overt wh-fronting

. This study reexamines the syntactic encoding of information structure embodied by Mandarin overt wh-fronting questions. The sentence patterns that this paper is concerned with are wh-questions containing one or more fronted wh-phrases surfacing in (i) a clause-initial position (in root or non-root contexts) or (ii) a position immediately following a topicalized subject (also in root or non-root contexts). Departing from previous literature that obscures the exhaustifying effect exerted by a clause-initial shi ‘be’, I propose a more fine-grained classification of the focus interpretations of this type of question: a bare wh-fronting question coerces a plain (non-exhaustive) contrastive focus or a mirative focus reading (when the wh-phrase is prosodically marked) of the wh-variable, and shi -marked wh-fronting questions are shown to enforce an exhaustive focus in the answer. These three types of focus-associated interpretations are treated as conventional implicatures following Bianchi et al. (2015).

Preposing the wh-phrase has been argued to serve the discourse function of focalizing the wh-variable (Hoh & Chiang 1990, Cheung 2008, Pan 2014) or topicalizing it (Xu & Langendoen 1985;Wu 1999;Yuan & Dugarova 2012).Cheung (2008) argues that such questions pattern with English clefted questions, regardless of the presence or absence of a clause-initial shi 'be'.Further, Pan (2014) claims that the presence of copula shi 'be' to the immediate left of the fronted wh-phrase is obligatory for a wh-fronting question to receive a cleft focus interpretation, while those unmarked by shi are treated as wh-topicalizations.This contrast is shown in (2).
COP which-one-CLF film Zhangsan most NEG like see 'Which movie is it that Zhangsan does not like at all?' (adapted from Pan (2014)) However, since (2b) is a clefted question that encodes an exhaustive focus, as indicated by the gloss, it remains unclear whether shi is a focus marker or an exhaustifying operator (Klinedinst & Rothschild 2011) like the English only that marks a specific type of focus.Presupposition failure, a characteristic teasing apart Mandarin wh-in-situ questions and whfronting questions in terms of their discourse requirements, has been mentioned by Wu (1999) and Cheung (2008).
(3) a.In For a wh-fronting question to be felicitous, three conditions must be met, as (3b) shows: Zhangsan must have eaten something; there must exist a pre-established set of alternatives that Zhangsan has potentially eaten in the context known to each interlocutor; and there must be at least one alternative in the set that was not eaten by Zhangsan.A2 is excluded in (3b) simply because the denial of the presupposition runs counter to the first condition.In contrast, since (3aQ), a default in-situ wh-question, can be asked out of the blue and does not presuppose the existence of the event expressed by the main predicate or a particular set of alternatives, A2 is absolutely congruent to address (3aQ).
The topicalization and focus approaches to Mandarin overt wh-fronting attempt to reveal the interpretive distinction between this construction and an in-situ counterpart.However, they tacitly reach a consensus on the view that forming a question with a fronted wh-phrase presupposes a pre-established closed set, of which one of its members is taken to supply a value to the wh-variable.These two analyses both relate the discourse-given set of alternatives quantified over by a fronted wh-phrase to the notion of 'D-linking' (Pesetsky 1987), but they vary in terms of their treatments of this notion.According to Wu (1999), the restricted set corresponds to some 'old information' introduced prior to the utterance of the question, though it is not clear why the answer to the question is already known by the speaker, but they are still requesting it.The displacement of the wh-phrase is hence to check the topic feature of the Top head in the left periphery (LP).Cheung (2008), on the other hand, adopts Kiss's (1997) definition of contrastive focus3 and demonstrates that Mandarin overt wh-fronting encodes a contrastive focus, the domain of which also ranges over a closed set of discourse-given alternatives.
In what follows, I will vindicate the focus approach to Mandarin overt wh-fronting with three diagnostics for focus: contrast, mirativity, and exhaustivity (van der Wal 2016), proposing that this syntactic reordering is specifically for marking these three types of focus as opposed to the new information focus (NIF)4 expressed by a wh-in-situ question.Before doing this, in section 2, I will first introduce the two designated positions where the displaced wh-phrases could surface.In section 3, I will demonstrate the syntactic nature of Mandarin overt wh-fronting by examining its behavior regarding locality.In section 4, a taxonomy of the focus interpretations of overt wh-fronting questions will be proposed.Section 5 will conclude this paper.

EX-SITU SENTENCE-INITIAL POSITION.
Mandarin overt wh-fronting prototypically targets the matrix LP.Questions formed in this way can be subcategorized into bare wh-initial questions, shi-marked 'wh-initial' questions, multiple questions with one wh-fronting, and multiple questions with multiple wh-frontings, as is shown in (4a-d).
(4) a. Bare wh-fronting question [CP[Shei] with who 3SG at-all-times NEG discuss EXP One possible translation: 'Which of the topics and with whom has he never discussed?'Mandarin overt wh-fronting is also able to target the LP of the embedded domain (see (5a)).This type of movement is referred to as 'partial movement' by Abels (2012) where the whphrase lands in an intermediate interrogative Spec-CP, as opposed to the cases of 'full movement' where the wh-phrase passes through all the intermediate landing sites and terminates at the matrix LP as in (5b).
(5) a. [IP Zhangsan xiang-zhidao [CP [na-suo xuexiao] Where a fronted wh-phrase can alternate with a topicalized subject in the sentence-initial position, I refer to 'pseudo' ex-situ clause-medial wh-structures.As such, when a subject immediately precedes a fronted wh-phrase landing in Spec-FocP, it has actually undergone topicalization to Spec-TopP, which is hierarchically higher than the fronted wh-phrase.This also applies equally to root or non-root domains (see ( 6)). ( 6 The topicality of the subject is verified by the failure of substituting it with anti-topic items (ATI, cf.Tomioka 2007) such as NPIs, quantificational NPs, or disjunctive NPs; consider (7ac).The ungrammaticality of (7d) corroborates that the subject preceding the displaced whphrase in an embedded clause is also a topic.

Locality.
In this section, I investigate the syntactic behavior of Mandarin overt wh-fronting regarding island constraints 6 , binding connectivity, and the Superiority effects.I argue that for wharg -fronting questions, island sensitivity mainly correlates with two factors: 1) the presence or absence of a resumptive pronoun taking up the gap; 2) whether the predicate that the wh-phrase is moved from is episodic or non-episodic7 .Wh-adjuncts are generally banned from extracting out of strong islands.In terms of the binding relations, the fronted wh-phrase reconstructs to its original position in short-distance questions for binding or to somewhere on the movement path above the embedded domain in long-distance questions.Multiple whfronting questions are shown to exhibit no Superiority effects.
3.1.ISLAND EFFECTS.It appears that Mandarin offers evidence for island effects from both overt and covert (or so-called wh-in-situ) wh-fronting questions.Specifically, we can observe similar argument-adjunct asymmetries in overt wh-fronting questions to those which Huang (1982) observed in wh-in-situ questions where the wh-phrase is treated as being covertly displaced to Spec-CP in order to obtain the quantificational force.Overtly extracting wharguments out of islands is allowed under specific conditions, while overt whadjunct-fronting is generally degraded.To illustrate, island effects are obviated in overt wharg-fronting questions as long as the wh-phrase is resumed by a pronoun when the predicate in the island encodes an episodic eventuality; when the predicate is non-episodic (such as psych verbs), the construction is always immune to island effects.This is exemplified by ( 8 Overtly extracting a wh-adjunct out of islands is banned regardless of the episodicality of the predicate.Crucially, the locality constraint on whadjunct-fronting does not give rise to the ungrammaticality of the sentence in narrow syntax but renders a failure for the extracted whadjunct to receive an embedded scope reading (see also Cheung (2008) We also find that the overtly displaced wh-phrase behaves as though it has reconstructed (to its canonical in-situ position) for the purpose of binding in a short-distance question like (10a).The interpretation of the reflexive pronoun ziji 'self' within the displaced phrase is referentially determined by Zhangsan, which is situated in a position that c-commands the unpronounced copy of ziji 'self' embedded in the unpronounced copy of the whole wh-phrase.In a long-distance question like (10b), however, the interpretation of the reflexive within the displaced wh-phrase cannot be determined either by referring to a backgrounded person that is not mentioned in the question at all or by reconstructing to its canonical position and referring to Lisi, which c-commands the gap.Instead, since the reflexive is most naturally construed as co-referring with laoshi 'teacher', the only possibility would be that it has reconstructed to a position along the displacement path lower than the subject of the main clause but higher than the embedded clause.Lisi read PFV One possible translation: 'Which book of the teacher himself made the teacher wonder that Lisi had read it?'3.3.SUPERIORITY.Since in Mandarin, more than one wh-phrase can be overtly fronted, we are interested in the question of whether the Superiority effects (Chomsky 1973) observed in questions derived by multiple wh-movements arise in Mandarin wh-questions featuring multiple wh-frontings.It turns out that Mandarin allows for superiority-obeying and superiority-violating multiple wh-frontings in both short-distance and long-distance questions, as this syntactic displacement is not clause-bounded (see (11a-b)).( 11 It can be concluded from the above observations that Mandarin overt wh-fronting questions involve syntactic displacement, and that syntactic reordering appears to entail a series of filler-gap dependencies: it is island-sensitive, and the wh-phrase reconstructs for binding purposes.Noticeably, the overt wh-fronting in Mandarin bears a close resemblance to the focus movement in Serbo-Croatian and Bulgarian multiple wh-questions proposed by Bošković (2002).This focus movement analysis solves the problem of the selective Superiority effects as it can be seen that only wh-phrases undergoing wh-movement are sensitive to Superiority while those undergoing focus movement are not.

Overt wh-fronting as focus movement.
In this section, I adopt the focus approach to Mandarin overt wh-fronting (Cheung 2008, among others) and propose a more fine-grained classification of the various focus interpretations of wh-fronting: mirative focus, plain (nonexhaustive) contrastive focus, and (shi-marked) exhaustive focus.The three types of focus vary in terms of the degree of contrast between the focused constituent and other discoursedependent alternatives.The more prominent or exact the alternatives are in a given context, the higher the degree of contrast.The degree of contrast, therefore, lowers from mirative focus to plain contrastive focus to exhaustive focus (Molnár 2002;Cruschina 2021).Further, I argue that these three focus interpretations are conventional implicatures associated with the wh-focus (see also Bianchi et al 2015 andCruschina 2021); thus, they are encoded by the fronting strategy and cannot be canceled by the speaker per se.

MIRATIVITY.
A mirative focus appears in contexts where the focused part of the utterance is taken to exceed the speaker's expectations.It also refers to information that is beyond the speaker's knowledge (DeLancey 1997;Skopeteas & Fanselow 2011).Aikhenvald (2012) extends the notion of mirativity to a set of values associated with the interlocutors or people involved in the conversation, such as 'sudden discovery, revelation or realization of information', 'surprise', 'unprepared mind', 'counter-expectation', and 'the newsworthiness of information'.In Rooth's (1992) alternative semantics terms, the mirative import can be paraphrased as 'there is at least one alternative in the context which is more likely than the uttered focus to make the presupposition true'.On the other hand, mirativity can also be expressed out of the blue, namely purely based on the interlocutors' background knowledge.In this case, the unexpectedness comes from 'contrast against expectation' (Cruschina 2011:119) instead of contrast against context-salient alternatives.
Mirativity is firstly singled out as a linguistic notion by DeLancey (1997), based on Slobin & Aksu's (1983) discovery of a Turkish verb suffix -miş applicable in various contexts denoting that the speaker is not a 'direct or fully conscious participant' of an event.This verbal suffix can either mark indirect evidentiality as (12a-b) appears to illustrate, or direct evidentiality conveying mirative construals as in (12c).(c) Surprise: The speaker hears someone approach, opens the door, and sees Kemal, a totally unexpected visitor.
Analogously, the contexts which license Mandarin wh-fronting constructions encompass the three situations where a Turkish mirative marker is adopted.If ( 13) is a question that Lisi asked Zhangsan, for example, the overt wh-fronting can be used to show the unexpectedness, from the speaker's perspective, of Wangwu's presence, even though this attitude is evoked by some inferential evidence, as (13a) makes clear.It can also be used to express that the speaker is not fully convinced by the hearsay, as (13b) shows.Aside from these cases exemplifying indirect evidentiality, the marked construction is also able to show the speaker's surprise at Wangwu's presence through direct observation, as (13c) explicates.A wh-fronting question raised in contexts like (13b-c) parallels with an echo question: it is a question that a speaker would raise without expecting an answer but instead to express their surprise at something previously known to them.Importantly, as has been mentioned previously, native speakers tend to put stress on the displaced wh-phrase in wh-fronting questions with a mirative import.
(13) Mandarin [Shei]i ni ganggang yu-dao le ti?/Ni [shei]i ganggang yu-dao le ti ?who 2SG just-now meet-arrive PFV 2SG who just-now meet-arrive PFV 'WHO did you meet just now?' (a) Inference: The speaker saw Zhangsan coming back from a lecture with Wangwu's coat in his hands, but the speaker did not see Zhangsan meeting Wangwu in the lecture.(b) Hearsay: The speaker was told that Zhangsan met Wangwu in the lecture, but this fact was beyond the speaker's expectations.(c) Direct evidence: The speaker saw that Zhangsan met Wangwu in the lecture, whereas Wangwu was supposed to be staying in the hospital at that time as he had suffered a fractured arm in a football game.Therefore, to express the unexpectedness of Wangwu's presence, the speaker double-checks by producing a wh-fronting question.
At this point, we are interested in the question of whether the mirative construal derived from the overt wh-fronting is encoded by this specific syntactic strategy.As proposed by Peterson (2017), two tests are relevant: the negation test and the cancellation test.The hypothesis of the negation test is that if mirativity is implicated by the Mandarin wh-fronting construction, then it is not part of the propositional content of the sentence.Thus, the negation of the main predicate will not affect the mirative meaning of the construction.Compare (14a) and (14b).
The mirative import coerced by the wh-fronting strategy in (14a) is evidently maintained in (14b) irrespective of the negated main predicate.Since the implication of (14b) is "I am surprised that you did not want to buy that car" rather than "I am not surprised that you wanted to buy that car", it suggests that the Mandarin overt wh-fronting strategy indeed encodes mirativity.
( Surprise, an attitude of the speaker, cannot be canceled by the speaker themselves.Thus, we hypothesize if we cannot cancel the surprise implication by adding a continuation like "but I am not surprised at all/I was actually expecting that…" after the question, then mirativity is conventionalized by the structure rather than a conversational implicature (Potts 2005: 28).As a result, (15b) bears out that cancellation of surprise is not possible with Mandarin whfronting constructions when the wh-phrase is stressed; thus mirativity is conventionalized by this sentence pattern.
(15) a. Wh-fronting question: [Na-bu-dianying]i Zhangsan zui xihuan kan ti? which-CLF-movie Zhangsan most like watch 'WHICH MOVIE did Zhangsan like to watch most?!' b.Wh-fronting question with a cancellation of surprise [Na-bu-dianying]i Zhangsan zui xihuan kan ti? which-CLF-movie Zhangsan most like watch #Qishi wo yidian ye bu yiwai.actually 1SG at-all also NEG surprise 'WHICH MOVIE did Zhangsan like to watch most?! #While actually, I am not surprised at all.' 4.2.CONTRAST AND EXHAUSTIVITY.In order for a focus to bear a plain contrastive construal, there must be at least one alternative in the presupposed set, distinct from the focused element, for which the predicate does not hold.Notably, a NIF lacks such a negative statement of alternatives as it merely represents non-presupposed information (Neeleman & Vermeulen 2012).Following this line of analysis, an exhaustive focus is identified as the sole member or the only subset of members of a presupposed set, to the exclusion of all the other alternatives for which the predicate potentially holds (Kiss 1998).The degree of contrast encoded by an exhaustive focus is lower than that of a plain contrastive focus because the identities of the alternatives that an exhaustive focus contrasts with are less specific or prominent.The most famous exhaustifying operator in English is only.Only forces an exhaustive listing of the constituent that it attaches to, be it a DP, a VP, or a PP, meanwhile negating all the other context-defined alternatives for which the predicate holds.Crucially, the identified referent does not have to be unique, namely the cardinality of the exhaustive subset ranges from one to many, as long as the subset represents an exhaustive identification of the referents in the context that fits the presupposition (Horn 1981;Kiss 1998).Consider example ( 16).The exhaustive semantics of the sentence comprises two components: an identification (see (16a)) of an entity (here, a T-shirt) among a set of alternatives presupposed in the context making the proposition 'John bought x in the market' true, and a denial (see (16b)) to the truth of 'John bought any alternatives other than a T-shirt in the market'.In probing the question of whether Mandarin bare wh-fronting questions encode a plain (nonexhaustive) contrastive focus or enable an exhaustive listing of the alternatives, Brunetti's (2004: 68)  This test renders similar results in Mandarin wh-fronting interrogatives.The fronted whphrases are compatible with haiyou 'also' or juran 'even' (see ( 18)), indicating this syntactic displacement strategy does not automatically encode an exhaustive focus interpretation of the wh-variable.
(18) Mandarin a. Haiyou [shei]i Zhangsan zuotian ye yu-dao le ti? also-have who Zhangsan yesterday also meet-arrive PFV One possible translation: 'WHO did Zhangsan also meet yesterday?!' b. [Shei]i Zhangsan zuotian juran yu-dao le ti? who Zhangsan yesterday even meet-arrive PFV One possible translation: 'WHO did Zhangsan even meet yesterday?!' In view of the discourse requirements of a plain contrastive focus and an exhaustive focus, this study devises question-answer pairs that fit specific contexts where the identity of the focus being questioned is explicit to both interlocutors.An example is shown in (19).A2: Ta mai le gangbi he bijiben.
3SG buy PFV pen and notebook 'He bought a pen and a notebook.A3: #[Gangbi/bijiben]i, ta mai le ti.
EXH pen and notebook 3SG buy PFV 'It was a pen and a notebook that he bought.'Before uttering the question, what is made salient to the interlocutors provided in ( 19) is a set of entities for which the predicate potentially holds: {a pen, a notebook, a highlighter}, while since Zhangsan failed to find a highlighter in the store, only two candidates make the proposition 'Zhangsan bought x' true -a pen and a notebook.For an information-seeking wh-in-situ question given in (19a), by either providing a partial/non-exhaustive answer like A1, or providing a complete/exhaustive answer like A2, the addressee is offering a true answer to the question.This is because the addressee is free to include as much information as they can or they want when answering a question (Groenendijk & Stokhof 1984).A1 and A2, with a mention of either or both of the entities that Zhangsan bought, also suffice to address a bare wh-fronting question, which is meant to be asked in order to mark the contrast between the focused element and other presupposed alternatives, as (19b) shows.
Besides, A3 is a congruent answer to question (19b) rather than (19a), as the information-structurally prominent object appearing in a derived position fulfills the discourse requirement imposed by a bare wharg-fronting question rather than that of a noncontrastive wh-in-situ question.The sentence-initial shi pattern given in A4 enforces an exhaustive interpretation of the focused element.This explains why it is only suitable to address a question in (19c) but not those in (19a-b), as the exclusion of all the other alternatives in the semantics of a shi-marked exhaustive focus is beyond what a simple wh-insitu question or a bare wh-fronting question requires.
In answering a shi-marked wh-fronting question like (19c), the addressee must exhaust what they know.Therefore, partial answers in A1 and A3 are infelicitous due to the lack of information irrespective of the sentence patterns they take.Both A2 and A4, however, congruently respond to the question albeit in distinct ways, as they contain the only two members within the set that make the proposition 'Zhangsan bought x' true.
So far, we can conclude with the following generalizations: (20) a.A Mandarin wh-in-situ question licenses a new information focus in the answer.b.A Mandarin bare wh-fronting question licenses a plain (non-exhaustive) contrastive focus or a mirative focus (when the wh-phrase is prosodically marked).c.A Mandarin wh-fronting question marked by shi enforces an exhaustive focus in the answer.

Concluding remarks.
To recapitulate, we have established that Mandarin overt whfronting parallels with A'-movement in the sense that it is sensitive to islands and the displaced wh-phrase reconstructs for binding.It is also borne out to be an instantiation of focus movement as it coerces three types of focus interpretations and does not exhibit the (7) a. *[TopP Meiyou-renheren]j [FocP shenme-dongxi]i tj mai ti le? WHAT does nobody buy?!' b. *[TopP Suoyou-ren]j [FocP shenme-dongxi]i [IP tj dou mai le ti ]? (Quantificational NP) everyone what-thing all buy PFV Intended: 'WHAT did everyone buy?!' c. *[TopP Zhangsan huo Lisi]j [FocP shenme-dongxi]i [IP tj mai le ti ]? (Disjunctive NP) Zhangsan or Lisi what-thing buy PFV Intended: 'WHAT did Zhangsan or Lisi buy?!' d. *Zhangsan xiang-zhidao [CP [TopP meiyou-renheren]j [FocP na-suo xuexiaowonders which of the schools nobody got into.' (10) a. Short-distance wh-question [CP [Zijij de na-ben shu]i [IP Zhangsanj jingchang kan ti]]?self SUB which-CLF book Zhangsan always read One possible translation: 'Which of his books does Zhangsan always read?' b.Long-distance wh-question [CP [Zijij de na-ben shu]i [IP laoshij xiang-zhidao [CP Lisi du le ti ]]]? self SUB which-CLF book teacher wonder Inference: The speaker sees Kemal's coat hanging on the front wall but has not yet seen Kemal.(b) Hearsay: The speaker has been told that Kemal has arrived but has not yet seen.Kemal.

( 16 )
John bought only a T-shirt in the market.a. <λx[John bought x in the market], a T-shirt, {a T-shirt, a bag, an apple ...}> b. ∀y[y ∈{ a T-shirt, a bag, an apple ...} & y≠ a T-shirt & ¬[John bought y in the market]]].
14) a. Context: Zhangsan and Lisi were attending a car exhibition.They glanced over a multitude of brands of cars.When they looked at Tesla's most expensive car, Zhangsan said he wanted to buy it.Lisi raised the question just to show his unprepared mind upon hearing what Zhangsan said.Zhangsan and Lisi were attending a car exhibition.Zhangsan found a car produced by his favorite brand and Lisi knew that it had been Zhangsan's dream to own it.Coincidentally, the car dealer was offering an exceptional discount for that car.Lisi believed that Zhangsan would buy it, whereas Zhangsan decided not to buy it.So Lisi raised the question to show his surprise at Zhangsan's decision.
PFV'WHICH CAR didn't you want to buy?!' 'even/also test' offers a clue.She claims that if a focused constituent in a displaced position can be modified by even or also, the fronting operation does not automatically give rise to an exhaustive focus reading.As the Italian examples in (17a-b) show, the fronted DPs are compatible with even and also, evidencing that the fronted constituents are 'exclusively' identified without exhaustion.