers. comm.)

In the long-distance condition, we find that foci may cross fronted modifiers, (17), while fronted modifiers may not cross foci (no example given here). The facts therefore show, once again, that the relative orders in the local and long-distance conditions are the same and that the local template doesn’t impose restrictions not already present in the long-distance condition.

(17)

  • Foc < Mod

  • QUESTA PROPOSTA credo che, rapidamente, tutti i deputati

  • THIS PROPOSAL I believe that rapidly all the representatives

  • hanno accettato.

  • have accepted

  • (Vieri Samek-Lodovici, pers. comm.)

As an aside, note that modifier movement is different from the other types of movement discussed here in that the latter allow unbounded movement in principle. Thus, in the pairwise comparisons any locality effect is due to the intervention, specifically, of the second member of the pair. This is not the case for fronted modifiers. As mentioned above, Rizzi (2004b) characterizes modifier movement as clause-bound. In (18a), for example, rapidamente ‘rapidly’ cannot be moved from the clause it modifies unless it is focal or topical (see also Benincà and Poletto 2004: 55, (6), and 72n3), although there is no obvious intervener here at all. A context that makes the modifier topical is provided in (18b). It is therefore not the intervention of a focus, specifically, that constrains modifier movement. The striking descriptive fact remains that the local ordering possibilities are no more restricted than the long-distance ones.6

(18)

  • a.

    Rapidamente, (*Gianni dice che) hanno risolto il problema.

    rapidly, (Gianni says that) they solved the problem

    (Rizzi 2004b:249n10, (ia))

  • b.
    • A:

      C’e quelche problema che hanno risolto rapidamente?

      is there a problem that they solved rapidly

    • B:

      Rapidamente, Gianni dice che hanno risolto il primo problema,

      rapidly Gianni says that they have solved the first problem

      ma non gli altri.

      but not the others

      (Rizzi 2004b:249n10, (ii))

To summarize, once the effects of locality are disentangled from the effects of the template, we find that the positioning of foci with respect to relative operators, topics, and fronted modifiers follows directly from locality interactions. There is no evidence in this domain that would motivate the existence of a template.

The Position of Mod

As we have seen, the positioning of fronted modifiers with respect to relative operators and foci is identical in the local and long-distance conditions. Locality theory can therefore explain these facts without the need to invoke additional templatic assumptions. Concerning the relative position of modifiers and topics, we observe that the template makes no restrictive claim at all, since it allows them in either order.

(19) Top > < Mod

  • a.

    Rapidamente, i libri, li hanno rimessi a posto.

    rapidly the books they put them in place

    (Rizzi 2004b:239, (49))

  • b.

    I libri, rapidamente li hanno rimessi a posto.

    the books rapidly they put them in place

    (Vieri Samek-Lodovici, pers. comm.)

Example (19b) suggests that topics may cross fronted modifiers, a supposition bolstered by the acceptable status of the long-distance case in (20). Example (19a) suggests that fronted modifiers may cross topics. This supposition must be true to allow (19a) but cannot be independently verified in the long-distance condition because of the general clause-bound nature of modifier movement discussed above.

(20)

  • I libri, credo che, rapidamente, li hanno rimessi a posto.

  • the books I believe that rapidly they put them in place

  • (Vieri Samek-Lodovici, pers. comm.)

As with relative operators and foci, the template adds no restrictions on modifier placement not independently needed to account for the nonlocal cases.

Interim Conclusions

So far, we have considered the pairwise interactions between relative operators, topics, foci, and fronted modifiers. When the possible and impossible orders in the local and long-distance conditions are compared, we see that there is no evidence for a templatic left-peripheral structure. It seems natural at this point to ask whether the template has any effect on the ordering of the elements under consideration that does not already follow from the pattern of the long-distance interactions.

To investigate this question, we have to move from pairs of elements to more complex patterns. In the discussion above we saw, on the basis of strictly cross-clausal interactions, that Rel, Foc, and Mod enter into the transitive, asymmetric order Rel < Foc < Mod and that Top may not cross Rel but is unordered with respect to the remaining two elements. A complete, ordered list of possible elements implementing these constraints and no others is given as the first line of (21). The second line is the order of specifiers that emerges from Rizzi’s template, (1).

(21)

  • graphic

  • graphic

It is a stark illustration of the paucity of truly independent evidence for the template that the two lines are almost identical. The only difference is the topic position between Foc and Mod. What (21) means is that, if locality were the only constraint impinging on the shape of the left periphery, we would expect there to be a topic position between the position for foci and that for fronted modifiers. Rizzi’s template entails that this position does not exist. We have thus identified a new potential test case for the existence or otherwise of templatic effects.

The question is probed empirically by the examples in (22) (Luigi Rizzi, pers. comm.; see also Benincà and Poletto 2004:54, (.5a), though the latter authors classify the phrase between the contrastive focus and the modifier as a second focus). The examples fail to support Rizzi’s specific template and are exactly as expected if locality alone shapes the left periphery of the clause.

(22)

  • a.

    graphic

    graphic

    graphic

  • b.

    IL TUO LIBRO, a Gianni, improvvisamente, gli hanno tirato

    YOUR BOOK to Gianni suddenly they threw

    graphic

    in faccia, non la sedia.

    in his face not the chair

To my mind, the facts discussed in this section strongly suggest that left-peripheral ordering arises as an effect of locality and that there is no templatic structure there at all. I call this hypothesis the locality approach to the left periphery. The existence of a position for Top between Foc and Mod is the approach’s first novel prediction. I hasten to add that the template could of course be enriched to accommodate the additional topic position. This would not strengthen but weaken the templatic approach, since it would remove a constraint. Notice also that the additional topic position is expected under the locality approach, while a templatic approach never gives rise to expectations of this sort; it merely provides a description of the facts.

Left-Peripheral Interrogative Syntax

We now turn to the syntax of left-peripheral interrogative elements. The situation for these turns out to be slightly more complicated than for the elements considered so far because some wh-elements move to the left periphery while others are externally merged there. As we will see, some of the complexities fall into place rather neatly under the locality approach but remain outside the reach of a templatic explanation.

Moved Wh-Phrases

We begin the discussion with wh-phrases that reach their left-peripheral positions through movement. In terms of the classic description, wh-movement is subject to weak (selective) and strong (absolute) islands. Relative clauses are strong islands; foci give rise to weak islands. The standard description of locality therefore entails the following cross-clausal interactions:7

(23) Cross-clausal locality interactions

  • a.

    *Wh < Foc

  • b.

    *Foc < Wh

  • c.

    *Wh < Rel

  • d.

    Rel < Wh

(23ac) are exemplified in (24ac), respectively; an example of (23d) was already given as (11).

(24)

  • a.

    ?*A chi pensi che QUESTO abbiano detto?

    to whom do you think that THIS they have said

    (Chiara D’Ippoliti, pers. comm.)

  • b.

    *QUESTO mi domando a chi hanno detto.

    THIS I wonder to whom they have said

    (Chiara D’Ippoliti, pers. comm.)

  • c.

    *A chii hai conosciuto [NP qualcuno [CP che folesse parlare ti]]?

    to whom have you met someone who wanted to speak

    (Cinque 1990:28, (79a))

From the cross-clausal interactions in (23), the locality approach derives the expectation that wh-phrases and foci cannot cooccur in the same clausal periphery, since whichever of the two precedes would have to have crossed the other in violation of either (23a) or (23b) (a locality account of the local incompatibility of elements was proposed as early as Chomsky 1977:92–93). According to Rizzi (1997), this expectation is borne out, (25).8 Unfortunately, we also derive the false expectation that relative operators and wh-phrases do cooccur within a single left periphery— but only in the order Rel < Wh. Examples of the type ‘the book {about which to whom/ to whom about which} Jack talked’ are as ungrammatical in Italian as they are in English.

(25) *Foc > < Wh

  • a.

    *A chi IL PREMIO NOBEL dovrebbero dare?

    to whom THE NOBEL PRIZE should they give

    (Rizzi 1997:298, (45a))

  • b.

    *IL PREMIO NOBEL a chi dovrebbero dare?

    THE NOBEL PRIZE to whom should they give

    (Rizzi 1997:298, (45b))

It will come as no surprise that the template in (1) sheds no further light on these questions. It encodes the incompatibility of foci and wh-phrases, albeit in a somewhat unsatisfactory way: by stipulating that foci and wh-phrases move to the same position (Spec,FocP). Furthermore—just like the pure locality approach advocated here—the template also sets up the false expectation that relative operators and wh-phrases can cooccur and that when they do, they do so in that order.

As in section 2, the data provide no independent evidence for the specific template. What remains a stipulation under both approaches is the local incompatibility of wh-phrases and relative operators with each other.9

Also in section 2, we saw that topicalization does not interact with focalization or modifier movement for the purposes of locality. The same lack of locality interactions also characterizes the relation between topicalization and wh-movement (Rizzi 2004b), (26).10

(26)

  • a.

    graphic

    I don’t know how you think that to Gianni we should talk to him

    (Rizzi 2004b:232, (27a))

  • b.

    graphic

    I don’t know to whom you think that these things we should say them

    (Rizzi 2004b:232, (27b))

Since topics can therefore cross wh-phrases and vice versa, the locality approach predicts that both orders are possible in the local case as well. This expectation is indeed borne out, (27).

(27) Wh > < Top

  • a.

    Mi domando, il premio Nobel, a chi lo potrebbero dare.

    I wonder the Nobel Prize to whom they could give it

    (Rizzi 1997:289, (14a))

  • b.

    ?Mi domando a chi, il premio Nobel, lo potrebbero dare.

    I wonder to whom the Nobel Prize they could give it

    (Rizzi 1997:289, (14b))

Finally, fronted modifiers can be crossed by wh-elements in both the long and the local conditions, while the modifiers themselves are characteristically sensitive. They do not cross wh-phrases locally or, given the clause-bound nature of the fronting involved, in the long-distance condition.

Int: On the Position of Base-Generated Wh-Elements

This section is based on observations in Rizzi 2001a, a paper in which Rizzi studies the positioning of the interrogative complementizer se ‘if ’ and the wh-phrases perché ‘why’ and come mai ‘how come’. The positioning of these elements differs markedly from that of other wh-phrases. The main difference lies in the fact that these elements may cooccur with foci and when they do, they precede them.11

We start the discussion with the complementizer se ‘if ’. Given the template in (21), which reproduces the relevant aspects of the structure proposed in Rizzi 1997, we might entertain two hypotheses about the cooccurrence and relative ordering of foci and interrogative complementizers: either (a) foci and interrogative complementizers can cooccur (the latter in Foc0, the former in Spec,Foc) and they occur in the order focus before interrogative complementizer or (b) foci and interrogative complementizers cannot cooccur because they would have to be hosted in the same phrase as head and specifier, but they are featurally incompatible. Neither of these expectations is borne out. Foci do cooccur with se ‘if ’ and when they do, the complementizer precedes the focus.

(28) Se < Foc

  • a.

    Mi domando se QUESTO gli volessero dire (non qualcos’ altro).

    I wonder if THIS they wanted to say to him (not something else)

    (Rizzi 2001a:289, (7a))

  • b.

    *Mi domando QUESTO se gli volessero dire (non qualcos’ altro).

    I wonder THIS if they wanted to say to him (not something else)

    (Rizzi 2001a:289, (7b))

Rizzi solves this problem by introducing a new position in the template, Int in (1).

If we consider the issue from the perspective of locality, expectations change. Recall that focus movement is sensitive to weak islands. In particular, foci cannot be extracted from indirect questions, including those introduced by se ‘if ’. From the weak-island sensitivity of focus movement, we deduce (29).

(29) *Foc < Se

Notice, however, that in contrast to the wh-phrases considered above, se ‘if ’ is base-generated in its surface position rather than being moved there. In the cases discussed above, we deduced the incompatibility of foci with wh-phrases from the fact that both were moved. Since se ‘if ’ is externally merged in the left periphery, a focus moving to a position below se ‘if ’ will not give rise to a locality violation, while a focus moving to a position above se ‘if ’ does. The situation is schematized in (30).

(30) Predicted interaction between base-generated se ‘if’ and focus

graphic

In other words, the observed cooccurrence and linear order of foci and se ‘if ’ is exactly as considerations of locality lead us to expect.12

According to Rizzi (2001a), perché ‘why’ and come mai ‘how come’ occupy the same position in the template that is occupied by se ‘if ’. Both elements therefore asymmetrically precede foci, (31) and (32). Rizzi accounts for the difference between regular wh-phrases and perché ‘why’ and come mai ‘how come’ by endowing ‘why’ and ‘how come’ with a special feature that allows them to occur in Spec,Int. This feature cannot be assigned to other wh-phrases.

(31) perché ‘why’

  • a.

    Perché QUESTO avremmo dovuto dirgli, non qualcos’ altro?

    why THIS we should have said to him not something else

    (Rizzi 2001a:294, (23a))

  • b.

    *QUESTO perché avremmo dovuto dirgli, non qualcos’ altro?

    THIS why we should have said to him not something else

    (Rizzi 2001a:294, (24a))

(32) come mai ‘how come’

  • a.

    Come mai IL MIO LIBRO gli ha dato, non il tuo?

    how come MY BOOK you gave to him not yours

    (Rizzi 2001a:294, (23b))

  • b.

    *IL MIO LIBRO come mai gli hai dato, non il tuo?

    MY BOOK how come you gave to him not yours

    (Rizzi 2001a:294, (24b))

The locality approach accounts for this behavior if both perché ‘why’ and come mai ‘how come’ can be externally merged in the left periphery. The idea that certain wh-elements including why and/or how come are externally merged in the left periphery of the clause they modify semantically is of course not new (see, e.g., Buell 2011, Ko 2005, Rizzi 1990, Starke 2001).

When it comes to interactions with topics and modifiers, se ‘if ’, perché ‘why’, and come mai ‘how come’ behave exactly like other wh-words. The lack of relevant interactions is illustrated in (33) for se ‘if ’ and topics.

(33) Whexternally merged > < Top

  • a.

    Non so se, a Gianni, avrebbero potuto dirgli la verità.

    I don’t know if to Gianni they could have said the truth to him

    (Rizzi 2001a:289, (9a))

  • b.

    Non so, a Gianni, se avrebbero potuto dirgli la verità.

    I don’t know to Gianni if they could have said the truth to him

    (Rizzi 2001a:289, (9b))

  • c.

    Mi domando se questi problemi, potremo mai affrontarli.

    I wonder if these problems we will ever be able to address them

    (Rizzi 2001a:289, (9c))

  • d.

    Mi domando, questi problemi, se potremo mai affrontarli.

    I wonder these problems if we will ever be able to address them

    (Rizzi 2001a:289, (9d))

Given that topicalization is not sensitive to wh-islands, this is as expected under the locality approach; the template does not add anything to the description of the facts.

To strengthen my case further, let me point out an observation in Rizzi 2001a that remains ultimately unexplained under Rizzi’s account. Come mai ‘how come’ and perché ‘why’ may appear in the left periphery of a clause higher than the one whose predicate they modify. A relevant example is given in (34a). The example is ambiguous between a reading that asks about the reason for saying and a reading that asks about the reason for resigning. For the latter reading, we may assume that perché ‘why’ has moved from the lower clause. The locality approach predicts that when perché ‘why’ moves, it should interact with foci as all other moved wh-phrases do: the two should be incompatible. This is indeed true. The addition of a focus in (34b) disambiguates the sentence; only the reason-for-saying reading remains. The derivation of this reading, of course, does not require movement of the wh-phrase.

(34) Moved perché ‘why’

  • a.

    Perché ha detto (a Gianni) che si dimetterà?

    why did he say (to Gianni) that he will resign

    (Vieri Samek-Lodovici, pers. comm., based on Rizzi 2001a:295, (27))

  • b.

    Perché A GIANNI ha detto che si dimetterà (non a Piero)?

    why TO GIANNI he said that he will resign (not to Piero)

    (Rizzi 2001a:295)

Under an approach like Rizzi’s where the appearance of perché ‘why’ and come mai ‘how come’ in Int depends on a feature (i.e., on an inherent lexical property of these items), the behavior in (34b) is unexpected. Why should movement block these elements’ inherent ability to appear to the left of foci in Spec,Int? On the locality approach, the key to the puzzle is a relational property of the elements in question: when they are externally merged in a position, they may cooccur with and precede foci without incurring any crossing violations; when they are moved, they are incompatible with foci for the same locality-derived reasons that other wh-phrases are.

With the additional assumption that the elements discussed here (se ‘if ’, perché ‘why’, and come mai ‘how come’) are externally merged in the left periphery of the clause they modify, the locality approach makes strikingly correct predictions. The facts might be describable in terms of a left-peripheral template—though with significant difficulty as perché ‘why’ and come mai ‘how come’ show—but they cannot be predicted.

Locality Behavior and Theories of Locality

What I have presented so far are sets of local ordering behaviors and sets of long-distance crossing behaviors—“locality behaviors.” On the assumption that locality is essentially a theory of which elements can cross which other elements (Relativized Minimality, Attract Closest), I have claimed that the nonlocal crossing behavior predicts the local ordering but not the other way around. I have not presented a theory of locality that would derive the set of crossing behaviors discussed here.

In this section, I sketch a theory of locality to fill this gap. It is a standard version of Relativized Minimality enriched with a nonstandard classificatory structure. This is the kind of theory whose existence was presupposed in the discussion of the previous sections.

Following among others Boeckx and Jeong (2004), Chomsky (1995), Rizzi (1990, 2001c, 2004b), and Starke (2001), I have assumed that syntactic locality should be handled in terms of a ban against likes crossing likes (Relativized Minimality). Such a ban requires a definition of what it means for one element to cross another element; it also needs to be supplemented by a categorization of items into classes to answer the question of what elements count as “likes.” The standard definition of crossing in syntax is in terms of c-command: in a movement chain of type τ, the head of the chain may not c-command any elements of type τ that (asymmetrically) c-command the foot of the chain.13

As mentioned, this needs to be supplemented by a suitable classification of moving elements or movements into structural types; that is, we need to specify what types τ ranges over in the definition above and how these types are constituted. The classification has to make sure that elements that do not interact with each other in terms of crossing are in orthogonal classes, and elements that block each other are in the same class. Rizzi’s (1990) classic formulation of Relativized Minimality approaches this issue by assuming three classes (A, Ā, and heads) that are orthogonal to each other.

As mentioned in the introduction, Rizzi (2004b:243, (61)) proposes the following refinement of this classification. Instead of a unified Ā-class, he suggests three different classes: Quantificational, Modifier, and Topic.

(35)

  • a.

    Argumental: person, number, gender, case

  • b.

    Quantificational: Wh, Neg, measure, focus, . . .

  • c.

    Modifier: evaluative, epistemic, Neg, frequentative, celerative, measure, manner, . . .

  • d.

    Topic

By and large, Rizzi’s three classes are orthogonal to each other, though some elements belong simultaneously to the modifier class and to the quantificational class: Neg and measure.

Starke (2001) makes an important addition to the logic of Relativized Minimality. According to him, syntactic elements are not only classified into orthogonal classes; rather, some classes have subclasses and superclasses (see also Boeckx and Jeong 2004). The construction of movement dependencies and the application of Relativized Minimality can then be understood in terms of the elsewhere or Pāṇini principle: the application of a more specific process preempts the application of a less specific one. Thus, an element that belongs only to a superclass will always move as a member of that superclass and this movement will be blocked by any intervener from that superclass. An element that belongs to a subclass, however, will be able to undergo the more specific rule of moving elements in that subclass and be able to circumvent blocking by elements in the superclass. Itself, it will block elements in the superclass and in the subclass.

The fact that Rizzi cross-classifies certain elements suggests arranging the relevant classes in a subclass-superclass relationship, (36). (The classification probably, though not necessarily, reflects the organization of features in a syntactic feature hierarchy (see Bianchi 2004, Boeckx and Jeong 2004, Starke 2001 for comparable ideas).) I will refer to Rizzi’s quantificational elements as Operators (Op).

(36)

graphic

This organization of features makes the reasonable claim that operators instantiate a particular kind of modification. The organization in (36) has three consequences of interest here. First, since foci and wh-elements are in the same class and since likes cannot cross likes, we derive the fact that moved foci and moved wh-elements do not cooccur. Second, since foci and wh-phrases are operators, they are also modifiers (by the logic of subclasses and superclasses).14 Therefore, foci and wh-phrases block the movement of fronted modifiers, deriving the fact that fronted modifiers cannot appear in front of foci or wh-phrases. Third, modifiers do not block the movement of foci and wh-phrases. Though foci and wh-phrases are modifiers, they are also, more specifically, operators. By the Pāṇini principle, when they move, they move as operators and this movement is not blocked by simple modifiers. We therefore derive that foci and wh-phrases must precede fronted modifiers. I will refer to the Modifier class henceforth as Mod.

Topics are correctly characterized in Rizzi’s classification as orthogonal to Mod and Op; as we have seen, neither Mod nor Op interacts for locality with topics. A more complete classification thus looks as follows, (37), where F is simply the class of all syntactic classes:

(37)

graphic

This leaves the issue of classifying relative operators (Rel). We have seen that Rel asymmetrically blocks movement of Mod and Op, suggesting that Rel is a subclass of Op. At the same time, we have seen that Rel asymmetrically blocks movement of topics, which suggests that Rel is also a subclass of Top. The complete classification needed to derive the order of elements in the Italian left periphery is given in (38).

(38)

graphic

The structure encodes that topics do not interact for locality with modifiers or operators but are asymmetrically blocked by relative operators.15 Modifiers are asymmetrically blocked by operators, which in turn are asymmetrically blocked by relative operators.16

One may wonder whether there is independent motivation for the particular classification given in (38). In particular, are there independent reasons to think of Rel as a subclass both of Top and of Op? The idea that Rel is a special kind of Op should not meet with much resistance. At least since Chomsky 1977, it has standardly been assumed that relative clause formation is an application of the more general process of wh-movement, which includes wh-question formation and focus movement. Cyclicity effects along the path of all types of wh-movement (Irish, McCloskey 1979; Kikuyu, Clements 1984; etc.) are usually taken as evidence in favor of classifying relative clause formation together with focalization and wh-movement. An additional reason to believe that Rel is a special case of Op comes from the observation that in some languages resumptive pronouns may occur in all wh-movement constructions while in other languages they are restricted only to relative clauses (see Boeckx 2001 for discussion and references). Indeed, there appears to be an implicational universal according to which every language that shows resumption in wh-questions also exhibits resumption in relative clauses but not the other way around. These facts strongly support the organization in (38).

Following the other branch of (38), the idea that there is a close connection between topicalization and relative clause formation has a tradition going back at least to Kuno’s (1976) argument that a particular element can be relativized if and only if it is possible to construe that element as the topic (more recently, comparable claims about a relation between topicalization and relativization have been made in Bianchi 2004:93–94 and Bayer and Salzmann 2010, both building on Rizzi’s (2001b) treatment of D-linking). The following paradigm from Williams 2011 also groups topicalization with relativization. The examples are particularly telling, as they show relative clause formation patterning not only with topicalization but also against wh-question formation. ((39aiii), (39biii), and (39diii) are from Williams 2011:168, (1e–f ), (1b) and (1a), and (1d) and (1c), respectively. (39ciii) are based on Williams’s (1g–h).)

(39)

  • a.

    Baseline

    • i.

      John is the mayor.

    • ii.

      The mayor is John.

  • b.

    Question formation

    • i.

      ?I wonder who is the mayor.

    • ii.

      I wonder who the mayor is.

  • c.

    Topicalization

    • i.

      John I think is the mayor.

    • ii.

      *John I think the mayor is.

  • d.

    Relativization

    • i.

      I met the man who is the mayor.

    • ii.

      *I met the man who the mayor is.

In (39ai) the subject precedes the predicate, while in (39aii) the predicate precedes the subject. According to Williams (2011), the order in (39aii) requires focus on John. Both orders allow questioning the subject, but when the predicate precedes the copula, relativization and topicalization of the subject are blocked, (39cii) and (39dii).

Considerations of this type lend independent plausibility to the classification in (38). If it is accepted, Relativized Minimality and the Pāṇini principle together derive the local and nonlocal ordering of phrases without the need to stipulate a specific left-peripheral template. The predictions include some very specific ones discussed above: (a) the locality approach correctly predicts the existence of a topic position between foci and fronted modifiers, (b) the locality approach correctly predicts the complex interaction between left-peripheral foci, scope, and the interrogative elements se ‘if ’, come mai ‘how come’, and perché ‘why’. Finally, (38) directly expresses the fact that Top is orthogonal to Op and Mod; a standard templatic structure with its linear order cannot do so.

Notice that the discussion here has centered exclusively on the relative positioning of phrases rather than heads in the left periphery—with the one exception of se ‘if ’. The relative order of heads in the left periphery and the relative order of phrases and heads has not been considered. These elements pose obstacles for the full reduction of the left-peripheral template to locality. For example, both complementizers che ‘that’ and di ‘of ’ may be crossed by moving elements. They should therefore be able to freely intersperse with left-peripheral material. As a matter of fact, che ‘that’ always precedes all the elements discussed here and di ‘of ’ always follows. At this stage of inquiry, these elements therefore require residual templatic stipulations. See Manzini 2010 and McCloskey 2006 for relevant discussion compatible with the approach taken here.

Conclusion

I have argued that Rizzi’s (1997, 2001a, 2004b) left-peripheral template is undermotivated as a theoretical construct. The same conclusion emerges from Haegeman 2010. Relativized Minimality combined with a suitable classificatory structure provides a more explanatory account of the relative positions and behavior of Rel, Int, Foc, Top, and for the most part Mod. The ordering predicted by unconfounded locality considerations alone can be compared with Rizzi’s (2004b) template in (40).

(40)

  • graphic

I have argued that the slight differences between Rizzi’s template and the “map” of the left periphery presented here are forced by the locality approach and represent descriptive improvements. The template itself adds nothing to the description of these facts. One of the two, locality or the template, will have to be retired from the theory by Occam’s razor. The following considerations force our hand. First, there is independent support for the classification in (38) but no independent support for the templatic structure. Second, the locality approach is more predictive (and hence more easily falsifiable) than the templatic approach. Third, even theories that have a template typically invoke Relativized Minimality, the idea that likes do not cross likes. It appears, then, that the left-peripheral template in (1) needs to be retired from the theory until a strong argument for it has been made.

It should be noted that Williams (2002, 2011) develops an account of facts like the relative ordering of Rel, Foc, and Mod where the correlation between local and nonlocal behavior follows from the assumption of level ordering (in more recent work called the F-clock) together with the Level Embedding Conjecture, the assumption that the timing of the embedding of a clause under a matrix predicate follows from the size of that clause. Williams’s account is the most principled version of letting Occam’s razor cut the other way in that locality derives from what corresponds to the template in the theory. Williams’s account—like the present one—fails to provide a ready answer concerning the status of the complementizer che ‘that’. Williams’s account potentially has some trouble with the fact that there are several positions for topics. This becomes a problem if it can be assumed, following Rizzi, that these topic positions are not distinct interpretively. A detailed comparison between the approaches will have to await another occasion.

The present article is silent on the question of whether movement to the left periphery targets the specifiers of dedicated functional heads or not. Such heads may still be assumed. They do not need to be ordered by selectional requirements but can be merged freely. Derivations where the heads are merged in the wrong order will be filtered out because the heads will then not be able to attract their appropriate specifiers without violating locality.

Notes

This article evolved from a homework assignment I set in PLIN3202: Readings in Syntax (University College London (UCL), 2008). I would like to thank the audiences at the NORMS workshop Root Phenomena and the Left Periphery (Tromsø, 2008), colloquia at Queen Mary University London (2008) and Essex University (2009), and at the second UCL/ Potsdam Workshop on Information Structure (London, 2010) for valuable input. I should also thank the two anonymous LI reviewers, Gisbert Fanselow, Liliane Haegeman, Caroline Heycock, Ad Neeleman, Andrew Nevins, Luigi Rizzi, and Vieri Samek-Lodovici for useful criticism, input, and encouragement. Special thanks go to Chiara D’Ippoliti, who spent countless hours as my informant in the summer of 2010.

1 An important background assumption in this discussion is a particular view of locality, namely, that locality is concerned with the possibility or impossibility of certain classes of elements crossing over certain other classes of elements—essentially, Relativized Minimality or Attract Closest.

2 Notice that possibility III in (6) is puzzling from the current perspective: locality must allow ▓ to cross ● to account for the local order, but must prevent this crossing in the long-distance case. If faced with this situation, we are virtually forced to assume that the violation in the long-distance case has a source different from ▓ crossing over ●. For a concrete example, see the discussion of the interactions between modifiers and topics in section 2.3 and related discussion in footnote 6.

3 It is an unfortunate terminological tangle that what Rizzi calls topicalization is called clitic left-dislocation by Cinque and what is called focalization by Rizzi is called topicalization by Cinque. I adopt Rizzi’s terminology except in a few cases where there is a good reason not to.

4 Example (8b) falls far short of forming a minimal pair with either (7b) or (8a). The reason is that the minimally paired example (i) is grammatical—but under the irrelevant construal called left-dislocation (LD) in Cinque 1990 and hanging topic in Benincà and Poletto 2004.

(i)

  • Il premio Nobel, questo é l’uomo a cui lo daranno senz’altro.

  • the Nobel Prize this is the man to whom they will undoubtedly give it

  • (Chiara D’Ippoliti, pers. comm.)

LD is distinguished from clitic left-dislocation (CLLD) by the inability of LD to appear in embedded contexts, the possibility of replacing the resumptive clitic with a full pronoun only in LD, and the impossibility of pied-piping the preposition introducing indirect objects only in LD—such pied-piping is obligatory in CLLD. On all these tests, structures like (i) behave like LD rather than CLLD (Chiara D’Ippoliti, pers. comm.). Thus, when the initial preposition is dropped in (8b), the example becomes unambiguously an LD structure and is grammatical, (ii).

(ii)

  • Gianni, ti parleró solo delle persone che senz’altro gli daranno il premio Nobel.

  • Gianni I will talk to you only about people who undoubtedly will give the Nobel Prize to him

  • (Chiara D’Ippoliti, pers. comm.)

  • Failure to properly distinguish LD from CLLD is presumably at the heart of Samek-Lodovici’s (2008) claim that CLLD is totally island-insensitive, a claim exemplified there ( p. 6, (9)) with an example structurally like (i).

5 Recall that I am assuming for the sake of the argument that Rizzi’s characterization of the phrase following the contrastively stressed focus as a topic on a par with the topic preceding the focus is correct. There is mounting evidence, however, that this is not quite correct. Thus, Belletti (2004), Benincà and Poletto (2004), Frascarelli and Hinterhölzl (2007), and Samek-Lodovici (2006, 2008) all agree that like focus-moved phrases and unlike clitic left-dislocated phrases, postfocal preverbal phrases do not give rise to obligatory clitic doubling with direct objects. Benincà and Poletto (2004) also argue that these phrases give rise to weak crossover effects, unlike clitic left-dislocated phrases. Benincà and Poletto (2004), Frascarelli and Hinterhölzl (2007), and Samek-Lodovici (2006) disagree regarding the exact information-structural properties of such phrases.

6Rizzi (2004b:249n10) tentatively suggests a way of deducing the locality of modifier movement from Relativized Minimality. If the deduction is workable, it will bring modifier movement fully in line with the other movements considered here.

7 Strictly speaking, we expect D-linked wh-phrases to be somewhat more liberal than non-D-linked ones in their ability to cross foci nonlocally and to cooccur with them locally. I ignore this complication in what follows.

8 The picture in the text is painted with a somewhat broad brush. A number of factors influence whether foci and wh-phrases may or may not cross each other and whether they may or may not cooccur. For example, Samek-Lodovici (2006) disputes the claim that foci cannot cooccur with wh-phrases locally based on corrective foci (i.e., altered repetitions of a question that the other interlocutor has misheard). In such contexts, both orders appear to be possible in the cross-clausal as well as the local condition, (i) (Chiara D’Ippoliti, pers. comm.). (In matrix questions, there are additional restrictions on what may appear immediately after the wh-phrase (see Rizzi 1997), a factor not properly controlled in Samek-Lodovici 2006.)

(i)

  • a.

    Mi chiedo a chi GIULIO l’ha prestata la macchina, non Giulia.

    I wonder to whom GIULIO has lent (it) the car not Giulia

  • b.

    Mi chiedo GIULIO a chi l’ha prestata la macchina, non Giulia.

    I wonder GIULIO to whom has lent (it) the car not Giulia

  • c.

    GIULIO mi chiedo a chi l’ha prestata la macchina, non Giulia.

    GIULIO I wonder to whom has lent (it) the car not Giulia

  • d.

    A chi pensi che GIULIO l’ha prestata la macchina, non Giulia?

    to whom do you think that GIULIO has lent (it) the car not Giulia

These examples illustrate the situation for subjects and indirect objects. When it comes to indirect and direct objects, the data appear to be somewhat more complex (see Rizzi 2001a). More empirical work is needed here, but there is no indication in the data I have seen that the local condition is more restrictive than the nonlocal one; that is, there are no data to favor a templatic over a locality approach.

D-linking of the wh-phrase further influences the acceptability of such cases in both the local and the long-distance conditions. D-linking also weakens the verb-second requirement in direct questions (Rizzi 1996:87n16).

9 The fact that relative operators can cross wh-phrases in the long-distance condition but cannot cooccur locally is an illustration of type II in (6). Type II is the case where an order is possible nonlocally but impossible locally. According to the logic summarized in (6), this is the state of affairs that an argument for a local template could be based on.

The full range of facts is as follows: (a) *Wh < Rel both locally and nonlocally; (b) *Rel < Wh only locally. Constraint (a) could be attributed to locality, as I am assuming here, while constraint (b) could be attributed to the template assigning the landing site for wh-movement a higher position than the landing site for relative operators: F0Wh c-commands F0Rel in the template. Of course, this is exactly the opposite of the order posited by Rizzi (1997, 2004b) but on grounds that I have argued to be insufficient. Another possible templatic approach to this local incompatibility might be to equate the landing site of relative operators and wh-phrases in the local template: F0Wh = F0Rel.

However, it seems plausible that the source of the incompatibility is semantic and not syntactic: questions are sets of propositions, semantically, while relative clauses are predicates. The combination of the two operators creates semantic objects with somewhat strange meanings and no known use. In addition, wh-in-situ languages bar relative operators and wh-operators from taking scope in the same CP, too. This further points to the conclusion that a semantic rather than a templatic constraint is at work.

10 Main clauses are somewhat more restricted than embedded clauses. According to Rizzi (1997), main clauses do not permit wh-phrases to precede topics, (i). Notice that the structure in (i) is a candidate for a templatic explanation according to the logic laid out in (6). Of course, Rizzi’s template encodes the facts in (i) of footnote 8, not those in (i) below. In fact, Rizzi (1997) argues the failure of Wh < Top in main clauses is part of a larger generalization (something akin to verb-second) by which even a nontopical subject cannot intervene between the wh-phrase and the verb. It appears, then, that there are independent interfering factors in main clauses.

(i) Main clauses: Top < Wh

  • a.

    A Gianni, che cosa gli dovremmo dire?

    to Gianni what should we say to him

    (based on Rizzi 1997:299, (47))

  • b.

    *Che cosa, a Gianni, gli dovremmo dire?

    what to Gianni should we say to him

    (based on Rizzi 1997:299, (47))

11 These statements will be qualified slightly below.

12 Bocci (2007:50, (32)) reports that focus movement to the right of conditional se ‘if ’ is degraded. This claim contrasts with Rizzi’s (2001a) judgments reported in (28a) for interrogative se ‘if ’. There is no contradiction here. Indeed, Haegeman (2010) discusses Bocci’s observation to support a movement analysis of conditional clauses. If indeed there is a contrast between interrogative and conditional se ‘if ’, this would be striking confirmation both for the general lines of a locality approach to the left periphery and for the details of the accounts of interrogative and conditional se ‘if ’ here and in Haegeman 2010, respectively. (I owe the observation in this footnote to Caroline Heycock ( pers. comm.).)

13 This move aligns syntactic locality closely with phonological locality (the line-crossing constraint of autosegmental phonology). The only substantive difference between the two is that intervention is defined linearly in phonology and hierarchically in syntax (Nevins 2010, Rizzi 2004b, Starke 2001).

14 A possible implementation is to view syntactic features as feature structures along the lines of Pollard and Sag 1994. The Mod feature would then have—as possibly atomic values—the nonquantificational items on Rizzi’s (2004b) list in (35c) and the structured value Op. Op in its turn would have the values Neg, Wh, Foc, . . . . The presence of the Op feature on an element asymmetrically entails the presence of the Mod feature on the same element.

On the assumption that modifier movement accesses the feature Mod and focus movement accesses the feature Op, intervention can be computed as in Boeckx and Jeong 2004 and Starke 2001, deriving the symmetric intervention effects between quantificational elements and the asymmetric intervention created by operators for modifiers.

15Abrusan (2007) and Fox and Hackl (2006) give a number of strong arguments for a semantic rather than a syntactic analysis of the standard set of weak islands. This would allow deleting the Op node from (38).

16 Clearly, this is no more than a first approximation. I argued above that several moved instances of Op block each other. The same is true for Rel. However, according to Rizzi (2004b), multiple modifiers do not block each other as long as they are moved to the front in an order-preserving fashion and multiple topics are possible in any order. These properties do not follow from the classificatory structure given here.

While Rizzi captures the difference between the unique Op and Rel on the one hand and the multiply-occurring Top and Mod on the other, by making the latter recursive (Top* and Mod*), the further difference between topics and modifiers does not follow from his template either.

References

Abrusan
,
Marta
.
2007
.
Contradiction and grammar: The case of weak islands
.
Doctoral dissertation, MIT, Cambridge, MA
.
Bayer
,
Josef
, and
Martin
Salzmann
.
2010
.
That-trace effects and resumption as the result of improper movement
.
Ms., University of Konstanz
.
Belletti
,
Adriana
.
2004
.
Aspects of the low IP area
. In Rizzi 2004a,
16
51
.
Benincà
,
Paola
, and
Cecilia
Poletto
.
2004
.
Topic, focus, and V2: Defining the CP sublayers
. In Rizzi 2004a,
52
75
.
Bianchi
,
Valentina
.
2004
.
Resumptive relatives and LF chains
. In Rizzi 2004a,
76
114
.
Bianchi
,
Valentina
, and
Mara
Frascarelli
.
2009
.
Is topic a root phenomenon?
Ms., University of Siena
.
Bocci
,
Giuliano
.
2007
.
Criterial positions and left periphery in Italian: Evidence for the syntactic encoding of contrastive focus
.
Nanzan Linguistics
1
:
35
70
.
Boeckx
,
Cedric
.
2001
.
Mechanisms of chain formation
.
Doctoral dissertation, University of Connecticut, Storrs
.
Boeckx
,
Cedric
, and
Youngmi
Jeong
.
2004
.
The fine structure of intervention in syntax
. In
Issues in current linguistic theory: A festschrift for Hong Bae Lee
, ed. by
Chungja
Kwon
and
Wonbin
Lee
,
83
116
.
Seoul
:
Kyungjin
.
Buell
,
Leston Chandler
.
2011
.
Zulu ngani ‘why’: Postverbal and yet in CP
.
Lingua
121
:
805
821
.
Chomsky
,
Noam
.
1977
.
On wh-movement
. In
Formal syntax
, ed. by
Peter
Culicover
,
Thomas
Wasow
, and
Adrian
Akmajian
,
71
132
.
New York
:
Academic Press
.
Chomsky
,
Noam
.
1995
.
Categories and transformations
. In
The Minimalist Program
,
219
394
.
Cambridge, MA
:
MIT Press
.
Cinque
,
Guglielmo
.
1990
.
Types of Ā-dependencies
.
Cambridge, MA
:
MIT Press
.
Cinque
,
Guglielmo
.
1999
.
Adverbs and functional heads: A cross-linguistic perspective
.
Oxford
:
Oxford University Press
.
Clements
,
George N
.
1984
.
Binding domains in Kikuyu
.
Studies in the Linguistic Sciences
14
:
37
56
.
Fox
,
Danny
, and
Martin
Hackl
.
2006
.
The universal density of measurement
.
Linguistics and Philosophy
29
:
537
586
.
Frascarelli
,
Mara
, and
Roland
Hinterhölzl
.
2007
.
Types of topics in Italian and German
. In
On information structure, meaning and form
, ed. by
Susanne
Winkler
and
Kerstin
Schwabe
,
87
116
.
Amsterdam
:
John Benjamins
.
Grimshaw
,
Jane
.
2000
.
Locality and extended projection
. In
Lexical specification and insertion
, ed. by
Peter
Coopmans
,
Martin
Everaert
, and
Jane
Grimshaw
,
115
135
.
Amsterdam
:
John Benjamins
.
Grimshaw
,
Jane
.
2005
.
Words and structure
.
Stanford, CA
:
CSLI Publications
.
Haegeman
,
Liliane
.
2010
.
The movement derivation of conditional clauses
.
Linguistic Inquiry
41
:
595
621
.
Ko
,
Heejeong
.
2005
.
Syntax of why-in-situ: Merge into [Spec,CP] in the overt syntax
.
Natural Language and Linguistic Theory
23
:
867
916
.
Kuno
,
Susumu
.
1976
.
Subject, theme, and the speaker’s empathy: A reexamination of relativization phenomena
. In
Subject and topic
, ed. by
Charles
N
.
Li, 417
444
.
New York
:
Academic Press
.
Manzini
,
Maria Rita
.
2010
.
The structure and interpretation of (Romance) complementizers
. In
The complementizer phase
, ed. by
E. Phoevos
Panagiotidis
,
167
199
.
Oxford
:
Oxford University Press
.
McCloskey
,
James
.
1979
.
Transformational syntax and model theoretic semantics: A case study in Modern Irish
.
Dordrecht
:
Reidel
.
McCloskey
,
James
.
2006
.
Questioning and questions in a local English
. In
Crosslinguistic research in syntax and semantics: Negation, tense, and clausal architecture
, ed. by
Raffaella
Zanuttini
,
Héctor
Campos
,
Elena
Herburger
, and
Paul H.
Portner
,
87
126
.
Washington, DC
:
Georgetown University Press
.
Nevins
,
Andrew
.
2010
.
Locality in vowel harmony
.
Cambridge, MA
:
MIT Press
.
Pollard
,
Carl
, and
Ivan A.
Sag
.
1994
.
Head-Driven Phrase Structure Grammar
.
Chicago
:
University of Chicago Press
.
Rizzi
,
Luigi
.
1980
.
Violations of the Wh-Island Constraint and the Subjacency Condition
.
Journal of Italian Linguistics
5
:
157
195
.
Reprinted in Luigi Rizzi, Issues in Italian syntax, 49–76. Foris: Dordrecht (1982). Page numbers refer to the 1982 version
.
Rizzi
,
Luigi
.
1990
.
Relativized Minimality
.
Cambridge, MA
:
MIT Press
.
Rizzi
,
Luigi
.
1996
.
Residual verb second and the WH Criterion
. In
Parameters and functional heads: Essays in comparative syntax
, ed. by
Adriana
Belletti
and
Luigi
Rizzi
,
63
90
.
Oxford
:
Oxford University Press
.
Rizzi
,
Luigi
.
1997
.
The fine structure of the left periphery
. In
Elements of grammar: Handbook in generative syntax
, ed. by
Liliane
Haegeman
,
281
337
.
Dordrecht
:
Kluwer
.
Rizzi
,
Luigi
.
2001a
.
On the position of Interrogative in the left periphery of the clause
. In
Current studies in Italian syntax: Essays offered to Lorenzo Renzi
, ed. by
Guglielmo
Cinque
and
Giampaolo
Salvi
,
287
296
.
Oxford
:
Elsevier North-Holland
.
Rizzi
,
Luigi
.
2001b
.
Reconstruction, weak island sensitivity and agreement
. In
Semantic interfaces: Reference, anaphora and aspect
, ed. by
Carlo
Cecchetto
,
Gennaro
Chierchia
, and
Maria Teresa
Guasti
,
145
176
.
Stanford, CA
:
CSLI Publications
.
Rizzi
,
Luigi
.
2001c
.
Relativized minimality effects
. In
The handbook of syntactic theory
, ed. by
Mark R.
Baltin
and
Chris
Collins
,
89
110
.
Oxford
:
Blackwell
.
Rizzi
,
Luigi
, ed.
2004a
.
The cartography of syntactic structures
.
Vol. 2
,
The structure of CP and IP
.
Oxford
:
Oxford University Press
.
Rizzi
,
Luigi
.
2004b
.
Locality and left periphery
. In
The cartography of syntactic structures. Vol. 3, Structures and beyond
, ed. by
Adriana
Belletti
,
223
251
.
New York
:
Oxford University Press
.
Samek-Lodovici
,
Vieri
.
2006
.
When right dislocation meets the left-periphery: A unified analysis of Italian non-final focus
.
Lingua
116
:
836
873
.
Samek-Lodovici
,
Vieri
.
2008
.
Topic, focus, and background in Italian clauses
.
Ms. (version 3), University College London
.
Starke
,
Michal
.
2001
.
Move dissolves into Merge: A theory of locality
.
Doctoral dissertation, University of Geneva
.
Szabolcsi
,
Anna
.
2006
.
Strong vs. weak islands
. In
The Blackwell companion to syntax
, ed. by
Martin
Everaert
and
Henk
van Riemsdijk
,
479
531
.
Malden, MA
:
Blackwell
.
Szabolcsi
,
Anna
, and
Marcel
den Dikken
.
2003
.
Islands
. In
The second Glot International state-of-the-article book
, ed. by
Lisa
Cheng
and
Rint
Sybesma
,
213
240
.
Berlin
:
Mouton de Gruyter
.
Williams
,
Edwin S
.
2002
.
Representation Theory
.
Cambridge, MA
:
MIT Press
.
Williams
,
Edwin S
.
2011
.
Regimes of derivation in syntax and morphology
.
London
:
Routledge
.