On the syntax and pragmatics of the ‘why + not + XP’ construction: a cartographic approach

2.1 The ‘why + not + DP’ construction, Sluicing and Stripping

A careful observation indicates that the ‘why + not + DP’ construction is not derived from Sluicing and Stripping and that the three constructions have different syntactic representations and properties.

First, in a typical example of Sluicing in English, there is no restriction on the type of wh-phrases that can occur in the structure, as shown in (6a); however, there is a strict restriction on the type of wh-phrases that can appear in the ‘why + not + DP’ construction; only the wh-phrase why (or how come) is acceptable while other wh-phrases are excluded, as in (6b).

Second, only matrix clauses can accommodate Stripping, while embedded clauses cannot, as in (7a) and (7b); however, the ‘why + not + DP’ construction can not only occur in matrix clauses but also in embedded clauses, as in (7c) and (7d).

In addition to these differences, the remnant in the ‘why + not + DP’ construction can be contrasted with a correlate in the antecedent clause or repeated from the antecedent clause, as in (8a) and (8b); in contrast, the remnant in Stripping can only be contrasted with a correlate in the antecedent clause, as in (8c), and the remnant in why-Stripping is repeated from the antecedent clause, as in (8d).

In this respect, the ‘why + not + DP’ construction behaves like Stripping and why-Stripping in terms of the properties of the remnant,^[1] but the ‘why + not + DP’ construction differs from Sluicing, Stripping and why-Stripping in one way or another and it is not a collective consequence of the three constructions.

2.2 The ‘why + not + DP’ construction and Sluice stripping

Another construction which resembles the ‘why + not + DP’ construction is called Sluice stripping (Nevins 2008), the derivation of which involves a wh-remnant and a non-wh-remnant, followed by a subsequent deletion of other constituents at PF. A typical example of Sluice stripping is shown in (9).

According to Nevins, the derivation of (9) involves the leftward movement of the wh-phrase who to [Spec, CP₁] and the rightward movement of the PP about phonology out of the TP containing it, followed by an adjunction of the PP to the CP₁ and a deletion of the TP at PF.

It appears as if the ‘why + not + DP’ construction and Sluice stripping share common properties in that both constructions contain a wh-remnant and a non-wh-remnant and both involve a deletion operation at PF. However, the two constructions differ in at least two aspects.

First, other wh-phrases are acceptable in Sluice stripping apart from the wh-phrase why, as in (10a), but the ‘why + not + DP’ construction can only accommodate the wh-phrase why (or how come) while other wh-phrases are not acceptable, as in (10b).

Second, Nevins proposed that Sluice stripping involves a rightward movement of the non-wh-remnant. One of the reviewers has rightly pointed out that, according to Rochemont (1986) and Kim (1997), focus movement of a constituent to the right is possible in English. Accordingly, it is appealing to assume that the DP also undergoes a rightward movement in the ‘why + not + DP’ construction. However, the rightward movement of a focused constituent is based on the assumption that the specifier of FocP is final (Kim 1997). In other words, the specifier is to the right of the head of FocP and the complement is to the left of the head. Besides, in Kim’s analysis of Sluicing in English, both CP and TP are head-initial which makes it difficult to explain why only the FocP is head-final while other phrases are head-initial.

In addition, a heavy DP can undergo rightward raising out of the TP in which it is contained (heavy DP shift), as in (11b). In contrast, it sounds weird for a light DP or a pronoun to do the same, as in (12b) and (12d). However, in the ‘Why + not + DP’ construction, both the heavy DP and the light DP can occur, as in (13).

Accordingly, I assume that the DP in the ‘why + not + DP’ construction involves a leftward movement.

2.3 An interim summary

After presenting the basic properties of the ‘why + not + DP’ construction, I argue that it is not derived from Sluicing, Stripping and Sluice stripping because it imposes strict restrictions on the type of wh-phrases and it has its unique properties and representations. Besides, I suggest that there is a syntax-pragmatics interface in the ‘why + not + DP’ construction which connects syntactic structure and the speech act it performs. The next section discusses the syntactic derivation and the syntax-pragmatics interface of the ‘why + not + DP’ construction and then the analysis is extended to similar constructions with PP, VP, AP, AdvP and even CP remnant as well as the construction with ‘why not’ only.

3.1 The theoretical framework

Cartography is a research program within the Government and Binding framework of syntactic theory. According to Cinque and Rizzi (2010), the purpose of the Cartographic approach is to draw maps of syntactic structures as precise and detailed as possible. Its research projects involve the study of functional categories, their content, number and order. Although the concept of Cartography was first proposed by Cinque (1999), Rizzi (1997) was the first explicitly cartographic study of the complementizer space, or the clausal left periphery (Shlonsky 2010). Rizzi (1997) proposed that the functions of complementizers can be divided into two systems, namely the force-finiteness system and the topic-focus system, with the former being the essential part while the latter being present in a structure only if needed. Therefore, the linear order of these two systems is as follows:

In addition, the speech act that is performed by making an utterance and the meaning which the utterance conveys are connected. In other words, syntax provides the basic framework that makes possible the performance of a speech act and the conveyance of meaning. Although in recent theories of syntax, topics related to speech act have been left to the study of semantics and pragmatics (Haegeman and Hill 2013; Hill 2007; Speas and Tenny 2003; Tenny 2006; Wiltschko and Heim 2016; Zu 2018), the research findings have contributed to the study of syntacticization of discourse. Performing a speech act involves both the addressor and the addressee. In the syntacticization of discourse, there should be a representation of the addressor and the addressee in syntax and this representation is structurally located above the actual utterance.

3.2 The syntax-pragmatics interface

On the basis of analyses of Miyagawa (2010, 2012, 2017, 2022), Oguro (2015), Ross (1970), Speas and Tenny (2003), Wiltschko (2014, 2017), it has been suggested that syntax encodes information that helps realize the performance of speech act. Therefore, I propose that a syntactic structure consists of a representation of the addressor and the addressee (Addressor Addressee Phrase, AAP) high above in syntax; there is a response layer (Response Phrase, RespP) below and a proposition is realized in the CP layer. The AAP hosts the representation of the addressor and the addressee; the RespP represents the addressor’s or the addressee’s response to a discourse, such as making a suggestion, seeking further information, making a complaint or taking an offer. It is the AAP and RespP that function as the locus of the illocutionary force of an utterance. In other words, they serve as the syntax-pragmatics interface and help realize the performance of a speech act. Oguro (2017) has suggested that interrogative sentences can be divided into plain conjectual questions, polite conjectual questions and information-seeking questions, with the latter involving both the addressor and the addressee. The addressee is expected to respond to the addressor in a linguistic communication. Accordingly, in the ‘why + not + DP’ construction, the addressor raises a constituent question and performs the speech act of seeking further information, making a suggestion or accepting an offer.

The syntax-pragmatics interface can also be applied to why-Stripping (Kim et al. 2021; Nakao et al. 2013; Ortega-Santos et al. 2014; Weir 2014; Yoshida et al. 2015, etc.) in that this construction also involves both the addressor and the addressee and that the addressor raises a question and performs a speech act of seeking further information from the addressee.

In the spirit of the Cartographic approach, CP can be further divided into Force Phrase (ForceP), Focus Phrase (FocP) and Finiteness Phrase (FinP). I follow Hofmann (2018), Ladusaw (1992), Laka (1990) and Potts (2002) in claiming that the negative marker not can project a polarity phrase (ΣP) above TP. This ΣP is headed by the negative marker not and it is sandwiched between ForceP and FocP. Therefore, the linear ordering of phrases relevant to the discussion can be represented as follows:

It is worth pointing out that when comparing examples (14) and (15) it is noticeable that the TopicP layer is missing. The reason is that TopicP is present in the structure only if needed (Rizzi 1997). According to my assumption, the hierarchical structure does not involve any topic, hence the absence of TopicP.

3.3 The overall structural computation

I agree with Yoshida et al. (2015) that the wh-phrase why is base-generated, specifically in [Spec, ForceP], thus typing the whole clause as being an interrogative (Cheng 1991). In addition, I suggest that the wh-phrase why is connected with the antecedent clause which serves as a presupposition for the former. Without an antecedent clause, the ‘why + not + DP’ construction will be ungrammatical, as the contrast shows in (16a) and (16b).

In the meantime, the wh-phrase why is also focus-associated with a constituent within its c-commanding domain. On the basis of Hofmann (2018), Ladusaw (1992), Laka (1990) and Potts (2002), I assume that the polarity phrase (ΣP) is headed by the negative marker not and it is sandwiched between ForceP and FocP, the idea of which is different from Merchant (2006a) and Weir (2014), which state that the negative marker not is directly adjoined to the right of the wh-phrase why. It has to be pointed out that the ‘why + not + DP’ construction introduces sentential negation rather than constituent negation. When the DP serves as a complement to the predicate verb, it can stay in-situ. Following Szendröi (2001), I assume that the conceptual-intentional interface can access both PF and LF and that both the syntactic and the prosodic representations feed into the conceptual-intentional system. Accordingly, the in-situ DP can be regarded as an information focus at PF (Kiss 1998; Kim 1997; Rochemont 1986), as represented by the DP a sofa in (17):

Although the DP a sofa has received an interpretation of information focus in its base position, (17) is not the structure under discussion. Whichever constituent is deleted at PF, be it a VP, TP or FocP, the DP a sofa will also be deleted, making it impossible to generate the ‘why + not + DP’ construction. On the basis of Gärtner (2002), I suggest that the DP in the ‘why + not + DP’ construction has undergone covert movement to [Spec, FocP]. Besides, when the lower copy, which is supposed to be pronounced, is contained within an ellipsis domain, ellipsis forces the highest copy to be pronounced instead due to a requirement that focus be overtly realized. Regarding the syntax-pragmatics interface, the wh-phrase why is then moved from [Spec, ForceP] to [Spec, RespP] where it can be focus-associated with a constituent in its c-commanding domain. The RespP further projects into AAP which relates both the addressor and the addressee with the discourse. The APP and the RespP serve as the syntax-pragmatics interface, performing a speech act of seeking further information or accepting an offer. Finally, as Focus Condition on Ellipsis (Merchant 2001) is satisfied, FinP is deleted at PF, thus deriving a well-formed ‘why + not + DP’ construction.

3.4 Evidence for full-fledged syntax at the ellipsis site and for the movement of the DP out of the ellipsis site

In the ‘why + not + DP’ construction, I argue that the DP undergoes covert movement. Specifically, the DP has moved to [Spec, FocP] and such a movement is triggered by the weak [+Focus] feature of the head of FocP.

Given that the DP in the ‘why + not + DP’ construction involves covert movement, such an operation shares syntactic properties with the remnant wh-phrase in English Sluicing, as the latter also undergoes movement according to the PF deletion analysis (Kim 1997; Lasnik 2001; Merchant 2001, 2004, 2006b, 2008, 2010, 2013, 2015, 2016; Ross 1969, etc.). Merchant (2001) argued that the wh-phrase in English Sluicing undergoes leftward movement from within TP to [Spec, CP], followed by a subsequent deletion of the complement of C, namely TP, at PF, thus stranding the wh-phrase. The remnant wh-phrase is closely-related to the correlate in the antecedent clause, showcasing the connectivity effect, such as the Case effect, the Binding effect and the Preposition-omission effect. Therefore, I assume that the DP in the ‘why + not + DP’ construction also displays connectivity effect with its correlate. Specifically, in this section, the Case effect is presented to provide evidence for full-fledged syntactic structure at the ellipsis site; the Binding effect and the Preposition-omission effect are presented to provide evidence for the movement of the DP out of the ellipsis site.

3.5 Syntactic functions of ‘why’ and ‘not’

The wh-phrase why and the negative marker not share one common feature, that is, both can be associated with a constituent in their c-commanding domain. Bromberger (1992) has pointed out that the wh-phrase why in an interrogative sentence can be focus-associated with any constituent in its domain. Specifically, the answer to an interrogative sentence is determined by which constituent is focus-associated with the wh-phrase why.

In (26), when the subject Lucy receives focus, a possible answer might be “Because Lucy is the only person who does not have a pencil”; when the object a pencil receives focus, we might respond “Because Lucy wants to draw a picture of the house with it”; likewise, when the VP borrow a pencil receives focus, a possible answer might be “Because Lucy does not want to buy a new one”.

However, wh-phrases other than why in an interrogative sentence cannot be focus-associated with a constituent in their c-commanding domain.

In (27), the answer to the question can only be about the place, the manner or the time about the event of Lucy borrowing a pencil. Therefore, only the wh-phrase why but not other wh-phrases can be focus-associated with a constituent in its c-commanding domain.

In a parallel fashion, the negative marker not also has the property of being associated with a constituent in its c-commanding domain and making it the target of negation.

According to the VP-internal subject hypothesis (Fukui and Speas 1986; Kuroda 1988; Koopman and Sportiche 1985, etc.), all subjects originate in VP and are moved to their surface position. Therefore, the subject Mary originates in VP and is moved to [Spec, TP], as in (28b). According to Haegeman (1995), the negative marker not is base-generated in [Spec, NegP] and the NegP is located in front of VP, thus taking negation scope over VP which contains the base-generated subject Mary. In (28), when the subject Mary receives focus, the sentence states the fact that it is not Mary who borrowed the book from the library; when the object the book receives focus, the sentence might mean it isn’t a book that Mary borrowed from the library; when the adjunct from the library receives focus, the sentence might mean the place Mary borrowed the book is not from the library and she might have borrowed it from one of her friends.

As both the wh-phrase why and the negative marker not share the property of being associated with a constituent in their c-commanding domain, both can be associated with a DP in the ‘why + not + DP’ construction.

3.6 Syntactic positions of ‘why’ and ‘not’

As English is a wh-movement language, wh-phrases, except in echoic sentences, undergo overt movement to the front of a sentence in order to type the clause as being an interrogative, as in (29).

The wh-phrase why is no exception, which means why is usually involved in overt movement and this helps explain the ambiguity of the following sentence.

(30) is ambiguous in that the wh-phrase why can be associated with both the matrix clause and the embedded clause. Specifically, the matrix clause interpretation is seeking for the reason why it is Mary who said that Lucy bought a book; the embedded clause interpretation is asking for the reason why Lucy bought a book from Mary’s narrative. The ambiguity has to do with the overt movement of the wh-phrase why which is moved from the embedded clause to the matrix clause.

Although the wh-phrase why can undergo overt movement in English, Ko (2005) and Yoshida et al. (2015) have proposed that the wh-phrase why can be base-generated in [Spec, CP]. I follow their idea in suggesting that the wh-phrase why is base-generated in [Spec, ForceP] in the ‘why + not + DP’ construction and this is supported by the following evidence.

First, the wh-phrase why in the ‘why + not + DP’ construction can not only be focus-associated with a constituent in the matrix clause but also with the one in the embedded clause.

As indicated in (31a) and (31c), the focus association of the wh-phrase why with the DP an apple in the embedded clause shows that the relation between them is not mediated by movement of the wh-phrase why from around the DP. Besides, Merchant (2001) has pointed out that long-distance why-Sluicing cannot escape a finite clause, as shown in (32).

The ungrammaticality of (32) shows a sharp contrast with (31c) in that the wh-phrase why in (31c) does not show such restriction, even though it is focus-associated with the DP in the embedded clause. Therefore, it is reasonable that the wh-phrase why does not move and it is base-generated in [Spec, ForceP].

Second, Collins (1991) stated that the wh-phrase why in sentences with quantifier phrases results in ambiguity.

In (33a), when the wh-phrase why is within the domain of the quantifier phrase everyone, the sentence can be interpreted as “for each person, there exists a reason, such that each person loves Mary for that reason”; when the quantifier phrase everyone is within the domain of the wh-phrase why, the sentence can be interpreted as “there exists one mutual reason, such that for all the people, they love Mary for that mutual reason”. It is worth noting that when the DP Mary receives focus and is focus-associated with the wh-phrase why, the ambiguity disappears, as in (33b), which can only be interpreted as “there exists one mutual reason, such that for all the people, they love Mary for that mutual reason”.

A similar case also appears in the ‘why + not + DP’ construction, as in (34).

If the assumption is on the right track, that is, the wh-phrase why is base-generated in [Spec, ForceP] which places it in a c-commanding position over the quantifier phrase everyone, then (34) can only be interpreted as “everyone loves Mary, but why is it not the case that everyone loves Lucy”.

Regarding the syntactic position of the negative marker not, Haegeman (1995) claimed that Negation Phrase (NegP) is a maximal projection headed by a phonetically or morphologically null negative element and that NegP is situated between TP and VP. The negative marker not is generated in [Spec, NegP] and is capable of making the following constituents, such as VP, PP or DP, a target for negation.

As the ΣP headed by the negative marker not is sandwiched between ForceP and FocP, there are two negative words in the sentence and this seems to result in a double negation problem,^[2] as in (35a). However, it turns out that the double negation problem does not exist and the sentence still has a sentential negation interpretation, as in (35b).

The fact that (35a) and (35b) are semantically identical is surprising because two negative expressions would normally lead to a positive interpretation. On the basis of Hofmann (2018), I argue that the negative marker not in the ΣP is closely related to the negative marker not in NegP, with the former being uninterpretable and the latter interpretable. Being uninterpretable, the negative marker not in the ΣP serves as a probe and searches in its c-commanding domain for a goal which can value its uninterpretable feature. The negative marker not in NegP, being interpretable, serves as an appropriate goal and values the uninterpretable feature on the probe. Ladusaw (1992) has suggested that negative expressions can be semantically vacuous in the licensing context of an instance of sentential negation. Accordingly, I assume that it is the uninterpretable negative marker that is semantically vacuous and that its uninterpretable feature is licensed by the semantically interpretable feature on the goal. As the connection between the negative marker not in the ΣP and the one in NegP is established, a sentential negation can be introduced. Let’s take (36) as an example.

(36a) and (36b) have the same semantic interpretation even though in the second conjunct there are two negative markers in the former while there is only one in the latter. According to my assumption, the higher negative marker not carries an uninterpretable feature and the lower negative marker not carries an interpretable feature. The higher not probes in its c-commanding domain for a goal. The lower not serves as an appropriate goal and the probe-goal agreement relation is established. As the uninterpretable feature is valued, the higher not becomes semantically vacuous and it is the lower not that enters into the semantic interpretation which introduces a sentential negation. Finally, the FinP which contains the interpretable lower not is deleted at PF, thus deriving (36a) which is semantically identical to (36b).

Having presented the syntactic positions of the wh-phrase why and the negative marker not, it is necessary to discuss which constituent is deleted and what the motivation is in the derivation of the ‘why + not + DP’ construction.

3.7 Deletion and motivation

Auxiliaries normally undergo T-to-C movement in matrix wh-questions in English, as shown in (37).

However, both the auxiliary and TP are prohibited from occurring in the ‘why + not + DP’ construction, as in (38a), and the reconstruction of the auxiliary and TP in the derived sentence results in its ungrammaticality as in (38b) and (38c).

The ungrammaticality of (38b) and (38c) can be explained by the assumption that the auxiliary in the ‘why + not + DP’ construction has undergone overt head movement from T to Fin and it is subsequently deleted at PF along with FinP, as in (39).

The question that needs to be answered is why FinP is deleted at PF and what triggers the motivation for deletion. Generally speaking, it is possible for a DP to stay in-situ in a sentence, as the DP a pear in (40a). If the DP a pear has undergone movement to the left periphery of TP and the ensuing constituents are maintained, the derived sentence will become ill-formed, as in (40b). The ill-formedness can be remedied only when all its ensuing constituents are deleted at PF, as in (40c).

In the derivation of the ‘why + not + DP’ construction, what is deleted is FinP and this deletion satisfies the Focus Condition on Ellipsis (Merchant 2001) which states that an XP α can be deleted only if α is e-given. The concept of e-givenness (Schwarzschild 1999) states that an expression E is e-given if and only if there is an antecedent A which entails E and which is entailed by E, modulo Ǝ-type-shifting.

As the wh-phrase why is base-generated in [Spec, ForceP] and it is closely related to the antecedent clause and the constituent in its c-commanding domain, when the antecedent clause has a positive interpretation, the DP in the second conjunct shows a contrast to its correlate, as in (41a); when the antecedent clause has a negative interpretation, the DP is a repetition of its correlate, as in (41b).

Since the negative marker not is base-generated in the ΣP which is higher up in the hierarchy than TP in the second conjunct, when the antecedent clause has a positive interpretation and there is no NegP within TP in the second conjunct, the negative marker not in the ΣP is interpretable. It enters into the semantic interpretation of the second conjunct and introduces a sentential negation, as in (42a); when the antecedent clause has a negative interpretation and there is a NegP within TP in the second conjunct, the negative marker not in the ΣP is uninterpretable and its feature is valued via agreement with the interpretable NegP goal. Accordingly, the higher not is licensed to become semantically vacuous (Ladusaw 1992) and the lower not enters into the semantic interpretation and introduces a sentential negation, as in (42b).

Regarding the spell-out of the DP in the second conjunct, Yoshida et al. (2015), in their analysis of why-Stripping, proposed that the focused phrase undergoes overt movement because ellipsis forces the movement of the remnant. In other words, if the focused phrase stays in-situ, it will also be deleted at PF. This is shown in the contrast of (43a) and (43b).

Bobaljik (1995) and Gärtner (2002) have suggested that if the higher copy of a given term is pronounced then we have the effect of overt movement and if the lower copy is pronounced then we have the effect of covert movement. However, I argue that the focused DP in the ‘why + not + DP’ construction undergoes covert movement from its base position to [Spec, FocP]. As the lower copy is contained within the ellipsis domain, as in (44a), ellipsis forces the highest copy to be pronounced due to a requirement that focus be overtly realized, as in (44b).

Having presented the role played by the wh-phrase why, the nature of the negative marker not and the spell-out of the DP, let’s come back and focus on the deletion and motivation.

For ease of exposition, I use A to represent the first part and E to represent the second part. Through Ǝ-type shifting, the focus-closure of A can be described as ‘Focus-closure (A) = Ǝx. Mary ate x’ which means there exists an x and that x refers to a fruit. Since the wh-phrase why serves as a bridge connecting the antecedent clause with the second part, the latter has a similar underlying structure ‘Mary ate a pear’. Accordingly, the focus-closure of E can be expressed as ‘Focus-closure (E) = Ǝx. Mary ate x’ which also means there exists an x and that x refers to a fruit. Given that A entails the focus-closure of E and E entails the focus-closure of A, the Focus Condition on Ellipsis is satisfied and FinP in the second part is licensed to be deleted at PF. Finally, the wh-phrase why moves from [Spec, ForceP] to [Spec, RespP] in order to convey the addressor’s response to a previous discourse.^[3]

The derivation of the ‘why + not + DP’ construction does not violate relevant syntactic constraints and it conforms to the economy principle in the process of syntactic computations, namely, ‘there can be no superfluous symbols in representations or superfluous steps in derivations (Chomsky 1995:92)’.

3.8 Analysis of other remnants in the ‘why + not + XP’ construction

If the analysis is on the right track, in addition to DPs, the ‘why + not + XP’ construction can also accommodate other types of phrases as their remnants, such as PPs, VPs, APs, AdvPs and CPs. Besides, it is also possible to have a construction where there is no constituent following the combination why not.