Syntactic analyses of discourse particles through the microvariation of Basque ote

In this article I look into dialectal data related to Basque discourse particles and I provide evidence that topics discussed in the literature on discourse particles can also be addressed within Basque microvariation. Four syntactically different types of the particle ote are examined: first, the general pattern found in all dialects where ote functions as a head and occupies a position in the tp-domain; second, the use typical of the eastern dialects where ote behaves as a maximal projection and is located in the cp-domain; furthermore, the configuration formed by wh -words and the discourse particle ote only found in the North-Eastern Basque; and, finally, a novel type of ote only used in the far east of the Eastern Basque (included the Souletin Basque, Amikuze’s variety and also the disappeared variety from the Roncalese valley) which functions as a sentence final particle occurring after all constituents of the clause and, apparently, conveying an intersubjective interpretation, unlike its canonical behaviour. Along with the syntactic data, phonological-prosodic and interpretational evidence also point to the same direction, i.e. to differentiate those four types of ote . This microvariation displayed by the particle ote allows us to compare the different syntactic statuses and positions claimed for discourse particles side by side in the same language; thus, the analysis of their separate properties attains a significant improvement.


Introduction
Two topics on the syntax of discourse or modal particles have dominated the literature lately:1 on one side, the debate on the syntactic status of these particles between those positioning in favour of considering discourse particles as heads and those claiming that particles are phrases occupying specifier positions.On the other side, the position of discourse particles in the syntactic structure has also received attention in the literature with some proposals locating them in the tp-domain (Coniglio 2007;Bayer 2009;Scherf 2017; among others), some others in the cp-domain (Kuong 2008;Coniglio & Zegrean 2012;Kuwabara 2013) and, finally, other studies claiming that they occupy a position in the Speech Act layer above cp (Munaro & Poletto 2002;Haegeman 2014;Corr 2016;Heim et al. 2016).This article examines the Basque discourse particle ote and its microvariation in eastern dialects 2 and discusses the issues mentioned above from a perspective internal to Basque data.First, the data found in those varieties bring up the question of whether the behaviour of discourse particles in Basque agrees with that described for heads, as has been claimed in the literature (Albizu 1991;G. Elordieta 1997;A. Elordieta 2001;Haddican 2008;Arregi & Nevins 2012;Lizardi-Ituarte 2017;Monforte 2018), or, rather, whether they should also be considered maximal projections (Etxepare 2010;Etxepare & Uria 2016).Secondly, I argue that ote occupies different syntactic positions depending on whether it functions as a head or a maximal projection: the head ote, as the other canonical particles, is base-generated in the tp-domain, whereas the weak adverb-like ote occurs in the cp-domain. 3Additionally, two other behaviours of discourse particles discussed in the literature are found in Basque concerning the discourse particle ote: the combination of the particle and wh-words creating a single constituent (Trotzke & Turco 2015) and the use of ote at the end of the utterance, similar to sentence final particles (Izutsu & Izutsu 2013).
The term particle, as pointed out by Paul (2015), has been used for those words which have not successfully been assigned their own particular category.Indeed, in languages such as German or Italian it has been debated whether discourse particles form their own category or they belong to the same category as their historically related counterparts, despite of their function in the proposition.For instance, some of the discourse particles in German derive from adverbs as doch, einfach, auch, eingentlich, wohl, bloβ and schon according to Meibauer apud Struckmeier (2014: 18); therefore, some authors (Abraham 1991;Jacobs 1991;Zimmermann 2004;Cardinaletti 2007) consider them a subclass of adverbs, as Struckmeier (2014) suggests. 4urthermore, the vagueness of their character makes it complicate not only to decide which words belong to the group of discourse particles, but also to establish their status within the generative framework.For instance, let us consider the following properties usually listed to describe discourse particles: a) they have a fixed order, unlike the majority of adverbs;5 b) they cannot occur in first position in V2 languages such as German; c) they behave mostly as clitics; d) they are not stressed;6 e) they are diachronically related to elements of other categories; f) they cannot be the only element as a reply to a question; g) they cannot be coordinated or modified; h) they cannot be topics or foci.Apart from the aforementioned ones, other general properties are also usually listed when describing discourse particles; for instance, that they cannot all occur in every kind of clause, i.e. they are clause-type dependant (Zimmermann 2011;cf. Rapp 2018), or that they can only be used in clauses containing full illocutionary force; hence, they can appear in root clauses and in certain embedded clauses (Coniglio & Zegrean 2012).Analyses on the syntactic nature of discourse particles take into account those properties as evidence in favour of their headness status.Likewise, these same properties (a-g) have been used to postulate that particles function as maximal projections.Indeed, the two hypotheses mentioned earlier (those positing that discourse particles can have x 0 or xp status) have been proposed for two well-studied languages such as German and Italian based on the previous properties (a-g).Hence, the same evidence leads to contradictory hypotheses regarding the syntactic nature of discourse particles.Some authors (Bayer & Obenauer 2011;Struckmeier 2014) claim that German discourse particles behave just as described above; therefore, they consider German particles to be heads, namely the head of the Particle Phrase located in the Middle Field.In a similar vein Coniglio (2008) concludes that Italian discourse particles are heads.Interestingly, he also examines German discourse particles and reaches the conclusion that these are maximal projections occupying a specifier position.
Coniglio is not the only one denying the head status to German particles; indeed, Cardinaletti (2011) supports this idea.Based on a previous work on pronouns (Cardinaletti & Starke 1999), she distinguishes three levels for adverbs: full adverbs, weak adverbs and clitic adverbs.7She claims that discourse particles in German are weak adverbs considering their syntactic and phonological behaviour.Let us briefly illustrate the main arguments Coniglio (2007) and Cardinaletti (2011) use in favour of the non-head status: a) discourse particles are closer to full words than clitic functional words concerning their phonological properties and prosody; b) if they were heads, they should block V raising, for instance, in German where they are supposed to occupy a position between tp and vP; or, otherwise, particles should move along with v; c) scrambled detps and pps can appear between two discourse particles, not an expected distribution if they were syntactic heads.Therefore, Cardinaletti (2011) and Coniglio (2007;2008) conclude that discourse particles in German occupy a specifier position where they remain since they are weak adverbs, i.e. because of the lack of some phonological properties.
So far, I have briefly reviewed the hypotheses on the syntactic status of discourse particles.In the Section 3 I will present data of the Basque language to conclude that discourse particles in that language behave as clitic-heads in their canonical pattern.Nevertheless, microvariation related to the discourse particle ote found in Eastern Basque shows that this particle can also behave as a weak adverb.This will be examined below in Section 4. Let us briefly exemplify both statuses:8 (1) Head status, Standard Basque Non utzi ote dut egunkaria?where leave part aux newspaper.abs'Where did I leave the newspaper?(I'm wondering)' 8 (2) Phrasal nature, eastern dialects Non utzi dut ote kazeta?where leave aux part newspaper.abs'Where did I leave the newspaper?(I'm wondering)' The debate on the syntactic position of discourse particles has been less controversial in the literature.Most of the analyses have looked into the distribution of particles along the clause, namely with respect to verbs, subjects, foci, topics and other particles.Such an approach is found in Scherf's (2017) work on Swedish particles: she examines the behaviour of particles, for instance, in contexts where the verb moves to cp or stays in situ, and in clauses containing strong and weak pronouns and she concludes that they occupy a high position in the tp-domain based on the fact that they always occur after the verb when this is moved to c 0 but higher than subjects which are claimed not to have moved from tp.German discourse particles have also been granted a similar position, i.e. high in the tp-domain (cf.Coniglio 2007;Bayer & Obenauer 2011;Struckmeier 2014) considering their distribution and restrictions along the clause.Nevertheless, this is not the only syntactic position where discourse particles can appear: the cp-domain has also been identified for some particles in Italian, Romanian and Asian languages such as Japanese (Kuwabara 2013;Del Gobbo, Munaro & Poletto 2015;among others).Not only syntactic distribution (namely at the beginning or the end of the clause) and strict order led into this conclusion, but also pragmatic arguments, since they modify the illocutionary force directly from that position, unlike those cases locating them in the tp-domain which need an lf movement of the particle to the cp-domain to account for its contribution.Finally, some works have rescued the idea of Speech Act Phrase following Ross's (1970) performativism and located the particles there.Authors following neoperformativism basically take the interpretational function of particles into consideration and how related they are to the subject or the addressee and the propositional content or the speech act; based on that, particles can be distributed along four phrases: (3) Additionally, authors such as Haegeman (2014), Speas and Tenny (2003) and Corr (2015;2016) put forward analyses in a similar direction.For instance, Corr (2015;2016) distinguishes two phrases related to the speech act of the clause: the Speech Act High Phrase (henceforth SAHighP) and the Speech Act Low Phrase (henceforth SALowP).The former is performative since it establishes discourse set-up and the speaker-addressee interaction; the latter is related to the speaker attitude and encodes modal and evaluative values (Corr 2015: 12).They both occupy positions above cp-domain, namely in the Speech Act phrase domain; also, they occur in a strict order: SAHighP is higher than SALowP.In this article I argue that these three domains are found in Basque regarding ote: the Middle Field (1), the cp-domain (2) and the Speech Act domain (4). 9 (4) Apart from these positions out of the clausal spine, Bayer and Trotzke (2015), Trotzke and Turco (2015) and Trotzke (2018) have related discourse particles to smaller syntactic enviroments such as Determiner Phrases or phrases containing wh-words.Setting the construction within discourse particles found in German (Trotzke 2018) aside, the latter has been found in typologically different languages such as German, Italian and Japanese (Munaro & Poletto 2002;Bayer & Trotzke 2015;Endo 2018).Also, the Basque particle ote occurs combined to wh-words: (5) Combination between wh-word and ote, North Eastern Basque Zer ote ari da haur hori?what part prog aux child that.abs'What is that child doing?('I'm wondering)' 10 9 Although I do not look into detail the interpretation of ote, it is usually translated as 'I'm wondering' following the traditional interpretation (de Rijk 2008).However, I do not translate it so in the case of the most restricted noncanonical ote ( §6), since it may have a different reading in this use.The contexts they appear are not alike and the fact that speakers from western dialects who lack this use refuse the particle even in its canonical position ( §6) lead us to think that.Therefore, I will not provide paraphrases for such use and mark its contribution by using 'ote'.
The paper is organised as follows: some properties of the Basque language will be briefly explained so that the reader can fully understand the data dealt with below; then in Section 3 the canonical behaviour of discourse particles in Basque will be examined and I will conclude that they behave as heads and occupy a position in the tp-domain; further, in Section 4 the non-canonical pattern of the particle ote found in eastern dialects is looked into and it is proposed that it has a maximal projection nature located in the cp-domain; in Section 5 the construction formed by a wh-word and the discourse particle ote is described and I argue that they form a single constituent, unlike the analysis claimed by Munaro & Poletto (2002)  As can be observed in the previous examples, finite verbs are for the most part analytic, i.e. composed of a morphologically independent lexical verb carrying aspectual information and an auxiliary form bearing tense, mood, and agreement with the arguments and, in some cases, also the addressee (Oyharçabal 1993; Alberdi 1994; Lizardi-Ituarte &  Munduate 2015). 11 Lexical and inflected verbs are usually adjacent; indeed, nothing can occur between them, such as adverbs: Zer ari ote da haur hori?what prog part aux child that.abs'What is that child doing?('I'm wondering)' As can be gathered from examples (5) and (i), ote conveys the same interpretation. 11Additionally, there are about 12 verbs which can be synthetic but only when the aspect is punctual (Laka 1996), for instance: (i) Leirek hori dakar.Leire.ergthat.absbring.prs.ind.3sg.erg.3sg.abs'Leire is bringing that.' These synthetic forms present the same information as periphrastic ones but realised in a single constituent.Also, they show similar syntactic behaviour as auxiliary verbs.
(9) *Xabier etxera etorri lehen da.Xabier.abshouse.all come before aux (Intended reading: 'Xabier came home before.')Nevertheless, the adjacency between lexical and auxiliary verbs12 can be broken in some contexts such as in negative (main) clauses:13 (10) Ez naiz Lindaura joan ez naiz.not aux Lindau.allgo 'I didn't go to Lindau.' Furthermore, Eastern Basque offers another context where this adjacency does not arise: whereas the standard procedure to form focal structures and wh-questions is by fronting both lexical and auxiliary verbs, in eastern dialects the inflected verbal form can be the only constituent moved next to (and immediately following) the focal element or whword, leaving the lexical verb in situ: As can be inferred from the previous examples, the cp field is head-initial in Basque.In fact, for this article I adopt the analysis put forth by Ortiz de Urbina (1999;2008) following the cartographic approach.Based on syntactic and morphological grounds, he sketches a structure where the finp and phrases below it are head-final and those phrases above it are head-initial. 14Additionally, he proposes that the target-position for wh-words, and hence the verb, is focp: Also, the focalised constituent and the verb move both to the cp-domain in questions, so that they are adjacent.In fact, nothing can intervene between the wh-word or focus and the verb in Basque: (15) Non (*sarritan) erosten du Liriak Riesling lehorra?where often buy.ipfvaux Liria.ergRiesling dry.abs 'Where does Liria often buy dry Riesling?'Once we have established some basic notions on Basque grammar, I will proceed to examine syntactically discourse particles in Basque in the next section, with special focus on the particle ote.
3 Canonical discourse particles in Basque: X 0 in the TP-domain The particle ote is usually grouped with other discourse particles, namely ahal, al, bide, ei and omen. 15Despite not being a long list, they can be classified into different groups: those conveying evidentiality (ei, omen), those expressing epistemicity (ahal, bide, ote) and the interrogative particle (al); for instance: ( 16) Txiki Iurretan lanean hasi omen da.Txiki.absIurreta.inwork.inbegin part aux 'Reportedly, Txiki began working in Iurreta.' Ba ahal dakizu arbola usteletik ez daitekela etor fruitu onik! cl part know tree rotten.ablnot can.compcome fruit good.prtt'You certainly know that a rotten tree cannot give good fruits!' (Larzabal 1992) These particles are traditionally considered to form a single syntactic group since they all are claimed to modify the illocutionary force and they behave in a similar manner.One shared property is that the canonical position of discourse particles in declarative sentences is between the lexical and the inflected verbs: (18) (*Ei) Marga (*ei) etxera (*ei) etorri (ei) da (*ei).part Marga.abspart house.allpart come part aux part 'Reportedly, Marga came home.'Another shared property is that all discourse particles form a morphonological word attached to the inflected verb (Arregi & Nevins 2012).Evidence of this relationship is found in contexts mentioned above ( §2) where the inflected verb is fronted, whereas the lexical verb stays in situ as in negative contexts ( 19) and focus contexts in eastern dialects (20); in such contexts the particle moves along with the inflected verb. 16Moreover, the fact that discourse particles cannot be found in clauses lacking the inflected verb reinforces the idea of their dependency on such verbs since discourse particles cannot be used if finite t does not occur (21 and 22).Additionally, forming a morphological word with the inflected verb implies that nothing can intervene between the particle and the inflected verb; this prediction is borne out: Txiki Iurretan lanean hasi omen (*berriro) da.Txiki.absIurreta.inwork.inbegin part again aux 'Reportedly, Txiki began working in Iurreta.' Indeed, if the particle stayed in situ after the lexical verb, the sentence would be ungrammatical: (24) *Jonek dia Jonek hori erran ote du? Jon.erg aux.part that.abs say part (Intended reading: 'Is it John who said that? (I'm wondering)') Furthermore, synthetic verbs are banned from the very first position and, therefore, an expletive morpheme ba-is required to precede the verb (Ortiz de Urbina 1989; 1994; A. Elordieta & Haddican 2018;among others).Even when a discourse particle appears before a synthetic verb in clause initial position, this does not count as a constituent and the use of ba-is still obligatory: (25) *(Ba) omen daki cl part know 'Reportedly, s/he knows it.' Concerning their phonological-prosodic properties, although the prosodic characteristics of particles in Basque should be studied deeply, at first glance it seems that they behave as follows: first, regarding the issue of whether they receive stress or not, they do show stress although this is not always located on the same syllable since it depends on the length of the inflected verb they are attached to.Let us briefly exemplify this with the particle ote: In some prosodic systems 18 the o-receives the stress if the inflected verb has two syllables (26a), but it is -te which receives it if the inflected verb has only one syllable (26b).Hence, Basque discourse particles do not have inherent stress.Moreover, it is the combination of particle and inflected verb which receives stress 19 and, therefore, they form a prosodic word. 20Interestingly, these particles may suffer apheresis in some contexts as the result of the position of the accent: ( 27) Barandiaran (1972 In addition, they cannot occur by themselves since they need to occur always attached to the inflected verbal form; this seems to be a common property of discourse particles cross-linguistically: (29) -Heldu da David Bostonera?-*Ei.arrive aux David.absBoston.allpart '-Did David arrive in Boston?-Reportedly.'So far, this pattern is attested in all particles; indeed, the only difference they present concerns their interpretation. 18The high degree of fragmentation of the Basque language also affects its prosodic pattern; this results into a system where some varieties have a [+2] system, whereas others [-2] and so on (Hualde 1997). 19One reviewer points out that the particle ote and the inflected verb daite both have stress in example (26a).This is true; however, they are not the same kind of stress: whereas the stress received by ote is the primary stress, the one received by daite should be considered the secondary stress, i.e. othédaitè (cf.Hualde 1997). 20In the Basque of Lekeitio (Hualde, Elordieta & Elordieta 1994: 57), the presence of the discourse particles ei or ete (ote in the western dialect) has an impact on the location of the stress provoking that the particle and the inflected verb constitute a new prosodic compound which differs from the one made up of the inflected and the lexical verb: (i) Gâur etorri-diras?today come-aux 'Did they come?' (ii) Gâur etorri ete di'ras?today come part aux 'Did they come?(I'm wondering)' Finally, it is an acknowledged fact that discourse particles in German or Italian (Thurmair 1989;Coniglio 2008;Cardinaletti 2011;Bayer 2012;Bayer & Struckmeier 2017) are historically related to other categories such as adverbs or conjunctions.This field has been recently explored in Basque linguistics (Lakarra 2019)  This hypothesis has improved our understanding of discourse particles in Basque since they show an impoverished nature (restricted syntax, bleached interpretation, weakened phonology and prosody), as expected for grammaticalised constituents, and occupy the canonical position of lexical verbs, namely preceding the finite verb. 22ased on their syntactic behaviour, Basque grammarians have considered discourse particles as heads which behave clitic-like and occupy a position in the tp-domain (G.Elordieta 1997;A. Elordieta 2001;Monforte 2018), since they always move along attached to the head of tp.Furthermore, the properties cross-linguistically claimed for discourse particles functioning as syntactic heads (Bayer & Obenauer 2011;cf. Gutzmann 2015) are also found in the Basque ones: a) they have a fixed order (18); b) they cannot occur in first position in v2 structures (25); c) they behave mostly as clitics (19, 20 & 24); d) they have no own stress (26); e) they are diachronically related to elements of other categories, in this case to verbs (30); and f) they cannot be the only element as a reply to a question (29) (see (Monforte 2019a;2020b) for further details). 23In addition to this, the fact that they move with the finite verb when this rises to the cp-domain (Coniglio 2007;2008;Scherf 2017) also reinforces the hypothesis that they function as heads.
Regarding to their syntactic position, particles in Basque also display a fixed order, since they always occur preceding the inflected verb, as observed in ( 16) and ( 17).Considering that the position they occupy in neutral contexts, i.e. between the lexical and finite verbs, and the fact that they derive from lexical verbs, it seems reasonable to locate their phrase between vP and tp.Moreover, their dependency on finiteness leads into this hypothesis.Let us illustrate this (remind that Basque is a head-final language).
(i) *Peter har väldigt antagligen köpt boken.Peter aux very probably bought book.abs.def(Intended reading: 'It is very probable that Peter has bought the book.')Also, they could not behave as focus because they cannot evoke alternatives.Thus, those restrictions usually consider to be caused by their syntactic nature should be attributed to their semantics.Following these ideas, such properties play no role in our current discussion. (31) To sum up, in this section I have demonstrated that discourse particles in Basque behave as heads and occupy a position in the tp-domain.Nevertheless, the particle ote has more than one grammar; as I will argue in the following section, this particle also behaves as a maximal projection occupying a position in the cp-domain.

Microvariation of ote in eastern dialects: XP in the CP-domain
Discourse particles have been proved above to function as heads based not only on their syntactic behaviour, but also on the data collected by testing the cross-linguistic properties of discourse particles.In this section I will provide evidence that ote can behave either as a head or as a weak adverb in Eastern Basque; in order to avoid confussion between both kinds of ote I will refer to the ote under exam now as ote2.
As observed before, discourse particles have a fixed position in the clause, i.e. preceding the inflected form.Nevertheless, data from eastern dialects goes against this statement, since ote2 occurs also in a postverbal position: (32) The fact that ote2 always occurs after the finite verb is not surprising since we are dealing with questions and those, as described above, always show the verb fronted in matrix questions.The same distribution is found in embedded contexts if they are wh-questions or polar questions containing a focal constituent.Nevertheless, it does not appear right after the finite verb, because the complementizer -(e)n (33 and 34), the interrogative mark -a (32) or the discourse particle bada ('then') may occur in between: Nor deitzen du bada ote Peiok egun guziz hain goizik?who call.ipfv aux part part Peter.ergday all.insso soon 'Who does Peter phone every morning so early?('I'm wondering)' However, ote2 also has a specific position in the syntactic structure, after the finite verb and the discourse particle bada ('then') and before the subject; in fact, appearing in other positions results into the ungrammaticality of the sentence: (36) Nor deitzen du (ote) Peiok (*ote) egun guziz (*ote)?who call.ipfv aux part Peter.ergpart day all.inspart 'Who does Peter phone every morning?('I'm wondering)' This agrees with the properties proposed by Cardinaletti (2011), i.e. that weak adverbs, unlike full adverbs, may only appear in a fixed position.
Following these authors, proof that the particle does not function as a clitic-head comes from the fact that the particle does not attach to the finite verb and form a prosodic word with it; in fact, ote2 shows an independent intonational contour.Let us illustrate this: (38) *Ez düa te eginahala egin?(cf.32) not aux.partpart effort.absdo (Intended reading: 'Didn't s/he do everything possible?') Concerning our topic, this is relevant because it suggests that ote and ote2 have separate phonological nature.
A similar situation is also attested in Austrian German (Coniglio apud Bayer & Obenauer 2011); those who claim that discourse particles in German behave as syntactic heads argue that the adverb vielleicht 'maybe' can be reduced to leicht in the Austrian variety but only when it functions as a discourse particle, i.e. as a head: It is an acknowledged fact that German discourse particles cannot occur in first sentential position; thus, vielleicht can occur in that position only if it had an adverbial reading and not a particle reading.Nevertheless, contrary to what has been stated in the literature, I would like to point out that this difference does not depend on the syntactic status of discourse particles (i.e.heads or maximal projections), but on their phonological nature, i.e. whether they are clitics or not.In fact, a x 0 or xp can function as a clitic and, hence, be phonologically reduced.
In addition to this, ote2 can occur in clauses lacking the inflected verb, unlike the one described in Section 3: (41) Non ote?where part 'Where?('I'm wondering)' (42) Moreno (2018) Pentsatzen dizü bihamonin oano harri hoi nola elkhiko ote.think.ipfvaux following.day.instill stone that.abshow leave.futpart 'The following day you keep thinking how they will remove that stone.'Moreover, another difference compared to the discourse particles described above is that they do not block the rise of the finite verb to cp and they do not move along together, as Coniglio (2008)  Based on this evidence, I conclude that ote in eastern dialects can behave not only as a head, but also as a maximal projection following Coniglio's (2007) terminology or as a weak adverb following Cardinaletti's (2011) terminology. 2424 Etxepare (2010) and Etxepare and Uria (2016) examine the evidential particle omen which also shows microvariation in the same dialects and claim a similar analysis in terms that omen can function as a head and as a phrase.Nevertheless, there are some differences between both particles.For instance, omen can be used as an answer to a question in eastern dialects; however, ote cannot (a restriction shared also by German discourse particles).This may suggest that ote is more restricted than omen, not only syntactically, but also semantically.However, the parallelism that both particles have separate syntactic statuses stands.
As for its syntactic position, we have seen that ote2 can occur in non-inflected clauses; moreover, it appears next to the focus of the sentences in those contexts which have suffered the ellipsis of tp: Non ote Aitor egon da? where part Aitor.absbe aux 'Where (may Aitor be)? ('I'm wondering)' (47) Berlinen ote Aitor egon da? Berlin.inpart Aitor.absbe aux 'In Berlin (may Aitor be)? ('I'm wondering)' This pattern cannot be accounted from the position located below tp proposed in Section 3. In fact, the sentential position and prosody of ote2 in such contexts do not agree with the pattern canonical discourse particles have and they are reminiscent of Rizzi's (2004) Modifier phrase.Rizzi details constituents occurring in such position as follows: a) adverbs in that position share the same prosody as topics, althougth they are not connected to the background and, therefore, they cannot be considered as topics; b) adverbs in that position modify the constituent they interact with.Concerning the intonation contour of ote2, it can be noticed that it goes higher at the end, similar to topics.As for the second property, note that Monforte (2020c) argues that, only if focp is lexicalised, ote2 is licensed in such context.In other words, according to this author ote is dependent on the activation of focp since it can only occur unless a focalised element has moved to such position (i.e.focp); a similar distribution is described for Italian and German by Munaro & Poletto (2002), Bayer & Obenauer (2011) and Egg & Mursell (2017).Taking the properties described for constituents appearing in Modifier phrase and for the particle ote in eastern varieties into account, there is enough resemblence to posit that ote behaves in a similar way to those adverbs claimed to occupy Rizzi's (2004) modp. 25Returning to the issue on the position of this kind of ote, the position of modp agrees with that expected for ote since it is above tp and below focp, but more importantly, in the cp-domain: To summarise, ote in eastern varieties functions not only as a head but also as a maximal projection as its phonologic-prosodic properties (independent prosodic word, intonational 25 Nevertheless, the particle ote2 is not a full adverb like those examined by Rizzi (2004) as its restriction to appear in first position or to move from its base-generated position suggest.In fact, Cardinaletti (2011) gathers those elements traditionally named as discourse or modal particles in the Germanic linguistics under the term weak or clitic adverbs based on their impoverished nature.Therefore, it is expected that weak adverbs such as ote have a more restricted syntactic behaviour or weakened phonology or functional interpretation compared to full adverbs.In our current case, ote would have a more reduced syntactic nature; this would explain its fixed position.
contour similar to topics, no apheresis) and its syntactic properties (fixed position, no movement attached to the verb even when the latter moves to cp-domain) suggest.Also, it occupies a position in the cp-domain, namely in Rizzi's (2004) Modifier phrase considering the resembled pattern between ote2 and Italian adverbs occurring in such position.

On an (un)usual context of discourse particles
Thus far, discourse particles in Basque have patterned mostly as described in the literature, i.e. they function as free constituents or similar to clitics by attaching to a specific word in the clause.Beside these patterns, there is another one which has recently received more attention in different languages, namely the 'wh-word particle' configuration; however, it does not seem as common as the behaviour of particles described above.This construction consists of a discourse particle combined with a wh-word, following the strict order 'wh-word -particle': (49) wh-word part / *part wh-word Combinations between wh-words and discourse particles are found cross-linguistically.This can be found in languages such as German (Abraham 1991 They have received distinct analyses: some posit that the wh-word and the particle form a single constituent (Bayer & Obenauer 2011;Bayer & Trotzke 2015;Endo 2018), whereas others (Munaro & Poletto 2002) claim that the particle maintains its position in the cpdomain and the adjacency between wh-word and particle is due to the lack of movement of the verb or the remmant constituents below it.I adopt the first hypothesis and, thus, put forward that wh-words and particles combine together in the base-generated position expected for the wh-word and that they form a single constituent.
Similarly, Basque also has this configuration, although it can only be found in north-eastern dialects (Trotzke & Monforte 2019) This construction arises not only by the combination between a wh-word and the particle ote, but also between a wh-word and the discourse marker ba(da): (56) Borda (2005) Nondik bada zetozen eskatu zien.where.ablpart come.compask aux 'S/he asked them where they were coming then from.'Note that the usual order in main questions is that of [wh-word + verb] and that no phrase can intervene between the wh-word and the verb, not even adverbs such as atzo 'yesterday', since the wh-word and the verb have been both fronted to the cp-domain and are in a [specifier-head]  Based on that evidence, the idea claimed in the literature that wh-words and discourse particles form a single constituent fits the Basque data, as the obligatory adjacency between the wh-word and verb suggests (cf.53 vs. 57) and, hence, the 'wh-word ote v' distribution can only be explained by assuming that 'wh-word ote' counts syntactically as a single element.
However, one should evaluate two alternative accounts available for this configuration in order to reinforce my analysis by showing that the two are not appropiate for our data: on the one hand, ote could be attached to the verbal complex, i.e. the merge between lexical and inflected verbs, instead of being adjacent to the inflected verb as in the canonical behaviour.In other words, [ ote [ v aux ]] would be the analysis, in contrast to the canonical pattern [ v [ ote aux]].On the other hand, analyses within the Italian languages claim that the particle occupies the same position in this kind of configuration as when it is not attached to another constituent (Munaro & Poletto 2002;Coniglio 2008);27 the difference between them is related to the movement of the wh-word or the whole cp: Therefore, in cases such as (58a) only the wh-word rises to the specifier position of the Functional phrase containing the particle (cf.59a), whereas in clauses such as (58b) it is the whole cp which moves to the specifier position of the Functional phrase (cf.59b).
Nevertheless, as the following data show, these approaches must be rejected, at least regarding the 'wh-word ote' combination.Note that in North-Eastern Basque the adjacency between the wh-word and the verb is not obligatory, at least in embedded contexts, as can be observed in (60); this makes them an appropriate context to test both approaches: First, the particle cannot occur before an in situ verbal complex and nothing can intervene between the wh-word and ote as shown in (61); however, if we alter the order so that the particle and the wh-word are adjacent, the sentence is grammatical (cf.60 and 62).Furthermore, these examples prove that ote is attached to the wh-word since in those cases the wh-word and the particle move along together to the target position of wh-words in the cp-domain and they appear contiguous irrespective of whether the verb also moves to the cp-domain (62) or not (60).Therefore, this discards the idea that ote merges with the complex verb rather it precedes the complex verb in these cases.
Additionally, this behaviour goes against the idea that the particle always occupies the same position and the different distributions ('wh-word ote v' and 'wh-word v ote') depend on which constituent moves to the specificier position of the phrase containing the particle (Munaro & Poletto 2002).In fact, movement of the verb to the cp-domain could be singled out as responsible for the two distributions in Basque: if the verb moved to the cp-domain, we would get the order of configurations described in Section 4 and illustrated here in examples (63a) and (64a) and, if the verb did not move to cp, then the construction dealt with in this section would arise (63b and 64b): However, such idea is not borne out by our data, since a) the movement of the verb does not alter the distribution in those contexts under exam in this section, given that the particle always follows the wh-word and precedes the verb whether the verb stays in situ 28 Munaro & Poletto (2002) claim that the distinct distributions lies in the movement of the wh-word to the specifier position of the phrase whose head is the particle or of the whole cp to the specifier position of such phrase.This does not seem to be the case of the Basque data since constituents appear after the moved verb and the particle in both cases; therefore, we would have to argue that those constituents have moved to a topp below the position of the particle, at least in the structure dealt with in Section 4. Since there is no evidence supporting such movements, I do not adopt the same syntactic structure proposed by Munaro & Poletto (2002); in fact, the main reason to evaluate their analysis is that superficially they have the same two distributions and they offer a single structure for both of them.However, this seems to be inappropriate in the case of the Basque particle ote.
(60) or is fronted (62) and b) the verb moves in both distributions 'wh-word ote v' and 'wh-word v ote' (65 & 66): (65) 'wh-word ote v' Non ote utzi dut kazeta non ote utzi dut? (cf.53) where part leave aux newspaper.abs'Where did I leave the newspaper?(I'm wondering)' (66) 'wh-word v ote' Non utzi dut ote kazeta non utzi dut? (cf.2) where leave aux part newspaper.abs'Where did I leave the newspaper?(I'm wondering)' Hence, this invalidates the account based on Munaro and Poletto's ( 2002) work for our data, i.e. that a single syntactic position and different derivations may account for the distinct superficial distribution.
Based on the idea that wh-words and ote form a single constituent, another analysis could also be suggested: the wh-word merges with the weak adverb ote which occupies the specifier position of a Modifier phrase when the former moves Spec-by-Spec to the cp-domain and then they (wh-word and ote) move to focp.Even though this is an elegant analysis for this construction, there are data to discard it.Note that the wh -word 'why' in Basque (but also in French, Italian and Spanish) behaves differently compared to the other wh-words since it does not need to be adjacent to the verb; thus, it has been proposed that the wh-word 'why' does not occur below tp and does not trigger movement to cp because it already occurs in cp, namely in a position above focp (Cecchetto & Donati 2012).In such case the wh-word could not adjoin to the particle by head movement to the Left Periphery; however, relevant data show that 'why' merges with ote: Based on this evidence, I conclude that the hypothesis which posits that wh-words and the particle are base-generated together and form the structure 'wh-word ote', or generally stated 'wh-word Particle', is appropriate for the analysis of configurations such as those described here.Indeed, prosodic data reinforces this approach, since the particle must be pronounced forming a prosodic unit with the wh-word. 29egarding its syntactic derivation, I adopt the analyses brought forward by Bayer & Trotzke (2015), Trotzke & Turco (2015) and Trotzke & Monforte (2019) that wh-word and ote combine together in a separate derivational workspace and merge in the position expected for the wh-word.Nevertheless, I propose that the combination 'wh-word ote' moves directly to its target position in the cp-domain following the canonical derivation for wh-questions in Basque; this is unlike previous hypotheses (Bayer & Trotzke 2015;Trotzke & Turco 2015;Trotzke & Monforte 2019), which claim that this combination needs to move to partp to take scope over the proposition by checking features with part 0 and, then, move to the cp-domain.The argument to discard this theoretically well-founded movement comes from the fact that zergatik ote ('why ote') could not fulfil it since, as stated above, it is base-generated in cp.Thus, I postulate that the particle takes scope over the proposition (and, hence, contributes to the interpretation of the clause) at the stage where the wh-word is inserted in cp, namely in focp (or intp in the case of 'why').
In conclusion, the microvariation found in Eastern Basque is not restricted to the issue of its syntactic status; the cross-linguistic combination between a wh-word and a particle is also found in those varieties, namely in North-Eastern Basque.As I proved above, syntactic evidence favours the analysis that wh-word and ote combine together in a separate derivational workspace, merge in the position expected for the wh-word and follows the canonical derivation of wh-questions in Basque, in contrast to the hypothesis claimed by Munaro and Poletto (2002) for a similar structure in Italian.

Ote as a sentence final particle
As remarked above, Basque is rich in microvariation not only among dialects, but also regarding subdialects.The particle ote has been a good example of this richness, since it shows differentiated behaviour in eastern dialects.Ote can function as a head, a weak adverb and it attaches to wh-words.Furthermore, there is a fourth distinguished syntactic use of ote in these varieties: the sentence final particle ote (henceforth 'sfp ote').This pattern, however, is used in a more reduced area, namely in the far east of the Eastern Basque (included the Souletin Basque, Amikuze's variety and also the disappeared variety from the Roncalese valley).Also, it is more restricted regarding the syntactic type of question it occurs: whereas ote appears in wh-questions, alternative questions and polar ones, this sfp ote has been only found in yes/no questions.Regarding its pragmatic type of question, the canonical particle ote (and ote2 too) is used in rhetorical and conjectural questions, hence, never in bona fide questions; however, questions containing sfp ote seem to demand a piece of information.Finally, it appears to have developed a separate interpretation since contexts containing it are not accepted by speakers of other areas, unlike the other kinds of ote.
We have seen in Section 4 that the particle ote can appear after the verb as in:30  Both examples show ote at the end of the clause, but I will argue that they do not occupy 3 1 the same position.First, syntax provides evidence that we are dealing with a different kind of ote.If we compare the following examples with those in Section 4, we observe that in (73) the particle appears sandwiched between the finite and lexical verb, as expected for constituents occupying a position between focp/polp 32 and tp in negative sentences; nevertheless, in ( 74) and ( 76) the particle occurs after all the constituents of the sentence even the lexical verb in negative sentences: (73) Also, if we change the position of ote and place it preceding the finite verb, as in the canonical pattern, further differences arise: the former (71) is pragmatically felicitous also for speakers interviewed and consulted from western and central dialects, whereas the latter (72) is not.For instance, the judgments of native speakers on examples such as (74) clearly reflect that sfp ote must have a distinct interpretation compared to the previous ones, since the use of the particle in its canonical behaviour (and, hence, its canonical interpretation) is rejected: (77) #Ez ote daukazia etxea sutan?not part have.parthouse.absfire.in(Intended reading: 'Don't have your house on fire?(I'm wondering)') 31 This question is used in the following context: some women are taking part in a sex course and one of them tells a story about her husband; then, another woman adds surprised that she did not know that she was married.The first woman replies that she was and it was pretty hard to sleep with an old man because he was always snoring.Then, the second woman asks her: 'Are you with a young man now ote?'.This is not a felicitous context for the use of the canonical ote in western and central dialects.This leads me to formulate the idea that these uses of ote may not convey the same interpretation; however, further research should be done on this topic in order to clarify this point.Consequently, this is a separate use of ote which seems to appear in what has been referred to as the Right Periphery (Izutsu & Izutsu 2013) and may occupy a position above cp.If we look at the data gathered in those dialects, it could be hypothesized that clauses as ( 74) and ( 76) may have originated this sfp ote by reanalysing the particle in modp as occupying a sentence final position.Let us illustrate this by taking the following sentence as an example: (78) Landart ( 2011) Ez nintzelakoz laborari izateko sortua ote? not aux.compfarmer be.to born part 'Because I wasn't born to become a farmer?('I'm wondering)' The ellipsis of the main predicate would have yielded this new use since the particle and the embedded clause (presumably in focp) occur adjacent.Let us briefly exemplify this.

(79)
Ez nintzelakoz laborari izateko sortua i ez zuen j ote enetzat estimurik?t i t j not aux.compfarmer be.to born not have part I.ben esteem.prtt'Wasn't he fond of me because I wasn't born to become a farmer?('I'm wondering)' Considering the hypothetical origin of sfp ote, it could be posited that the sfp ote in (74-76) remains in the same position as its source ote2, i.e. [spec, modp], and the proposition below it moves to [spec, focp].Nevertheless, to propose the same syntactic position and almost identical derivation to distinct structures exemplified in ( 73) and ( 74) seems conflictive since they do not share the same sentential distribution in all contexts and they do not convey the same contribution to the sentence, as argued above based on the fact that they are not interchangeable.Although the interpretation of this kind of ote is beyond the scope of the aim of this paper, based on the context of these examples, a first conclusion could be presented: unlike the interpretation of ote in the other cases, this has an intersubjective meaning which apparently matches the fact that it appears in utterance-final position (Izutsu & Izutsu 2013;Haselow 2015) since the speaker seems to expect a specific response or reaction from the addressee by using the particle ote at the end of the utterance.In fact, this use reminds of those containing usual elements which can be easily considered as sfp such as ja and oder in German (Izutsu & Izutsu 2013: 230), but in English (Richards apud Izutsu & Izutsu 2013: 232) or, even more language specific as e in Basque (Lizardi-Ituarte 2019); for instance: Lizardi-Ituarte (2019) Ez esan inori e. not tell anyone.datpart 'Don't tell anyone e.' Since such pragmatic markers are considered to occupy a position above cp, I adopt the same analysis for the kind of ote under question in this section.Following del Gobbo, Munaro & Poletto (2015), particles in the cp-domain are attractors, unlike particles occurring in the tp-domain; I extend this property to particles in the cp-domain and above it.Thus, the particle triggers the movement of the whole proposition to the specifier position of a functional phrase whose head is occupied by the sfp ote.
Having said that, I would like to briefly discuss two issues.On the one side, it would be interesting to specify the phrase it occupies having in mind those separate phrases differentiated in the neoperformative approach: following Corr's (2015;2016) analysis, there is SAHighP related to the discourse set-up and speaker-addressee relation and also SALowP related to the speaker attitude, evidentialy and epistemicity commented above in Section 1.Although it is necessary to deepen the interpretation of the sfp ote first in order to avoid misanalyses, I sketch the distribution of the sfp ote considering that its contribution is related to speaker-addressee interaction and, therefore, occupies the position SAHighP, as represented below (81).On the other side, I propose that the sfp ote has a head status since it attracts the phrase containing the whole proposition to its specifier position, property not available for xps; also, crossinternal and typological evidence such as those proposed for e in Basque (Lizardi-Ituarte 2019) or for a in Mandarin Chinese (Del Gobbo et al. 2015) reinforces that hypothesis. 33Let us illustrate the derivation of an example as ( 72): (81) Therefore, the sfp ote occupies the head position of the SAHighP based on its function in the clause, closer to those elements marking the discourse set-up and, since those particles are attractors, the whole proposition (i.e.forcep) rises to the specifier position of the SAHighP.
To summarise, ote shows a fourth separate behaviour in a reduced domain of the East Basque area syntactically and pragmatically distinct from the ones previously discussed; this ote always appears in a sentence final position and it conveys an intersubjective interpretation.Based on its syntactic and interpretational properties, I have concluded that it functions as a head and occurs in a position above cp; furthermore, the particle attracts the whole proposition to its specifier position.

Conclusion
Syntactic microvariation offers new insights into a deeper study of phenomena discussed in other languages from distinct research approaches.The discourse particle ote was examined through these lines considering its microvariation and, hence, this provided us the oportunity to observe the behaviour of those phenomena identified cross-linguistically within the same language.First, I looked into the canonical use of ote, i.e. its behaviour in non-inflected contexts and in those where the finite verb moves along the sentence, and 33 One could point out that the evolution of the syntactic status of ote seems inconsistent, since it turns from a head one into a phrasal one and further from a phrasal one into a head one again.Nevertheless, I follow the hypothesis put forth in Monforte (2020a) that the ote2 does not derive directly from the canonical ote (dealt in Section 3), but from a paratactic clause (Schoonjans 2012;Van Bogaert & Leuschner 2015).This grammaticalization process is not unknown in the Basque language since other particles or epistemic adverbs seem to share a similar evolution; for instance: (i) antza denez ('as it seems') > antza ('apparently') (ii) behar bada ('if it is neccesary') > beharba(da) ('maybe') (iii) bada ('if it is') > ba(da) ('then') (see Lizardi-Ituarte 2017 for its syntactic analysis) This new resulted ote would have further grammaticalised into a head when speakers reanalysed it as argued in Section 6 regarding Example (80).its phonologic-prosodic properties; based on that, I concluded that discourse particles in Basque have head status and that they occupy a position in the tp-domain.Second, I have turned into the data gathered from eastern dialects where the particle ote can function not only as a head but also as a maximal projection.As for its syntactic position, I have argued that it occupies the specifier position in Rizzi's (2004) Modifier phrase in the cp-domain considering its occurrence in contexts affected by the ellipsis of tp, its dependency on the lexicalization of focp and also typological evidence.Furthermore, I have presented data from eastern dialects which gives evidence of two distinguished uses: the 'wh-word Particle' configuration, also found in German, Italian, Dolomitic Ladin or Japanese, and the utterance final ote presumably conveying intersubjectivity.Regarding the former, I have suggested that they are combined together at the stage the wh-word is inserted into the proposition and form a single constituent which moves along the sentence, namely to the focp following the canonical derivation of wh-questions in Basque.Concerning the latter, its sentential distribution and distinct interpretation yield the idea that this ote should be considered a sentence final particle which occupies a head position in a phrase above cp; this sfp ote attracts a phrase containing the whole proposition to its specifier position deriving the word order displayed in this use, i.e. the particle at the end of the utterance.As can be perceived, the discourse particle ote has a high degree of microvariation and, consequently, this provides fine-grained insights and evidence into the study of discourse particles discussed in other languages.

(
Canonical ote b.Noncanonical ote2 Consequently, it cannot be phonetically reduced, i.e. ote > *te, in contrast to what can happen when ote behaves as a head (cf.28), as confirmed by speakers when asked for this issue based on examples such as (32-34):

2 Some notions on Basque grammar
19) Ez al duzu egunkaria erosi ez al duzu?not part aux newspaper.absbuy 'Didn't you buy the newspaper?' .absbuy part and other.absthrow part aux.comp ask aux 'I asked whether s/he bought this one and threw away the other one.' this and the particles ahal, ei, omen and ote have been claimed to originate from verbs; for instance:21 but not.comp.knowI.erg mermaid those.abswhat were.comppart 'But I don't really know what those mermaids may have been.'Eta gero Jainkoa zertako dugun ote samur, estonatuko gira gu! and then God.abs why have.comppart soft astonish.futaux we 'And then we'll be astonished why God may be hard on us.' expects for particles with head status (see 35).Finally, instances of the innesive case marker -(a)n attached to the particle ote [ot(h) e+an > othian] or the partitive case marker -(r)ik attached also to it [ot(h)e+rik > otherik] provide further evidence in favour of a maximal projection nature:As can be observed in the previous examples, the behaviour of ote2 is closer to that one expected for adverbs, as adverbs can function similar to nouns in some contexts (e.g. the adverb 'tomorrow') and admit case marking (e.g.biharrik 'tomorrow.prtt').For instance: not know I.erg how part do.ipfv can aux.comp'I don't know how it could be done.' not know where part leave aux.comp newspaper.abs'I don't know where I may have left the newspaper.' However, some examples are not as clear regarding the position of ote, since it could be considered to occupy a position in the right periphery: