https://doi.org/10.1515/9783110726077-014 Karlos Cid Abasolo and Marta Carretero 14 Evidentiality in Basque 14.1 Introduction This chapter concerns evidentiality in Basque, focusing on the expressions cor- responding to the following crosslinguistic hyperlexemes: the inferential evidentials MUST and SEEM, the reportive evidentials SAY, (BE) LIKE and ACCORDING TO, and the visual evidential SEE. The description, based on naturally occurring examples, shows that Basque has a number of idiosyncratic evidentials, but the expression of evidentiality in this language is essentially similar to that of some Indo-European languages such as English and Spanish in several respects: syntactic variety, coverage of different modes of gathering the evidence, poly- semy of many evidentials and a strong semasiological relationship with epis- temic modality. 14.1.1 Structure of the chapter The chapter is organized as follows: the remainder of Section 14.1 describes the corpora used for the account of the expressions given in the following sections, summarizes the typological features of Basque and offers a panoramic view of evidentiality in this language. Section 14.2 concerns the inferential evidentials irudi and eman, both meaning ‘seem’, and behar ‘must’. Section 14.3 covers the reportive evidentials esan ‘say’ and arabera ‘according to’, the hearsay particles omen and ei ‘it is said’, and a few other expressions of hearsay. Section 14.4 pertains to perceptual evidentiality, concretely to the perceptual verb ikusi ‘see’ and related expressions. Section 14.5 covers a number of indirect indifferent evidentials with the meaning ‘as it seems, apparently’. Section 14.6 contains remarks on the diachrony of some of the expressions studied in previous sec- tions. Section 14.7 sums up the main conclusions. 14.1.2 The corpora This article has been elaborated with the aid of four corpora, from which exam- ples have been cited in order to illustrate many of the phenomena described: the first is Ereduzko Prosa Gaur1 (‘Exemplary Prose Today’), which contains || 1 URL: http://www.ehu.eus/euskara-orria/euskara/ereduzkoa/, accessed December 5, 2018. 586 | Evidentiality in Basque approximately 25 million words, of which 13 are extracted from quality books and 12 from the newspapers Berria and Herria. The second is Egungo Testuen Corpusa2 (‘Corpus of Current Texts’). The third is Euskal Klasikoen Corpusa (‘Corpus of Basque Classics’).3 The other corpus was created by one of the au- thors, Karlos Cid Abasolo. The sources of this corpus, fully specified in the Ap- pendix, are of four types: – contemporary Basque literature; – examples from older written texts, most of them obtained from a website run by Susa publishers and from the Orotariko Euskal Hiztegia – Diccionario General Vasco (henceforth OEH-DGV) (Michelena and Sarasola 1987–2005); – contemporary written journalistic texts, on paper and digitized; – spoken examples from Euskal Telebista 1 (ETB1, public TV channel that broadcasts in Basque). Throughout the article, the examples obtained from the corpora are cited with the corresponding source. The examples with no citation of a source have been constructed by the authors or by an informant. 14.1.3 A brief typological characterization of Basque Basque has the peculiarity of not being genealogically related to any other liv- ing language according to the present state of knowledge. This section provides a general characterization of the main typological features of Basque, geared at a better understanding of the description of evidential expressions carried out in the following sections: 1) The Basque dominant order of clausal constituents is SOV. In Basque, like in other SOV languages, the verb complements usually precede the verb, and noun complements (except for demonstrative determiners and adjective phrases, the numeral bat ‘one’ and, in the Biscayan dialect, the number bi ‘two’) precede the head in noun phrases. 2) Basque is an agglutinating language, in which the definite article is af- fixed to the noun (as in mendi-a ‘the mountain’, literally ‘mountain-the’). Ac- cording to Sareko Euskal Gramatika (‘Basque Grammar Online’), postpositions may be bound or free.4 Bound postpositions are affixed to noun phrases, as is || 2 URL: http://www.ehu.eus/etc/, accessed January 15, 2019. 3 URL: http://www.ehu.eus/ehg/kc/, accessed January 15, 2019. 4 “Postposizioa eta postposizio sintagma”, Sareko Euskal Gramatika (SEG), Euskara Institu- tua, EHU, www.ehu.eus/seg. Introduction | 587 the case of the inessive postposition -n (mendi-a-n: ‘mountain-the-in’, ‘in the mountain’); free postpositions occur after the Noun Phrases that they modify, as in bila (‘in search of’): etxe bila (‘house-in search of’, i.e. ‘in search of a house’). 3) Basque has case marking. Traditional Basque grammars have proposed many cases: for example, Euskal Gramatika. Lehen Urratsak-I, published by Euskaltzaindia (the Royal Academy of the Basque Language) in 1991, proposes 15 cases (pp. 216–217). In the more recent Sareko Euskal Gramatika, whose ap- proach will be adopted in this paper, these cases have been reduced to four: absolutive, ergative, dative and genitive. This grammar considers that all the other devices which were formerly considered as cases are actually postposizio itsatsiak (‘bound postpositions’). 4) Basque is a morphologically ergative language. Ergativity is reflected in nominal morphology: the absolutive case (zero-marked) signals the subjects of intransitive verbs and the objects of transitive verbs; the subjects of transitive verbs take the ergative case (suffix -k). In verbal morphology, there is no ergativity strictu sensu, but split ergativity. However, we will not deal with this issue in order not to deviate excessively from the purposes of the present paper. 5) The morphological agreement markers that can appear in the Basque verb are three: absolutive, ergative and dative. The combination of these gram- matical cases determines four types of verb agreement patterns: absolutive, as in joan naiz ‘I have gone’ (n = ABS1s); absolutive + dative, as in gustatzen natzaio ‘I please him/her’ (n = ABS1s and o = DAT3s); absolutive + ergative, as in ikusi naute ‘they have seen me’ (n = ABS1s and te = ERG3PL); and absolutive + dative + ergative, as in eman dizkizute ‘they have given them to you’, (zki = ABS3PL, zu = DAT2s and te = ERG3PL). 6) Basque has synthetic and periphrastic verbs. Synthetic verbs are only a few, and some of them are currently not in common use. An example of a syn- thetic verb is noa ‘I am going’ (n ‘ABS1s; oa is the VL). In periphrastic verbs, the main verb precedes the auxiliary in affirmative sentences (joan naiz ‘I have gone’, joan ‘gone’ + naiz ‘I am’), but this order is reversed in negative clauses (ez naiz joan ‘I have not gone’). The auxiliary of periphrastic verbs may be omitted under certain circumstances, as in coordinate copulative structures. 7) Due to its rich verbal morphology, Basque is a pro-drop language. 8) Basque also differs from other European languages in that its complementizers are not free words but suffixes adhered to verbs. Comple- mentizers have semantic values associated with declarative vs. interrogative contrast, factivity, polarity, mood or evaluation of propositional content. Some examples are -ela, -enik and -en.5 || 5 An extensive account of complementizers is Artiagoitia and Elordieta (2016). 588 | Evidentiality in Basque 9) Basque has a few morphemes that indicate the syntactic class to which lexical items belong, such as -tu (one of the morphemes marking participles) or -ki (one of the morphemes marking adverbs). 10) According to Koldo Zuazo (p.c.), a dialectologist and professor at the University of the Basque Country, nowadays there are five dialects: three of them belong to the Spanish zone of the Basque Country: Western (Biscayan), Central (Gipuzkoan) and Navarrese, and two to the French zone: Navarrese- Lapurdian and Souletin. There was a sixth dialect, Eastern Navarrese, now ex- tinct. On the other hand, since 1968, there is a standard version of Basque, euskara batua (literally ‘unified Basque’). 14.1.4 A panoramic view of evidentiality in Basque As point of departure we define evidentiality as a domain in which information source is specified. The speaker/writer chooses different means that allow to indicate (or narrow down) the cognitive or communicative basis of a given proposition. This is along the lines of many key references in the literature (Wil- lett 1988; Aikhenvald 2004; Wiemer and Stathi 2010; Boye 2012; Marín-Arrese 2013 inter alia). Throughout this chapter, it will be shown that, in spite of its typological difference, the expression of this category in Basque resembles that of other languages studied in this volume in a number of respects. Firstly, Basque has evidential expressions of all the basic types in terms of mode of gathering the evidence (inferential, reportive, perceptual and indirect indiffer- ent). Secondly, most of these expressions are derived from non-evidential ex- pressions, and are polysemous in present Basque. A third similarity lies in the range of syntactic realizations of evidential expressions: there are particles such as omen or ei ‘it is said’ or bide ‘apparently’, constructions with different forms of lexical verbs such as irudi or eman ‘seem’, nouns such as irudi ‘appearance’, zurrumurru ‘rumor’, berri, albiste ‘news’ or antz ‘appearance’, and constructions such as the noun behar ‘obligation, necessity, duty’ in some of the cases where it, accompanied with an auxiliary, forms a compound predicate. A fourth similarity between Basque and other languages, which will be treated here at greater length, is the strong relation between evidentiality and epistemic modality, understood as the expression of the estimation of the chances for a proposition to be or become true. On the one hand, some of the evidential expressions studied here have both epistemic modal and evidential features; therefore, we believe that they should be considered as epistentials in the sense of Faller (2002) and Lampert (2015). In order to illustrate epistentiality, Introduction | 589 we will mention one of the meanings of English must, as in ‘There is a smell of smoke: something must be burning in the kitchen’, where an evidential mean- ing of indirect evidence coexists with an epistemic modal meaning of high probability. Examples of Basque epistentials are behar, which is a close equiva- lent to English must or Spanish deber (de) (see Section 14.2.2.), and the expres- sions of indirect indifferent evidentiality described in Section 14.5, which have non-cancellable meanings of indirect evidence and of lack of total certainty. In their turn, the particles omen and ei also display a feature of lack of total cer- tainty, but with the status of a Generalized Conversational Implicature (GCI), since it can be cancelled by the context (see Section 14.3.2.). On the other hand, Basque also has many expressions not included here because their epistemic meaning is considered to predominate over their evidential meaning. Some of these expressions are, among many others (Jendraschek 2003; King 2009): – The participle with suffix -ko followed by an auxiliary in the present tense, which is equivalent to will + INF in English or the future tense in Spanish; – Some occurrences of the participle with suffix -ko followed by an auxiliary in the past tense, whose nearest expressions in English and in Spanish are would + INF and the conditional tense respectively (e.g. ‘He would be about forty when I first met him’); – Lexical verbs such as uste izan ‘think’ or sinetsi ‘believe’; – Adverbs such as segurki, ziurki ‘certainly’, seguru aski, seguru asko, ziur aski ‘probably’ or agian, beharbada, akaso, apika ‘perhaps’. Apart from the fact that there is a mutual implicative relationship between epis- temic modality and evidentiality (Jendraschek 2003: 18–19), epistemic modal expressions can occur with explicit evidence in favor of (and, in some cases, against) the truth of the proposition, which brings them nearer to evidential expressions. This is the case of (1), an example where the proposition euria egingo duela ‘that tomorrow it will rain’ is qualified by ziurki ‘certainly’ and also supported by the evidence ‘those clouds’, and of (2), an example of epistemic modality expressed by -ko (the /k/ undergoes voicing in /g/ before a nasal con- sonant) with the past auxiliary: (1) Hodei hori-ek pents-araz-ten d-i-da-te cloud that-ERG.PL think-CAUS-IPFV AUX (ABS.3SG-VL-DAT.1SG-ERG.3PL) bihar ziur-ki euri-a-Ø egin-go du-ela. tomorrow certain-ADVS rain-ART-ABS do (PFV)-FUT AUX.PRS.3SG-COMP ‘Those clouds make me think that, certainly, tomorrow it will rain.’ (2) Bere adin-eko ume guzti-ek legez, 3SG.GEN age-GLOC.SG child all-ERG.PL like 590 | Evidentiality in Basque bi urte-Ø inguru izan-go zituen orduan. two year-ABS about have (PFV)-FUT AUX.PST.3SG then ‘Like all the children of his age, he would then be about two years old.’ [K. Uribe: Mussche. 2012: 125. e-book] We may state that Basque provides evidence about the crosslinguistic signifi- cance of a general category of epistemicity that covers evidentiality and epis- temic modality, as proposed in Boye (2012). The following descriptive sections provide additional justification for this statement. 14.2 Inferential expressions This section deals with a number of constructions that express inferential evidentiality in Basque, namely constructions with the verbs irudi ‘seem’ and eman (whose main meaning is ‘give’), and constructions with behar, a noun that literally means ‘obligation, necessity, duty’ but has an evidential extension. 14.2.1 irudi and eman ‘seem’ The closest forms to the crosslinguistic hyperlexeme SEEM are the verb irudi ‘seem’ (and its dialectal variety iduri), and the verb eman (emon in the Biscayan dialect), whose main meaning is ‘give’. The evidential extension of eman can be accounted as a conceptual metaphor (Lakoff and Johnson 1980), namely as a transfer from the concrete domain of a person giving an object to someone to the abstract do- main of a situation giving evidence to the speaker.6 Both verbs (irudi and eman) belong to the ergative-absolutive pattern; that is, they require a constituent in the ergative case marked as subject of a transitive verb, and another constituent in the absolutive case marked in the same way as direct objects, although it is a ‘predica- tive complement’ in the sense of Huddleston and Pullum (2002: 217).7 Irudi and eman may be constructed with a predicative complement or with a complement clause. The absolutive marker in the verbal morphology of the two verbs is always third person singular (the absolutive plural morpheme -it- is not admitted). Ex- amples of evidential irudi and eman are (3) and (4): || 6 The same metaphor occurs with the Spanish verb dar ‘give’, as in Me da que David se ha recuperado de su enfermedad (lit. ‘It gives me that David has recovered from his illness’). 7 The predicative complement is illustrated by intelligent in ‘Jane is/seems intelligent’ and ‘Harry considers Jane intelligent’. Inferential expressions | 591 (3) Musika hon-ek Bach-en-a-Ø dirudi-Ø. music this-ERG Bach-GEN-ART-ABS seem.PRS-3SG (4) Musika ho-nek Bach-en-a-Ø ema-ten du. music this-ERG Bach-GEN-ART-ABS seem-IPFV AUX.PRS.3SG ‘This music seems like Bach.’ The evidential meaning of both verbs coexists with an epistemic element of lack of certainty whether the proposition is true or false, which has the status of a GCI, in the sense that it is communicated by default but may be cancelled or blocked under certain contextual conditions. For example, in the case of (3)– (4), the utterance triggers by default the implicature that the speaker does not have knowledge that this music is by Bach nor knowledge that this music is not by Bach. However, this implicature may be cancelled, e.g. by a later statement that the proposition is true, as in (5), (5) Musika hon-ek Bach-en-a-Ø dirudi-Ø, eta music this-ERG Bach-GEN-ART-ABS seem.PRS.3SG and Bach-en-a-Ø da. Bach-GEN-ART-ABS be.PRS.3SG ‘This music seems like Bach, and is by Bach.’ or by a later statement that it is false, such as ‘But actually it is by Herbst’. In other cases, the GCI is blocked (i.e. not even communicated), e.g. in a context where the speaker and the addressee know beforehand that the music is by Herbst. When irudi and eman govern a complement clause, they may be placed at the beginning of the sentence. When irudi occurs in its synthetic form, it usually appears with the prefix ba- as a morphological reinforcement of synthetic verbs (6); however, if it occurs after the complement clause it may appear without this prefix (7). (6) Badirudi/Ematen du musika hau-Ø Bach-en-a-Ø it seems music this-ABS Bach-GEN-ART-ABS de-la. be.PRS.3SG-COMP (7) Musika hau-Ø Bach-en-a-Ø de-la dirudi/ematen du. music this-ABS Bach-GEN-ART-ABS be.PRS.3SG-COMP it seems ‘It seems that this music is by Bach.’ The demonstrative pronoun is in absolutive case in (6) and (7), since it is the subject of a copular verb; however, in (3), (4) and (5) it is in the ergative case, 592 | Evidentiality in Basque since it is the subject of the evidential verbs irudi and eman, which function with the ERG-ABS pattern cited above. In its turn, the complement clause carries as complementizer the verbal suffix -ela or -enik. The distribution of both suffix- es is far from clearcut, due to diachronic and diatopic variations. However, for many present-day speakers of western and central Basque dialects, -ela ex- presses a higher degree of certainty than -enik. That is, both suffixes display a similar difference to that signaled in Spanish by the choice of the indicative or the subjunctive mood, which may be seen in the Spanish translations of the subordinate verbs in (8) and (9): (8) Ez dirudi musika hau-Ø Bach-en-a-Ø de-la. NEG seem.PRS.3SG music this-ABS Bach-GEN-ART-ABS be.PRS.3SG-COMP (Spanish) ‘No parece que esta música es (IND) de Bach.’ ‘It does not seem that this music is by Bach.’ (9) Ez dirudi musika hau-Ø Bach-en-a-Ø de-nik. NEG seem.PRS.3SG music this-ABS Bach-GEN-ART-ABS be.PRS.3SG-COMP (Spanish) ‘No parece que esta música sea (SUBJ) de Bach.’ ‘It does not seem that this music is by Bach.’ The subject of the complement clause may also be raised; with this construc- tion, it takes ergative case and is governed by the main verb (i.e. the verb of seeming). Irudi and eman do not admit a correlate structure to English seem followed by infinitive (‘This music seems to be by Bach’). The raised construc- tion, of which (10) is an example, was analyzed from a generative perspective by Artiagoitia (2001a, b, 2003: 653–656). (10) Orain-go tanta-k ema-ten du lodi-tzen now-GLOC drop.ART-ERG seem-IPFV AUX.PRS.3SG swell-IPFV ari ze-la. PROG AUX.PST.3SG-COMP lit. *‘The drop now seems that it was swelling.’ [A. Lertxundi: Paper-festa. 2012: 232] Subject raising is also possible when the Subject is not third person singular, since the verb irudi or eman takes the person mark corresponding to the subject of the subordinate verb, which is third person plural in (11): (11) Orain-go tant-ek ema-ten dute lodi-tzen now-GLOC drop-ERG.PL seem-IPFV AUX.PRS.3PL swell-IPFV ari zire-la. PROG AUX.PST.3PL-COMP lit. *‘The drops now seem that they were swelling.’ Inferential expressions | 593 Concerning person, Artiagoitia (p.c.) made an informal inquiry with example (12) about the possibility for subject raising with the first or second person. Most informants considered (12) ungrammatical, but a few others considered it grammatical. Consequently, we may state that subject raising is virtually re- stricted to the third person. (12) %Asko-Ø irakur-tzen duzu-la dirudi-zu. a_lot-ABS read-IPFV AUX.PRS.2S-COMP seem.PRS-2SG lit. *‘You seem that you read a lot.’ Subject raising with irudi and eman may well be motivated by stylistic factors. Let us consider (13): (13) Baina amaiera-rik-Ø ez dute-la dirudite-n but end-PARTIT-ABS NEG have.PRS.3PL-COMP seem.PRS.3PL-REL beste batzuk-Ø (…) others-ABS lit. *‘But others who seem that they do not have an end (…)’ [TV, ETB1, Programme “Tribuaren kideak” (‘The Members of the Tribe’). Stated by show host Kike Amonarriz, February 19, 2013] The choice of subject raising was, in all probability, due to the occurrence of irudi in a relative clause. In Basque, relative clauses with the relativizer suffix -n occur at the left of the head of the Noun Phrase: the verb dirudite (‘seem’-3PL) occurs between the complement clause (amaierarik ez dutela ‘that they do not have an end’) and the head. This construction has the disadvantage that it sug- gests a false syntactic relationship between the two elements (beste batzuk is not the notional subject of diruditen, but of dute ‘have’-3PL). However, in (14), which is the continuation of (13), eman occurs but the corresponding construc- tion with raised subject (15) was not chosen, probably because eman is not with- in the scope of a relative clause: (14) Buka-tu-ko dir-ela ema-ten du. finish-PFV-FUT AUX.PRS.3PL-COMP seem-IPFV AUX.PRS.3SG ‘It seems that they will finish.’ (15) Buka-tu-ko dir-ela ema-ten dute. finish-PFV-FUT AUX.PRS.3PL-COMP seem-IPFV AUX.PRS.3PL *‘(…) they seem that they will finish.’ With irudi and eman, inferences may be derived from different kinds of sources: 594 | Evidentiality in Basque – Perceptual: (16) Sofan esertzeko duzun moduagatik, oso nekatua dirudizu. ‘You seem very tired, by the way you are sitting on the sofa.’ – Cognitive, as in (17), where the inference is based on the writer’s knowledge that certain results of Basque pelota games often occur in boring matches: (17) Emaitzari erreparatuta, badirudi zer gozatu handirik ez zuela izan pilo- talekura joandako jende andanak. ‘Given the final result, it seems that the great amount of people who went to the pelota court did not really enjoy themselves.’ [https://www.berria.eus/paperekoa/1554/026/001/2011-12-27/gainera- hala-jokatzen-badute.htm] – Discourse-based: (18) Txostenak dioenez, Eusko Jaurlaritzak zabortegiari emandako ingurumen baimenak Europako zuzentarauak betetzen zituen. ‘According to what the report says, the environmental authorization granted to the landfill by the Basque Goverment complied with European directives.’ [https://www.berria.eus/paperekoa/2086/012/002/2020-09- 05/jaurlaritzak-zuzen-bete-zuen-verterrekiko-kontrola-ikuskaritza-baten- arabera.htm] (19) The Washington Post eta The New York Times egunkariek diotenez, Michael Forest Reinoehl Lacey herrian hil zen. ‘According to what the newspapers The Washington Post and The New York Times say, Michael Forest Reinoehl died in a village called Lacey.’ Irudi and eman are sometimes accompanied with expressions used for introduc- ing explicit evidence, such as those mentioned below, among others: a) the noun modu ‘way’ with the cause bound postposition -gatik (moduagatik ‘by the way’):8 (20) Zure jokatzeko moduagatik, badirudi ez duzula presiorik sentitzen. || 8 “Motivative” is the term used by basqueology (Hualde and Ortiz de Urbina 2003: 742, among others). Inferential expressions | 595 ‘By the way you play, it seems that you do not feel pressure.’ [http://paperekoa.berria.eus/kirola/2013-02- 21/024/001/ez_goaz_kezkatuta_uste_osoa_dugu_emaitza_on_bat_lortuk o_dugula.htm] b) gauzak horrela ‘given the state of things’: (21) Gauzak horrela, ez dirudi beteko dutenik Nafarroako AHT osoa egiteko jarritako epemuga: alegia, 2015. urtea. ‘Given the state of things, it does not seem that they will meet the dead- line set for doing all Navarra’s TAV, that is to say, the year 2015.’ [https://www.berria.eus/paperekoa/0/010/002/2012-08-07/nafarroan- ahtko-lanen-20-martxan-daude-loturak-zehazteko-badaude-ere.htm] c) erreparatu ‘notice, realize’, kontuan hartu ‘take into account’, etc. These expressions, which have a meaning of conditionality (‘taking into account, if we take into account’…), contain a participle, which takes the adverbial suffixes -ta, -ik or -z: (22) Ikusleen erantzuna kontuan hartuta badirudi asmatu egin dela. ‘Taking into account the spectators’ responses, it seems to have suc- ceeded.’ [http://www.diariovasco.com/20080724/alto-deba/mairuaren-alardea- aztertu-berritzeko-20080724.html] Finally, it must be noted that irudi is also a polysemous noun, whose meaning may be concrete (‘figure’), as in (23) or abstract (‘opinion, image, look’) (24)– (26): (23) Berun-ez-ko irudi hori-Ø apur-tu egin da. lead-INS-DER figure this-ABS break-PFV do (PFV) AUX.PRS.3SG ‘This figure of lead has broken.’ (24) Ni-re irudi-z/irudi-ko, lor-tu egin-go duzu. 1SG-GEN opinion-INS/GLOC manage-PFV do (PFV)-FUT AUX.PRS.2SG ‘In my opinion, you will manage.’ (25) Politikari horr-ek irudi on-a-Ø du. politician that-ERG image good-ART-ABS have.PRS.3SG ‘That politician has a good image.’ The noun irudi may express perceptual evidence, as in (26): 596 | Evidentiality in Basque (26) Bere irudi-a-Ø ikus-i-ta, Bilbo-ko-a-Ø 3SG.GEN look-ART-ABS see-PFV-ADVS Bilbao-GLOC-ART-ABS de-la esan-go nu-ke. be.PRS.3SG-COMP say (PFV)-FUT AUX.1SG-CON ‘After seeing her appearance, I would say s/he is from Bilbao.’ 14.2.2 behar ‘must’ In Basque, the closest item to the hyperlexeme MUST is the noun behar ‘obliga- tion, necessity, duty’,9 accompanied by the auxiliary for transitive verbs. This construction resembles English must in that it has an evidential meaning of indirect evidence together with an epistemic modal meaning of high probabil- ity, thus being best considered as epistential (see Section 14.1.4.). Behar may express the following meanings, apart from the epistential: 1) Need of an object or commodity (‘need, require’): (27) Diru-a-Ø behar dut. money-ART-ABS AUX.PRS.1SG ‘I need money.’ 2) Deontic necessity (‘have to, must’), with a participle before behar. This meaning is very common: (28) Irabaz-i behar dut. win-PFV AUX.PRS.1SG ‘I have to win/I must win.’ 3) Near future, with a participle of a verb before behar. As an expression of future, behar is intentional: it only occurs with states or events to be carried out intentionally by a volitional agent. (29) Zer-Ø har-tu behar duzu? what-ABS take-PFV AUX.PRS.2SG ‘What are you going to have?’ || 9 The noun behar has dialectal variants: in the past, the /h/ was aspirated in all the dialects, but nowadays this aspiration is lost in all the dialects but Souletin. The first vowel may be pronounced /e/ or /i/, depending on the areas. Inferential expressions | 597 The epistential meaning is the most recent of all the meanings of behar (see Section 14.6). A fact related to this recency is that most of the epistential exam- ples found have as main verbs the copular verbs izan (‘be [permanent]’) and egon (‘be [transitory]’). Egon may occur with the Subject in the ergative case or in the absolutive case. (30) is an example with the Subject in ergative case (suf- fix -k in Anek ‘Ane’) agreeing with the transitive auxiliary du (ERG.3S-ABS.3S): (30) Anek lur jota egon behar du, eta hobe dugu bakean utzi. ‘Ane must be distressed and we’d better leave her in peace.’ (31) is an example of egon with absolutive case (morpheme Ø), where the Sub- ject is itxaso gaisto xamarra ‘quite rough sea’ instead of the ergative case (which would be itxaso gaisto xamarrak); the absolutive case prompts an existential interpretation (Rezac et al. 2014): (31) Itxaso gaisto xamarr-a-Ø egon bear du. sea rough quite-ART-ABS be (PFV) AUX.PRS.3SG ‘There must be quite rough sea.’ [R. Illarramendi: Testimonio Falsoa. 1918: 5]10 In certain cases, izan and egon occur in construction with other verbs, for ex- ample in the passive (egina izan ‘be made’; asmatua izan ‘be invented’, etc.), as in (32)–(33). Other verbs may also appear, although they are less usual (34). It may be seen that in (32)–(34) the Subjects are also in absolutive case: pastel hori, Artzai Arkadi hau and egotia occur instead of the corresponding ergatives pastel horrek, Artzai Arkadi honek, and egotiak, even though the auxiliary is du (with mark of ERG3s). (32) Pastel hori-Ø Mikel-ek egin-a izan cake that-ABS Mikel-ERG make (PFV)-ART be (PFV) behar du. AUX.PRS.3SG ‘That cake must be made by Mikel.’ (33) Artzai Arkadi au-Ø Bergili-k bera-k asma-tu-a shepherd Arkadi this-ABS Virgil-ERG himself-ERG invent-PFV-ABS izan behar du ziurr-ik asko. be (PFV) AUX.PRS.3SG sure-PARTIT much ‘This shepherd Arkadi must have been invented by Virgil himself, surely.’ [A. Ibiñagabeitia: Bergiliren Idazlanak osorik. 1966: 44] || 10 Examples (31), (33), and (34) were found in the entry behar izan of the OEH-DGV, Volume 4. 598 | Evidentiality in Basque (34) Ikara-tu bear du bakarrik ego-ti-a-Ø! frighten-PFV AUX.PRS.3SG alone be-NMLZ-ART-ABS ‘It must frighten to be alone!’ [R. Illarramendi: Testimonio Falsoa. 1918: 18] Nevertheless, we have also found examples such as (35), in which the Subject (ark ‘that’) takes the ergative case (-k). (35) Onezkero neska zirzill ark, nik aditu nubenez, Madrill-aldeko komenturen batean sar-tu-a izan bear du moja-Ø enter-PFV-ART be (PFV) AUX.PRS.3SG nun-ABS ‘That shabby girl, according to what I heard, must have already become a nun in a convent in the Madrid area.’ [V. Mocoroa: Damuba... garaiz, edo Bertuteric gabeco aberastasuna iñoiz ez da eguizco zoriontasuna. 1897: 21] Behar with epistential meaning is extremely rare in negative clauses. For in- stance, the meaning of negated epistemic necessity, with both external and internal negation (as in ‘John need not know English’ or ‘John cannot know English’, respectively) is most often conveyed by other epistemic modal or evi- dential expressions, some of which are studied in this article. Therefore, behar in negative clauses nearly always has a different meaning from the epistential one. With regard to aspect, epistential behar has no formal restrictions, but it is clearly more frequent with imperfective aspect than with perfective aspect. The participle of the main verb may also appear after behar. This word order differs from the other in that the participle is backgrounded and the focus (which, in Basque, is placed immediately before the verb) is given extra empha- sis. In (36), the focality of ona ‘good’ is highlighted and the participle izan (‘be’) is backgrounded; this difference in communicative importance is weakened in the correlative construction (37), even if the focus is still ona. (36) Neska-Ø irakur-tzen ari de-n liburu-a-k girl.ART-ABS read-IPFV PROG AUX.PRS.3SG-REL book-ART-ERG on-a-Ø behar du izan. good-ART-ABS ‘The book that the girl is reading must be good.’ [A. Lertxundi: Paper-festa. 2012: 263] (37) Neska-Ø irakur-tzen ari de-n liburu-a-k girl.ART-ABS read-IPFV PROG AUX.PRS.3SG-REL book-ART-ERG on-a-Ø izan behar du. good-ART-ABS Inferential expressions | 599 Another feature that must be noted is that the copular verbs izan or egon can be omitted when there is an explicit complement governed by the omitted verb, which allows its retrieval. In (38), the complement is Siberiako izotzpean ‘under Siberia’s ice’, with inessive bound postposition governed by the omitted verb egon. In (39), the nominalizing suffix -tear ‘about to’ joined to the verb iritsi ‘arrive’ is also governed by omitted egon: (38) Poeta-ren gorpu-a-k Siberia-ko izotz-pe-a-n [egon] poet.ART-GEN corpse-ART-ERG Siberia-GLOC ice-under-ART-IN be (PFV) behar du. AUX.PRS.3SG ‘The poet’s corpse must [be] under Siberia’s ice.’ [A. Lertxundi: Paper-festa. 2012: 203] (39) Seme-a-k iris-tear behar du. son-ART-ERG arrive-about AUX.PRS.3SG ‘The son must be about to arrive.’ [E. Rodriguez: Katu jendea. 2010: 62] This omission may provoke ambiguity between the epistential meaning and the meaning ‘need’, but the linguistic or situational context most often disambigu- ates the meaning of the utterance, as in (40), (40) Txori bat-Ø behar du. bird one-ABS AUX.PRS.3SG Possible meanings: ‘S/he needs a bird’ or ‘It must be a bird’. [A. Lertxundi: Paper-festa. 2012: 275] where the previous stretch of discourse, which says “Zerbait solidoa erori da kedarrarekin batera. Puska beltz bat” (‘Something solid has fallen at the same time as the soot. A black piece’), prevents possible ambiguity in favor of the epistential reading. Finally, a constraint of the behar construction is that it cannot occur in sub- ordinate clauses. The evidential basis of the behar construction, like that of English must, is restricted to immediate evidence, most often gathered in a time close to the speech moment. In this sense, its evidential basis may be contrasted with that of the Basque future tense, a linguistic device not included in this paper due to its comparably stronger epistemic modal component. King (2009: 473) describes the epistemic modal meaning of the Basque future tense as “probability, conjecture, and in general various shades of epistemic possibil- ity”. The difference between the epistemic meanings of behar and the Basque future tense is, therefore, analogous to that between epistemic deber (de) and the future of probability in Spanish, and between epistemic must and will in 600 | Evidentiality in Basque English (Coates 1983; Carretero 1998). Coates states that must, in its most normal usage, “conveys the speaker’s confidence in the truth of what he is saying, based on a logical process of deduction from facts known to him (which may or may not be specified)” (Coates 1983: 41), whereas with will “the speaker’s confi- dence is not based on a process of logical inference. Instead it is based on com- mon sense, or on repeated experience” (Coates 1983: 177). Even though Coates arguably pushes the difference too far, since virtually all epistemic qualifica- tions may be considered to involve a process of logical inference, the difference undoubtedly exists, and is applicable to the construction with behar and the Basque future tense. The contrast may be seen in (41) and (42): (41) Maria-k hogeita1 hamar2 urte-Ø izan behar Maria-ERG thirty1,2 year-ABS have (PFV) ditu orain. AUX.PRS.3SG now ’Maria must be thirty years old now’. (42) Maria-k hogeita1 hamar2 urte-Ø izan-go ditu orain. Maria-ERG thirty1,2 year-ABS have (PFV)-FUT AUX.PRS.3SG now ‘Maria will be thirty years old now.’ This coexistence of the epistemic modal and evidential meanings has led some scholars to distinguish between two different meanings, or readings, of must and other modal auxiliaries or similar expressions in other languages (De Haan 2000; Cornillie 2009; Mortelmans et al. 2009; Alonso-Almeida 2010, among others). With regard to behar, Zubeldia (2008: 166) gives an example, cited here as (43), (43) Unai-k etxe-a-n behar du. Unai-ERG house-ART-IN AUX.PRS.3SG ‘Unai must be at home’ and states that, depending on the context, behar may have an epistemic modal interpretation (‘Unai will certainly be at home’) or an evidential inferential one, i.e. it may be based on some evidence, as when the light is on and hence the speaker infers that Unai is at home. We argue, however, that both the epistemic modal and the inferential meaning of behar are present in all its occurrences in which it does not have the other non-epistemic meanings mentioned above. That is to say, we consider that these cases of behar should not be divided into either epistemic modal or evidential, but should all be considered as both, i.e. as epistential. Depending on the concrete example, the basis on evidence or the expression of high probability may have more or less relative prominence, but Inferential expressions | 601 this contextual factor is not sufficient to qualify behar as polysemous as far as epistemic modality and evidentiality is concerned. The difference between epis- temic modal and evidential meanings cannot be proved by some classic tests (Quine 1960; Cruse 1986), which would demonstrate a difference between the epistential and the deontic meanings: – The coordination of epistential and deontic meanings produces a zeugma: (44) *Maria-k hogeita1 hamar2 urte-Ø izan behar Maria-ERG thirty1,2 year-ABS have (PFV) ditu eta oraintxe bertan etorr-i. AUX.PRS.3SG and right now come-PFV lit. *‘Maria must be about thirty and come here right now.’ – An example can easily be created with epistential behar followed by a statement that would be contradictory with the meaning of obligation: (45) Ane-k orain zine-a-n egon behar du, Ane-ERG now cinema-ART-IN be (PFV) AUX.PRS.3SG bulego-a-n egon behar ba-lu ere. office-ART-ABS be (PFV) COND-3SG also ‘Ane must be in the cinema now, although she should be in the office’ It must also be noted that, due to the strong evidential component of behar mentioned above, it often occurs with explicit evidence in favor of the epistemic qualification. As for the mode of knowing, there are no restrictions. The evi- dence may be: – Perceptual: (46) Leihatilak lurrinez estalita zeuden eta argi zimel bat baino ez zen sartzen aurpegia zurbildu eta zorroztu egiten ziona; oso itsusia egon behar nuen. ‘The windows were covered with vapour and there went in only a dim light, which made his face pale and sharp; I must have appeared very ugly.’ [R. Saizarbitoria: Kandinskyren tradizioa. 2003: 84] (47) Ahotsagatik eta hitz egiteko moduagatik adinekoak izan behar zutela pentsatu nuen. ‘By their voice and their way of speaking I thought they must be of a certain age.’ [A. Lertxundi: Berria journal, 29-3-2005] – Cognitive, as in (48), where the main basis of the inference is the narrator’s knowledge that only males have the health issue in question: 602 | Evidentiality in Basque (48) Beren hormonekin zerikusia duen zerbait behar du izan, emakumeoi ez baitzaigu horrelakorik gertatzen. ‘It must be something related to their hormones, because to us women that does not happen.’ [R. Saizarbitoria: Kandinskyren tradizioa. 2003: 82] – Discourse-based: the inference being derived from report (49) or from hear- say ((35) above, cited again as (50)). (49) Baliteke, dena den, gertaeretara nolabaiteko hurbiltze poetikoa izatea; honela balitz ere, 1650. urtea baino lehenagokoa izan behar du, zeren eta bakarrik urte horretara arte bait dugu egilea – poemaren sarreran aipatzen den bezala – Arteako benefiziadu. ‘In any case, it may be that there is a sort of poetic approach to events; even though it were so, it must be dated before the year 1650, because the author, according to the citation at the entry of the poem, is beneficiary from Artea only until that year.’ [B. Urgell: “Egiaren kantaz: I. testua eta iruzkinak”. 1986: 79] (50) Onezkero neska zirzill ark, nik aditu nubenez, Madrill-aldeko komenturen batean sartuba izan bear du moja. ‘That shabby girl, according to what I heard, must have already be- come a nun in a convent in the Madrid area.’ [V. Mocoroa: “Damuba... garaiz, edo Bertuteric gabeco aberastasuna iñoiz ez da eguizco zoriontasuna”. 1897: 21] In (51), the evidence in favor of the proposition qualified with behar is provided by an epistemically modalized clause expressed by participle + -go (FUT), fol- lowed by the causal prefix bait- ‘because’ and the past auxiliary zituen: (51) Udaberri-a-Ø izan behar zuen, 14 edo 16 gradu-Ø spring-ART-ABS be (PFV) AUX.PST.3SG or degree-ABS egin-go bait-zituen. do (PFV)-FUT because-AUX.PST.3SG ‘It must have been spring, because there would be about fourteen or six- teen degrees.’ [J. Sarrionandia: Moroak gara behelaino artean? 2010: 11] 14.3 Reportive expressions This section concerns a number of linguistic devices used to express reportive evidentiality. The evidence may consist of concrete reports, or else of rumor and Reportive expressions | 603 hearsay. Some expressions, such as omen or ei ‘it is said’, may be used in both cases, while others can only be used with hearsay evidence. 14.3.1 esan ‘say’ and arabera ‘according to’ Basque has verbs of saying, such as the equivalent of Engl. say, which is esan,11 as well as other verbs such as adierazi ‘declare’ or azaldu ‘explain’ (Jendraschek 2003: 48). For reasons of space, this paper will exclude direct and indirect re- ported speech, which is often considered as lying outside the category of evidentiality (Boye 2012; Chojnicka 2012, among others). However, in order to deal with the evidential function of esan we have to specify that, when it occurs as a verb introducing a complement clause, the complementizing suffix is -ela added to the subordinate verb. The sentence word order may be OVS, SVO or SOV (examples 52–54): (52) Errepide-a-n gizon bat-Ø hil de-la road-ART-IN man one-ABS die (PFV) AUX.PRS.3SG-COMP dio Ane-k. say.PRS.3SG Ane-ERG ‘Ane says that a man has died on the road.’ (53) Ane-k dio gizon bat-Ø hil de-la Ane-ERG say.PRS.3SG man one-ABS die (PFV) AUX.PRS.3SG-COMP errepide-a-n. road-ART-IN (54) Ane-k errepide-a-n gizon bat-Ø hil Ane-ERG road-ART-IN man one-ABS die (PFV) de-la dio. AUX.PRS.3SG-COMP say.PRS.3SG The three orders are possible when the Object is a complement clause, although the second is the most frequent one in spoken Basque. Esan may also be used with the suffix -enez ‘according to’, formed by the relative subordinating suffix -n and the instrumental bound postposition -z. The suffix sequence -enez is specialized in declarative verbs. With this construction, || 11 The variety esan is used in western and central dialects (Biscay and Gipuzkoa), and the variety erran in eastern dialects. 604 | Evidentiality in Basque the second clause is not a complement clause but the main clause, and conse- quently the complementizer -ela does not appear on its verb: (55) Ane-k dio-en-ez, gizon bat-Ø hil Ane-ERG say.PRS.3SG-REL-INS man one-ABS die (PFV) da errepide-a-n. AUX.PRS.3SG road-ART-IN ‘According to [what] Ane says, a man has died on the road.’ An alternative to -enez is the free postposition arabera ‘according to’,12 which requires the genitive case (-en) for the relative verb dioena (dio ‘says’ + relativizer -en + definite article -a): (56) Ane-k dio-en-a-ren arabera, gizon bat-Ø Ane-ERG say.PRS.3SG-REL-ART-GEN according to man one-ABS hil da errepide-a-n. die (PFV) AUX.PRS.3SG road-ART-IN ‘According to what Ane says, a man has died on the road.’ Arabera may also occur without the verb esan, in which case it governs the genitive for the noun or noun phrase mentioning the source of the report: (57) Ane-ren arabera, gizon bat-Ø hil da Ane-GEN according to man one-ABS die (PFV) AUX.PRS.3SG errepide-a-n. road-ART-IN ‘According to Ane, a man has died on the road.’ When the source is imprecise (hearsay), the verb esan may appear in the follow- ing constructions: with the impersonal form (esaten da ‘it is said’ (58–59)) and with third person plural (diote or esaten dute ‘they say’ (60–61)): (58) Esa-ten de-n-ez, Cristiano Ronaldo-k ez say-IPFV AUX.PRS.3SG-REL-INS Cristiano Ronaldo-ERG NEG daki nor-Ø ze-n Saramago-Ø. know.PRS.3SG who-ABS be.PST.3SG-COMP Saramago-ABS ‘According to what is said, Cristiano Ronaldo does not know who Sara- mago was.’ (59) Esa-ten da Cristiano Ronaldo-k ez daki-ela say-IPFV AUX.PRS.3SG Cristiano Ronaldo-ERG NEG know.PRS.3SG-COMP || 12 Arauera is a dialectal variety of arabera. Etymology of arabera: arau ‘rule’ + era ‘manner’. Reportive expressions | 605 nor-Ø ze-n Saramago-Ø. who-ABS be.PST.3SG-COMP Saramago-ABS ‘It is said that Cristiano Ronaldo does not know who Saramago was.’ (60) Diote-n-ez, Cristiano Ronaldo-k ez daki say.PRS.3PL-REL-INS Cristiano Ronaldo-ERG NEG know.PRS.3SG nor-Ø ze-n Saramago-Ø. who-ABS be.PST.3SG-COMP Saramago-ABS ‘According to what is said, Cristiano Ronaldo does not know who Sara- mago was.’ (61) Diote-n-a-ren, arabera, Cristiano Ronaldo-k ez say.PRS.3PL-REL-ART-GEN according to Cristiano Ronaldo-ERG NEG daki nor-Ø ze-n Saramago-Ø. know.PRS.3SG who-ABS be.PST.3SG-COMP Saramago-ABS ‘According to what is said, Cristiano Ronaldo does not know who Sara- mago was.’ 14.3.2 Hearsay: the particles omen and ei Omen and ei are two synonymous particles with a meaning of hearsay. The source of information cannot be explicit in the clause in which they occur, even though it may be inferred from the previous linguistic context or from the extralinguistic context (Korta and Zubeldia 2014: 417–418). Omen13 occurs in the Navarrese and Gipuzkoan dialects, and in the northern dialects; ei occurs in the Biscayan dialect. Both particles are well documented, especially omen, but the resources for documenting the use of ei increased from the second half of the 19th century onwards. Both particles may also be used as nouns. Some observations regarding the diachrony of omen and ei as particles and as nouns are included in Section 6. Omen as a noun means a) ‘fame, reputation, opinion about a person’; b) ‘hon- our, homage, remembrance’. The first meaning is in common use in present-day Basque (62). The second meaning is currently limited to two cases: the construc- tion omen egin (literally ‘do homage’) and the use with the instrumental postpo- sition -z (omenez). Omenez governs genitive case (-en); -en omenez means ‘in honour of, in homage to, as remembrance of’ (63): || 13 Other variants are emen (Gipuzkoan), umen and ümen (Souletin), emon (Roncalese) and men (Upper Navarrese). 606 | Evidentiality in Basque (62) Lehenagotik saia-tu-a nintzen June-ren omen-a-ren before try-PFV-ART AUX.PST.1SG June-GEN reputation-ART-GEN kontra egi-ten. against do-IPFV ‘I had already tried before to go against June’s reputation.’ [E. Rodriguez: Katu jendea. 2010: 147] (63) Ane-ren omen-ez egin-go dugu bihar-ko Ane-GEN homage-INS do (PFV)-FUT AUX.PRS.1PL tomorrow-GLOC afari-a-Ø. dinner-ART-ABS ‘We’ll do tomorrow’s dinner in homage to/in honor of/as a remembrance of Ane.’ As a noun, ei means ‘fame’ as does omen, but its use is very restrictive and not firmly established. For example, the clause with ei in (64) means ‘that the fame of the Basques has arrived from too far’. However, the author also used two other synonyms of ei, namely ospe and entzute (which do not occur as particles, but only as nouns), without which the sense of ei as a noun would probably not have been clear for the reader. (64) Urrun-egi-tik el-du da-la euskaldun-en ei far-too-ABL arrive-PFV AUX.PRS.3SG-COMP Basque-GEN.PL fame edo ospe edo entzute-a-Ø. or fame or fame-ART-ABS ‘The fame, reputation or renown of the Basques has arrived from too far.’ [Euskalzale journal. 1897: 135a] An example of omen as an evidential particle is (65): (65) Cristiano Ronaldo-k Saramago-ren liburu guzti-a-k-Ø Cristiano Ronaldo-ERG Saramago-GEN book all-ART-PL-ABS irakurr-i omen ditu. read-PFV AUX.PRS.3SG ‘It is said that Cristiano Ronaldo has read all the books by Saramago.’ The meaning of omen and ei is a source of disagreement. Zubeldia (2011: 5, Note 5) cites many sources, such as Larramendi (1854), Lafitte (1962), Trask (1981), Wilbur (1981) or de Rijk (2008), which characterize both particles in terms of two mean- ings, hearsay and lack of total certainty. This position is also supported by the Royal Academy of the Basque Language, henceforth Euskaltzaindia (1987: 515): Reportive expressions | 607 The modality of these particles [omen and ei] has two interrelated sides: on the one hand, they express that what is said has been heard from someone else, and on the other, that the speaker has no total certainty that this is true. […] Therefore, these particles are used when the speaker wishes to signal that s/he does not subscribe to what has been said. In fact, everything that has not been seen is not qualified, without any reason, with the par- ticles omen and ei. Someone who has learned something will not necessarily say ‘It is said that Paris is large’, although s/he has never been in Paris. These particles are not used for facts accepted with no caveats. The speaker uses them when s/he wants to hedge his / her position. (here and in the following translations are ours)14 Later, Euskaltzaindia (1993: 446) supports this position even more firmly: “The modality expressed by the particles omen and ei displays two sides: that the speaker heard what s/he states from someone else, and that s/he is not sure at all about the veracity of what s/he states.”15 By contrast, the OEH-DGV only mentions the reportive value for both parti- cles;16 it characterizes their meaning as indicating that the information ex- pressed has its origin in other persons or sources. This view is shared by Zubeldia’s in-depth semantic-pragmatic studies on omen (Zubeldia 2010, 2011, 2013; Korta and Zubeldia 2014),17 which state that the speaker’s/writer’s lack of certainty may be absent. As evidence in favor of this proposal, she applies the cancellability test (Grice 1975), according to which (66) and (67), but not (68), are grammatical (Zubeldia 2013: 114):18 (66) Euria ari omen du, baina nik ez dut uste euririk ari duenik. ‘It is said that it is raining, but I do not think that it is raining.’ (67) Euria ari omen du, baina ez du euririk ari. ‘It is said that it is raining, but it is not raining.’ || 14 Original: “Partikula hauen modaltasunak bata bestearekin loturiko bi alderdi ditu: a) esaten dena besteri entzuna dela adierazten dute batetik eta b) egia denentz ezin duela hiztunak erabat ziurtatu. […] Hiztuna esandakoaz jabe egiten ez dela markatu nahi denean erabiltzen dira, beraz, partikula hauek. Izan ere, ikusi ez den guztia ez da, besterik gabe, omen edo ei partikulen pean jartzen. Zerbait ikasia denak ez du nahitaez ‘Paris handia omen da’ esango, Parisen sekula izan ez bada ere. Besterik gabe onartzen direnentzat ez dira partikula hauek erabiltzen. Hiztunak bere jarrera ñabartu nahi duenean erabiltzen ditu.” 15 Original: “Omen eta ei partikulek adierazten duten modaltasunak bi aurpegi erakusten dizkigu: hiztunak esaten duena besteri entzuna dela eta esaten duen horren egiatasunaz ez dagoela batere ziur.” 16 http://www.euskaltzaindia.eus/index.php?lang=eu. 17 Zubeldia’s works are restricted to omen, but her treatment of omen is also valid for ei. 18 Similar sentences are cited in Korta and Zubeldia (2014: 405–406). 608 | Evidentiality in Basque (68) *Euria ari omen du, baina inork ez du esan euria ari duela. *‘It is said that it is raining, but nobody has said that it is raining.’ In order to confirm these grammaticality judgments, we passed the three sen- tences on to two speakers of Basque, who were to decide if the sentences were possible or impossible in everyday language. Both informants agreed with Zubeldia that (66–67) are possible, in that there is no contradiction between the contents of the two clauses, while in (68) contradiction does exist and hence this sentence is impossible. However, our two informants commented that they had found it difficult to imagine a suitable context in which (66) and (67) would be uttered; they suggested a context in which two persons are in a bed and one of them, still half asleep, might utter these sentences while listening to the in- formation transmitted by the radio, just before getting up to look at the window in search of conclusive evidence. Omen (and ei) can also be used when the speaker knows that the proposi- tion is true (Korta and Zubeldia 2014: 408–409), as in the example from our data cited as (69), where the writer is Jon Kortazar, a professor of Basque literature and therefore an expert on the matter about which he writes. His use of omen can hardly be interpreted as lack of certainty about what he states; rather, it seems due to his wish to avoid an authoritative tone in a generic statement. (69) Ipuingintza-k ez omen du nobela-k short story.ART-ERG NEG AUX.PRS.3SG novel.ART-ERG beste ospe-rik-Ø, ipuingintza-Ø ez omen as much as fame-PARTIT-ABS short story.ART-ABS NEG da nobela-Ø beste sal-tzen. AUX.PRS.3SG novel.ART-ABS as much as sell-IPFV ‘It is said that the short story does not have as much fame as the novel, it is said that the short story does not sell as much as the novel.’ [J. Kortazar: “Ipuina haragi egin zenean”. El País, 2-7-2011 http://elpais.com/diario/ 2001/07/02/paisvasco/994102811_850215.html] Therefore, the epistemic meaning of omen / ei has the status of a GCI, which is communicated by default but may be cancelled or blocked: For this reason, I consider ‘omen’ as an evidential particle, and the epistemic content would surge (when it surges) in certain utterances, as a generalized conversational implicature […] Therefore, we cannot totally associate ‘omen’ with modality; it does not express the speaker’s degree of certainty with its meaning. By its meaning ‘omen’ would Reportive expressions | 609 be evidential, and only in certain cases it acquires an epistemic nuance, when this nuance is acquired. (Zubeldia 2011: 27)19 Within the cases in which the GCI operates, the communicated degree of cer- tainty may vary depending on inferences invited by the context. There are many cases in which certainty may well be interpreted as almost total, and omen and ei serve mainly to indicate that the information was collected by means of re- ports of the very person(s) concerned. In these cases, of which (70) is an exam- ple, omen is best not translated if the target language does not have an equiva- lent particle to omen or other grammatical devices to express evidentiality: (70) Goiz partean Madame Kontxexik laguntzaile guztiekin hitz egin zuen. Bati ez omen zitzaion patea atsegin, besteak hepatitis pasa berria omen zuen eta pekatu larria omen zuen osasunarentzat patea jatea, beste batek linea gorde nahi omen zuela eta ez zuen egunean gaileta bat eta mahats ale pare bat besterik jaten… ‘In the morning, Madame Conchesi spoke with all her assistants. One did not like pâté (+omen), another had suffered from hepatitis (+omen) and considered (+omen) eating pâté as a serious sin against health, another, since he wanted to keep fit (+omen), would not eat but a biscuit and two grapes a day…’ [A. Lertxundi: Paris de la France-ko pateen kasua. 1989: 59–60] A case of clearly weaker degree of certainty is (71), where the topic is Gernika’s bombardment (and hence its distance in time). The use of omen implies that there were no witnesses of the conversation between the man and the girl, or at least not witnesses who remember it accurately. The author, therefore, signals a certain distance with regard to the credibility of the conversation, which could be considered as lack of commitment. (71) Gernika-Ø erre-tan egoa-la, neskato har-ek Gernika-ABS burn-IPFV be.PST.3SG-when girl that-ERG esan ei Ø-eu-tso-n: say (PFV) AUX (ERG.3SG-VL-DAT.3SG-PST) ‘When Gernika was burning, it is said that that girl told him: (…).’ [B. Enbeita: Bizitzaren joanean. 1986. http://www.armiarma.com/ emailuak/elkar/enbeita10.htm] || 19 Original: “Hori dela eta, partikula ebidentzialtzat dut ‘omen’, eta eduki epistemikoa elkarrizketa inplikatura orokortu gisa sortuko litzateke, sortzen denetan, esaldi batzuetan (…) Ezin lotuko dugu, hortaz, ‘omen’ modalitatearekin bete-betean; ez du hiztunaren ziurtasun neurririk adierazten esanahiz. Ebidentziala litzateke esanahiz ‘omen’, eta kasu batzuetan bakarrik hartzen du ukitu epistemikoa; hartzen duenetan.” 610 | Evidentiality in Basque Other similar cases, where the commitment to the truth of the proposition is al- most irrelevant, are omen’s frequent occurrences in certain kinds of fiction, such as jokes, tales or legends. For example, omen occurs frequently in the jokes by the bertsolari20 Pernando Amezketarra (1764–1823) (72). Another example is the be- ginning of Juan Manuel Etxebarria’s legend on Biscay’s highest hermitage (73): (72) Pernando-Ø negu-a-n Aduna-ra joa-ten omen zan Fernando-ABS winter-ART-IN Aduna-ADL go-IPFV AUX.PST.3s bere ardi-akin. his sheep-GEN.PL ‘It is said that in winter Fernando used to go to Aduna with his sheep.’ [G. Mujika: Pernando Amezketarra. 1925) https://klasikoak.armiarma.eus/idazlanak/M/Mujika Pernando042.htm] (73) Behin bat-en, Gorbeia-ko mandazain bat-Ø Arratia-ko once one-IN Gorbea-GLOC muleman one-ABS Arratia-GLOC feria-ra joan ei zan San Inazio bezpera-n. fair-ADL go (PFV) AUX.PST.3s Saint Ignatius Eve.ART-IN ‘It is said that, on a certain occasion, a muleman from Gorbea mountain went to Arratia fair on Saint Ignatius’ Eve.’ [https://bizkaie.biz/1462833842969] In contrast to the degree of certainty, the reportive meaning is context- independent and belongs to the semantics of omen and ei. Alcázar (2010: 146– 147) cites two examples of omen from the Basque translation of the Spanish Consumer Eroski magazine, which refer to facts and which are not cases of reportive evidentiality ((74–75); the addition of “[it’s said]” is ours). These ex- amples sound non-idiomatic in both spoken and written Basque, and omen preserves its meaning in them, so that they do not provide evidence against the semantic status of the reportive meaning (although they do illustrate the diffi- culties that particles like omen pose to translators): (74) Nik neuk euskailu kimikoz baliatuz segituko omen dut, ene postprodukzioa digitala izanagatik. ‘I think [it’s said] I will continue using chemical support, in spite of my postproduction being digital.’ || 20 ‘Bertsolari’ is a person who sings verses, most often improvized. The Basque tradition of bertsolarism, which started before the use of written language, has continued up to the present time. Reportive expressions | 611 (75) Ikuspegi horretatik, neu ikasten saiatzen naizenean, ikasleei egokitzen ahalegintzen naizenean bakarrik irakasten omen dut. ‘From that perspective, [it is only] when I myself try to learn, only when I make an effort to adapt to my students, [it’s said] that I teach.’ Alcázar (2010: 147) also points out that omen can be used as a disclaimer, but he does not provide any examples that show that its reportive value is totally lost. Another fact about omen and ei worth mentioning is that, although they have been translated in the previous examples as ‘it IS said that’, and this prac- tice will be also followed in the remaining examples in this subsection for the sake of simplicity, omen and ei differ from the English construction in that they are not marked for time. Consequently, when the main verb is in the past, there is indeterminacy whether omen and ei mean it is said or it was said: they simply qualify a proposition about the past without explicitly ascribing the communi- cative evidence to the present or to the past. This issue deserves further re- search, but will not be treated in this paper for reasons of space.21 With regard to the status of the evidential value of omen and ei, Korta and Zubeldia (2014: 399–405) convincingly state that omen is not an illocutionary operator and that it contributes to the truth conditions of the proposition, by means of what they call the ‘assent-dissent test’ and the ‘scope test’: both tests show that, in negative clauses with omen, not only the reported content but also the evidential content may be within the scope of a negation operator, as in the example cited here as (76), where omen lies under the scope of ez da egia ‘it is not true’: (76) Ez da egia-Ø euri-a-Ø ari omen NEG be.PRS.3SG true.ART-ABS rain-ART-ABS PROG du-ela. AUX.PRS.3SG-COMP ‘It is not true that someone else stated that it is raining.’ Korta and Zubeldia provide further evidence for their position: omen can occur within the scope of communication predicates such as esan ‘say’ and knowledge predicates such as kontuan hartu ‘take into account’. However, despite belonging to the content of a proposition, omen and ei cannot be directly under the scope of negation: they can be used to express ‘it is not true that it is stated that’, but not ‘it is not stated that’, whose equivalent in Basque is the impersonal form with the verb esan ‘say’ ez da esaten. || 21 Carretero (2019: 292–293) addresses the evidential qualification of propositions about the past with visibly and similar adverbs. 612 | Evidentiality in Basque (77) Ez da esa-ten euri-a-Ø ari du-ela. NEG AUX.PRS.3SG say-IPFV rain-ART-ABS PROG AUX.PRS.3SG-COMP ‘It is not stated that it is raining.’ We must make it clear, however, that omen and ei belong to the content of a proposition, but in their turn they have scope over a proposition, in agreement with the requirement that evidential expressions have propositions within their scope (Boye 2012). For example, arriskutsua omen da expresses a proposition that contains the meaning of omen, ‘it is said that it is dangerous’, and omen in its turn has scope over the proposition expressed by arriskutsua da ‘it is danger- ous’. The relation between the two propositions is graphically rendered as ‘it is said that [it is dangerous]’: omen belongs to the proposition expressed within double quotes and has scope over the proposition expressed within square brackets.22 The use of omen/ei has certain restrictions, which are not shared by esan (Zubeldia 2013: 121–123; Alcázar 2010: 133–134): – The use of omen/ei is restricted to declarative clauses: they cannot occur in interrogative, exclamative or imperative structures. – They always have clausal scope; that is, they cannot focalize on a single constituent of the clause. – They cannot appear in certain types of subordinate clauses: conditional, purpose and subjunctive complement clauses. To these three restrictions, we would add another: – Omen and ei cannot be modified by adjuncts expressing circumstances of any type (time, etc.). However, esan in impersonal form does not have this restriction. For example, in (78), the time adverb beti ‘always’ affects esan ohi da ‘it is usually said’; therefore, omen / ei would not have been possible in this case. (78) Ingelesari buruz, kasu, beti esan ohi da haren silaba bakarreko hitz multzo ikaragarria berealdiko abantaila dela pop musika egiteko. ‘Concerning English, for example, it is always said that its immense quantity of monosyllabic lexical items is a great advantage for compos- ing pop music.’ [H. Cano: “Agente bikoitz baten aitorpenak”. 2016: 129] With regard to the clausal position of omen and ei, the range is as follows: || 22 Boye (2012: 198) provides a similar analysis for the evidential adjective evident in the exam- ple It is evident that Bob is ready. Reportive expressions | 613 a) With periphrastic verbs: a.1.) affirmative clauses: omen and ei are mostly placed between the main verb and the auxiliary verb, as may be seen in the examples quoted above, as in (65), cited again as (79): (79) Cristiano Ronaldo-k Saramago-ren liburu guzti-a-k-Ø Cristiano Ronaldo-ERG Saramago-GEN book all-ART-PL-ABS irakurr-i omen ditu. read-PFV AUX.PRS.3SG ‘It is said that Cristiano Ronaldo has read all the books by Saramago.’ However, the OEH-DGV displays some examples of omen such as (80), all from French-Basque dialects (eastern dialects), with omen behind the compound verb, i.e. behind the main verb and the auxiliary verb. (80) Berant ibil-i dira omen.23 be late-PFV AUX.PRS.3PL ‘They have been late, it is said.’ a.2) negative clauses: omen and ei occur between the negation (adverb ez) and the auxiliary verb. (81) C. Ronaldo-k ez omen ditu Saramago-ren C. Ronaldo-ERG NEG AUX.PRS.3SG Saramago-GEN liburu guzti-a-k-Ø irakurr-i. book all-ART-PL-ABS read-PFV ‘It is said that Cristiano Ronaldo has not read all the books by Saramago.’ b) With synthetic verbs: b.1.) affirmative clauses: omen and ei are placed before the verb: (82) Cristiano Ronaldo-k ba24 omen daki nor-Ø Cristiano Ronaldo-ERG know.PRS.3SG who-ABS ze-n Saramago-Ø. be.PST.3SG-COMP Saramago-ABS ‘It is said that Cristiano Ronaldo knows who Saramago was.’ b.2.) negative clauses: omen/ei are placed between the negation and the verb. || 23 Instead of “berant ibili omen dira”. 24 As we stated about example (6), the prefix ba- functions as morphological reinforcement of synthetic verbs. 614 | Evidentiality in Basque (83) Cristiano Ronaldo-k ez omen daki nor-Ø Cristiano Ronaldo-ERG NEG know.PRS.3SG who-ABS ze-n Saramago-Ø. be.PST.3SG-COMP Saramago-ABS ‘It is said that Cristiano Ronaldo does not know who Saramago was.’ (84) Ane-Ø ez omen dator. Ane-ABS NEG come.PRS.3SG ‘It is said that Ane is not coming.’ In certain cases, the verb is omitted, especially in the case of izan with the meanings ‘be [permanent]’, ‘be [existential]’ and ‘have’ [possessive]’: (85) Euskaldun guzti-a-k-Ø apustuzale-a-k-Ø omen [dira]. Basque all-ART-PL-ABS fond_of_bets-ART-PL-ABS be.PRS.3PL ‘It is said that all Basques [are] fond of bets.’ (86) Mundu-a-n ez omen halako-rik-Ø. world-ART-IN NEG something/somebody like that-PARTIT-ABS ‘It is said that in the world (there is) nothing/nobody like that.’ (87) Buru-rik-Ø ez du-en-a-k txapel head-PARTIT-ABS NEG have.PRS.3SG-REL-ART-ERG beret beharr-ik-Ø ez omen. need-PARTIT-ABS NEG ‘It is said that he who does not have a head does not (have) need of a beret.’ However, omen and ei are not totally identical as far as position is concerned. They differ in that only omen can also occur parenthetically after the clause over which it has scope, as in example (80) above. Mujika characterizes these cases as follows (Mujika 1988: 470): In my opinion, these cases are parentheticals, in which the particle appears ‘outside the Verb Phrase’ and not properly ‘behind it’. This accounts for the pause between the verb and the parenthetical, which is the very particle. This is a resource available in the [Basque] language in order to confer more emphasis or strength to a given element of the sentence. But it cannot be said that the particle is placed after the verb inside the Verb Phrase, but it lies outside it, thus being an emphatic parenthetical.25 || 25 Original: “En mi opinión, se trata de incisos en los que la partícula aparece “fuera del sintagma verbal” y no propiamente “detrás”. De ahí la pausa que se hace entre el verbo y el Reportive expressions | 615 However, the same reference states that omen is never focal when it occurs close to the verb (Mujika 1988: 467). The same may well be stated for all the occur- rences of ei. This different informational prominence of omen depending on its position reminds of the opposition between two critical forces, iconicity and information structure, which Nuyts (2001: 263–270) identifies as the reason for the syntactic variation of epistemic modal expressions in languages. In our view, the study of this opposition can also be applied to evidential expressions. Iconicity pushes evidential expressions to a position outside the clause, in ac- cordance with the fact that evidentials are operators that have propositions under their scope. By contrast, information structure pushes evidentials to non- initial positions inside the clause, in accordance with the tendency of evidential expressions to be non-focal. With the caveat that Nuyts (2001) is based largely on English and Dutch (even though he mentions other modern languages such as German and French and European languages in older periods such as Old English and Middle Dutch), we may tentatively state that these forces also apply to the different positions of omen. The intra-clausal positions described above are chosen when the information-structural factor (i.e. no-focality) is considered as a priority, while the parenthetical position is chosen when the most im- portant force is iconicity, i.e. when language users lay emphasis on omen as a qualifier of the proposition lying under its scope. As further evidence for this non-focality in intra-clausal position, omen in written language is always an independent word, but in spoken language it constitutes an accentual unit with the verb, which has phonetic repercussions. A rule of Basque phonetics is: [s] → [ts] / n ___ That is to say, the sibilant predorsal fricative undergoes affrication when pre- ceded by [n] (and also by [r] or [l]).26 Compare an example: (88) Egin zuen. (‘S/he did it.’) → ‘egintzuen’ [egín swèn] → [egíntswèn] Since omen ends with /n/ and many past auxiliary verbs start with /s/, affricates are produced in spoken language. This kind of affrication is captured in many texts where omen and the past auxiliary started with /s/ occur as a single word || inciso, que no es otro que la partícula. Se trata de un recurso del que dispone la lengua para dar mayor énfasis o fuerza a algún elemento de la oración. Pero no se puede decir que vaya detrás del verbo dentro del mismo sintagma verbal, sino que se halla fuera de este, constituyendo un inciso enfático.” 26 This affricatization is part of a larger-scale phonetic change towards neutralization. 616 | Evidentiality in Basque with a t between them (the way to reflect the affrication of the fricative sibilant), in order to reflect spoken language: (89) Ederki afal-tzen omen-tzuen. very well dine-IPFV it is said-AUX.PST.3SG ‘It is said that she dined very well.’ [Eusko-Folklore Journal, 1957: 81. http://www.euskomedia.org/PDFAnlt/ munibe/1957081096.pdf] 14.3.3 Other expressions of hearsay Basque also has several expressions with complement clauses pointing to the existence of rumors. Some of the complex predicates that are constructed with a complement clause (whose verb takes the complementizer suffix -ela) are nouns (zurrumurrua, hotsa ‘rumor’, berria, albistea ‘news’, etc.) accompanied by verbs such as zabaldu ‘spread’, ibili ‘walk’: (90) Zurrumurru-a-Ø dabil Cristiano Ronaldo-k ez rumor-ART-ABS walk-PRS.3SG Cristiano Ronaldo-ERG NEG daki-ela nor-Ø ze-n Saramago-Ø. know.PRS.3SG-COMP who-ABS be.PST.3SG-COMP Saramago-ABS ‘It is rumored that Cristiano Ronaldo does not know who Saramago was.’ The noun is often omitted, and the meaning of hearsay is expressed with ad- verbs such as bala-bala (‘from mouth to mouth, profusely’): (91) Bala-bala dabil Cristiano Ronaldo-k ez profusely walk-PRS.3SG Cristiano Ronaldo-ERG NEG daki-ela nor-Ø ze-n Saramago-Ø. know.PRS.3SG-COMP who-ABS be.PST.3SG-COMP Saramago-ABS ‘It is rumored that Cristiano Ronaldo does not know who Saramago was.’ Another structural possibility is the function of the complement clause as noun complement instead of a verb complement, with one of the hearsay nouns cited above (‘the rumor that Cristiano Ronaldo…’). In this case, the complement clause is placed on the left of the head noun (which agrees with the Basque order of constituents of the noun phrase, as stated in Section 14.1.3.1), and the complementizer suffix -ela is followed by the derivative suffix -ko, which con- verts verbal complements into nominal complements (dakiela ‘that he knows’ turns into dakielako). The subject may precede the predicate, as in (92), or the predicate may precede the subject (93). Perceptual expressions | 617 (92) Cristiano Ronaldok Saramago nor zen ez dakielako zurrumurrua bala-bala dabil. The rumor goes from mouth to mouth that Cristiano Ronaldo does not know who Saramago was.’ (93) Bala-bala dabil Cristiano Ronaldok Saramago nor zen ez dakielako zurru- murrua. ‘The rumor goes from mouth to mouth that Cristiano Ronaldo does not know who Saramago was.’ The difference between (90)–(91) and (92)–(93) is structural, not semantic; on the other hand, it must be noted that (93) is preferable over (92), since it is easy for speakers/writers to code and for addressees to decode, for the reason that the shorter Verb Phrase (bala-bala dabil) precedes the long subject noun phrase (Saramago nor zen ez dakielako zurrumurrua). 14.4 Perceptual expressions This section deals with expressions specifying perceptual evidence. The analy- sis is restricted to the uses agreeing with Anderson’s (1986: 274) second criteria, i.e. to the uses in which the expressions are not the main predication of the clause. Accordingly, the verb ikusi ‘see’ and other verbs or perception will be excluded when they function as matrix predicates, as in (94), where ikusi is constructed with a complement clause with the suffix -ela as direct object. Ikusi can also govern a two-argument construction, with a noun phrase functioning as a direct object in the absolutive case, and a verbal nominalization (-tze) with the inessive bound postposition -n, as in (95): (94) Jon-Ø kale-a-Ø zeharka-tzen ari ze-la Jon-ABS street-ART-ABS cross-IPFV PROG AUX.PST.3SG-COMP ikus-i dut. see-PFV AUX.PRS.1SG ‘I have seen that Jon was crossing the street’. (95) Jon-Ø kale-a-Ø zeharka-tze-n ikus-i dut. Jon-ABS street-ART-ABS cross-NMLZ-IN see-PFV AUX.PRS.1SG ‘I have seen Jon cross/crossing the street.’27 || 27 According to Arteatx (2012: 409), the difference marked by the feature [± progressive] does not appear to exist in Basque, where the subordinate verb always takes the morpheme -t(z)en. 618 | Evidentiality in Basque By contrast, other constructions with ikusi are included here, such as ikusten dudanez ‘from what I see’ (formed by ikusten dut ‘I see’ + -nez; see Section 14.3.1), ikusten denez ‘according to what is seen’ (formed by ikusten da ‘it is seen’ + -nez): (96) Ikus-ten duda-n-ez, zapata distiratsu-a-k-Ø see-IPFV AUX.PRS.1SG-REL-INS shoe brilliant-ART-PL-ABS moda-n daude. fashion.ART-IN be.PRS.3PL ‘From what I see, brilliant shoes are fashionable.’ (97) Ikus-ten de-n-ez, festa itzel-a-Ø see-IPFV AUX.PRS.3SG-REL-INS party unbelievable-ART-ABS da. be.PRS.3SG ‘According to what is seen, it is an unbelievable party.’ With these constructions, evidentiality is always indirect: in (96)–(97) above, a perceptual (visual) source is combined with an inferential source. The visual meaning of ikusi may also be metaphorically used to communicate inferential- conceptual or inferential-reportive evidentiality: – Inferential-conceptual: (98) Ikus-ten duda-n-ez, kalkulu hau-ek-Ø see-IPFV AUX.PRS.1SG-REL-INS calculation this-PL-ABS akats-ez josita daude. error-INS full be.PRS.3PL ‘From what I see, these calculations are full of errors.’ – Inferential-reportive: (99) Ikus-ten de-n-ez, jende-a-Ø kexa-tu see-IPFV AUX.PRS.3SG-REL-INS people-ART-ABS complain-PFV egin da. do (PFV) AUX.PRS.3SG ‘Apparently, people have complained.’ These expressions also have the epistemic meaning of lack of total certainty, which is a non-cancellable semantic feature of ikusten dudanez and ikusten denez. There are also constructions with nouns or adjectives of sight, which have evidential uses, such as guztien bistara ‘at the sight of everyone’ (guzti-en ‘all’- GEN.PL; bista-ra ‘sight’-ADL); begi hutsez ‘by the naked eye’ (begi ‘eye’; huts-ez ‘mere’-INS); lehen begi-kolpean ‘at first sight’ (lehen ‘first’; begi ‘eye’; kolpe-a-n Indirect indifferent expressions | 619 ‘blow’-ART-IN); guztien agerian ‘at the sight of everyone’ (guzti-en ‘all’-GEN.PL; ageri-a-n ‘discovered, manifest’-ART-IN): (100) Guzti-en bista-ra, Messi-Ø jokalari ikaragarri-a-Ø all-GEN.PL sight-ADL Messi-ABS player impressive-ART-ABS da. be.PRS.3SG ‘At the sight of everyone, Messi is an impressive player.’ (101) Lehen begi-kolpe-a-n, ordenagailu hau-ek-Ø kalitate first eye blow-ART-IN computer this-PL- ABS quality on-eko-a-k-Ø dira. good-GLOC-ART-PL-ABS be.PRS.3PL ‘At first sight, these computers are good quality.’ The evidentiality of these expressions is indirect: in (100)–(101) a visual source is combined with an inferential source. In the cases of begi hutsez and lehen begi-kolpean, there is an epistemic GCI of non-total certainty, which may be cancelled, as in (102): (102) Lehen begi-kolpean, ordenagailu hauek kalitate onekoak dira, baina haiekin zenbait egunetan lan egiten baduzu, konturatzen zara ez direla hain onak. ‘At first sight, those computers are good quality, but if you work with them for a few days you realize that they are not so good.’ 14.5 Indirect indifferent expressions This section concerns expressions of indirect indifferent evidentiality, i.e. where the evidence is indirect and the main source is not clear. In some of their occur- rences, the inferential or reportive source is explicitly expressed or implicated by the linguistic or situational context. 14.5.1 Adverbial clauses meaning ‘as it seems’ Basque has a wide range of expressions equivalent to Engl. as it seems. To start with, there are adverbial clauses with the verbs irudi and eman (see Section 14.2.1), which occur in adverbial clauses. The following three constructions are equivalent: 620 | Evidentiality in Basque – Dirudienez ‘as it seems’, formed by dirudi ‘seems’ + suffix -enez ‘according to’ (see Section 14.3.1). – Iduri duenez ‘as it seems’, formed by iduri du ‘seems’ + suffix -enez; – Ematen duenez ‘as it seems’, formed by ematen du ‘seems’ + suffix -enez. These three constructions are illustrated in (103): (103) a. Dirudi-en-ez, Begoña-Ø eta Jon-Ø seem.PRS.3SG-REL-INS Begoña-ABS and Jon-ABS ederki molda-tzen dira. very well get-IPFV AUX.PRS.3PL b. Iduri duenez, Begoña eta Jon ederki moldatzen dira. c. Ematen duenez, Begoña eta Jon ederki moldatzen dira. ‘As it seems, Begoña and Jon get on very well.’ Three authentic examples are (104)–(106), in which the main source of evidence is not known: for instance, we do not know whether the main basis of the jour- nalist who uttered (104) lies in what he was told by sources such as the neigh- bours or the police, or in the inferences he made when he visited the place of the incidents: (104) Bi kotxe-Ø erre dira Irun-en, dirudi-en-ez two car-ABS burn (PFV) AUX.PRS.3PL Irun-IN seem.PRS.3SG-REL-INS nahita piztutako sute-engatik. deliberately lightened fire-MOT.PL ‘In Irun two cars have been burnt, apparently by intentional fires.’ [Web of the public Basque Radio-television (EITB). http://www.eitb.eus/eu/albisteak/gizartea/osoa/808465/bi-kotxe-erre- dira-irunen-nahita-piztutako-suteengatik/] (105) Ematen duenez bakoitzak nahi duena irakurtzen du besteek idazten dutenetik. ‘As it seems, everyone reads what they want, from what others write.’ [http://zuzeu.eus/2013/11/26/herritarron-eredu-hirueledunaren-alde/] (106) Dirudienez lantzen duen estetika oso deigarria zaie epaimahaikoei, eta dirudienez ez soilik mahai horretakoei. ‘As it seems, the aesthetics that he uses is very showy to the members of the jury and, as it seems, not only to those of that jury.’ [Jon Kortazar, “Munduari bira”, El País, 22-10-2007. http://elpais.com/diario/2007/10/22/paisvasco/1193082006_850215.html] Basque also displays other constructions with the same meaning: Indirect indifferent expressions | 621 – Antza denez/itxura denez, formed by the nouns antz ‘appearance’ or itxura ‘aspect’ with definite article (-a) and absolutive case (Ø) plus denez (da ‘is’ + -enez); itxura may occur on its own, with the instrumental bound postposi- tion (itxuraz), or before batean (bat ‘one’ plus the definite article -a and the inessive bound postposition -n): itxura batean. (108) is an example from the corpus. (107) Antza denez/itxura denez/itxuraz/itxura batean, Begoña eta Jon ederki moldatzen dira. ‘As it seems, Begoña and Jon get on very well.’ (108) Antz-a-Ø de-n-ez, ontzi-a-k-Ø appearance-ART-ABS be.PRS.3SG-REL-INS crockery-ART-PL-ABS garbi-tze-a-Ø gusta-tzen za-i-o wash-NMLZ-ART-ABS like-IPFV AUX (ABS3SG-VL-DAT3S) ‘As it seems, she likes washing up.’ [R. Saizarbitoria: Martutene. 2012] – Agidanez, composed by the noun agi ‘phenomenon, event’, the verb da ‘is’, from izan ‘be-permanent’, and the suffix -enez. Agi izan is a verb whose meaning is ‘appear, manifest oneself as’, but is most frequently used in the fossilized adverbial form agidanez. Aidanez is an alternative form, which has lost the intervocalic plosive voiced /g/, which is a usual phenomenon of spoken Basque.28 Another alternative form is agidanean, where the bound postposition is inessive instead of instrumental: (109) Agidanez/aidanez/agidanean, Begoña eta Jon ederki moldatzen dira. ‘As it seems, Begoña and Jon get on very well.’ In all these examples, the evidential expression may also appear at the end of the utterance (110), as an afterthought. This position is the most habitual one for antza when it appears on its own, without denez (111). An example from the corpus is (112): (110) Begoña eta Jon ederki moldatzen dira, agidanez/antza denez. (111) Begoña eta Jon ederki moldatzen dira, antza. ‘As it seems, Begoña and Jon get on very well.’ || 28 The word aidanez was popularized even in written language due to the title of Anjel Lertxundi’s novel Hamaseigarrenean aidanez (‘At the sixteenth time, as it seems’), published in 1983. 622 | Evidentiality in Basque (112) Haize-a-ren portaera-Ø alda-tzen ari da, wind-ART-GEN behavior.ART-ABS change-IPFV PROG AUX.PRS.3SG agidanez. ‘The behavior of the wind is changing, as it seems.’ [http://www.argia.com/argia-astekaria/2181/haizearen-portaera- aldatzen-ari-da-agidanez] The occurrence of these expressions in initial and final position and detached from the rest of the clause indicates that their clausal position is determined by Nuyts’ (2001) principle of iconicity, that is, their clausal position highlights that they lie outside the structure of the clause and have the proposition indicated by the clause under its scope. Apart from the evidential meaning, these adverbial expressions have an ep- istemic meaning: they cannot be used when the speaker/writer knows that the proposition is either true or false. In other words, the speaker/writer communi- cates lack of total certainty of the truth (or of the falsity) of the proposition. In this respect, they are different from omen and ei. 14.5.2 The particle bide The evidential particle bide has a homonymous noun, which means ‘path, way’. The online dictionary Elhuyar29 offers the following translations for the particle bide into Spanish: según parece, al parecer, por lo visto; parece (que), se dice (que); probablemente, seguramente ‘as it seems, apparently; it seems (that), it is said (that), probably, certainly’. The OEH-DGV offers probablemente, segura- mente, al parecer ‘probably, certainly, apparently’, as well as deber de (epistem- ic ‘must’). As for the source of evidence, the coexistence of reportive and infer- ential expressions in the translations offered is an argument for its classification as an indirect indifferent evidential. This wide range of translations also attests the strong association between epistemic modality and evidentiality; the mean- ing of bide is best characterized as inference that 1) derives from evidence and 2) does not result in total commitment; the second element leads to the considera- tion of bide as an epistential in a similar way to behar. Besides, the OEH-DGV example etorri bide da, with the Spanish translation debe de haber venido ‘s/he must have come’ suggests a complementary distribution between both devices: bide is used with perfective aspect as well as with imperfective aspect, while || 29 https://hiztegiak.elhuyar.eus. Indirect indifferent expressions | 623 behar tends to occur with imperfective aspect (see Section 14.2.2). For the sake of simplicity, bide will be translated in the examples as ‘apparently’. (113) is a constructed example, parallel to those constructed for the adverbial expres- sions; (114) is an example from the corpus: (113) Begoña eta Jon ederki moldatzen bide dira. ‘Apparently, Begoña and Jon get on very well.’ (114) 1980ko hamarkada-n, drog-ek har-tu zuten 1980-GLOC decade.ART-IN drug-ART.PL.ERG raze-PFV AUX.PST.3PL Euskal Herri-a-Ø, eta horr-ek ekarr-i bide Basque Country-ART-ABS and that-ERG bring-PFV zuen hies-a-Ø. AUX.PST.3SG AIDS-ART-ABS ‘In the decade of the 80s, drugs razed the Basque country to the ground, and that, apparently, brought AIDS’ [Berria journal, 25-8-2012. P://www.berria.esu/paperekoa/1519/007/ 001/2012-08-25/ prebentzioaren_inguruko_heziketa_da_hiesari_aurre_ egiteko_giltza.htm] Bide, like omen and ei, occurs sometimes with esan, in which case a report is qualified as indirect evidence: (115) “Ez itzul-i” esan bide z-i-o-n. NEG come back-PFV say (PFV) AUX (ERG.3SG-VL-DAT.3SG-PST) ‘“Don’t come back” he apparently told her.’ [R. Saizarbitoria: Hamaika pauso. 1995: 45] Bide has the same meaning as the evidential adverbials described in Section 14.5.1, and also shares the epistemic feature of lack of certainty. However, bide differs from the adverbials in that its possible syntactic positions are the same as those of ei, and also the same as those of omen with the difference that bide never occurs in final position (see Section 14.3.2). Therefore, its position varies depending on the synthetic or periphrastic kind of the verb and on the positive or negative polarity of the sentence. The syntactic position of bide is, therefore, not determined by iconicity but by information structure, concretely by its non- focality. In contrast, the fact that bide does not belong to the syntactic structure of the clause is not highlighted. 624 | Evidentiality in Basque 14.5.3 The adverb nonbait The adverb nonbait ‘as it seems’ has great positional mobility. Nonbait tends to be placed in initial or final position, detached from the clause by a pause in spoken language and by a comma in written language (116–117). (118) is an example from the corpus. It may also occur as a parenthetical in medial posi- tion, as in (119). (116) Nonbait, Begoña eta Jon ederki moldatzen dira. ‘As it seems, Begoña and Jon get on very well.’ (117) Begoña eta Jon ederki moldatzen dira, nonbait. ‘Begoña eta Jon get on very well, as it seems.’ (118) Nonbait, infernuak badik bere grazia. ‘Apparently, hell has its grace.’ [TV, ETB1, spoken example heard from singer Amaia Zubiria in the pro- gramme “Bisitaria”, 8 March 2012] (119) Begoña eta Jon, nonbait, ederki moldatzen dira. ‘Begoña and Jon, as it seems, get on very well.’ Nonbait, therefore, resembles the adverbials described in Section 14.5.1 and differs from bide in that its clausal scope is highlighted, so that its syntactic positions are determined by the principle of iconicity. However, in contrast to the adverbs, nonbait is not derived from a noun or verb of appearance; its mean- ing is derived from a locative meaning ‘somewhere, in some place’ (it must be noted that non ‘where’ is inessive). Like all the indirect indifferent expressions described above, nonbait has a non-cancellable semantic feature of lack of total certainty, so that it may be semantically characterized as epistential. 14.6 Remarks on diachrony This section includes a number of observations about the origin of some of the expressions analyzed in the preceding sections. With regard to behar ‘must’, its epistemic meaning is relatively recent: in the OEH-DGV, the epistemic meaning appears last among the meanings listed in 14.2.2, and the earliest attested ex- amples date from the beginning of the nineteenth century; since then, its use has been growing in the North and in the South of the Basque-speaking region. The development of this epistemic meaning agrees with the path from deontic Remarks on diachrony | 625 to epistemic meaning, widely attested in diachronic studies of modal expres- sions in many languages (Bybee et al. 1994; Hansen and De Haan 2009). This path may well coalesce with the influence of Spanish and French, both of which have polysemous deontic – epistemic modal auxiliaries. The diachrony of omen has still not been included in the Euskal Hiztegi Historiko eta Etimologikoa (‘Basque Historical and Etymological Dictionary’) (Lakarra et al. 2019). Up to date there is no agreement in the unfinished Dicci- onario Etimológico Vasco (‘Basque Etymological Dictionary’) by Agud and Tovar (1989–1995), which collects the largest number of omen etymologies proposed until 1995. Most experts agree on the Latin origin of this word; among the Latin proposals for etymology, the most widely accepted is omen ‘augur, presage’, but there are also nomen ‘name’, which also has the sense ‘fame, reputation’, and lumen ‘light, resplandor’. The use of omen and ei as nouns, mentioned in Section 14.3.2, and their evo- lution to particles might well be related to the Spanish expression es fama que (literally ‘it is fame that’), now dated. Two examples from the literature are (120) and the more recent (121): (120) Este libro, señor compadre, tiene autoridad por dos cosas: la una, porque él por sí es muy bueno, y la otra, porque es fama que le compuso un discreto rey de Portugal. [Miguel de Cervantes: Don Quixote. 1605. Part I, Chapter 6] ‘This book, gossip, is considerable upon two accounts; the one, that it is very good in itself; and the other, because there is a tradition that it was written by an ingenious kind of Portugal.’ [Translation by Charles Jarvis. 1742] (121) Recordé que es fama entre los etíopes que los monos deliberadamente no hablan para que no los obliguen a trabajar y atribuí a suspicacia o a temor el silencio de Argos. [Jorge Luis Borges, ‘El inmortal’, 1947] ‘I recalled that it is generally believed among the Ethiopians that mon- keys deliberately do not speak, so that they will not be forced to work; I attributed Argos’ silence to distrust or fear.’ [Jorge Luis Borges. “The immortal.” Translation by Andrew Hurley. 1988] As for bide, the OEH-DGV does not specify whether there is any diachronic rela- tion between the particle and the noun. In contrast, Mujika (1988: 477) does state that the evidential particle may well derive from the honomymous noun, through the conventionalization of some expressions such as bide dela ‘as is supposed’ (etymologically bide + da ‘is’ + suffix COMP -ela) or eratu ni bide naiz ‘I must have gone mad’ (etymologically eratu ‘get mad’ + ni ‘I’ + bide + naiz ‘am’), found in texts dated from the early 17th century, as collected in Michelena (1964). 626 | Evidentiality in Basque 14.7 Conclusions This paper provides a detailed analysis of the Basque expressions correspond- ing to the crosslinguistic evidential hyperlexemes SEEM, MUST, SAY, ACCORDING to and SEE. Through the analysis of these expressions, the paper has shown that Basque displays a wide array of expressions of evidentiality. These expressions are shown to have a high degree of idiosyncrasy, motivated by the peculiar etymology of some of them, the subtypes of evidence with which they can be used, the syntactic type, their clausal position (which may be mainly motivated by iconicity or by information structure), and their relationship with epistemic modality. For instance, behar, bide or nonbait have both evidential and epistem- ic semantic features, while in omen and ei have an evidential semantic feature and an epistemic Generalized Conversational Implicature. The linguistic richness of the category of evidentiality in Basque leads us to recommend further research on individual evidential expressions along the lines of the existing research on omen, which would increase knowledge of the Basque language and provide valuable data for crosslinguistic studies on evidentiality. Acknowledgments: This research has received support from the research project The expression of evidentiality and modality in English and other European lan- guages: Crosslinguistic perspectives (EUROEVIDMOD), Ref. FFI2011-23181, fund- ed by the Ministerio de Economía y Competitividad (Spanish Ministry of Econo- my and Competitiveness). We thank Xabier Amuriza for his patience and tireless help as an informant, and Xabier Artiagoitia for a thorough review of an earlier version of this manu- script. The remaining shortcomings and inconsistencies are our sole responsi- bility. Abbreviations in glosses (not included in LGR) ADL adlative (bound postposition) ADVS adverbial suffix DER derivative suffix GLOC genitive-locative (bound postposition) IN inessive (bound postposition) MOT motivative (bound postposition) PARTIT partitive (article) VL verb lexeme References | 627 References Agud Manuel & Antonio Tovar. 1989–1995. Diccionario Etimológico Vasco (7 vol.). Anejos del Anuario del Seminario de Filología Vasca “Julio de Urquijo”, nº13, 19, 24, 26, 30, 33, 37. San Sebastian: Diputación Foral de Guipúzcoa. Aikhenvald, Alexandra Y. 2004. Evidentiality. Oxford: Oxford University Press. Alcázar, Asier. 2010. Information source in Spanish and Basque: A parallel corpus study. In Gabriele Diewald & Elena Smirnova (eds.), Linguistic realization of evidentiality in Europe- an languages, 131–156. Berlin & Boston: De Gruyter Mouton. Alonso-Almeida, Francisco. 2010. The value of may as an evidential and epistemic marker in English medical abstracts. Studia Anglica Posnaniensia 46(3). 59–73. Anderson, Lloyd B. 1986. Evidentials, paths of change, and mental maps: Typologically regular asymmetries. In Wallace Chafe & Johanna Nichols (eds.), Evidentiality: The linguistic cod- ing of epistemology, 273–312. Norwood, NJ: Ablex. Arteatx, Iñigo. 2012. Perception verb complements in Basque. In Urtzi Etxeberria, Ricardo Etxepare & Myriam Uribe-Etxebarria (eds.), Noun phrases and nominalization in Basque: Syntax and semantics, 397–435. Amsterdam & Philadelphia: John Benjamins. Artiagoitia, Xabier. 2001a. Irudiak eta emaileak. Anuario del Seminario de Filología Vasca “Julio de Urquijo” 25(1). 29–64. http://www.ehu.eus/ojs/index.php/ASJU/article/view/ 8815. Artiagoitia, Xabier. 2001b. Seemingly ergative and ergatively seeming. In Julia Herschensohn, Enrique Mallén & Karen Zagona (eds.), Features and interfaces in Romance. Essays in honor of Heles Contreras, 1–22. Amsterdam & Philadelphia: John Benjamins. Artiagoitia, Xabier. 2003. Complementation (noun clauses). In José Ignacio Hualde & Jon Ortiz de Urbina (eds.), A grammar of Basque, 634–710. Berlin & New York: De Gruyter. Artiagoitia, Xabier & Arantzazu Elordieta. 2016. On the semantic function and selection of Basque finite complementizers. Kasper Boye & Petar Kehayov (eds.), Complementizer se- mantics in European languages, 379–412. Berlin & Boston: De Gruyter Mouton. Azkarate, Miren, Xabier Kintana & Xabier Mendiguren Bereziartu (dir.). 1996. Elhuyar euskara- gaztelania/castellano-vasco hiztegia. Usurbil: Elhuyar. https://hiztegiak.elhuyar.eus. Boye, Kasper. 2012. Epistemic meaning: A crosslinguistic and functional-cognitive study. Berlin & Boston: De Gruyter Mouton. Bybee, Joan L., Revere Perkins & William Pagliuca. 1994. The evolution of grammar: Tense, aspect and modality in the languages of the world. Chicago: University of Chicago Press. Carretero, Marta. 1998. On the intraclausal verbal present high probability in Spanish and English. Multilingua 17(4), 329–260. Carretero, Marta. 2013. Review of Boye (2012). Linguistics 51(6). 1401–1410. Carretero, Marta. 2019. Evidentiality in adverbs of manner of perceivability. The case of English manifestly, noticeably, patently and visibly. Functions of Language 26(3). 275–307. Chojnicka, Joanna. 2012. Reportive evidentiality and reported speech: Is there a boundary? Evidence of the Latvian oblique. In Aurelija Usonienė, Nicole Nau & Ineta Dabašinskienė (eds.), Multiple perspectives in linguistic research on Baltic languages, 170–192. Newcas- tle Upon Tyne: Cambridge Scholars. http://www.cee-ities.uni-bremen.de/Reportive_evi- dentiality_and_reported_spe.pdf. Coates, Jennifer. 1983. The semantics of the modal auxiliaries. London: Croom Helm. Cornillie, Bert. 2009. Evidentiality and epistemic modality. On the close relationship between two different categories. Functions of Language 16(1). 44–62. 628 | Evidentiality in Basque Cruse, D. A. 1986. Lexical semantics. Cambridge: Cambridge University Press. De Haan, Ferdinand. 2000. Evidentiality in Dutch. In S. S. Chang, L. Liaw & J. Ruppenhofer (eds.), Proceedings of the Twenty-Fifth Annual Meeting of the Berkeley Linguistics Society, 74–85. Berkeley: Berkeley Linguistics Society. Euskal Herriko Unibertsitatea/Universidad Del País Vasco: Sareko Euskal Gramatika, http://www.ehu.eus/seg/. Euskaltzaindia. 1987. Euskal Gramatika. Lehen urratsak-II. Bilbao. www.euskaltzaindia.eus. Euskaltzaindia. 1993. Euskal Gramatika. Laburra: Perpaus Bakuna. Bilbao. www.euskalt- zaindia.eus. Euskaltzaindia. 1999. Euskal Gramatika. Lehen urratsak-V (Mendeko perpausak-I). Bilbao. www.euskaltzaindia.eus. Faller, Martina. 2002. Semantics and pragmatics of evidentials in Cuzco Quechua. Stanford University Ph.D. dissertation. Grice, H. Paul. 1975. Logic and conversation. In Peter Cole & Jerry L. Morgan (eds.), Syntax and semantics 3: Speech acts, 41–58. New York: Academic Press. Hansen, Björn & Ferdinand De Haan (eds.). 2009. Modals in the languages of Europe. A refer- ence work. Berlin & New York: De Gruyter. Hualde, José Ignacio & Jon Ortiz de Urbina (eds.). 2003. A grammar of Basque. Berlin & New York: De Gruyter. Huddleston, Rodney & Geoffrey K. Pullum. 2002. The Cambridge grammar of the English lan- guage. Cambridge: Cambridge University Press. Jendraschek, Gerd. 2003. La modalité épistémique en basque. München: LINCOM Europa. King, Alan R. 2009. Modality in Basque. In Björn Hansen & Ferdinand De Haan (eds.), Modals in the languages of Europe. A reference work, 457–486. Berlin & New York: De Gruyter. Korta, Kepa & Larraitz Zubeldia. 2014. The contribution of evidentials to utterance content: Evidence from the Basque reportative particle omen. Language 90(2). 389–423. Lafitte, Pierre. 1962. Grammaire basque (Navarro-Labourdin littéraire). Édition revue et corri- gée. Bordeaux: Delmas (re-edition: 1979, San Sebastian: Elkar). Lakarra, Joseba, Julen Manterola & Iñaki Segurola. 2019. Euskal Hiztegi Historiko-Etimologikoa (EHHE-200). Bilbao: Euskaltzaindia. Lakoff, George & Mark Johnson. 1980. Metaphors we live by. Chicago: University of Chicago Press. Lampert, Günther. 2015. Sense activation triggering in English epistentials: Attention distribu- tion, contextual modulation of meaning, and categorization issues. Cognitive Semantics 1. 77–103. Larramendi, Manuel de. 1854. Grammaire de la langue basque: d’après celle du P. Manuel de Larramendi intitulée El impossible vencido/par S.G. Blanc. Lyon. Paris: S.H. Blanc et Cie. Marín-Arrese, Juana I. 2013. Stancetaking and inter/subjectivity in the Iraq Inquiry: Blair vs. Brown. In Juana I. Marín-Arrese, Marta Carretero, Jorge Arús Hita & Johan van der Auwera (eds.), English modality: Core, periphery and evidentiality, 410–445. Berlin & Boston: De Gruyter Mouton. Michelena, Luis. 1964. Textos arcaicos vascos. Madrid: Minotauro. Michelena, Luis & Ibon Sarasola (dir.) 1987–2005. Orotariko Euskal Hiztegia/Diccionario General Vasco. Bilbao: Euskaltzaindia, etab. I. A-Ama (1987). II. Ame-Asd (1989). III. Ase- Bapuru (1989). IV. Bar-Berp (1990). V. Berr-Daf (1992). VI. Dag-Erd (1992). VII. Ere-Fa (1992). VIII. Fe-Gub (1995). IX. Gud-Jab (1996). X. Jad-Kop (1997). XI. Kor-Maj (1998). XII. Mak-Oal (1999). XIII. Oam-Peb (2001). XIV. Ped-Sua (2002). XV. Sub-Usu (2004). XVI. Ut- Zuz (2005). www.euskaltzaindia.eus. References | 629 Mortelmans, Tanja, Kasper Boye & Johan van der Auwera. 2009. Modals in the Germanic lan- guages. In Björn Hansen & Ferdinand De Haan (eds.), Modals in the languages of Europe. A reference work, 11–69. Berlin & New York: De Gruyter. Mujika, José Antonio. 1988. Partículas modales de la flexión verbal. Anuario del Seminario de Filología Vasca “Julio de Urquijo”, 20(2). 453–478. http://www.ehu.eus/ojs/index.php/ ASJU/article/view/7993 Nuyts, Jan. 2001. Epistemic modality, language and conceptualization. A cognitive-pragmatic perspective. Amsterdam & Philadelphia: John Benjamins. Quine, Willard V. O. 1960. Word and object. Cambridge, MA: MIT Press. Rezac, Milan, Pablo Albizu & Ricardo Etxepare. 2014. The structural ergative of Basque and the theory of case. Natural Language & Linguistic Theory. Online publication date: 15-May- 2014. http://www.umr7023.cnrs.fr/sites/sfl/IMG/pdf/ERG-Complete12f.pdf. Rijk, Rudolf P. G. de. 2008: Standard Basque: A progressive grammar. Vol. 1: The grammar. Cambridge, Massachusetts: MIT Press. Trask, Robert L. 1981. Basque verbal morphology. In EUSKALTZAINDIA, Euskalarien nazioarteko jardunaldiak (Iker-1), 285–304. Bilbao. http://www.euskaltzaindia.net/dok/ikerbilduma/ 6777.pdf. Wiemer, Björn & Katerina Stathi. 2010. Introduction. The database of evidential markers in Euro- pean languages. A bird’s eye view of the conception of the database (the template and prob- lems hidden beneath it). STUF/Language Typology and Universals 63(4). 275–285. Wilbur, Terence H. 1981. Basque syntax. In EUSKALTZAINDIA, Euskalarien nazioarteko jardunal- diak (Iker 1), Bilbao, 169–186. http://www.euskaltzaindia.net/dok/ikerbilduma/6777.pdf. Willett, Thomas. 1988. A cross-linguistic survey of the grammaticalization of evidentiality. Studies in Language 1(1). 51–97. Zubeldia, Larraitz. 2008. Ebidentzialitatearen eta modalitatearen arteko harremanaz. Gogoa 8(2). 159–176. http://www.ehu.eus/ojs/index.php/Gogoa/article/view/3520. Zubeldia, Larraitz. 2010. “Omen” partikularen azterketa semantikoa eta pragmatikoa. UPV/EHU (University of the Basque Country) Ph.D. dissertation. http://www.euskara.eus- kadi.eus/appcont/tesisDoctoral/PDFak/Larraitz_Zubeldia_TESIA.pdf. Zubeldia, Larraitz. 2011. ‘Omen’ partikula: ziurtasunak eta ziurtasunik ezak. Gogoa 11(1). 1–32. http://www.ehu.eus/ojs/index.php/Gogoa/article/view/6648. Zubeldia, Larraitz. 2013. (Non)-determining the original speaker: Reportative particles versus verbs. Research in Language 11(2). 103–130. Appendix (Sources of the corpus created by Karlos Cid Abasolo) Texts from contemporary Basque literature (20th and 21st centuries) Arrieta, Bertol, Alter Ego, Susa, 2012. Cano, Harkaitz, “Agente bikoitz baten aitorpenak”, in Anjel Lertxundi, Demagun ehun urte barru. San Sebastian: EIZIE, 2016. 630 | Evidentiality in Basque Enbeita, Balendin, Bizitzaren joanean, Elkar, 1986. Etxaniz, Nemesio, Lur berri billa, 1967, Izarra. Ibiñagabeitia, Andima, Bergiliren Idazlanak osorik: Unai-Kantak eta Alor-Kantak, 1966. Illarramendi, Ramón, Testimonio falsoa. Jostirudi negargarri eta pozgarria. Egintza bat eta iru kuadroetan moldatua, Altuna, 1918. Iraizoz, Policarpo, Yesu Kristo gure Yaunaren bizia, Capuchinos, 1934. Lertxundi, Anjel, Paris de la France-ko pateen kasua, Erein, 1989. Lertxundi, Anjel, Paper-festa, Alberdania, 2012. Mocoroa, Valeriano, Damuba... garaiz, edo Bertuteric gabeco aberastasuna iñoiz ez da eguizco zoriontasuna, Francisco Muguerza, 1897. Narbaitz, Pierre, San Frantses Jatsukoa, Izarra, 1966. Rodriguez, Eider, Katu jendea, Elkar, 2010. Saizarbitoria, Ramon, Hamaika pauso, Erein, 1995. Saizarbitoria, Ramon, Kandinskyren tradizioa – La tradición de Kandinsky, Atenea, 2003. Saizarbitoria, Ramon, Martutene, Erein, 2012. Sarrionandia, Joseba, Moroak gara behelaino artean?, Pamiela, 2010. Uribe, Kirmen, Mussche, Susa, 2012. Sources of older written texts Euskalzale, 1897. Eusko-Folklore, 1957. Mujika, Gregorio, Pernando Amezketarra, Iñaki Deuna, 1925. Urgell, Blanca, “Egiaren kantaz : I. testua eta iruzkinak”. Anuario del Seminario de Filología Vasca “Julio de Urquijo”, 20:1, 1986, 75–149. http://www.ehu.eus/ojs/index.php/ASJU/ article/view/7725 Web pages http://www.argia.eus/ http://www.armiarma.com http://berria.eus https://bizkaie.biz/ http://www.diariovasco.com http://www.eitb.eus http://elpais.com http://www.euskaltzaindia.eus http://www.euskomedia.org https://klasikoak.armiarma.eus http://zuzeu.es