24xd-raw [⌂ EvenYrbk]

Yue Xing Marcel den Dikken

Ā-dependencies in conditionals: The view from Mandarin

1. The explananda Mandarin Chinese conditional ‘if’-clauses are introduced by ru (1-a), optionally combined with guo (1-b).1 (1)

a.

b.

ru ni bu gei ta yi-ci jihui, if 2SG NEG give 3SG one.CL chance (na) ta hui hen shangxin. then 3SG will very upset ‘If you don’t give him a chance, he will be very upset.’ ru guo ni bu gei ta yi-ci jihui, if 2SG NEG give 3SG one.CL chance (na) ta hui hen shangxin. then 3SG will very upset ‘If you don’t give him a chance, he will be very upset.’

The two versions differ with regard to their tolerance of focus and topic fronting within the conditional clause. ‘Bare’ ru is compatible with focus on the subject (2a) but allows focus fronting of non-subjects only marginally, regardless of whether the focused constituent is placed to the right or to the left of the subject, as in (2b), (2c). However, ‘bare’ ru strongly resists topic fronting, see (3).2

1

2

‘Bare’ ru and ru guo are often used interchangeably to introduce conditional clauses in contemporary written Chinese. While ‘bare’ ru conditionals can be traced back to ancient China and remain prevalent in present-day written Mandarin, the use of ru guo conditionals is more common in contemporary everyday conversations. Throughout the paper, all grammaticality diacritics are based on acceptability judgments we obtained from native Mandarin speakers born and raised in various parts of Mainland China. A reviewer considers the strings in (3) to be grammatical. We currently lack a full understanding of the variation in the judgments on (3) and their implications for the analysis of ‘bare’ ru conditionals. This matter warrants further investigation in future research. The asterisks in (3) represent the intuitions of the first author of this paper and the pool of native speakers consulted by the first author.

Xing & Den Dikken, Ā-dependencies in conditionals: The view from Mandarin 2

(2)

a.

b.

c.

(3)

a.

b.

ru lian NI dou bu gei ta yi-ci jihui, … if even 2SG all NEG give 3SG one.CL chance ‘If even YOU don’t give him a chance, …’ ?ru ni lian YI-CI JIHUI dou bu gei ta, … if 2SG even one.CL chance all NEG give 3SG ‘If you don’t even give him ONE CHANCE, …’ ?ru lian YI-CI JIHUI ni dou bu gei ta, … if even one.CL chance 2SG all NEG give 3SG ‘If even ONE CHANCE you don’t give him, …’

ru zhe-ci

jihui ni gei ta, if this.CL chance 2SG give 3SG ta hui haohao zhenxi. 3SG will well cherish

‘If this chance you give him, he will cherish it.’
ru caipiao

ta zhong-le, if lottery 3SG win-PRF ta hui qing women chi fan. 3SG will invite 1SG eat meal

‘If the lottery he wins, he will invite us for dinner.’

By contrast, ru guo conditionals are consistently tolerant of focus, for subjects and non-subjects alike (4a), (4b), and (4c). And ru guo conditionals allow topic fronting to exactly the same extent that topicalization is permitted in clauses that serve as complements of matrix factive predicates. We see this by comparing (5) to (6), the latter featuring the factive matrix predicates zhidao ‘know’ and fouren ‘deny’. 3 (4)

a.

b.

c.

3

ru guo lian NI dou bu gei ta yi-ci jihui, … if even 2SG all NEG give 3SG one.CL chance ‘If even YOU don’t give him a chance, …’ ru guo ni lian YI-CI JIHUI dou bu gei ta, … if 2SG even one.CL chance all NEG give 3SG ‘If you don’t even give him ONE CHANCE, …’ ru guo lian YI-CI JIHUI ni dou bu gei ta, … if even one.CL chance 2SG all NEG give 3SG ‘If even ONE CHANCE you don’t give him, …’

An anonymous reviewer finds (5b) grammatical and has the same judgment for (6b). The native speakers consulted for this paper find both ungrammatical. While we have yet to identify the root cause of this variation in topicalization tolerance across the Mandarinspeaking community, the thing to note is the parallelism in the judgments for (5b) and (6b), which is robust. This parallelism is consistent with the structural analysis proposed in (21).

Xing & Den Dikken, Ā-dependencies in conditionals: The view from Mandarin 3

(5)

a.

b.

(6)

a.

b.

ru guo zhe-ci jihui ni gei ta, if this.CL chance 2SG give 3SG ta hui haohao zhenxi. 3SG will well cherish

‘If this chance you give him, he will cherish it.’
ru guo

caipiao ta zhong-le, if lottery 3SG win-PRF ta hui qing women chi fan. 3SG will invite 1SG eat meal

‘If the lottery he wins, he will invite us for dinner.’

Xiaoming zhidao zhe-ci jihui laoban gei ta le. Xiaoming know this.CL chance boss give 3SG PRF ‘Xiaoming knows that this chance the boss has given him.’

Xiaoming fouren

caipiao ta zhong le. Xiaoming deny lottery 3SG win PRF

‘Xiaoming denies that the lottery he has won.’

These data raise the following set of questions: i) what is the syntax of ‘bare’ ru conditionals, and how does it explain the facts in (2)–(3)? ii) what is the syntax of ru + guo conditionals, and how does it explain the facts in (4)–(5)? iii) what causes ru guo conditionals to pattern with cases of factive clausal complementation?

2. The hypotheses The main hypotheses that we advance to explain the facts laid out in section 1 are the ones listed below. The conjunction of hypotheses H4 and H5 is, to the best of our knowledge, original to this paper. H1

H2

The syntax of conditional clauses involves the establishment of an operator–variable (Ā) dependency for a null conditional operator (henceforth OpCond) in the conditional clause. The dependency involving non-subject focus fronting marginally interferes with the operator–variable dependency for OpCond within the same clause, due to the fact that, although the two Ā-dependencies do not cross each other, they are not properly nested.

Xing & Den Dikken, Ā-dependencies in conditionals: The view from Mandarin 4

H3

H4

H5

Subject focalization with lian … dou ‘even’ averts interference with conditional formation, in one of two ways: a) in-situ focus, or b) properly nested Ā-dependencies. Mandarin conditionals featuring ru + guo have a bi-clausal syntax, with guo as the predicate of the matrix clause. OpCond builds the dependency with its variable within the ru guo clause, unhindered by focus or topic fronting in the clause embedded under guo. The predicate guo ‘fulfill/realize’ is factive, hence ru guo conditionals tolerate topic fronting within the clause in the complement of guo to the same extent as other factive predicates do.

3. The syntax of conditionals Conditional constructions are correlative constructions (see Iatridou 1991; Bhatt and Pancheva 2006; Haegeman 2006). English conditional if is the complementizer that introduces the relative clause. We assume the same is true for its Mandarin translation equivalent, ru.4 In addition to a complementizer, the left periphery of the antecedent clause also features an operator. In both English and Mandarin conditionals, the relative operator (OpCond) is silent. The consequent clause optionally includes an overt correlative particle — English then or Mandarin na. (7)

a. b.

[Antecedent=RelCl OpCond if you do not give him a chance], … [Antecedent=RelCl OpCond ru ni bu gei ta yici jihui], … [Consequent=CorrelCl (then) he will be very upset] [Consequent=CorrelCl (na) ta hui hen shangxin]

An important assumption that we will exploit in our discussion of Ādependencies inside the antecedent clause of conditional constructions is that the conditional operator OpCond binds a variable on the edge of the constituent representing the antecedent clause’s complete argument structure — i.e., vP. This is schematized in (8). (8)

[Antecedent OpCond … [TP DPSubject [T′ T [vP tCond [vP … ]]]]] [Consequent (then) … ]

Our rationale for assuming that the variable bound by OpCond is adjoined to vP is the following. The antecedent clause of a conditional construction presents a condition under which the consequent holds. This condition involves the fulﬁlment of a proposition — more particularly, the proposition expressed by the predication structure of the antecedent clause. In conditional if-clauses, the 4

Nothing fundamentally changes if this assumption should turn out to be false and ru is instead to be treated as the exponent of the relative operator (OpCond).

Xing & Den Dikken, Ā-dependencies in conditionals: The view from Mandarin 5

minimal syntactic constituent representing this proposition is the vP, within which the predication relation between the predicate and its subject is established. The optimal locus for the placement of the variable bound by OpCond in conditional if-clauses is a position adjoined to vP.5

4. The syntax of focus fronting A focus is in an information-structural relationship with presupposed material, with the presupposition representing given information and something in the focused constituent providing new information. In focus-fronting constructions, this information-structural relationship correlates with a syntactic relationship between the focus and the presupposition, mediated by a head that, by general consensus, is called ‘Foc’. The focus (‘XP’ in the structure in (9)) is associated with a variable inside the constituent in the complement of the Foc (‘YP’ in (9)).6 We assume that for non-wh foci, the connection between the fronted focus and the variable inside YP is indirect: the variable is bound by an operator within YP; the operator–variable dependency established within YP creates a predicate for the focused XP in SpecFocP.7 (9) 5

[FocP [XP Focus] [Foc′ Foc [YP Opi … vbli …]] Support for the hypothesis that the conditional operator binds a variable adjoined to the minimal predication structure of the antecedent clause comes from with-absolutes, which have conditional semantics but do not feature a verbal predication structure and, concomitantly, lack the complementizer if. In (ia), the with-phrase introduces the condition under which the consequent holds, similarly to (ib), featuring a conditional if-clause as the antecedent. In (ia), the antecedent has a non-verbal predication structure, not included in a projection of T. The only syntactic constituent corresponding to a proposition within the antecedent is thus the bare predication structure [John as/being our leader]. It is to this predication structure (a small clause, with as/being as its head) that the variable bound by OpCond is attached in (ia). (i)

6

7

a. [Antecedent with [John as/being our leader]], [Consequent we will deﬁnitely win] b. [Antecedent if [John is our leader]], [Consequent we will deﬁnitely win]

Extending the analysis of (ia) to semantically equivalent (ib) and other conditional if-clauses is straightforward, with OpCond consistently binding a variable adjoined to the minimal predication structure. An alternative approach insisting on TP being the locus of the variable in conditional if-clauses cannot carry over to with-absolutes, which do not contain a TP. Thus, the analysis in (8) is more inclusive of the two logical possibilities. We will henceforth assume that the conditional operator is always engaged in an Ā-dependency with a variable adjoined to the minimal predication structure of the antecedent clause. That foci bind a variable is clear from the fact that focalization gives rise to Weak Crossover Effects (Chomsky 1977; Lasnik and Stowell 1991). We leave wh-foci aside altogether in this paper, for the simple reason that wh-dependencies cannot be established within conditional clauses (for obvious reasons: SpecCP is occupied by OpCond already).

Xing & Den Dikken, Ā-dependencies in conditionals: The view from Mandarin 6

The operator binding the variable within YP is in a position syntactically local to the variable and high enough in YP to deﬁne YP as a predicate of the focus. For English-type languages, where the operator within YP is silent, it is often difﬁcult to tell precisely where the operator is located. But for Mandarin, it appears that the YP-internal operator is actually overt in the case of ‘even’ focus: as shown in (10), the element dou is an integral part of Mandarin lian … dou ‘even’ focus constructions 8 , where dou is always to the left of the verb, separable from the verb only by the preverbal negation particle bu.9 The lianphrase must occur to the left of dou. This follows if the lian-phrase occupies the position labeled ‘XP’ in (9), and dou is contained in the constituent labeled ‘YP’. (10)

a.

b.

ni lian YI-CI JIHUI dou 2SG even one.CL chance all ‘You don’t even give him ONE CHANCE.’ lian YI-CI JIHUI ni dou even one.CL chance 2SG all ‘Even ONE CHANCE you don’t give him.’

bu NEG

gei ta. give 3SG gei ta. give 3SG

Giannakidou and Cheng (2006), in their analysis of Mandarin free-choice items such as nage xuesheng dou ‘whichever student, any student’, call dou ‘all’ a maximality operator. Badan (2008: 12) applies this approach to dou to ‘even’ focus constructions, assuming that dou ‘operates over the set of alternatives, closes the domain and gives the maximal set of these alternatives, i.e., it maximizes the set of the presuppositions’. The idea that dou in Mandarin ‘even’ focus constructions maximizes the set of presuppositions is readily mapped onto the structure in (9), with dou serving as ‘Op’. We take dou to be occupying exactly the same surface position in (10a) and (10b) alike — a position adjoined to vP, the minimal predication structure representing the presupposition.10 We 8

9

10

As a matter of fact, dou ‘all’ is more essential to ‘even’ focus in Mandarin than is the element lian, usually glossed as ‘even’ (and elsewhere in Mandarin expressing ‘connecting, including’; Xing 2004): lian is often omissible in Mandarin ‘even’ focus constructions, but dou must always be there. We have no speciﬁc analysis of bu ‘not’ to offer here, but assume (analogously to Dutch niet, German nicht, the negation particles of the Scandinavian languages, and possibly English full not as well) that it is an adverbial modiﬁer adjoined to vP — i.e., not the lexicalization of the head or speciﬁer of NegP. Why bu cannot occur to the left of dou still remains a question; it is entirely possible that a further development of the path containment approach laid out in section 5 can cover the relative hierarchical positions of dou and bu on the edge of vP, but we must leave execution of this development to a future occasion. A reviewer suggests an alternative structural possibility, where the sentence subject in (10a) and (10b) is realized in TopP; and that may challenge the claim that dou occupies the same surface position in both cases. Building on Rizzi (2005), Xing (2023) argues that the sentence-internal subject in constructions of type (10b) lacks the D-linking property

Xing & Den Dikken, Ā-dependencies in conditionals: The view from Mandarin 7

will also assume that dou is linked to the variable inside YP, which in turn is linked to the focused XP. Though the position of dou is constant, the size of YP is different in (10a) and (10b): in (10a), YP is no larger than vP, while in (10b), YP contains the structural subject and hence must be minimally as large as TP. This gives us the structures in (11). (11)

a. b.

[TP ni [T′ T [FocP [XP lian YICI JIHUI]i [Foc′ Foc [YP=vP Opi = dou [vP … [VP … vbli …]]]]]]] [FocP [XP lian YICI JIHUI]i [Foc′ Foc [YP=TP ni [T′ T [vP Opi = dou [vP … [VP … vbli …]]]]]]]]

With this elementary syntax of focus fronting and Mandarin lian … dou ‘even’ constructions in place, we are now ready to return to the facts in (2). We will begin by addressing the marginality of the examples in (2b) and (2c) in section 5.1. In section 5.2, we subsequently discuss why subject focalization is harmless in conditional clauses.

5. Focalization in ru conditionals 5.1. Non-subject focus As we argued in section 4, focus fronting of non-subjects involves the maximality operator dou adjoined in a position on the edge of vP, regardless of whether the focused constituent stands before or after the subject. The trace of the conditional operator, OpCond, is likewise adjoined to vP — for recall that OpCond consistently binds a variable adjoined to the minimal predication structure of the antecedent clause. The question of the relative positions of the two vP-adjuncts is not perfectly straightforward to answer. What is certain is that the variable bound by OpCond is introduced in its adjunction position to vP via External Merge. For the maximizing operator dou, it is less clear whether it is externally or internally merged with vP. The maximizing operator dou of Mandarin free-choice whitems (cf. Giannakidou and Cheng, 2006) is likely merged in the immediate locality of the wh-constituent. Assuming that the dou of ‘even’ focus constructions likewise starts out life local to the focus variable, dou gets to its vP-adjoined position via Internal Merge, which takes place after all External Merge involving vP has been completed. dou is then the highest element in vP, associated with being a topic. We interpret this as evidence supporting the claim that this subject occupies the structural subject position in SpecTP, consistent with the Extended Projection Principle. The sentence-initial subject in (10a), which displays the D-linking property, neither affects the analysis of dou adjoining to the edge of vP nor the structural interpretation of the marginality of the examples with non-subject focalization in (2b) and (2c).

Xing & Den Dikken, Ā-dependencies in conditionals: The view from Mandarin 8

higher than the trace of the conditional operator. The result is schematized in (12). (12)

? [Antecedent OpCond … [vP douFocus [vP tCond [vP … vblFocus …]]]]

The (OpCond, tCond) and (douFocus, vblFocus) dependencies are not properly nested, which causes this structure not to perfectly respect Fodor’s (1978) Nested Dependencies Constraint in (13) and Pesetsky’s (1982) Path Containment Condition in (14). (13)

Nested Dependencies Constraint (Fodor 1978: 448(40)) If there are two or more ﬁller–gap dependencies in the same sentence, their scopes may not intersect if either disjoint or nested dependencies are compatible with the well-formedness conditions of the language.

(14)

Path Containment Condition (Pesetsky 1982: 309(94)) If two paths overlap, one must contain the other.

But the violation incurred in (12) is mild because the two dependencies do not strictly cross either, since dou and tCond are on the edge of the same structural domain. This accounts for the fact that the examples in (2b) and (2c) are only marginally deviant. 5.2. Subject focus Focalization creates a bifurcation of the utterance into the focus and the presupposition. The syntax establishes this bifurcation by creating a FocP with the focus in its speciﬁer position and the constituent expressing the presupposition in the complement of Foc. In the case of non-subject focalization, the complement of Foc must be turned into a predicate for the focus by the application of operator movement inside YP, as in (9). But the subject of the clause is syntactically licensed in SpecTP, outside the minimal constituent representing the presupposition. So a bifurcation of the utterance into the focus and the presupposition is created for subject foci without any operator movement being necessary: subjects are in principle amenable to in-situ focalization, as in (15a), where the quantiﬁer dou is not undergoing any movement: it is Externally Merged on the edge of vP, local to the NP-trace of the subject.11 In (15a), the T-head serves as the mediator of the focus–presupposition relation between the subject and the vP, whereas v is the mediator of

11

In cases of ‘even’ focalization on non-subjects, it is not necessary for dou to be adjacent to the focus. But in subject focus constructions, dou must be adjacent to the nominal subject

Xing & Den Dikken, Ā-dependencies in conditionals: The view from Mandarin 9

the thematic predication relation between the VP and its external argument. In (15a), the chain (OpCond, tCond) is the only movement-derived operator–variable dependency in the structure. Path containment is trivially satisﬁed, therefore. But even if the focused subject moves from SpecTP into the high left periphery, as in (15b), the focus-movement dependency is properly nested within the (OpCond, tCond) dependency, in keeping with the Nested Dependencies Constraint. (15)

a. b.

[Antecedent OpCond … [TP FOCUS=DPSubject [T′ T [YP=vP douFocus [vP tCond [vP tSubject ]]]]]] [Antecedent OpCond … FOCUS=DPSubject [YP douFocus … [TP tFocus [T′ T [vP tCond [vP tSubject ]]]]]]

Hence, regardless of whether the focused subject is in situ in SpecTP (as in (15a)) or in the high left periphery (as in (15b)), (2a) is correctly predicted to be grammatical.

6. The syntax of topic fronting A topic is in an information-structural relationship with a comment, with the topic representing given information and the comment providing information about the topic that must be at least partially new. In topic-fronting constructions, this information-structural relationship correlates with a syntactic relationship between the topic and the comment, once again mediated by a functional head (‘Top’), with the topic in its speciﬁer position and the comment in the complement, as depicted in (16). YP again contains an empty category associated with the fronted constituent. And as in the case of non-wh foci, the connection between the fronted constituent and the empty category is not movement-derived — in topic–comment constructions, the empty category inside YP linked to the topic-XP is a resumptive pronominal proform (pro). (16)

[TopP [XP Topic]i [Top′ Top [YP proi … ]]]

We will conﬁne our discussion of topic fronting in conditional clauses to cases of contrastive topicalization. In Dutch, contrastive topics, when associated with an overt resumptive element in YP, necessarily see this resumptive element fronted to the edge of YP: (17)

a.

Jan, die ken ik niet (Piet wél). Jan DEM know I NEG Piet AFF ‘Jan, I don’t know (Piet, I do).

focus. This follows from (15a) on the assumptions i) that no verb movement to T takes place and ii) that dou must be the highest element in vP.

Xing & Den Dikken, Ā-dependencies in conditionals: The view from Mandarin 10

b.

?Jan, ik Jan I

ken die niet know DEM NEG

(#Piet wél). Piet AFF

While the demonstrative die can (somewhat marginally) stay in clause-internal position when contrastive Piet wél is not included in (17b), contrastive topicalization forces the resumptive element to be fronted, as in (17a). The portion of the clausal structure that the Top-head connects to the fronted topic is necessarily large. In contrastive topic-fronting constructions of the type in (16), YP cannot be as small as TP: the resumptive proform proi in the left periphery of YP would be locally bound by XPi in that case, causing a violation of the binding theory. Contrastive topic fronting constructions require YP to be a local domain for binding, to protect the proform from local binding by the topic. The exact size and category of YP are of no immediate concern to us — rather, what is important is that in constructions involving the fronting of a contrastive topic, a) a resumptive element linked to the topic is undergoing leftward movement to a high position within YP and, b) YP is larger than TP.12 (18)

[TopP [XP c-Topic] [Top′ Top [YP proi … [TP … ti … ]]]]

That in contrastive topic-fronting constructions, YP must be large and feature Ā-movement of the resumptive element linked to the topic helps us account for the fact that contrastive topicalization is impossible in conditional if-clauses. We will show this next.

7. Topicalization in ru conditionals Topic fronting to a pre-subject position is uniformly ungrammatical in Mandarin conditional clauses introduced by ‘bare’ ru, as in (3).13 Mandarin is similar to English in this regard: the versions of the English prose translations of (3a) and (3b), with this chance and the lottery placed immediately to the right of if, are also ungrammatical. The ungrammaticality of the examples in (3) is unsurprising. The derivation of these sentences is schematized in (19). It is clear at a glance that the Ādependencies involving OpCond and the resumptive proform associated to the topic intersect, in violation of the Nested Dependencies Constraint.

12

13

The fact that non-contrastive topicalization does not force the fronting of the resumptive element leads to the expectation that it should not interfere with the OpCond – variable dependency — and indeed, non-contrastive topic fronting is grammatical in conditional ifclauses in English. This is a clear indication that it would not be accurate to say (as does Badan 2008) that sentence-initial lian-phrase in Mandarin ‘even’ focus constructions is a topic.

Xing & Den Dikken, Ā-dependencies in conditionals: The view from Mandarin 11

(19)

[Antecedent OpCond … [TopP Topic [Top′ Top [YP proTopic … [TP DPSubject [vP tCond

[vP … tTopic …]]]]]]]

With the facts in (2) and (3) under control, what is left to explain is that in the presence of the element guo, the focus and topic facts change quite markedly.

8. The bi-clausal syntax of ru guo conditionals The element guo in (1b), repeated as (20) (with an enriched gloss and prose translation), goes back to a predicate translatable as ‘fulfill, realize’.14 Taking this ancestry seriously, we analyze ru guo conditionals as having a bi-clausal antecedent clause, with the dependency between OpCond and its trace established wholly within the guo clause, as in (21).15 (20)

ru guo ni bu gei ta yi-ci jihui, if fulfill 2SG NEG give 3SG one.CL chance (na) ta hui hen shangxin. then 3SG will very upset ‘If it is true/the case that you don’t give him a chance, he will be very upset.’

(21)

[Antecedent OpCond … [TP ‘it’ø [T′ T [vP tCond [vP guo ‘fulfill’ [CP C(%Topic/ ✓ Focus) … ]]]]]]

In (21), the Ā-dependency between OpCond and its trace is established wholly within the matrix clause headed by guo ‘fulfill, realize’. As a consequence, focalization and topicalization within the conﬁnes of the clause embedded under guo can never interfere with the dependency between OpCond and tCond.

14

An example illustrating the verbal nature of guo in classical Chinese is provided in (ⅰ):

(ⅰ) wen zhi, xinran gui wang. wei guo, xun bing zhong. hear PRON, happily return go not fulfill soon ill died ‘He heard of this story, and planned to go and find it, but was taken ill and died before he could fulfill his wish.’ (extracted from Tao Yuanming’s “The Peach Colony” from the East Jin Dynasty, translation provided by Lin Yutang) 15

We assume that the guo clause of ru guo conditionals is a full-ﬂedged TP, primarily because we subscribe to the view that ru is a complementizer, which categorially selects a TP. But for an understanding of the facts in (4)–(6), it does not actually matter whether the portion of the structure deﬁned by guo is as large as (21) takes it to be: ultimately, all that matters is that guo deﬁnes a predication structure in its own right, hosting the trace of the conditional operator and containing the clause within which focus or topic fronting takes place.

Xing & Den Dikken, Ā-dependencies in conditionals: The view from Mandarin 12

9. Focalization and topicalization in ru guo conditionals Viewed against the background of the syntax of ru guo conditionals in (21), the grammaticality of the focus fronting constructions in (4) is now entirely straightforward: focus fronting takes place within the complement of guo, so the paths deﬁned by the syntax of focus movement (in the clause embedded under guo) and the syntax of conditional formation (in the clause whose predicate is guo) could not possibly intersect. But by the same logic, topic fronting in ru guo conditionals also does not give rise to path containment problems. So, what remains to be accounted for is the mixed pattern exhibited by (5): while (5a) is grammatical, (5b) remains as bad as the corresponding example without guo. We had already pointed out in section 1 that the (un)acceptability of topic fronting in ru guo conditionals systematically tracks the (un)acceptability of topic fronting in the complement of factive matrix predicates. This parallel between ru guo conditionals and factive complement clauses now falls out directly from the analysis of ru guo conditionals in (21). Guo, treated as the matrix predicate of the bi-clausal structure of the antecedent clause of ru guo conditionals, is itself factive. It is an important asset of the analysis of ru guo conditionals proposed in this paper that it directly predicts the exact match in acceptability between (5a) and (6a), and between (5b) and (6b).

10. Closing remarks In this paper, we have provided an analysis of the restrictions imposed on focus and topic fronting in Mandarin conditional clauses, embedded within a principled syntax of conditionals, focus fronting and topic fronting. The Mandarin facts provide clues (in particular, the distinction between ru and ru guo, and the use of the quantiﬁer dou in ‘even’-focus constructions) that would not be as readily available in more densely studied languages such as English and Dutch. References Badan, Linda. 2008. The even-construction in Mandarin Chinese. In: Redouane Djamouri, Barbara Meisterernst, and Rint Sybesma (eds.). Chinese linguistics in Leipzig. Paris: EHESS-CRLAO. 101–116. Bhatt, Rajesh and Roumyana Pancheva. 2006. Conditionals. In: Martin Everaert and Henk van Riemsdijk (eds.). The Blackwell Companion to Syntax, vol. Ⅰ. Blackwell Publishing Ltd. 638–687. https://doi.org/10.1002/9780470996591.ch16 Chomsky, Noam. 1977. On Wh-movement. In: Peter Culicover, Thomas Wasow and Adian Akmajian (eds.). Formal Syntax. New York: Academic Press. 71–132. Fodor, Janet Dean. 1978. Parsing strategies and constraints on transformations. Linguistic Inquiry 9(3): 427–473.

Xing & Den Dikken, Ā-dependencies in conditionals: The view from Mandarin 13 Giannakidou, Anastasia and Lisa Lai-Shen Cheng. 2006. (In)definiteness, polarity, and the role of wh-morphology in free choice. Journal of Semantics 23(2): 135–183. https://doi.org/10.1093/jos/ffl001 Haegeman, Liliane. 2006. Conditionals, factives and the left periphery. Lingua 116(10):1651– 1669. https://doi.org/10.1016/j.lingua.2005.03.014 Iatridou, Sabine. 1991. Topics in conditionals. PhD dissertation. Massachusetts Institute of Technology. Lasnik, Howard, and Tim Stowell. 1991. Weakest crossover. Linguistic Inquiry 22(4): 687– 720. Pesetsky, David Michael. 1982. Paths and categories. PhD dissertation. Massachusetts Institute of Technology. Rizzi, Luigi. 2005. On some properties of subjects and topics. In Proceedings of the XXX Incontro di Grammatica Generativa: 203-224. Xing, Janet Zhiqun. 2004. Grammaticalization of the scalar focus particle lian 連 in Mandarin Chinese. Journal of Historical Pragmatics 5(1): 81–106. https://doi.org/10.1075/jhp.5.1.05xin Xing, Yue. 2023. Mirative focalization: A case study of Mandarin lian DP dou construction. Linguistics Beyond and Within (LingBaW) 9(9): 238-250. https://doi.org/10.31743/lingbaw.17027

Yue Xing Eötvös Loránd University & Hungarian Research Centre for Linguistics yingxue037@gmail.com Marcel den Dikken Hungarian Research Centre for Linguistics & Centre of Linguistics, University of Lisbon dmarcel@nytud.hu