Question particles in Basque ∗

This article proposes separate analyses for the question particles -a and al in Basque, which occur only in polarity questions, within the framework of Generative Grammar. I will propose that the former one, used in the eastern dialects, is the head of FiniteP and that the latter one, used in the central dialects, occupies the head of Particle Phrase located between TP and the CP field. I provide the following evidence in support of this dual analysis: 1) -a can be used with ote but no other particle can appear at the same time with al; 2) al is compatible with allocutivity but -a is not; 4) and, finally, al can be used in embedded clauses, whereas -a cannot. The fact that -a is not allowed to occur in indirect questions and that it is incompatible with the allocutive verbal paradigm shows that it is in complementary distribution with the head of CP and, therefore, that -a occupies such a head; on the other hand, the impossibility of al to appear with other particles suggest that they must occur in the same position and, since it can appear in embedded questions and with allocutive forms, it does not occupy the head of the CP, but the head of a phrase below.

Recent research (Li 2006, Paul and Pan 2017, Scherf 2017) has differenciated two positions for particles: Li (2006) distinguishes an outer position and an inner position.This author provides evidence for this distinction from Cantonese, Mandarin and Wenzhou, since in these languages more than one particle can occur in the same sentence: 1 Similarly, Scherf (2017) claims that in Swedish there are two kind of modal particles ocurring in differentiated positions: (2) Nog P har has väl P Peter Peter köpt bought boken?book.ART 'Peter definitely has bought the book, hasn't he?' In the view of these distinctions, the proposals to locate question particles across languages can be reduced to two: -Those which claim that interrogative particles occupy the head of CP or some other position in the Left Periphery (for Italian Rizzi 2001, for Mandarin Li 2006, Paul and Pan 2017, for Catalan Prieto and Rigau 2007, for Cantonese Kuong 2008, typologically Bailey 2013, for Japanese Kuwabara 2013), i.e., that they occupy an outer position.
As I hope to show in this article, Basque data confirm that there are typologically two positions for question particles.
In what follows, I shall be analysing question particles in as much detail as space permits focusing on their syntactic characteristics and putting aside the semantic contributions of these particles to the sentence.First, I will give a brief explanation of some aspects of Basque grammar dealt with in this paper; second, I will analyse the features which differentiate the two particles from each other; afterwards, in section 3, the weaknesses of previous analyses will be pointed out: one related to interpretation, the impossibility of -a occurring in embedded clauses and with allocutive forms and -a's compatibility with ote.Finally, section 4 will argue that, given the characteristics examined previously in this paper, the particle -a is the head of the Finite Phrase and the particle al is located in a separate Particle Phrase.Section 5 summarizes the main conclusions of the article.

A brief introduction to question particles in Basque
In order to create a polarity question in Basque, speakers from eastern dialects may use the question particle -a and speakers from the central dialect may use the question particle al. 23 These particles cannot occur together, since they belong in two different dialectal areas which are not in contact: 4   (3) Nehor anybody Nevertheless, polarity questions do not obligatorily require the presence of these particles in order to be acceptable, since Verb-Auxiliary fronting (Ortiz de Urbina & Hualde 1989& Hualde , 1993) ) and prosody mark a proposition as a question (Zuazo 2014); in fact, the data collected for this research confirms their optionality with some clarifications: in Central Basque, speakers resort to al as the preferred and prestigious way to ask polarity questions (cf.Hack 2014); in eastern dialects, the use of -a in yes/no questions is more a matter of sociolinguistic or generational variation (younger speakers use it less often than older speakers).Further research will result in more accurate data.

3
In this work and due to space limitations I will consider all the so-called question particles as true, pure question particles, although they are not obligatory in order to form acceptable polarity questions in most languages, as Xiang (2012)  Nevertheless, this rare example or hapax may after all turn out not to be relevant because of its isolated character and because the occurrence of both particles can be explained considering that the speaker was a native of the eastern dialects and studied in full immersion Basque schools, so the particle -a could be the particle adquired at home and al the one learned at the school.Below, I will shortly describe some basic grammatical issues of modal particles in Basque.

Modal particles in Basque
Traditionally, grammarians (de Rijk 1969, Euskaltzaindia 8 1987, Albizu 1991, Elordieta  1997) have grouped these particles, al and ote, together with others which seem to have a similar distribution; in the case of the particle -a few grammars have mentioned it and, when it has been cited (Azkue 1923, Laffite 1944), it has been labeled as a interrogative marker, separated from modal particles.It is the normative grammar of the Royal Academy of the Basque Language (Euskaltzaindia 1987: 486) who lists both of them, interrogative markers and modal particles, together under the term "elements attached to the inflected form": omen/ ei, ote, bide ahal/al and -a.
Furthermore, recent research (Haddican 2006, Monforte 2015) has led to a new classification considering semantics and contextual differences: (10) a. Evidential particles: omen/ ei b.Epistemic particles: bide and ote 6 In contrast to the particle al, it is not clear whether -a can occur in disjunctive clauses, since eastern dialects speakers do not always refuse such sentences.In what follows, I will briefly explain why these particles have been previously grouped together.
Although written separately, they usually function as proclitics to the finite verb.Evidence of this relationship is found in the following contexts (Euskaltzaindia 1987, Ortiz de Urbina 1993, Hualde et al. 2003).
In negative contexts the inflected verb is fronted; if a particle occurs, it moves along with the inflected form: (  (Euskaltzaindia 1987: 511) As regards the phonology, some morphonological changes occur between particles and auxiliaries, such as the so-called Sandhi rule and vowel change: 9 Nevertheless, Etxepare (2010: 91)

[daa P
→ dea] portuges?Portuguese 'Are there any Portuguese people here?'Therefore, based on the presented data, I propose that particles and inflected verbs form a morpho-phonological-word, in the terms stated in Elordieta (1997).
Before starting the analysis of these question particles, I will give a brief explanation of some aspects concerning the syntax of Basque Left Periphery in (2.3).

2.
3. An outline of assumptions made in the CP field in Basque Rizzi (1997) claims that two phrases can be differentiated in the Complementizer Phrase, one related to the illocutionary force (Force Phrase) and another one to the finiteness (Finite Phrase); along with the two phrases above, Rizzi also proposes two other phrases connected with the information structure, i.e.Topic Phrase and Focus Phrase.Ortiz de Urbina (1999b), following Rizzi (1997), proposes for the Basque language that polarity operators occur in Force Phrase, Wh-words and foci in Focus Phrase and complementizers in the Finite Phrase.He claims that, for instance, the Basque complementizer -(e)n, a suffixal complementizer attached to the inflected form, is a head expressing the finiteness morphology of the sentence, since complementizers such as -(e)n and -(e)la, are only found in tensed sentences.
Although phrases in Basque are claimed to be head-final (de Rijk 1969), headmovement to the left is attested in some contexts such as negation, question formation and focalization.Since the target of these movements is taken to be either CP (Ortiz de Urbina 1989), ΣP (Laka 1990), or FocusP (Uriagereka 1995, Ortiz de Urbina 1999a), the said target phrase has been considered sentence initial; however, Ortiz de Urbina (1999b) proposes that the split CP should be further articulated as follows (Ortiz de Urbina 1999b): 'They asked who John gave the book to.' (Ortiz de Urbina 1999a: 183) Therefore, not all phrases located in the Left Periphery are head-first: Ortiz de Urbina (1999b) claims that FiniteP is final and phrases above it are sentence initial.The head final character of FinP explains the canonical sentential position of complementizers at the end of the sentence: 'Was it Michael who went home?' In ( 21) we can see that main and auxiliary verbs (joan da) are fronted, adjacent to the focalised element (Mikel); as proposed by Ortiz de Urbina (1999a), in Basque the focalised element triggers movement to [Spec, FocP]; once the inflected form is lexicalized by the lexical verb, the complex V 0 -T 0 /I 0 unit moves to Foc 0 , in order to match features between the operator of the focal element and the T 0 /I 0 .Therefore, since there is movement to FocP, CP is split into the phrases detailed in (17).

Basque question particles: their syntactic characteristics
Below, I will introduce some characteristics of -a which make it different from al and the rest of particles.As stated before, I will not consider the semantic contribution of these particles; therefore, all characteristics mentioned above are related to their syntactic behaviour.However, I would like to indicate that these particles, -a and al, do not seem to be clause-typing ones, since they are not obligatory in order to form acceptable polarity questions, as said in footnotes 2 and 3; hence, I consider them as pragmatic markers.
Nevertheless, further research must be conducted to clarify their character.

Particles' sentential position
Previously, (2.2),I have pointed out that particles occur attached to the inflected form.As most grammars describe, modal particles are prefixed to the inflected verb: (22) Lurdes Lurdes etxera home.ADL joan go bide P da AUX 'Lourdes seems to have gone home' (Euskaltzaindia 1987: 502) Nevertheless, that does not hold true for both question particles.Indeed, the particle -a, unlike al and the others, appears suffixed to the auxiliary: (23) Hemen here bizi live dea?AUX.P 'Does s/he live here?' (Euskaltzaindia 1987: 500) Therefore, the particle -a shares its sentential position with complementizers such as -(e)la(ko) or -(e)n, since they also occur suffixed to the inflected form: (24) Galdetu asked du AUX (ea) ea egitea doing merezi deserve du-en AUX-C 'He has asked whether doing it is worth it' (Ortiz de Urbina 1999b: 181) As will be shown in (3.3) and (3.4), the particle -a could be closer to complementizers than other modal particles, concerning its syntactic behaviour and distribution.

Compatibility with other particles
Although the normative grammar of the Royal Academy of the Basque Language postulates the contrary, 10 there is evidence that the particle -a is compatible with the particle ote in the same proposition, as shown in ( 25 (Camino 2009: 193) This piece of data strongly suggests that the particles -a and ote are compatible both syntactically and semantically. 11Consequently, since -a is compatible with the epistemic ote, I propose that -a has no other function than reinforcing that the question is of the yes/no type.
On the other hand, the particle al cannot occur in the same sentence with any other particles: ( Typologically there is no such restriction.In fact, more than one particle can occur in the same sentence in languages such as Catalan (Prieto & Rigau 2007), Mandarin, Cantonese and Wenzhounese (Li 2006, Kuong 2008), Japanese (Kuwabara 2013), German (Bayer and Trotzke 2015) and Swedish (Swerf 2017) .

11
The presence of a particle reinforcing or marking a polarity question and another one conveying an 'I wonder' effect is well found in other languages such as Lillooet Salish, Thompson Salish, Gitksan (Littell et al 2010).The incompatibility between al and ote could be due to the fact that they occupy the same syntactic position or because they have similar semantics; nevertheless, I favour the syntactic approach, since, on the one hand, al is incompatible not only with ote, but also with the other particles such as omen, ei and bide and, on the other hand, because ote and al do not have the same semantics: whereas ote reconducts the addressee of the question to the speaker himself, al is compatible with a bona fide question.

Compatibility with allocutive forms
In Basque, when one is speaking to someone whom one would address with the pronoun hi (or zu in eastern dialects such as Souletin and Low Navarrese varieties), the finite verb requires the addition of an allocutive morpheme which corresponds to the overt addressee, which is in fact not an argument selected by the verb (see Haddican 2015) 12 .As show in (29), the auxiliary agreees with the subject (prefix n-, 1st sg.abs) but it also agrees with the addressee (-k 2nd sg.alc.mas and -n 2nd sg.alc.fem).These morphemes, -k and -n, are not an argument selected by the verb: (29) Barakaldora Barakaldo.ALL etorri come nauk/-n AUX.1SG.ABS.2SG.ALC.M/2SG.ALC.F 'I've come to Barakaldo.'Oyharc ¸abal (1993) shows that in classical Basque and in the eastern dialects the allocutive cannot occur if the particle -a is present.Therefore, even if the context leads to its use, the neutral form will be used:.-a? -P 'Does your mother know that?'This restriction is not related to the fact that the sentence is a yes/no question; although it is a generally acknowledged fact that questions in eastern dialects are a barrier for the allocutive forms (Alberdi 1994), we can find polarity questions inflected with the allocutive form in recent works (Coyos 1999: 205 (Thikoipe 2009: 39) As stated above, -a cannot appear in contexts such as ( 31) and (32) where the allocutive morpheme is present, although they are polarity questions.
However, the central particle al can attach to an allocutive auxiliary form: (33) Jaiatekuak festivals.ART al P dituk AUX.ALC oiek?those 'Are those ones for festivals?' (Altzaga 1888) Therefore, the two question particles pattern differently with respect to their compatibility with the allocutive paradigm.

Occurrence in embedded questions
Finally, these particles do not behave similarly regarding their occurrence in indirect questions. 13On the one hand, -a cannot appear in embedded sentences.This particle and complementizers both seem to occupy the same position: The occurrence of the particle in both the main and the embedded clauses provides evidence to limit the field of use of the particle al, since each particle in example (36) has scope over one clause. 1413 If we observe particles from other languages such as ka in Japanese, ma in Mandarin and si in Italian, we realise that not all of them behave similarly in embedded questions: some (the Japanese and the Italian particles) must appear in indirect questions, others (the Mandarin and Cantonese particles) are not possible in such contexts.Therefore, we could separate particles into two groups regarding their possibility to occur in embedded questions.
14 As stated by Li (2006: 30), the Mandarin particle ma has not an embedded question reading, although it occurs adjacent to the embedded clause; it will be always interpreted as a matrix yes/no question: Question particles in Basque Isogloss 2018, 4/1 39

Interim conclusions
In view of the evidence at hand, I conclude that al and -a do not behave similarly; consequently, I propose that they occupy separate positions in the syntactic structure.
The following section discusses the analyses put forward so far within the Generative framework and dismisses some aspects of these hypotheses on further empirical grounds.

Previous analyses
In this section I will present the grammatical issues which neither of the analyses proposed previously in Basque grammar can solve entirely; the analyses under examination are the following: Albizu 1991, Elordieta 1997, Elordieta 2001, Haddican 2008, Arregi & Nevins 2012.I classify the issues found in two groups: 1) the negation-particle scope issue and 2) other syntactic issues.

Who c-commands whom
It is an acknowledged fact that in positive sentences particles c-command the proposition, since they contribute to the interpretation of the whole proposition.However, in negative contexts negation also must c-command the whole proposition, therefore, there is a conflict between particles and negation.
As Haddican (2008) and Etxepare (2010) point out, in Basque the particle has scope over the negation and, therefore, over the whole proposition.Consider the following example: (37) Ez not omen P zuen AUX urik water.PTV topatu find 'Reportedly, s/he didn't find water.'(Haddican 2008: 72) The interpretation of this example is that "it is said that s/he didn't find water"; this proves that the particle c-command the negation because it adds evidentiality to the negative sentence.If the negation c-commanded the particle, the interpretation would be "it is not true that it is said that s/he found water".However, this is not the interpretation speakers get.The next example further clarifies the relationship between particle and negation: ( said not to have been born in Oiartzun, but s/he's said not to have been born outside Oiartzun.'(Haddican 2008: 72) The interpretation of this example is not acceptable in terms of the pragmatics, although it is grammatically correct: speakers are given the impression of a person who was born nowhere.The reason why the continuation of ( 38) is pragmatically infelicitous is because, according to Haddican (2008), this suggests that the particle c-commands the negation.Indeed, if negation c-commanded the particle, the interpretation would be totally normal, i.e. it is not the case that s\he is said to have been born in Oiartzun but it is not the case that s\he is said to have been born outside Oiartzun.Nevertheless, once again, this is not the interpretation speakers get.Therefore, particles c-command negation.
Based on this piece of data, Haddican (2008) and Etxepare (2010) propose that particles are heads of a ModalP located between PolarityP and IP/TP inside which negation is generated; therefore, in these analyses particles c-command negation as just claimed.Even though negation moves to PolarityP, i.e. over the phrase hosting particles, it reconstructs to its TP-internal position and, hence, we get the scope facts right.
The analyses of Elordieta (1997) and Elordieta (2001) also claim that particles occupy the head of ModalP; however, this phrase is located between IP/TP and VP and not on the top of the proposition.Therefore, regarding the scope over the whole proposition, neither Elordieta (1997) nor Elordieta (2001) meet the criteria of the scope domain of the particles over IP/TP. 15 Concerning the negation, the two analyses diverge: Elordieta (1997) suggests that it occupies a position below IP/TP, whereas Elordieta (2001)

. [VP]]]]]] (Elordieta 2001)
Regarding the fact that particles c-command negation, Elordieta (1997) explains the hierarchical relationship between particles and negation; Elordieta (2001), however, does not, since in this proposal particles are in any case located below negation and, as shown in (37), particles being c-commanded by negation would lead to an incorrect interpretation.
Therefore, these analyses cannot totally explain the scope showed by particles since ModP is located below IP/TP. 1615 Elordieta (1997) explains the scope domain of the particles over the whole proposition by LF-raising of the particle to CP.

16
The hypotheses of Albizu (1991) and Arregi & Nevins (2012) do in principle meet the criteria of the scope domain of particles over IP/TP: However, they do not discuss the position of negation in the syntactic structure; therefore, they cannot be tested according to this parameter.In conclusion, question particles must occupy the highest position on the syntactic structure, higher than the negation.17

Other syntactic issues
Now I will examine the hypotheses previously cited from a syntactic perspective.This will provide the basis to elaborate and build a new analysis.AUX.P 'Might the belly of the Miñarri mountain be full of salt (I wonder)?(Camino 2009: 193) None of the five proposals we are evaluating here could admit both particles in the same sentence because they provide the same head for all of them; therefore, the particles -a and ote should be assigned different heads.

Restriction of the particle -a in embedded questions
Similarly, none of these five analyses can explain why the particle -a cannot appear in indirect questions.If it behaved like the rest of the particles, we would expect it to occur in embedded questions.The usual explanation says that the incompatibility with the complementizers lies in the suffix character of -a.Nonetheless, as I have proposed above, if the particle -a is the head of the CP, or the Finite Phrase, the issue of the incompatibility with the complementizers is immediately solved, since both the particle and the complementizers will be occupying the same position.

Incompatibility of -a with allocutive forms
Considering that allocutive forms are excluded from any sentence where C is filled by a lexical element or its trace such as relative clauses, subjunctive complement sentences, indirect and direct questions, Oyharc ¸abal (1993) and Miyagawa (2012) claim that there must be a relationship between allocutivity and complementizers.Therefore, they propose that complementizers block the presence of allocutive forms, because allocutivity can only arise when an allocutive operator moves from [Spec, TP] to [Spec, CP] and C 0 is empty to be available for the operator (Oyharc ¸abal 1993) or when an allocutive probe, bearing uninterpretable agreement features, occupies the head of CP, which accounts for the complementary distribution between complementizers and allocutivity (Miyagawa 2012).
This hypothesis would explain 1) that -a can neither occur with allocutive forms nor 2) arise in embedded sentences.
Therefore, as in 3.2.2., none of these five hypotheses can explain why the particle -a cannot appear with allocutive forms.

Interim conclusion
To sum up all the information presented so far in this section, we should consider the following before proposing the syntactic position of the particles: a) Particles c-command negation; in fact, their scope domain is over IP/TP.b) Allocutivity (Oyharc ¸abal 1997, Miyagawa 2012) and complementizers occupy C 0 (Ortiz de Urbina 1999b); none of them is compatible with the particle -a, but they both are with al.Therefore, -a seems to occupy the same position as allocutivity and complementizers.c) Evidence from morphological, syntactic and prosodic fields proves that particles and auxiliaries merge; indeed, they form a morphological word (Elordieta 1997).
Therefore, since these particles have been shown not to behave similarly in some aspects, I conclude that they do not constitute a single group i.e. they seem to form two different kinds of particles (cf.Li 2006 on outer and inner particles) and they occupy separate syntactic positions.In the following section I will bring more arguments for assigning them different syntactic positions.

Towards a new analysis of al and -a
5.1.A new proposal for the particle al I propose that al occupies the head of the Particle Phrase (Bayer & Obenauer 2011), and that so do the rest of the particles except -a, since more than one of these particles cannot occur at the same time.I suggest that the Particle Phrase is located between the Finite Phrase and the Inflectional or Tense Phrase due to the following two considerations: a) With the exception of -a, which may be considered as a way to reinforce the polarity question type, the particles al, bide, omen/ei and ote do not mark the clause type; 18 in fact, typologically the occurrence of particles depends on the clause type (Coniglio & Zegrean 2010, Cardinaletti 2011, Haegeman 2014).b) As I have indicated the particle al and the complementizer -(e)n are compatible in the same clause.If we look at the order of the constituents, i.e.V-Part-Aux-C, the complementizer 19 is the last one to appear.Indeed, if the particle were hierarchically higher, there would be a problem with the interpretation because the particle would ccommand the complementizer.
For the time being, I will consider that all the particles, except -a, occur in the same head, i.e. the head of the Particle Phrase.
18 For instance, it would not be grammatically correct to ask a question using the particle ote alone, e.g.[Paulek nori liburua eman ote dio?] (Paule to whom the book gave, I'm wondering) although this particle only occurs in questions; other syntactic operations are required to form a question such as fronting of the Verb and Auxiliary and movement of the Wh-word in the case of  According to Cinque (1999: 56) complementizers always mark the limit of the periphery where particles occur.

particles in Basque
Isogloss 2018, 4/1 43 5.1.1.Syntactically analysing polarity questions containing the particle al In this section I will present my analysis for the particle al and prove that it solves the problems of the previous analyses.
Direct yes/no questions.First, let us consider a standard yes/no question containing the particle al, such as (42): (42) Ikusi See al P dezu AUX poligono industrial-area hori? that 'Have you seen that industrial area over there?' (Txurruka & Urbieta 2003: 148) The derivation proceeds as in (43).I suggest that the particle has a [uTense*] feature which is strong (Adger 2003) and T 0 has a [Tense] feature; after checking features, the strength of the particle triggers movement of T 0 to Part 0 .Additionally, T 0 has a [uQ*] feature transferred to Part 0 once both heads merge.Therefore, since it is a polarity question, the presence of a strong [Q] feature in C 0 triggers movement of Part 0 to that location so that they can check features.As stated in (2.2), the CP will be split only if FocusP and TopicP are activated; since in this example there is no topic, no focal element, the CP field is not articulated, so the target phrase of Part 0 is C 0 .Finally, al does not satisfy the lexical necessity of the finite T 0 , so, it attracts the nearest lexical head under its c-command, i.e. the Verb head moves to C 0 constituting the following head: [V 0 [[Part 0 -T 0 ]']]: Consequently, this analysis accounts for the syntactic relationship and the hierarchical position between the particle and the auxiliary.Embedded questions.Previously I have stated that the particle al and the complementizer -(e)n are compatible, as shown in (44).Now I will further illustrate that this analysis can account for the use of both particle and complementizer.Consider the following example: P.AUX 'They reportedly asked whether they had seen two young people leaving by horse as fast as the wind.' (Azurmendi 1996) First, since TopicP is activated in this example, the CP field is articulated as detailed in (17), repeated here for convenience: Therefore, as can be observed, the presence of the complementizer does not prohibit the occurrence of the particle al if they are located in C 0 /Fin 0 and Part 0 respectively.The restriction of the particle al .Finally, I will provide evidence that this analysis explains the incompatibility of the particle al with any other particles, as shown in (27)(28) and here in ( 47 As with the previous case, I will only examine the part concerning the restriction of the particle al.In this case the reason for its not occurring with other particles is that they all occupy the same position; therefore, if ote occurs, al cannot appear: Up to this point I have presented a new analysis of the particle al; below, I will proceed with the analysis of the particle -a.

particles in Basque
Isogloss 2018, 4/1 45 5.2.A new proposal for the particle -a I propose that the particle -a is located in the head of the CP or FiniteP (Rizzi 1997).Below I will present more evidence for this position: a) In other languages (Catalan, Japanese, Mandarin) Q-particles have been located mainly in ForceP (or the phrases immediately around the ForceP), which is claimed to mark the clause type (Li 2006, Prieto & Rigau 2007, Kuong 2008, Bailey 2013and Kuwabara 2013).In fact, this site is generally assigned to the complementizers in other languages, but not in Basque.According to Ortiz de Urbina (1999b) Basque complementizers occur in FinP.Consequently, it seems reasonable to propose the head Fin 0 as the position of the particle -a.
b) Secondly, in Basque the particle merges with the auxiliary; in Italian, however, the Q-particle does not merge with the auxiliary, i.e. se does not attract the inflected form in order to be adjacent and can occur even in non-inflected contexts (Rizzi 2001).We obtain also proof of the merging between the particle and the auxiliary in Irish, since the copulative verb merges with the particle and changes its form in this language: Moreover, considering that the particle and auxiliary merge, if -a occupied Force 0 , it would be difficult to explain how they merge in such contexts as in (50) since -a would appear on its own in Force 0 : 20 In Basque the focalised element trigers movement to [Spec, FocP]; whereas the complex V 0 -T 0 /I 0 unit moves to Foc 0 , in order to match features between the operator of the focal element and the T 0 /I 0 .
ad hoc morphological rule could be claimed in order to amend this incorrect order, i.e. move the Force 0 to Foc 0 at FF so that particle and inflected verb are adjacent.Nevertheless, I suggest that the derivation is more natural by locating the particle in Fin 0 , considering that the auxiliary would merge with the particle while triggering movement to Foc 0 : 21 In this case, 22 apparently, the particle -a has not merged with the auxiliary, but the focalised constituent: NIK-a 23 .If -a were located in Force 0 , an ad-hoc morphological movement would be necessary; however, as illustrated in (54), the particle and the focus merge as a result of silencing the constituents below the Finite Phrase: Considering this evidence, I conclude that the particle -a occupies the head of the Finite Phrase (or the CP) differently from the rest of the particles.5.2.1.Syntactically analysing polarity questions containing the particle -a: Now I will present my analysis for the particle -a and, as previously done in (5.1.1)for al, prove that it solves the problems of the preceding analyses.Direct yes/no questions.After proposing a new position for -a, now I will explain the derivation of a yes/no question containing the particle -a step by step: (55) Nahi want (d)uka AUX.P borroka?fight 'Do you want to fight?' (Etxamendi 2010(Etxamendi [1989]]: 74-75) Following Adger (2003), I theorise that the auxiliary has a [uclause type*] feature which is strong and -a has a [Q] feature; after checking features, the strength of the particle triggers movement of T 0 to C 0 .Finally, -a does not satisfy the lexical necessity of T 0 , so, 22 Other similar structures have been found in eastern dialects (Larrasquet 1931, 1935, Estornes 1985): (1) Baia?Considering this behaviour, -a may be thought to be a focus particle such as desu in Japanese (Kuwabara 2013) or b'a or ay'a in Somali (Saeed 2000); however, -a does not appear in other contexts than polar questions, never in declarative clauses containing a focal element, and it do not always occur in polar questions with focal elements.Consequently, this analysis explains the syntactic relationship and the hierarchical position between the particle and the auxiliary.
Embedded questions.As noted above, -a and complementizers are in complementary distribution.Therefore, in an embedded question, since the complementizer -(e)n appears in the Fin 0 following Ortiz de Urbina (1999b), this prevents the occurrence of -a: If -a were in the Part 0 , there would be no impediment for -a to appear in this kind of sentence.However, we know it cannot; therefore, this analysis explains its restriction.Polarity questions where two particles occur: -a and ote.Finally, I will prove that this analysis explains the compatibility of -a with other particles such as ote, as shown in ( 59 The particle -a occupies the head of FinP, since the presence of a focalised element activates the split CP; and, on the other hand, ote is located in the head of PartP.Therefore, as shown in the diagram, there is no constraint for these particles to occur in the same proposition: (60) ForceP In addition to this, in this example it can be observed that if the particle were in Force 0 , the inflected verb and -a would not merge, unless an ad hoc movement of -a to Foc 0 were employed.

Summary
In this paper I have examined two particles which occur in polarity questions in Basque, the eastern particle -a and the central particle al. Considering that they have a different behaviour in three relevant aspects (they are not used in the same way in embedded questions, with allocutive forms and with respect to other particles), I propose that they each belong to a separate set of particles (Hagstrom 2004, Kuong 2008) and that they occupy separate syntactic positions: -a occurs in the head of CP or FinP and al appears in the head of Particle Phrase.This last phrase is located between FinP and TP, and the other (epistemic and evidential) particles (ote, omen/ei, bide) occupy the same position.
4.2.1.Compatibility of the particles -a and ote As previously shown in examples (25 & 26) and (41) below further corroborates, -a and the epistemic ote can appear in the same sentence:
Rizzi (1997)t John gave the book to Michael.' On the other hand, asRizzi (1997)states, the CP will be articulated only if necessary; therefore, in this paper CP in main clauses will not split into ForceP and FinP, unless FocusP or TopicP are activated (see Kuwabara 2013 for a similar proposal for Japanese).