Skip to content
BY 4.0 license Open Access Published online by De Gruyter Mouton May 10, 2022

The Case and Finiteness Restrictions of Italian Relative che

Giuseppe Rugna ORCID logo
From the journal Probus


This paper addresses the restrictions of Italian relative che ‘what’ to finite environments and direct case gaps. While the standard analysis takes these restrictions to follow from the C status of che, this paper argues for an alternative approach, according to which che is a DP, on a par with other interrogative and relative elements. Specifically, it is argued that relative che is identical to interrogative che and relative cui ‘what.obl’ in the narrow syntactic derivation, up to the point of transfer. Realization of relative che is then blocked at ext with oblique case gaps by the more specific cui, along the lines of the Elsewhere Principle. The status of cui as a specialized relative element is also discussed. The restriction to finiteness for relative che is treated as an instance of a more general phenomenon that precludes bare DPs from occurring at the edge of infinitival relatives. Here I adopt and extend (Richards, Norvin. 2010. Uttering trees. Cambridge, Massachusetts: MIT Press) Distinctness Theory to account for the facts of Italian.

1 Introduction

Italian che ‘what’ can appear in interrogatives, headed relative clauses (RCs) and complement clauses (along with a multitude of other morphosyntactic environments, which will not be discussed here). As is well known, these uses of che do not share the same distribution. The distribution of interrogative che (che+q) and relative che (che+rel) essentially varies along two dimensions: finiteness and case.

Che+q can appear in both finite (1a) and non-finite clauses (1b), it can be embedded under prepositions, i.e., it licenses direct as well as oblique case gaps (1c),[1] and it can also function as a determiner taking a nominal complement (1d). On the other hand, che+rel can appear only in finite clauses (2a), and is incompatible with oblique case gaps (2c). In infinitival clauses che+rel is replaced by the prepositional complementizer da (2b). For oblique gaps in RCs, Italian makes use of two elements specialized for headed relatives: cui ‘what.obl’ and Det + qual- ‘the which’. See Cinque (1978, 1982, 2008 for further details on the distribution of relativizers in Italian. Finally, (3) shows use of che as a finite sentential complementizer.

Che fai?
What do-2sg
‘What are you doing?’
Che fare?
What do-inf
‘What to do?’
Di che parli?
Of what speak-2sg
‘What are you talking about?’
Che libri leggi?
What books read-2sg
‘What books do you read?”
L’uomo che hai sposato
The man what have-2sg married
‘The man you married’
L’uomo *che / da sposare
The man what/from marry-inf
‘The man to marry’
L’uomo *di che / di cui / del quale parli
The man of what / of what-obl / which.masc-pl speak-2sg
‘The man you are talking about’
Penso  che andrò in vacanza
Think-1sg what go-fut-1sg in vacation
‘I think that I’ll go on vacation’

Despite their syncretic morphophonological basis, the standard analysis takes che+q and che+rel to be two distinct elements instantiating different categories: while the former is a D(P), the latter spells out C, a functional head intrinsically specified for finiteness and other ‘clause-typing’ (Cheng 1991) properties (or Force, Chomsky 1995, Rizzi 1997). Accordingly, che+rel would be a version of C that is specified as [+finite], da as [−finite], etc. The restriction to finite clauses has constituted a pivotal argument for the treatment of relative and sentential che (as well as English that, French que etc.) as instantiations of the same functional category (cf. Cinque 1978, 1982; Kayne 1976). In this paper, I will refer to the hypothesis that che+q and che+rel are categorially distinct elements as the C hypothesis.

While the C hypothesis has essentially remained unchallenged for several years, the debate over the categorial status of (relative) complementizers has recently been revitalized (e.g., Baunaz and Lander 2018; Kato and Nunes 2009; Kayne 2014; Manzini and Savoia 2003; Poletto and Sanfelici 2018; Rinke and Aßmann 2017; Roussou 2020a; Sportiche 2011, among others). The main point of contention is that the C hypothesis offers no insight into the syncretism between complementizers and pronominal elements. Thus the syncretism between che+q and che+rel, for instance, is argued to require an explanation, and should not be reduced to cross-categorial homophony or to the result of grammaticalization (cf. Poletto and Sanfelici 2018, 2019; Roberts and Roussou 2003; Roussou 2020b; van Gelderen 2009). The rejection of the D/C dichotomy is explicitly argued for in Manzini and Savoia (2003) (and further elaborated in subsequent work: Manzini 2012, 2014a; Manzini and Savoia 2005, 2011). These authors propose that there is a single che in the Italian lexicon, of category D. Furthermore, Kayne (2014), Poletto and Sanfelici (2018) argue that the theoretical distinction between pronominal elements and complementizers lacks empirical support. The hypothesis defended by this line of research that che+q and che+rel are categorially non-distinct will be referred to here as the D hypothesis.

The D hypothesis has the advantage of accounting for the syncretism between che+q and che+rel in a simple and appealing way. However, it also effectively reinstates the issue of their different morphosyntactic distribution. Specifically, two problems remain unaddressed under the D hypothesis, namely (i) the inability of che+rel to license oblique case gaps (henceforth, the case problem),[2] and (ii) the restriction to finite contexts for che+rel (henceforth, the finiteness problem). While I follow Manzini and Savoia in assuming that che+rel is a DP like che+q, I depart from them in that I assume that these elements are identical only in narrow syntax. I argue that their different properties emerge after transfer, and specifically at ext, as a result of their different derivational history.

The present paper is organized as follows.

In Section 2, I discuss the empirical properties standardly assumed as diagnostics for the C/D dichotomy. I show that these properties are not reliable diagnostics for determining the categorial status of Italian che+rel. I further discuss the D hypothesis of Manzini and Savoia and conclude that it leaves unanswered the case and finiteness problems.

In Section 3, I address the case problem by adopting a form of the D hypothesis. The solution I propose is that the realization of che+rel is blocked at the mapping with phonology (ext) by the more specific cui. I moreover discuss the theoretical status of specialized relative pronouns such as cui.

In Section 4, I address the finiteness problem, which I include within the larger issue of the conditions that prohibit bare DPs from occurring at the edge of infinitival relatives. Here I follow Richards (2010) in accounting for the unavailability of che+rel in infinitival relatives as deriving from a violation of his Distinctness Condition.

Finally, Section 5 concludes the discussion.

2 Pronouns, Complementizers and the D Hypothesis

2.1 On the Differences Between Pronouns and Relative Complementizers

It is standardly assumed that relativizers can belong to different morphosyntactic categories. Thus for instance English relative who and which are assumed to be DPs, whereas relative that spells out the category C (to which sentential complementizers are also standardly assumed to belong). This is the position I refer to as the C hypothesis, which is motivated on the basis of four empirical properties that are assumed to classify relativizers into pronominal elements (DPs) and relative complementizers (Cs) (with arguments tracing back to Klima 1964, Kayne 1975, 1976): (a) animacy restrictions (b) case-marking; (c) compatibility with prepositions; and (d) sensitivity to the finiteness of the clause. Examples (4)–(7) illustrate this for English. (4) shows that the animacy restriction applies to relative pronouns who and which, but not to that; (5) shows that pronoun who can be case-marked, unlike that; (6) shows that who and which can be embedded under Ps, in contrast to that; and (7) shows that wh-pronouns can be licensed in infinitival contexts, unlike that. The different morphosyntactic behaviour of who and which on the one hand and of that on the other is thus explained away as a difference in categorial status, according to this standard position.

The man *which / who / that John saw
The table which / *who / that John broke
The man whose wife / *that’s wife John saw
The man with whom / *with that John spoke
The chair on which / *on that John sat
The man with whom to speak
*The man that (to) see

Insofar as properties (a–d) are real diagnostics for determining whether a given element is a D(P) or a C, however, it is never made quite explicit (to the best of my knowledge) how these properties should theoretically follow from a difference in categorial status. In other words, it is unclear why D elements should be subject to a different syntax vis-à-vis C elements. It is nonetheless implicit in the literature that the distribution of properties (a–d) is related to the assumption that Cs are functional heads strictly connected with the TP-layer of the sentence. This strict connection would then render C elements incompatible with properties (a–c), the hallmark of (pro-)nominal elements; on the other hand, Cs would be intrinsically specified for finiteness, and therefore display property (d).

Be that as it may, properties (a–c) have been challenged as diagnostics for the D/C dichotomy on empirical grounds (see in particular Kayne 2014, Poletto and Sanfelici 2018). Poletto and Sanfelici (2018) argue that Old Italian varieties featured relativizers that show a mixed behaviour with respect to properties (a–c). For instance, the Old Ligurian relativizer che falls into the C category according to property (c) (since incompatible with Ps), while at the same time this element showed sensitivity to the animacy of the antecedent in subject relatives, thus behaving as a DP with respect to property (a). As shown in (8) (from Poletto and Sanfelici 2018: 275), the agreeing relativizer chi was employed with a [+human] extracted subject (8a); otherwise, the default form che (also used as a sentential complementizer) was selected (8b).

questa femena chi m’ à spanyunto questo inguento adosso
this woman rel cl.1sg.acc has Spread this unguent
‘This woman that spread this unguent on me’.
receveyva tuto zo che era dayto a Criste
received-3sg all that rel was given to Christ
‘He received all that was given to Christ’

Conversely, Poletto and Sanfelici (2018) show that an element generally labeled as a DP, qual- ‘which’, could remain uninflected in relative constructions. This is the case of Old Neapolitan, where quale remained invariable regardless of the properties of its antecedent, as illustrated in (9) (from Poletto and Sanfelici 2018: 281). Old Neapolitan quale is comparable to Italian che+rel in this respect, even though the latter is standardly taken to be a C.

Amico quale te si’[…]
friend rel you are
‘Friend that you are […]’
Haverno facte cose quale mai tentarono fare
have-3pl done things rel never tried do
‘They did things that they never tried to do’

In fact, it can be argued that properties (a-c) are inconclusive for determining the categorial status of relativizers even in Italian. The distribution between che+rel and the relative wh-pronoun cui, for instance, only differs with respect to the case of the gap in the RC. As shown in (10)–(11), neither che nor cui display property (a), i.e., restrictions to animacy.

L’uomo che Gianni vede
The man what G. sees
‘The man who Gianni saw.’
L’uomo con cui Gianni parla
The man with what.obl G. speaks
‘The man with whom Gianni speaks.’
Il libro che Gianni legge
The book what G. reads
‘The book that Gianni reads.’
Il libro su cui Gianni studia
The book on what.obl G. studies
‘The book from which Gianni studies.’

With regards to property (b), it is possible to consider cui to be case marked as an oblique. This is suggested by its distribution and by the affix –ui, which could mark non-nominative case in older stages of the language (cf. egli ‘he.NOM’ vs. lui ‘he.ACC/OBL’; see Benincà 2010 on some uses of cui in Old Italian; cf. Poletto and Sanfelici 2019). While cui is pronominal according to property (b), it cannot be excluded that it is simply an oblique form of che+rel (as traditional (pre-generativist) grammars would have it). In fact, this assumption might account in a simple way for the distribution of che+rel versus cui in modern Standard Italian. Specifically, the incompatibility of che+rel with Ps (property (c)) could be attributed to the fact that the Italian lexicon contains a more specific element (cui), which would block the spell-out of che+rel with oblique case gaps (along the lines of the Elsewhere/Subset Principle; Halle 1997; Kiparsky 1973). This is essentially what I will propose in Section 3.

Furthermore, as pointed out in Roussou (2020a: fn. 2), compatibility of Ps with purported C elements seems to be subject to cross-linguistic variation. The following example from Italian featuring sentential che is a case in point:

Gianni era favorevole a che ti sposassi
Gianni was favorable to what cl.2sg married-subj-2sg
‘Gianni was in favor of you getting married.’

Note however that this possibility is restricted to just the preposition a ‘to’ in Italian, which suggests that it is an idiosyncratic lexical rule rather than a generalized property. Still, the fact that it is possible at all to combine prepositions with elements standardly labeled as C indicates that property (c) cannot be used as a reliable categorial diagnostics.[3] Let me also point out that the C hypothesis, while compatible with data like (2c) or (13a) below, does not force the empirical facts upon us. Even maintaining that che+rel is a C, it would be unclear why che could not lexicalize both a relative DP and C (cf. 13b, where C is silent). Hence, the incompatibility of Ps with che+rel (or equivalent) does not necessarily speak in favor of che+rel belonging to the category C: something additional would need to be said to account for the unavailability of sentences like (13a).

*L’uomo con che parlo spesso
The man with what speak-1SG often
‘The man I often speak with’
L’uomo [CP [PP con che] [C <che> [parlo spesso]]].

The empirical evidence thus does not offer conclusive support for the argument that properties (a–c) cut across Ds and Cs. What about property (d)?

Sensitivity to finiteness prima facie appears to be a real discriminating property of elements standardly labelled as Cs. Consider the sentences in (14). In the infinitival relative (14a), che cannot be licensed, as opposed to che in the interrogative (14b) or cui in (14c). It is in particular the restriction to finite environments that has bolstered the hypothesis that elements such as che+rel instantiate elements of a different categorial nature than, e.g., their interrogative counterparts (cf. Cinque 1978, 1982; Kayne 1976). The standard reasoning is that if che+rel were a pronoun like che+q or cui, it would remain unexplained why its distribution is constrained by the finiteness of the clause, since pronouns do not seem to be subject to such a constraint.

*L’uomo che vedere
The man what see-inf
‘The man to see.’
Non sa che fare
neg know.3sg what do.inf
‘He/she doesn’t know what to do.’
L’uomo a cui affidarsi
The man to what.obl rely-inf-refl
‘The man to rely on.’

However, there are reasons for being skeptical about a strict correlation between complementizers and finiteness.

First, consider the data in (15) (from Manzini 2012: 311), from Paulilàtino (Sardinia), which features inflected infinitives. In (15a), the prepositional complementizer is compatible with both the inflected and the non-inflected infinitive. However, in (15b) ki (also used as a finite sentential complementizer) is also compatible with the inflected infinitive. These data then suggest that, whatever categorial status we attribute to elements like ki/che, these elements are not sensitive to the finiteness of the clause, but rather to the expression of nominal properties on T (i.e., its inflection).[4]

… innantis ˈεnn-εrε-(nε) ˈiɕɕɔɕɔ
… before to come-inf-(3pl) they
… innantis ki ˈεnn-εrε-nε ˈiɕɕɔɕɔ
… before that come-inf-3pl they
‘… before they came’

Second, as Manzini and Savoia (2005: 489) discuss, there are contexts in Italian that can license the relativizer che in non-finite environments. This is the case of light-headed relatives with an interrogative interpretation (20). If che were specified as a [+finite] C, one would counterfactually predict the use of che in (20) to be ungrammatical.

?Non sa quello che dire
NEG knows that what say-INF
‘He doesn’t know what to say’

Third, it is not the case that pronominal elements do not show sensitivity to the finiteness of the clause. For instance, some very formal registers of Italian allow the use of Det + qual- in restrictive headed relatives with direct case gaps, provided that the verb in the RC is in the subjunctive mood (16a) (cf. Cinque 1978, 1982). Indeed, use of the indicative mood leads to severe deviance if Det + qual- is employed in such restrictive relatives (16b).

I  cittadini i quali abbiano riscontrato problemi…
The citizens the which have-subj-2sg found problems
‘The citizens who might have had problems…’
??/* I cittadini i quali hanno riscontrato problemi…
The citizens the which have-ind-2sg found problems
The citizens who might have had problems…’

The contrast in (16) points towards the conclusion that pronominal elements, too, can be sensitive to the finiteness of the clause. In this particular case, Det + qual- would be sensitive to the mood specification on T.

Finally, in languages such as French or English, wh-pronouns cannot be licensed in infinitival relatives unless they respect certain morphosyntactic requirements: they cannot occupy the edge of the relative as bare DPs (17)–(19), but must be either non-overt or embedded within larger phrases (19).

A man (*who) to marry.
A show (*which) to enjoy.
*A man whom to dance with.
*A brush which to paint with.
A man with whom to dance.
A brush with which to paint.

This asymmetry is well-known and has been extensively discussed in the literature (see references cited in Section 4). However, what is failed to be recognized is that the alleged relative ‘complementizers’ are barred under the same conditions that bar pronominal elements. Suppose that che+rel projects a DP. If so, its restriction to finite environments should not come as a surprise given the comparable restriction of other bare DPs at the edge of infinitival relatives. As I argue in Section 4, an account of che+rel’s restriction to finite clauses should be amenable to the same treatment, whichever it may be, that is offered to account for cases such as (17)–(19). The assumption that che+rel is specified as [+finite] does not help achieve a comprehensive generalization of the factors that preclude bare DPs from occurring at the edge of infinitival relatives.

All of the above then suggests that property (d) cannot ascertain whether a given element belongs to C or D. These considerations aside, it seems clear that in headed RCs che+rel is restricted to finite contexts. The question that arises (to be addressed in Section 4) is why this should be the case if che+rel is not a [+finite] C.

2.2 The D Hypothesis

In the previous section I argued that the categorial distinction between pronominal elements and relative complementizers rests on dubious empirical premises (cf. Kayne 2014; Poletto and Sanfelici 2018). There is another empirical aspect of the C hypothesis that has been subject to criticism in recent years (see Manzini and Savoia 2003 among others), namely that such a hypothesis falls short of accounting for the systematic syncretism between pronominal elements and the purported C elements across Indo-European.[5] In particular, it can be argued that the C hypothesis lacks explanatory value insofar as such syncretism is treated in terms of cross-categorial homophony or as the result of grammaticalization (cf. Baunaz and Lander 2018; Manzini and Savoia 2011). Moreover, the question arises as to why a functional specification that can host verbs (in, e.g., V2, T-to-C movement, etc.) would need to be lexicalized via an independently available pronominal element (Manzini and Savoia 2011: 14–5). It is uncontroversial that che+q as well as relative pronouns and their equivalents in wh-movement languages undergo Internal Merge to the CP-layer of the sentence. If che+relis assumed to belong to a non-nominal category subject to different syntactic operations, our theory would be missing an overarching generalization.

Manzini and Savoia (2003 et seq.) among others criticize this aspect of the C hypothesis. These authors take the syncretism between the different uses of che at face value, treating che as the very same lexical item in every context in which it appears. In particular, che is taken to be a DP capable of introducing a variable at LF, essentially on a par with a lambda operator (cf. Arsenijević 2009). The difference between the different uses of che would lie in the type of variable that che can bind. Specifically, che as an interrogative or relative element introduces a variable that ranges over individuals;[6] as a sentential complementizer, che introduces a variable that ranges over propositions/possible worlds. This analysis has as an effect that there can now be a single che in the lexicon, with its interpretation being contextually determined.

The D hypothesis defended in Manzini and Savoia (2003 et seq.) is intuitively preferable over the C hypothesis on grounds of simplicity and explanatory power. As it stands, however, it is problematic in that it leaves unaddressed the case and finiteness problems mentioned in Section 1 (cf. fn. 4). Nonetheless, given the empirical problems faced by the C hypothesis, in this paper I attempt to solve these issues under some form of the D hypothesis. The major point of departure I take from Manzini and Savoia’s proposal is that I assume a model of grammar where the lexicon is distributed across different modules (as in Distributed Morphology; e.g., Arregi and Nevins 2012; Halle and Marantz 1993). Under this model, I argue that elements like che+q and che+rel are identical only for a portion of the derivation: until the point of transfer. More specifically, I take che+q and che+rel to instantiate the same (featurally indistinguishable) DP in the pre-syntactic lexicon and under the manipulation of the narrow syntactic component (with Manzini and Savoia). However, after transfer, che+q and che+rel effectively become different elements as a result of the different syntactic operations that they undergo in narrow syntax. After transfer, and specifically at ext, these different elements could potentially be assigned different phonological exponents, which I argue is the case with the oblique forms of che+q and che+rel (/ke/ vs /kui/).

3 Addressing the Case Problem

3.1 Preliminary Assumptions

This section addresses the case problem, i.e., the inability of che+rel to license oblique case gaps, under the assumption that it is a DP. We are interested in providing an account of the pattern in (22)–(23):

Di che / *di cui / *del quale parli?
Of what / of what.obl /
‘What are you talking about?’
L’uomo *di che / di cui / del quale parli
The man of what / of what.obl / speak.2sg
The man you are talking about’

There are in fact two sides to this issue. One the one hand, we may wonder why che+q may be embedded under prepositions/license oblique case gaps, while che+rel may not. On the other hand, we may also ask why cui or Det + qual- cannot be used at all in interrogatives. These issues are particularly acute in a generative theory of grammar, since it is not at all clear why wh-DPs should fall into construction-specific paradigms if the grammar is not driven by the notion of ‘construction’ (Chomsky 1981: 7). Moreover, abstracting away from the categorial status of che+rel, relative and interrogative wh-elements have long been assumed to be subject to the same syntax (Chomsky 1977). It should therefore come as a surprise that empirically there exist such specialized wh-elements as e.g. Italian cui, French dont, Hungarian amit, Russian kotorij, etc. To my knowledge, the question of why some elements can be specialized for the (headed) relative construction has not been addressed in the literature.[7] Against this backdrop, the case problem may actually be placed within the broader issue of why wh-elements show paradigmatic morphosyntactic properties (which issue includes, among others, the finiteness problem). Note in this regard that the C hypothesis would have to treat as separate issues the incompatibility of che+rel with oblique case gaps and the unavailability of cui in interrogatives. Under the D hypothesis, these issues may potentially be amenable to a unifying treatment, which I attempt to develop in 3.2. Before doing so, however, it is important to spell out a few core assumptions.

First, I assume that wh-elements are not pre-syntactically specified for the syntactic environment in which they are to be inserted (via features such as e.g. [+interrogative] or [+relative]). This assumption is motivated by the theoretical desideratum of avoiding construction-specific statements in the grammar (cf. Wiltschko (1998)). Moreover, it is entailed by the Minimalist Program insofar as some form of the Inclusiveness Condition holds (which “bars introduction of features that are not inherent to lexical items”, Chomsky et al. (2019: 237); see however Bocci et al. 2021). Since interrogativity and relativity can reasonably be taken to be complex properties determined at the interface with semantics and pragmatics, I assume that such properties do not play a role in the featural make-up of DPs. As such, the notation +REL/+Q or the terms ‘interrogative’, ‘relative’ etc. are merely conventional and will have no theoretical import in the analysis to be developed.

Second, I assume that wh-elements have internal structure (cf. Barbiers et al. 2009; Boef 2013). I assume this structure to be quite minimal. Focusing on argumental wh-elements,[8] I take these elements to be Ds merged with a collection of φ-features. The assumed underlying structure is represented in (24).


The D head encodes the wh-feature, which is what gives wh-elements their semantic flavor. We may assume that the wh-feature signals the property of introducing a variable at the interpretive interface (e.g. Nishigauchi 1990), essentially like Heimian indefinites (Heim 1982; cf. Postma 1994; see however Caponigro 2003 for a different view). As far as I can tell, nothing crucial hinges on this assumption. What is important is that wh-elements have a semantic core that is syntactically represented, and invariant across constructions.

The other ingredient of wh-elements, the φP (cf. Déchaine and Wiltschko 2002), encodes the usual features of gender, number, person and animacy. I assume that these features may restrict the range of the variable to be bound (e.g. Heim 2008; Heim and Kratzer 1998: 244). For instance, if the animacy feature is specified as [+human], the range of the variable is restricted to [+human] entities.

The syntactic structure is illustrated as a bundle of features in (25a). For the sake of simplicity, we may avoid representing each individual φ-feature, and adopt instead the reduced but equivalent version of (25a) in (25b).

[+wh, +φ: [[gender: ] [number: ] [person: ] [animacy: ]]]
[+wh, +φ: […] ]

In short, I take relative and interrogative argumental wh-DPs to be structurally identical: the null hypothesis.[9]

Nevertheless, the empirical evidence shows that interrogative and relative wh-DPs can be different from a morphophonological point of view (e.g., che+q vs cui). The question that arises, then, is how the grammar can classify wh-DPs into construction-specific paradigms. In this paper I assume that the distinction between interrogative and relative DPs arises at ext on the basis of the Agree operation.

Consider the sentences in (26)–(27). In the interrogative (26), the spell-out of the wh-DP as who signals that the range of the variable to be bound is restricted to [+human] entities. In the relative (27), on the other hand, who does not serve the same semantic purpose. It rather seems to act as a bound variable whose range is restricted by the nominal head man. The spell-out of the DP as who seems to be demanded by a purely formal morphophonological requirement of the English grammar, which is inconsequential for the C-I interface. Support for this conclusion comes from dialects of English where what can be used as a relative pronoun to refer to [+human] antecedents (cf. (28), from Edwards 1993: 228).

Who did it?
The man who did it
The girl what’s coming over

In (26), I assume that who enters the derivation with the animacy feature pre-syntactically valued as [+human]. In the case of relative pronouns such as who in (27), I follow Kratzer (2009), Landau (2015) among others and assume that these elements (along with different types of bound variables, e.g., PRO, reflexives) enter the derivation as minimal pronouns: they are Ds with an unvalued set of φ-features.[10] Moreover, following Rooryck and Vanden Wyngaerd (2011), minimal pronouns are related to their antecedent via the operation Agree, whereby they also acquire the antecedent’s φ-features. In other words, I assume that relative DPs are born without values for φ-features, which are valued under Agree with the antecedent (cf. Furuya 2017).[11] The ultimate shape of the wh-DP can be determined at ext, according to language-particular rules, as notoriously advocated by DM.

According to an anonymous reviewer, the assumption concerning a pre-syntactically valued who (interrogative) and a syntactically valued who (relative) implies two separate entries for who: one for interrogative who and one for relative who. This is partly correct. The reviewer is correct in claiming that we are assuming two separate entries for who. However, these two entries are not implied by our assumption that φ-features can be valued either pre-syntactically or syntactically. Even if we assumed that who entered the derivation with a pre-syntactically valued [+human] feature in both interrogatives and RCs, we would still have to assume two separate entries for who. Consider why. As discussed, the empirical evidence shows that grammars can classify wh-elements into construction-specific paradigms. The information about which paradigm a given DP belongs to is not, we assume, part of the featural make-up of DPs; rather, it must somehow be acquired from the syntactic context. Hence, whatever the nature of this information, there is a sense in which we can end up with distinct bundles of features after transfer, e.g., one bundle containing the information that the DP is in an interrogative clause and another bundle containing the information that the DP is in a RC. Therefore, at ext Standard English would still map the exponent /hu:/ onto two (minimally) different bundles of features (thereby generating two different entries), irrespective of whether or not the relative DP enters the derivation with a valued [+human] feature. Note that this distinct (interrogative/relative) mapping is necessary also in order to rule out what from occurring in RCs: the ext component of Standard English does not allow the mapping of /wɒt/ onto a relative DP to take place (though this mapping is allowed in some dialects, cf. (28)).

Under our approach, then, the difference between interrogative and relative wh-DPs would lie in the way in which they reach the interfaces. Interrogative wh-DPs can reach the interfaces with pre-syntactically valued φ-features, as in the case of the animacy feature of English interrogative who, Italian interrogative chi, etc. On the other hand, relative wh-DPs enter the derivation as minimal pronouns and must establish Agree with their antecedent, as roughly sketched in (29).[12]


[man[+human] [CP [+wh, φ: ∅] …]

(pre-Agree configuration)


[man[+human] [CP [+wh, φ: {+human}] …]

(post-Agree configuration)

Consequently, interrogative and relative wh-DPs are structurally identical in the narrow syntactic derivation; however, they differ after transfer, in that relative wh-DPs have undergone Agree. We may therefore tentatively propose that it is the establishment of such an Agree relation that makes a wh-element count as ‘relative’.[13]

The assumptions I made so far can be surmised as in (I–IV):

  1. Construction-specific features are not part of the featural make-up of DPs.

  2. Argumental wh-elements are phrases comprising a D head and a collection of φ-features, which may restrict the range of the variable.

  3. Relative wh-DPs enter the derivation as minimal pronouns, i.e., with an unvalued set of φ-features.

  4. Relative wh-DPs are defined by their entering an Agree relation with their antecedent.

3.2 On the Case Problem and Specialized Relative Elements

Let us now return to the issue of Italian che. If che+rel is born with unvalued φ-features (as per (III)), then it is featurally identical to che+q in narrow syntax. This is because che+q does not specify any restriction on the range of the variable (cf. Manzini 2014b: 192). As shown in (30), a felicitous answer to an interrogative with che+q can refer to both human and non-human entities. Assuming that (i) the answer to an interrogative provides the value for the variable, and that (ii) φ-features restrict the range of the variable, we can conclude that che+q reaches the interfaces with unvalued φ-features.[14]

Che vedi?
What see.2sg
‘What do you see?’
Un uomo / Un orso
A man / A bear

Therefore, che+rel and che+q enter the derivation as the same minimal pronoun, represented as the feature-bundle in (31) (ø represents lack of values). This hypothesis is in line with Manzini and Savoia’s (2003 et seq.) proposal that che+rel and che+q constitute one and the same element.

[+wh, +φ: [ø]]

However, I depart from Manzini and Savoia in that I assume that che+rel and che+q are identical only in narrow syntax (before transfer takes place). As discussed above, relative DPs differ from interrogative DPs in that the former reach the interfaces having established Agree with their antecedent. Thus, assuming that the mapping to morphophonology (ext) can “see” whether Agree has taken place, I propose that relative DPs may potentially be assigned a different morphophonological exponent from their interrogative counterparts (cf. Rooryck and Vanden Wyngaerd 2011 for a similar proposal applied to the empirical domain of reflexives). Italian happens to employ the same exponent when binding direct case gaps, as shown in the lexical entries in (32). (32a) shows the lexical entry for che+q: a wh-element that reaches the interfaces with unvalued φ-features (+φ: [ø]); (32b) shows the lexical entry for che+rel: a wh-element that reaches the interfaces with φ-features valued under Agree (I use the notation {v} to indicate valuation under Agree). In both cases, the same exponent /ke/ is assigned at ext.

[+wh, +φ: [ø]] /ke/ (che+q)
[+wh, +φ: {v}] /ke/ (che+rel)

However, syncretic exponents are not required in principle. Indeed, empirically we observe that languages can have different exponents for what in Italian would be che+q and che+rel. This is the option selected, for instance, by several dialects of Italian (Manzini and Savoia 2003, 2005), as well as standard varieties of English (what vs that). Some insight into this variation can be gained under the assumption that a wh-DP may be assigned a different phonological exponent at ext depending on whether it has undergone Agree in the derivation.[15], [16]

This assumption analogously accounts for why some languages use elements specialized for the headed relative construction: they use a different exponent for a wh-element that has undergone Agree. Abstracting away from the issue of Det + qual- (to which I return below), I propose that this is why Italian has the specialized element cui for oblique case gaps in headed relatives: oblique che+q and oblique che+rel have different phonological exponents (cf. (33) and (34)). Specifically, the exponent /kui/ is assigned to a wh-DP that (i) has established Agree in the derivation; and (ii) bears oblique case (+obl). On the other hand, /ke/ is assigned to a wh-DP that (i) has not established Agree in the derivation; and (ii) bears no specifics for case. This latter point accounts for why che+q licenses direct as well as oblique case gaps (see however fn. 1).

[+wh, +φ: [ø]] /ke/ (che +q )
[+wh, +φ: {v}, +obl] /kui/ (cui)

The assumption that the exponent /kui/ is assigned to a wh-DP that has established Agree in the derivation can account for why it is specialized for the relative function. Interrogatives do not feature nominal antecedents; hence, wh-elements cannot establish Agree in these contexts. This predicts that cui will never occur in an interrogative in Standard Italian.[17]

Furthermore, if cui is specialized as +obl, we can begin to understand why che+rel cannot license oblique case gaps, assuming that some form of the Elsewhere Principle (Kiparsky 1973) plays a role at ext. In general terms, the Elsewhere Principle (or the Subset Principle in DM; Halle 1997) states that a more specific rule apply over a more general one. For the case at hand, this means that /kui/ will be realized at ext whenever the feature-bundle [+wh, +φ: {v}, +obl] reaches the interface. Realization of /kui/ represents a more specific rule than the general /ke/, since unlike the latter it also specifies case. Thus, realization of /ke/ is blocked in relative contexts with oblique cases by the more specific /kui/.

The hypothesis that cui blocks realization of che+rel with oblique cases raises the question of why cui does not block realization of Det + qual ‘the which’ in the same contexts. As mentioned in previous sections, Det + qual is grammatical with both direct and oblique cases (cf. the restrictive relatives in (35) and the appositive relatives in (36)). This distribution indicates that Det + qual- isn’t specified for case, just like che+rel. Therefore, by the Elsewhere Principle cui should block realization of Det + qual in oblique contexts, since more specific. This is not the case, however: cui and Det + qual- are in free distribution in oblique contexts (cf. (23) supra).[18]

I cittadini i quali abbiano riscontrato problemi…
The citizens the which have.subj.2sg found problems
‘The citizens who might have had problems…’
Una malattia contro la quale lotta da anni
A sickness against the which fight.3sg since years
‘A sickness that he’s been fighting for years.’
L’uomo, il quale Gianni disprezza …
The man, the which G. despises …
The man, who Gianni despises …
Le nuove proposte, sulle quali il governo lavora da mesi …
The new proposals, on which the government works from months
‘The new proposals, on which the government has been working for months…’

Abstracting away from the issue of the determiner in Det + qual-,[19] suppose that qual- realizes a different structure than che+rel and cui. In particular, suppose that qual- has an additional functional layer that encodes its D(iscourse)-linked (Pesetsky 1987) character, as in (37). Indeed, note that qual- as an exponent realizes an inherently D-linked element in interrogatives, as does English which, German welch-, etc. If this particular semantics is encoded in a dedicated functional head, as in cartographic approaches (e.g., Rizzi and Cinque 2016), it could be assumed that (37) is realized by a specialized exponent (38).

[+wh, +D-linked, +φ: {+sg}] /kʷale/

Note that this hypothesis is not in conflict with the minimal pronoun approach that we are advocating. The relative wh-DP would still enter the derivation with unvalued φ-features. The only difference from the other relative wh-DP would be the presence of an additional functional layer encoding the D-linked semantics.

If this is accepted, Det + qual- would not be in competition with cui (nor with che+rel). A solution to the case problem can therefore be obtained by simply assuming that realization of /ke/ is blocked at ext by the more specific /kui/.

An anonymous reviewer suggests an alternative analysis of the che/cui alternation in RCs. According to the reviewer, the proposal made here is problematic in that cui should be expected to show up also in interrogative clauses containing oblique case gaps in a [che [NP]] configuration, assuming that che and its NP complement enter into Agree as in RCs. Assuming a raising analysis of RCs (e.g., Bianchi 1999; Kayne 1994) the reviewer instead proposes that the realization of oblique che+rel as cui must be associated with the raising step of the derivation (cf. Kato and Nunes 2009; Poletto and Sanfelici 2019 for this type of ‘movement-triggered allomorphy’ approach), as sketched in (39).

ho affidato l’incarico [a che studente]
I have assigned the task [to what student]
[a che studente] ho affidato l’incarico <[a che studente]> (wh-movement)
Lo studente [a che <studente>] ho affidato l’incarico <[a che studente]> (Raising)
Lo studente a cui ho affidato l’incarico (movement-triggered allomorphy)

However, the problem noted by the reviewer disappears if DP-internal agreement is kept separate from DP-external agreement (cf. fn. 13). Note that the alternative proposed by the reviewer must assume the raising derivation as the only available derivation for RCs (restrictive and non-restrictives), which as known raises some non-trivial issues (pertaining to e.g. the derivation of anti-reconstruction effects or the derivation of non-restrictive relatives). While I cannot exclude that Raising could be active in the derivation of (some) RCs, I believe there are reasons to suspect that Raising cannot be at the basis of the che/cui allomorphy (the same could argued to be the case of Matching, though I leave discussion of this issue to a future occasion).

4 Addressing the Finiteness Problem

4.1 The Finiteness Problem Reframed

In the previous section I argued that che+relis a DP entering the derivation as a minimal pronoun, identical to che+q and cui in the narrow syntactic derivation up to the point of transfer. This hypothesis seemingly exacerbates the issue of why che+relis restricted to finite contexts (i.e., the finiteness problem), whereas e.g. che+q and cui are not. In particular, the question arises as to why the same wh-DP should show restriction to finiteness in some syntactic environments but not in others. I think that some light on this issue can be shed once we consider that the finiteness problem does not affect just che+rel or other relative ‘complementizers’ across languages, but generally affects DPs under the same conditions. In other words, che+rel is not very special with respect to the restriction to finiteness. As a matter of fact, one would expect the unavailability of che+rel in infinitival relatives to be the case.

As mentioned in Section 2, wh-elements are generally excluded from occurring bare in languages such as French and English. The situation of Italian mirrors the one observed in English and French; see (40)–(43), adapted from Sportiche (2011: 100–1).[20] Note that sentences such as (40) become grammatical when the wh-DP is embedded within a larger phrase (43). Note further that the preposition in an English relative must be pied-piped along with the wh-DP to the clausal edge (43c–d).

*un uomo che/il quale invitare.
*un homme qui inviter.
*a man who to invite.
*qualcosa che/la quale fare.
*quelque chose quoi faire.
*something which to do.
*il momento quando dormire.
*le moment quand dormir.
*the moment when to sleep.
un bambino di cui/del quale parlare.
un enfant de qui parler.
a child of whom to speak.
*a child whom to speak of.

The generalization that can be gleaned from (40)–(43) is that bare wh-DPs cannot sit at the edge of infinitival relatives. If we take che+rel to be a DP, we can therefore reframe the finiteness problem as the issue of what grammatical factor(s) is responsible for barring bare DPs from occurring in such positions.

This issue has extensively been debated in the literature (e.g., Chomsky 1977; Douglas 2016; Hasegawa 1998; Kayne 1976; Law 2000; Pesetsky and Torrego 2006; Sportiche 2011, among several others). To my knowledge, however, there still lacks consensus over which account is best suited to handle these facts. Of course, given its scope, this debate cannot be settled within the context of this paper. My more modest goal here is to include the finiteness problem within this larger debate, and suggest a potential solution for its treatment by extending Richards’ (2010) Distinctness Theory (DT) to the facts of Italian.

4.2 Richards’ (2010) Distinctness Theory

The gist of Richards’ (2010) DT is that the derivation of a sentence crashes if a linearization statement containing two non-distinct elements, i.e., of the form <α, α>, is generated. Crucially, the derivation crashes only if <α, α> is generated within the same Spell-Out domain ― the complement of the phase head that is sent to transfer. The derivation does not crash if two non-distinct elements belong to different Spell-Out domains, which is the case when there is an intervening phase head between the two. Moreover, Richards takes CP, v*P, PP and KP to be phases, but crucially not DP. Richards further assumes that what α amounts to may be subject to parametrization. For English, he assumes that α might simply amount to a syntactic label. Consequently, according to DT, English cannot linearize <α, α> if (i) α = X(P); and (ii) <α, α> is generated within the same Spell-Out domain.

Richards (2010: 34–8) thus argues that DT may be responsible for the ungrammatical status of infinitival relatives in English such as (42a). In particular, (44a) is barred because the determiner merged with the relative head bears the same label as the wh-pronoun and both elements belong to the same Spell-Out domain (an assumption to which we return momentarily) (cf. (44b)).

*The man whom to see.
[ DP the man [ CP [ DP whom] C [to see]]].

On the other hand, a wh-element with a pied-piped preposition is allowed to occur in an infinitival relative (45a) because the preposition is a phase head, according to Richards. This latter assumption ensures that the wh-element is part of a different Spell-Out domain than the upper determiner (45b).

The man with whom to dance.
[ DP the man [ CP [ PP [ P with [ DP whom]]] C [to dance]]].

It is crucial for this type of account that infinitival relatives do not contain an intervening phase head between the lower and the upper DP. On the other hand, this phase head must be present in tensed relatives. Richards assumes that this is the case, following Bianchi’s (1999) configurations for (non-)finite relatives (cf. also Douglas 2016). Abstracting away from the specifics of Bianchi’s (1999) analysis (which, as Richards notes, are not crucial for his account), we can sketch the structural configuration for tensed and infinitival RCs as in (46) and (47), respectively (irrelevant details omitted, the boldfaced C1 is a phase head).[21] According to (46) and (47), the wh-DP and the upper DP are separated by a phase boundary in tensed relatives (C1); hence, the two DPs are part of two different Spell-Out domains and no Distinctness effects arise. On the other hand, in infinitival relatives the upper DP and the wh-element are not separated by an intervening phase head. This may potentially result in violations of Distinctness.

[DP the man [C1P C 1 [C2P [DP who] C2 [she married]]]]
*[DP the man [C1P [DP who] C 1 [to marry]]]

The account proposed in Richards (2010) can be extended straightforwardly to the case of Italian for our purposes. Suppose that Distinctness in Italian, like in English, is concerned with labels.[22] If so, the ungrammaticality of (48) is derived from the same conditions that bar (44) in English, i.e., the linearization statement <dp, dp> is generated within the same Spell-Out domain. Likewise, the grammaticality of (49)–(50) is amenable to that of (47): the wh-DP cui/Det + qual- is part of the complement of the phase head a ‘on’ (lit. ‘to’) and is therefore in a different Spell-Out domain than the upper determiner l(o) ‘the’.

*L’uomo che vedere
‘The man to see.’
[dp l’uomo [cp [dp che] c [vedere]]].
L’uomo a cui affidarsi
‘The man to rely on.’
[dp l’uomo [cp [pp [p a [dp cui] c [affidarsi]]].
L’uomo al quale affidarsi
‘The man to rely on.’
[dp l’uomo [cp [pp [p a [dp il quale] c [affidarsi]]].

A potential problem for this analysis may be raised by a more stylistically marked use of cui, whereby cui can be licensed on its own, i.e., without a preposition, as illustrated in (51). In these cases, it may be sufficient to assume that cui is embedded under a non-overt PP/KP, as in (52).

L’uomo cui chiedere consiglio
‘The man to ask for advice’
[dp l’uomo [cp [pp [p <a> [dp cui]]] C [chiedere consiglio]]]
[dp l’uomo [cp [KP [K [dp cui]]] C [chiedere consiglio]]]

Finally, a further potential problem may be raised by infinitival relatives with bare Det + qual (cf. fn. 20). An example of this construction is given in (53) (from Cinque 1982: 282).

?Cercavo una ragazza la quale poter invitare alla cerimonia di inaugurazione
I was looking for a girl to be able to invite to the inauguration ceremony’

A solution to this problem may be obtained by assuming, with Richards (2010), that Distinctness is subject to parametric variation. Suppose that the formal register to which this (highly marked) construction belongs constitutes an I-language (relatively) distinct from Italian. If so, it can be proposed that this register has a different setting for the parameter that dictates what the properties relevant for Distinctness are. I leave it to future research, however, to determine exactly which properties these may be.

5 Conclusions

This paper discussed the restriction to finiteness and case for che+rel under the D hypothesis. I argued that such restrictions, while constituting the central argument in favor of the C hypothesis, can be accounted for assuming that che+rel is a DP, on a par with other interrogative and relative wh-elements.

More specifically, I argued that che+rel is identical to che+q and cui in the narrow syntactic derivation, until the point of transfer. After transfer, and specifically at ext, wh-DPs may potentially be assigned different phonological exponents depending on whether they have undergone Agree in the derivation. This assumption moreover accounts for why cui is specialized for the headed relative construction. In particular, the exponent /kui/ is assigned to a wh-DP that has undergone Agree in the derivation and is marked as +obl. Assuming that some form of the Elsewhere Principle holds at ext, the unavailability of che+rel with oblique case gaps is thus accounted for via the assumption that realization of /kui/, being a more specific rule, blocks realization of /ke/ in such contexts. Clearly, such language-particular analysis cannot be extended to account for the incompatibility of Ps with relativizers in others languages, like English that, where other explanations must be sought under the D hypothesis (see Kayne 2014; Seppänen 1997).

With regards to the restriction to finite contexts, I argued that this restriction is not unique to che+rel, but generally applies to bare DPs that occur at the edge of infinitival relatives. As such, the problem of the unavailability of che+rel in infinitival relatives can (and should) be treated as another instance of the larger issue of what precludes bare DPs from occurring in such positions. This problem is notoriously a difficult one to solve, especially so within current minimalist frameworks. Nonetheless, I attempted to show a potential solution for its treatment by extending Richards’ (2010) analysis of English infinitival relatives to the facts of Italian.

Corresponding author: Giuseppe Rugna, Università degli Studi di Firenze, Florence, Italy, E-mail:

Previous versions of this paper were presented virtually at the Paris Denis-Diderot University (01/03/2021), at The Nominal approach to Clausal Complementation (University of Patras, 27/05/2021), and at the Encuentro De Lingüística Formal 2 (Benemérita Universidad Autónoma de Puebla, 09/09/2021). My thanks to the audiences at those events for comments and discussion. I am also very grateful to two Probus anonymous reviewers for their comments and suggestions, which have significantly improved the quality of this paper. I am indebted to Lena Baunaz, Ludovico Franco, and Maria Rita Manzini for their comments on a previous draft, as well as to Anna Roussou, Caterina Donati, Davide Mocci, Leonardo Maria Savoia and Serena Crocchi for invaluable suggestions and discussion. All errors are my own.


Arregi, Karlos & Andrew Ira Nevins. 2012. Morphotactics: Basque auxiliaries and the structure of spellout. Dordrecht: Springer.10.1007/978-94-007-3889-8Search in Google Scholar

Arsenijević, Boban. 2009. Clausal complementation as relativization. Lingua 119. 39–50.10.1016/j.lingua.2008.08.003Search in Google Scholar

Baier, Nico. 2015. Adjective agreement in Noon: Evidence for a split theory of noun-modifier concord. In Thuy Bui & Deniz Özyldiz (eds.), Proceedings of NELS 45, vol. 1, 67–80. Amherst, MA: GLSA.10.3765/exabs.v0i0.3013Search in Google Scholar

Barbiers, Sjef, Olaf Koeneman & Marika Lekakou. 2009. Syntactic doubling and the structure of wh-chains. Journal of Linguistics 45. 1–46.10.1017/S0022226709990181Search in Google Scholar

Baunaz, Lena & Eric Lander. 2018. Syncretisms with the nominal complementizer. Studia Linguistica 72(3). 537–570.10.1111/stul.12076Search in Google Scholar

Benincà, Paola. 2010. Headless relatives in some Old Italian varieties. In Roberta D’Alessandro, Ledgeway Adam & Ian Roberts (eds.), Syntactic variation. The dialects of Italy, 55–70. Cambridge: Cambridge University Press.Search in Google Scholar

Bianchi, Valentina. 1999. Consequences of antisymmetry: Headed relative clauses. Berlin: Mouton de Gruyter.10.1515/9783110803372Search in Google Scholar

Bjorkman, Bronwyn & Hedde Zeijlstra. 2019. Checking up on (φ-)Agree. Linguistic Inquiry 50(3). 527–569.10.1162/ling_a_00319Search in Google Scholar

Bocci, Giuliano, Valentina Bianchi & Silvio Cruschina. 2021. Focus in wh-questions. Natural Language & Linguistic Theory 39. 405–455. in Google Scholar

Boef, Eefje. 2013. Doubling in relative clauses: Aspects of morphosyntactic microvariation in Dutch. Utrecht: LOT Doctoral dissertation.Search in Google Scholar

Brandt, Patrick & Eric Fuß. 2014. Most questionable pronouns: Variation between das- vs. was-relatives in German. Linguistische Berichte 239. 297–329.Search in Google Scholar

Caponigro, Ivano. 2003. Free not to ask: On the semantics of free relatives and Wh-words cross-linguistically. UCLA. Doctoral dissertation.Search in Google Scholar

Caponigro, Ivano & Maria Polinsky. 2008. Almost everything is relative in the Caucasus. In Tova Friedman & Satoshi Ito (eds.), Semantics and linguistic theory (SALT) XVIII, 158–175. Ithaca, NY: Cornell University.10.3765/salt.v18i0.2474Search in Google Scholar

Carlo Cecchetto & Caterina Donati. 2015. Subject Intervention in free relatives. In Elisa Di Domenico, Cornelia Hamann & Simona Matteini (eds.), Structures, strategies and beyond, 369–392. Amsterdam: Benjamins.10.1075/la.223.17cecSearch in Google Scholar

Cheng, Lisa Lai-Shen. 1991. On the typology of wh-questions. Cambridge, Massachusetts: MIT Doctoral dissertation.Search in Google Scholar

Chomsky, Noam, Ángel Gallego & Dennis Ott. 2019. Generative grammar and the faculty of language: Insights, questions, and challenges. Catalan Journal of Linguistics 0. 229–261.10.5565/rev/catjl.288Search in Google Scholar

Chomsky, Noam. 1977. On Wh-movement. In Peter Culicover, Thomas Wasow & Akmajian Adrian (eds.), Formal syntax, 71–132. New York: Academic Press.Search in Google Scholar

Chomsky, Noam. 1981. Lectures on government and binding: The Pisa lectures. Dordrecht: Foris Publications.Search in Google Scholar

Chomsky, Noam. 1995. The minimalist program. Cambridge, Massachusetts: MIT Press.Search in Google Scholar

Chomsky, Noam. 2001. Derivation by phase. In Michael Kenstowicz (ed.), Ken Hale: A life in language, 1–52. Cambridge, Massachusetts: MIT Press.Search in Google Scholar

Cinque, Guglielmo. 1978. La sintassi dei pronomi relativi ‘cui’ e ‘quale’ nell’italiano moderno. Rivista di Grammatica Generativa 3. 31–126.Search in Google Scholar

Cinque, Guglielmo. 1982. On the theory of relative clauses and markedness. Linguistic Review 1. 247–294.10.1017/CBO9780511554261.003Search in Google Scholar

Cinque, Guglielmo. 2008. Two types of nonrestrictive relatives. In Bonami Olivier & Patricia Cabredo Hofherr (eds.), Empirical issues in syntax and semantics, vol. 7, 99–137. Paris: CNRS.Search in Google Scholar

Collins, Chris & Erich Groat. 2018. Distinguishing copies and repetitions. Manuscript. (accessed 01 march 2021).Search in Google Scholar

Déchaine, Rose-Marie & Martina Wiltschko. 2002. Decomposing pronouns. Linguistic Inquiry 33. 409–442.10.1162/002438902760168554Search in Google Scholar

Douglas, Jamie Alexander. 2016. The syntactic structures of relativisation. Cambridge: University of Cambridge Doctoral dissertation.Search in Google Scholar

Edwards, Viv. 1993. The grammar of southern British English. In James Milroy & Lesley Milroy (eds.), Real English: The grammar of English dialects in the British Isles, 214–38. London: Longman.Search in Google Scholar

Furuya, Kaori. 2017. (Under)specification of the person feature in relative clauses. Acta Linguistica Academica 64(2). 281–311.10.1556/2062.2017.64.2.6Search in Google Scholar

Gelderen, Elly van. 2009. Renewal in the left periphery: Economy and the complementizer layer. Transactions of the Philological Society 107(2). 131–195.10.1111/j.1467-968X.2009.01216.xSearch in Google Scholar

Halle, Morris & Alec Marantz. 1993. Distributed morphology and the pieces of inflection. In Kenneth Hale & Samuel Jay Keyser (eds.), The view from building 20: Essays in linguistics in Honor of Sylvain Bromberger, 111–176. Cambridge, Massachusetts: MIT Press.Search in Google Scholar

Halle, Morris. 1997. Distributed Morphology: Impoverishment and fission. In Benjamin Bruening, Yoonjung Kang & Martha McGinnis (eds.), Papers at the interface. MIT working papers in linguistics, vol. 30, 425–449. Cambridge, Massachusetts: MITWPL.10.1075/cilt.202.07halSearch in Google Scholar

Hasegawa, Iroshi. 1998. English infinitival relatives as prepositional phrases. English Linguistics. Journal of the English Linguistics Society of Japan 15. 1–27.10.9793/elsj1984.15.1Search in Google Scholar

Heim, Irene. 1982. The semantics of definite and indefinite noun phrases. Amherst: University of Massachusetts Doctoral dissertation.Search in Google Scholar

Heim, Irene. 2008. Features on bound pronouns. In Daniel Harbour, David Adger & Susana Béjar (eds.), Phi theory: Phi-features across modules and interfaces, 35–56. Oxford: Oxford University Press.Search in Google Scholar

Heim, Irene & Angelika Kratzer. 1998. Semantics in generative grammar. Oxford: Blackwell.Search in Google Scholar

Kato, Mary & Jairo Nunes. 2009. A uniform raising analysis for standard and nonstandard relative clauses in Brazilian Portuguese. In Jairo Nunes (ed.), Minimalist essays on Brazilian Portuguese, 93–120. Amsterdam/Philadelphia: John Benjamins.10.1075/la.142.07katSearch in Google Scholar

Kayne, Richard. 1975. French syntax. Cambridge, Massachusetts: MIT Press.Search in Google Scholar

Kayne, Richard. 1976. French relative ‘que’. In Hensey Fritz & Marta Lujan (eds.), Current studies in Romance linguistics, 255–299. Washington, D.C.: Georgetown University Press.Search in Google Scholar

Kayne, Richard. 1994. The antisymmetry of syntax. Cambridge, Massachusetts: MIT Press.Search in Google Scholar

Kayne, Richard. 2014. Why isn’t “this” a complementizer? In Svenonius Peter (ed.), Functional structure from top to toe: The cartography of syntactic structures, vol. 9, 188–231. Oxford: Oxford University Press.10.1093/acprof:oso/9780199740390.003.0007Search in Google Scholar

Kayne, Richard. 2019. On complementizers and relative pronouns in Germanic vs. Romance. Manuscript. New York: New York University. (accessed 12 february 2021).10.1093/oso/9780198841166.003.0016Search in Google Scholar

Kiparsky, Paul. 1973. ‘Elsewhere’ in phonology. In Kiparsky Paul & Steven Anderson (eds.), A Festschrift for Morris Halle, 93–106. New York: Holt, Rinehart and Winston.Search in Google Scholar

Klima, Edward Stephan. 1964. Studies in diachronic transformational syntax. Cambridge, Massachusetts: Harvard University Doctoral dissertation.Search in Google Scholar

Kratzer, Angelika. 2009. Making a pronoun: Fake indexicals as windows into the properties of pronouns. Linguistic Inquiry 40(2). 187–237.10.1162/ling.2009.40.2.187Search in Google Scholar

Landau, Idan. 2015. A two-tiered theory of control. Cambridge, Massachusetts: MIT Press.10.7551/mitpress/9780262028851.001.0001Search in Google Scholar

Law, Paul. 2000. On relative clauses and the DP/PP adjunction asymmetry. In Artemis Alexiadou, Paul Law, André Meinunger & Chris Wilder (eds.), The syntax of relative clauses, 161–199. Amsterdam: John Benjamins.10.1075/la.32.05lawSearch in Google Scholar

Manzini, Maria Rita. 2012. The status of complementizers in the left periphery. In Lobke Aelbrecht, Liliane Haegeman & Rachel Nye (eds.), Main clause phenomena, 297–318. Amsterdam: Benjamins.10.1075/la.190.13manSearch in Google Scholar

Manzini, Maria Rita. 2014a. The Romance k-complementizers. In Svenonius Peter (ed.), Functional structure from top to toe the cartography of syntactic structures, vol. 9, 148–187. New York: Oxford University Press.10.1093/acprof:oso/9780199740390.003.0006Search in Google Scholar

Manzini, Maria Rita. 2014b. Grammatical categories: Strong and weak pronouns in Romance. Lingua 150. 171–201.10.1016/j.lingua.2014.07.001Search in Google Scholar

Manzini, Maria Rita & Leonardo Savoia. 2003. The nature of complementizers. Rivista di Grammatica Generativa 28. 87–110.Search in Google Scholar

Manzini, Maria Rita & Leonardo Savoia. 2005. I dialetti italiani e romanci. Morfosintassi generativa [Italian and Romansh dialects. Generative Morphosyntax], vol. 1. Alessandria: Edizioni dell’Orso.Search in Google Scholar

Manzini, Maria Rita & Leonardo Savoia. 2011. Grammatical categories. Cambridge: Cambridge University Press.10.1017/CBO9780511974489Search in Google Scholar

Nishigauchi, Taisuke. 1990. Quantification in the theory of grammar. Dordrecht: Kluwer.10.1007/978-94-009-1972-3Search in Google Scholar

Norris, Mark. 2014. A theory of nominal concord. Santa Cruz: University of California Doctoral dissertation.Search in Google Scholar

Pesetsky, David. 1987. Wh-in-situ: Movement and unselective binding. In Eric Reuland & Alice ter Meulen (eds.), The representation of (in)definiteness, 98–129. Cambridge, Massachusetts: MIT Press.Search in Google Scholar

Pesetsky, David & Esther Torrego. 2006. Probes, goals, and the nature of syntactic categories. In Yukio Otsu (ed.), Proceedings of the Seventh Tokyo Conference on Psycholinguistics, 25–60. Tokyo: Hituzi Syobo Publishing Company.Search in Google Scholar

Pesetsky, David & Esther Torrego. 2007. The syntax of valuation and the interpretability of features. In Simin Karimi, Vida Samiian & Wendy Wilkins (eds.), Phrasal and clausal architecture: Syntactic derivation and interpretation, 262–294. Amsterdam: John Benjamins.10.1075/la.101.14pesSearch in Google Scholar

Poletto, Cecilia & Emanuela Sanfelici. 2018. On relative complementizers and relative pronouns. Linguistic Variation 18(2). 265–298.10.1075/lv.16002.polSearch in Google Scholar

Poletto, Cecilia & Emanuela Sanfelici. 2019. On the relative cycle: The case of P+che relative clauses from Old to Modern Italian. In Miriam Bouzouita, Anne Breitbarth, Lieven Danckaert & Elisabeth Witzenhausen (eds.), Cycles in language change, 177–198. Oxford: Oxford University Press.10.1093/oso/9780198824961.003.0010Search in Google Scholar

Postma, Gertjan. 1994. The indefinite reading of WH. In Reineke Bok-Bennema & Crit Cremers (eds.), Linguistics in the Netherlands, 187–198. Amsterdam/Philadelphia: John Benjamins.10.1075/avt.11.19posSearch in Google Scholar

Preminger, Omer. 2011. Agreement as a fallible operation. Cambridge, Massachusetts: MIT Doctoral dissertation.Search in Google Scholar

Richards, Norvin. 2010. Uttering trees. Cambridge, Massachusetts: MIT Press.10.7551/mitpress/9780262013765.001.0001Search in Google Scholar

Rinke, Esther & Elisabeth Aßmann. 2017. The syntax of relative clauses in European Portuguese. Extending the Determiner Hypothesis of relativizers to relative que. Journal of Portuguese Linguistics 16(4). 1–26.10.5334/jpl.172Search in Google Scholar

Rizzi, Luigi. 1997. The fine structure of the left periphery. In Liliane Haegeman (ed.), Elements of grammar: Handbook of generative syntax, 281–337. Dordrecht: Kluwer.10.1007/978-94-011-5420-8_7Search in Google Scholar

Rizzi, Luigi. 2020. Che and weak islands. In Ludovico Franco & Paolo Lorusso (eds.), Linguistic variation: Structure and interpretation, 599–608. Boston/Berlin: Walter de Gruyter.10.1515/9781501505201-029Search in Google Scholar

Rizzi, Luigi & Guglielmo Cinque. 2016. Functional categories and syntactic theory. Annual Review of Linguistics 2. 139–163.10.1146/annurev-linguistics-011415-040827Search in Google Scholar

Roberts, Ian & Anna Roussou. 2003. Syntactic change: A minimalist approach to grammaticalization. Cambridge: Cambridge University.10.1017/CBO9780511486326Search in Google Scholar

Rooryck, Johan & Guido Vanden Wyngaerd. 2011. Dissolving binding theory. Oxford: Oxford University Press.10.1093/acprof:oso/9780199691326.001.0001Search in Google Scholar

Roussou, Anna. 2010. Selecting complementizers. Lingua 120. 582–603.10.1016/j.lingua.2008.08.006Search in Google Scholar

Roussou, Anna. 2020a. Complement clauses: Case and argumenthood. In Ludovico Franco & Paolo Lorusso (eds.), Linguistic variation: Structure and interpretation, 609–631. Boston/Berlin: Walter de Gruyter.10.1515/9781501505201-030Search in Google Scholar

Roussou, Anna. 2020b. Some (new) thoughts on grammaticalization: Complementizers. In András Bárány, Theresa Biberauer, Jamie Douglas & Sten Vikner (eds.), Syntactic architecture and its consequences I: Syntax inside the grammar (Open Generative Syntax 9), 91–111. Berlin: Language Science Press.Search in Google Scholar

Seppänen, Aimo. 1997. Relative that and prepositional complementation. English Language and Linguistics 1. 111–133. in Google Scholar

Sportiche, Dominique. 2011. French relative Qui. Linguistic Inquiry 42. 83–124.10.1162/LING_a_00029Search in Google Scholar

Vries, Mark de. 2002. The syntax of relativization. Utrecht: LOT Doctoral dissertation.Search in Google Scholar

Wiltschko, Martina. 1998. On the syntax and semantics of (relative) pronouns and determiners. Journal of Comparative Germanic Linguistics 2. 143–181.10.1023/A:1009719229992Search in Google Scholar

Zeijlstra, Hedde. 2012. There is only one way to agree. The Linguistic Review 29. 491–539.10.1515/tlr-2012-0017Search in Google Scholar

Published Online: 2022-05-10

© 2022 Giuseppe Rugna, published by De Gruyter, Berlin/Boston

This work is licensed under the Creative Commons Attribution 4.0 International License.