Yes or No, or how to answer a negative question

so that ne cannot retain its negative force. I will show that despite the absence of Genitive of Negation in Czech, the phenomenon of expletive negation is relevant and allows to account for the distribution of A and B answers to negative questions. Abstract: This paper deals with answers to negative yes-no questions, focusing on data from Czech. It is shown that answering particles can express both positive and negative answers to negative questions, but that their choice is not free. Several pieces of evidence are discussed in order to show that the use of the particles depends on the interpretation of negation in the question, which can either be expletive or true. This semantic distinction is furthermore tightly linked to the syntactic position of the negation, according to which we distinguish between negative interrogative clauses and negative declarative clauses used as questions. An analysis in terms of absolute and relative polarity is proposed to account for the mixed behaviour of answering particles: particles express absolute polarity in answers to interrogative questions, whose polarity is open, and relative polarity in answers to declarative questions, whose polarity has been already specified.


INTRODUCTION
The aim of this paper is to investigate the status of particles in answers to negative questions. A yes-no question is a question that asks to identify whether P or not P is true (Hamblin 1973), and is minimally answered by an answering particle. Answers to yes-no questions have given rise to a large amount of literature since Pope (1972), who gives a detailed description of the types of answers across languages. She distinguishes two major systems: truth-based systems, in which the particles confirm or disconfirm the true value presupposed by the question, like in Japanese in (1), and polarity-based systems, in which particles express positive or negative polarity, like in Spanish in (2).
(1) a. The same particles are used to answer both positive and negative questions, although some languages "reinforce" the positive particle to negative questions, e.g. the conjunction but before yes in Czech, see (3), and some languages use a specific particle for positive answers to negative questions, e.g. si instead of oui in French, see (4).
( Although the behaviour of particles has been described for many languages (see Laka 1990;Martins 1994;Holmberg 2001;Holmberg 2012;Farkas 2010;Krifka 2012 a.o.), it has been hardly observed that particles could be used in both ways within a same language.
So, Holmberg (2001; notes that negative questions with not in English can be confirmed by both yes and no, see (5), because not is ambiguous between sentential and VP-negation. In other words, he argues that yes only confirms negative questions when not is interpreted as VP-negation, thus when the question is rather affirmative that truly negative. The use of the particles is however not free. It is argued that it depends on the interpretation of the sentential negation in the question, which can be either true or expletive (cf. Brown and Franks 1995 for Russian negative questions). This semantic distinction is furthermore tightly linked to the syntactic position of the negation, according to which we can distinguish between negative interrogative clauses and negative declarative clauses used as questions (cf. Gunlogson 2001). An analysis in terms of absolute and relative polarity (Farkas 2010) is then proposed to account for the mixed behaviour of answering particles: particles express absolute polarity in answers to interrogative questions, whose polarity is open (Holmberg 2001), and relative polarity in answers to declarative questions, whose polarity has been already specified.
The paper is structured as follows. Section 2 describes two types of answers to negative questions in Czech, showing a mixed behaviour of answering particles. Section 3 deals with negative questions; several pieces of evidence are discussed in order to show that we have to distinguish between questions with expletive negation and questions with true negation and that this distinction accounts for the distribution of answering particles.
Section 4 focuses on expression of the polarity in the answers; it is proposed that particles express absolute or relative polarity depending on the polarity of the question. Section 5 sums up the paper.

TWO TYPES OF ANSWERS TO NEGATIVE QUESTIONS
To show that both answering particles in Czech can confirm and disconfirm a negative question, we need to specify that Czech (like many other languages, e.g. Basque, Finish, Portuguese, Irish) can reply a question by using use the finite verb of the question, either alone or in combination with a particle, see (7). The positive verb always indicates a positive answer and the negative verb a negative answer. Verbal answers will therefore be used throughout the paper to clearly state the polarity of the answer.  yes neg-are no are 'Aren't the parents at home?' 'They are not.' // 'They are.' The puzzle we are dealing here is that the question 'Aren't the parents at home?' can be answered in two ways, which I call type A and type B answers and which are summarized in table 1 below. In type A answers, ne means 'they are not', while ano means 'they are'.
In contrast, in type B answers, ne means 'they are', while ano means 'they are not'. The particles appear, thus, to be ambiguous; they may also combine with both negative and positive verb. The interpretation of each answer is however clearly given by the verb, meaning that the combination of the particle and the verb cannot be considered as redundant. 2

Negative answers
Positive answers Type A answers ne + neg-V (ale) ano + verb Type B answers ano + neg-V ne + V Furthermore, B answers are generally judged inacceptable as replies to questions containing an initial verb, meaning that the question's form matters for the choice of the answer (see section 3 for more details): (10) Neposlal jsi mu ten dopis? *Ano (neposlal). / Neposlal. // *Ne (poslal). / Poslal. neg-sent be.2sg to-him the letter yes (neg-sent) neg-sent no (sent) sent 'Didn't you send him the letter?' Importantly, however, B answers can be used as replies to negative questions with negative polarity items like ani (not-one/not-any), see (11). Since any must be licensed by sentential negation (see section 3.3 for more on NPIs), negative questions confirmed by B answers must contain sentential negation. They therefore differ from negative questions with not in English that allow confirmation by yes only with VP-negation reading, see (5) above.
John (neg-)invited not-one schoolmate yes not-one // no (he) some invited 'Didn't John invite any of his schoolmates?' The difference between A and B answers cannot thus be due to the contrast between sentential and VP-negation, but it must nevertheless be linked to the negation. The hypothesis I explore in the following section is that in some negative questions, negation loses its negative force and behaves likes expletive negation. Negative questions with expletive negation behave like positive questions, whose polarity I assume to be open (following Holmberg 2001). Thus, they are answered by type A answers, in which particles indicate absolute (positive or negative) polarity. In contrast, negative questions with true negation behave like negative assertions, whose polarity is fixed. Thus, they are answered by type B answers, in which particles indicate relative polarity, i.e.
(dis)agreement with the polarity of the question.

NEGATIVE QUESTIONS
The main claim of this section is that in some questions, negation is not true (from the truth conditional point of view), but expletive. The term 'expletive' (or 'pleonastic') negation usually refers to a phonologically overt negative morpheme that lacks negative semantic content (see Espinal 1992  More interestingly, Brown & Franks (1995) and Abels (2002) observe that negative questions with the interrogative particle li in Russian do not license NI-phrases, which would suggest that they also contain expletive negation. However, the same questions license Genitive of Negation. Moreover, questions without particle li license both NIphrases and Genitive of Negation. To account for this asymmetry, Brown & Franks (1995:266) claim that "certain independently motivated grammatical principles, such as Rizzi's (1990) Relativized Minimality, conspire to prevent negation from having negative force [...] in Russian Yes/No questions". In their analysis, negation in li-questions is forced to be pleonastic, because a Yes/No operator in Spec-CP prevents ne to be licensed by the Negation Operator (adjoined to CP) when ne (cliticized on V) moves to C. I pursue here the view that negation can be rendered expletive by the interrogation. In other words, the interrogative operator blocks the negative operator to bind ne when Verb(Neg)-moves to C, so that ne cannot retain its negative force. I will show that despite the absence of Genitive of Negation in Czech, the phenomenon of expletive negation is relevant and allows to account for the distribution of A and B answers to negative questions.

Word order
Main questions in Czech do not contain an overt interrogative C. Still, assuming that syntactic clause-type (declarative, interrogative, exclamative) and illucutionary act (assertion, questioning, exclamation) can be dissociated (Beyssade & Marandin 2006), we can distinguish two types of questions. Czech is a SVO language in which interrogative clauses, are formed by V-fronting (and wh-fronting in case of wh-questions), see (14).
Interrogative clauses are typically used as questions. But declarative clauses with (not only) SVO order can also be used to express questioning. In the following discussion, I will show that negative interrogatives and negative declarative clauses used as questions differ with respect to negative presupposition, combinability with evidential and modal adverbials, licensing of NPIs, and also the answers they receive. given Mary nom children dat cakes acc who dat given Mary nom cakes acc 'Did Mary give children some cakes?' / 'To whom did Mary give some cakes?'

Negative presupposition
Negative interrogatives (with non-focal intonation 5 ) are used to elicit information, to solicit an opinion, or to make a polite request for action (see Gunlogson 2001 for detailed distribution). They do not convey any presupposition. Questions in (15)  However, the biased interpretation in (16) is not obligatory, which explains that these questions can be also answered by type A answers, as shown in (16') (see section 3.4, ex. (24) for analysis). The ambiguity of declarative questions can be avoided by using final rising intonation for neutral interpretation (like in interrogative clauses) and rising intonation followed by final fall for biased interpretation.

Evidential and modal adverbs
The fact that only negative declarative questions can express a negative bias is confirmed by their compatibility with adverbs expressing modality or evidentiality, like určitě ('of course'), zřejmě ('apparently'), jistě ('surely'), see (17). These adverbs are considered as bias markers' and cannot appear in interrogative clauses in English neither, see (18) (cf. Huddleston 1994).
'You are probably not from here?' (18) a. #Did they certainly agree with that? b. #Have they apparently left? c. #Are you probably from here?

Negative polarity items (NPIs)
Three types of negative polarity items can be distinguished in Czech: NI-phrases (nikdo 'nobody', nic 'nothing', etc., žádný N 'no N'), weak NPIs (vůbec 'at all', sebemenší 'the slightest') and strict NPIs (ani jeden 'not one'), see Dočekal (2016 The contrast observed above needs to be more specified. First, NI-phrases are ungrammatical in interrogative clauses in (20a) and (21a) below, but they are acceptable in (21b), where they are used with focal interpretation. Brown & Franks (1995) argue however that negation cannot be expletive with focal interrogation because the verb doesn't raise to C (recall that expletive negation in questions is triggered in contexts with Verb(Neg)-raising to C). I claim that the verb in (21b) actually doesn't raise to C but stays in T, whose specifier is occupied by a null expletive subject (because of the EPP feature).
The focused subject stays in Spec-vP. The fact that an overt expletive subject (v)on ('he') can co-occur with the focused subject in (21c) makes this claim plausible.
neg.needs nobody / somebody help 'Does anyone need help?' 7 The expletive status of negation can be supported by the behaviour of PPI. The PPI in the sentence is interpreted (as expected) only with the wide scope with respect to negation, while the PPI in the sentence (ii) with V(Neg)-raising is not. I thank Mojmír Dočekal for this observation. Finally, recall that that negation in declarative questions can also have expletive interpretation. This suggests that the verb movement to C can be covert; consequently, we obtain type A answers to negative declarative questions: (24) Doktor není v ordinaci? => LF: Není doktor t v ordinaci? Ne (není). the doctor isn't in his office?
isn't doctor in his office no (he isn't)

Summary
Negative questions in Czech can be expressed either by using interrogative or declaratives clauses. These two types of clauses behave differently with respect to several properties (presupposition, adverbs, NPIs, see table 2) that can be explained if the sentential negation in interrogative clauses looses its negative force (thus becomes expletive) by virtue of its movement to the C. Importantly, the distribution of type A and B answers described in section 2 also follows from the distinction between true and expletive negation.

BACK TO ANSWERS
It is generally assumed (since Laka 1990) that answering particles are generated in the polarity projection (ΣP, PolP) in the CP domain, because they express positive or negative polarity. The concept of the polarity must be nevertheless made clearer in order to account for the mixed behaviour of answering particles in Czech.
I propose to distinguish, following Farkas (2010), between absolute and relative polarity or rather between absolute and relative value of the polarity. In declarative clauses, polarity is a feature that has two absolute values: positive [+] and negative [-].  what do you think, should(n't) I get my hair cut

Expressing polarity in answers to interrogative clauses
As noted by Brown & Franks (1995), negative interrogatives must thus be endowed with some specific communicative and pragmatic value. Negative interrogatives in Czech are actually considered as more polite or less direct than positive ones. 9 I suggest the following explanation for such a politeness effect. 10 A polar question asks the addressee to choose between two alternatives, positive or negative (P or not P), but refusing or saying no can be perceived as rude, and therefore difficult to express. By using the negative form of the question, the speaker presents the negative alternative as acceptable and consequently allows the addressee to express his refusal more easily. It is precisely because negation has lost its truth-conditional role that it can play such a role at the discourse level (see Groenendijk & Stokhof 1997).

Polarity in answers to declaratives clauses
I have said above that declarative clauses have their polarity feature specified. The value of the polarity feature in negative declarative clauses is negative. Likewise, negative declarative clauses used as questions have a negative polarity value. That's why they convey a negative presupposition (see section 3.1). The answer confirms or disconfirms 9 See Leech (2014:167) for negation as a "a strategy to express a degree of polite indirectness". 10 A similar effect can be observed with Negation Raising. The sentences Nemyslím, že s tím bude souhlasit ('I don't think he will agree with that') and Myslím, že s tím nebude souhlasit ('I think he will not agree with that') are semantically (i.e. the negation is interpreted in the embedded clause), but only the raised negation adds a politeness effect to the sentence interpretation. this negative presupposition. Consequently, the particles only indicate relative polarity, i.e.
relation between the polarity of the question and that of the answer. 11 (29) Oni ten návrh nepřijali [-] ? [ PolP Ano [-,-] [ Pol' [-] [ XP ø]]] (= nepřijali) they didn't accept the proposal yes (= they didn't accept the proposal) [ PolP Ne [-,+] [ Pol' [-] [ XP ø]]] (= přijali) no (= they accepted the proposal) Importantly, this analysis can be extended to positive questions conveying a positive presupposition, as in (30). But there will be no difference at surface between answers expressing absolute polarity and those expressing relative polarity. In both (25) above and (30) below, the positive answer will be expressed by "ano + positive verb", and the negative answer by "ne + negative verb". The mixed behaviour of the particles can therefore be only observed with negative questions.
(30) Oni ten návrh přijali [+] ? [ PolP Ano [+,+]  Particles are therefore not ambiguous, but their felicitous use depends on the felicitous interpretation of the question. This can be resolved by using specific prosody schemes to indicate biased or focused interpretation. Finally, the proposed analysis predicts that the mixed behaviour of particles can be observed in languages in which negation raising to a particular C leads to its expletive interpretation and in which declarative clauses can be used as questions. This prediction seems comes true e.g. for Russian and Spanish (see Gruet-Skrabalova 2014), as shown in (31) and (32)

CONCLUSION
In this paper, I have dealt with answers to negative yes-no questions, focusing on data from Czech. I have shown that answering particles can express both positive and negative answers to negative questions, but that their distribution depends on the semantic and syntactic properties of questions. I have argued that negation in questions loses its negative force when it moves to the interrogative head C and behaves thus as expletive negation.
Consequently, I have distinguished between two types of negative questions: (i) negative interrogative clauses that contain an initial negative verb and expletive negation, and that do not convey a negative presupposition, and (ii) negative declarative clauses used as questions, that contain a non initial negative verb and true negation, and that convey a negative presupposition. Following this syntactic and semantic distinction, I have shown that negative interrogatives receive the same answers as positive interrogatives (yes in case of positive answer and no in case of negative answer), while negative declaratives used as questions are confirmed by yes (corresponding to a negative answer) and denied by no (corresponding to a positive answer). I have explained this distribution by proposing that the particles express absolute polarity in answers to interrogative questions, whose polarity is open and must be fixed by the particle, and relative polarity in answers to declarative questions, whose polarity has been already specified.
12 These examples come from my Russian and Spanish informants.