Abstract
Basque is the only non-Indo-European language in western Europe. This fact, and particularly its ergative alignment, make its morphosyntactic structure and its verb different from those of Standard Average European. However, the massive and prolonged influence which Basque has received first from Latin and later from Romance has conditioned the layout of the analytic vps (the open type) in a very curious way. Since Basque synthetic verbs have a template of the type S-vb.root for intransitives and O-vb.root-A for transitives, as opposed to vb.root-A/S for any kind of verb in sae, lexical borrowing of verbs from Latin was impossible. A solution arose when the old periphrastic resultative perfect was grammaticalized in Late Latin as the primary expression of the perfect. This form distinguished intransitive and transitive verbs, so it served as an entry point for Latin and Romance verbal lexicon into Basque, by means of autochthonous auxiliaries.
1 Introduction1
Basque has several unique features among the languages of western Europe. It is one of the few which survived the Romanization process. It is actually the only one of non-Indoeuropean filiation. From the point of view of the inner linguistic structure, it is the only language with an ergative alignment in its morphology, with Object (= O) agreement within the finite verb, and with Agent-Object-Verb (= aov) as the unmarked order of constituents within the clause. Despite all these differences, and although it is technically a language isolate, Basque has been anything but linguistically isolated in the course of the last twenty centuries. In the lexicon, estimations about the relative amount of loanwords vary, but I believe it is reasonable to agree with Trask (1997: 249) that “the number of native Basque words […] are now probably outnumbered in ordinary speech by words of Latin or Romance origin, though not by a large margin”. As for the grammatical structure, even if Basque has gone its own way due to its structural differences, that way has run parallel to the one of neighboring Romance and more generally of Standard Average European (= sae),2 as shown in the following sections.
The purpose of this paper is precisely to describe how the pervasive and prolonged influence of Latin and Romance upon Basque has profoundly shaped the morphosyntactic structure and the verbal system of this language, even though this is an ergative language with a template of finite verbs totally different from that of sae, a feature which has blocked borrowing by direct route (from finite verb to finite verb). As shown, the periphrastic perfect is the cornerstone of the whole process.
In Section 2, I will explain from when on Basque starts to be attested, as well as the nature of these attestations in each period. I will also describe what the nature of the Latin and Romance influence upon Basque has been like, and what chronology is assumed for its prehistory according to the standard terminology. Section 3 is a brief description of the general features of Basque, focusing particularly on its ergative alignment, and Section 4 a summary of its verbal morphology and its Tense-Aspect-Mood (= tam) system.
From then on, I will make an attempt at an explanation of how the different analytic vps have been formed in the course of its prehistory, starting from the contact with Late Latin, whose perfect appears to have been the starting point for the creation of the analytic perfect of Basque (Section 5). Next, I will expound how Basque transformed an originally progressive periphrasis into an analytic present following the model of the perfect, as proposed by Mounole (2008) (Section 6), and perhaps a deontic periphrasis into an analytic future by the same procedure (Section 7). In Section 8, I will argue that, despite all this, two-piece vps did not arise through contact with Latin and Romance, but that Basque most probably already had an analytic vp construction comprising the verbal root plus another set of auxiliaries. Section 9 is a chronological summary of the developments set out and/or proposed for Basque, and Section 10 a concluding remark.
2 Sources, the Nature of Latin and Romance Influence, Periodization
The beginning of a Roman influence of a considerable degree upon the territory which today constitutes the Basque Country, including the Autonomous Basque Community, Navarre, and the French Basque Country – the latter belonging today to the Department of the Pyrénées-Atlantiques –, can be set, perhaps, in the 1st century bc, as a result of the Sertorian War. From then on, the territory corresponding to the modern Basque Country was Romanized, and, while this Romanization was deeper in the plains like the Ebro Valley than in the mountainous regions closer to the Atlantic Ocean, all Basque dialects do show evidence of massive lexical borrowing covering practically every semantic realm (see Trask, 1997: 259–261).
As for the linguistic sources, after the person and divinity-names engraved in inscriptions during the 1st-3rd centuries ad (the so-called Aquitanian language, with a very high proportion of evidence coming from the upper Basin of the Garonne river), we have to wait until the 10th century to find a considerable amount of onomastic material again. Most of this is related to the administration of monasteries, which flourish in the Kingdom of Pamplona. Later in the course of the Lower Middle Ages, some other sources like codes of laws, censi, and many others also emerge. Yet all kinds of sources are written in Latin until the first half of the 13th century, and mainly in Romance – be it Castilian, Navarrese-Aragonese, Occitan, or even French – from then on.
This is an important point for our discussion: unlike most neighboring Romance languages, Basque was never used, during the Middle Ages, as a written language. This makes our sources few and far between. Moreover, except for a couple of short sentences dating from the beginning of the 11th century, the evidence is mainly reduced to some lexicon and some few cases of the noun declension. Even more important for the issue at stake, this has prevented the occurrence of any influence of Ecclesiastical Latin upon the syntax and verbal morphosyntax of Basque.
Even when a Basque (mostly religious) written tradition begins in the 16th and 17th centuries, in the great majority of works which are translations of other religious texts – Christian doctrines, catechisms, devotional works, etc. –, the source language is Spanish or French, not Latin. In the case of the translation of the New Testament published by the Calvinist priest Ioanes Leizarraga in 1571, in recent years it has been convincingly shown that he must have mainly used one or two French translation(s) made in the early 60s within the Calvinist tradition, probably without even having an occasional look at the Vulgata.
Before going into the content, I consider it necessary to make a remark about the chronology and the chronological terminology which is currently employed in Basque diachronic research. Since Mitxelena (1981/1987), it is generally assumed that modern Basque dialects do not date back to Antiquity and that some coincidences between their geographic distribution and the tribe division made by Ptolemy in the 2nd century ad and some other authors are purely incidental. Instead, Mitxelena proposed that modern dialects may have had their starting point approximately in the 5th-6th centuries, simultaneously with a series of events like: a) the disappearance of all pre-Roman languages in neighboring areas except Basque itself, which suddenly became a linguistic island surrounded by (pre-)Romance languages, b) the socio-political upheavals typical of Late Antiquity, like the Bagaudas’ riots – simultaneous in turn with the peak of Germanic invasions (5th century) –, some of which took place within the territory of the Vascones or in their surroundings, and c) from the 6th century onwards, the hostility of Visigoths first from the south, and of Franks later from the north. All those factors must have brought about some kind of common linguistic awareness which perhaps fostered a koiné. This is what is referred to as Old Common Basque (= ocb) in later research.
3 General Features of Basque: Is it Morphologically Ergative?
Basque is an agglutinative language, which marks grammatical relations by means of case suffixes. Its non-marked word order is theoretically aov, although this is not strict (in fact, in many corpus analyses, the opposite is true). Its word order is mostly patterned according to the Modifier-Head type: Verb-Auxiliary, Genitive-Noun, Relative. Clause-Noun, and Noun-Postposition (= VAux, gn, RelN, and NPo), but also Noun-Adjective (= NAdj, traditionally considered an order of the Head-Modifier type, although this correlation has been disputed). There are also attestations of ng and NRel (albeit much fewer than those of gn and RelN), particularly in old texts.
Basque is, in principle, morphologically ergative. However, the number of unergative verbs is large enough for some analysts to consider it a split-S language (whatever the terminology). Nobody has succeeded, though, in establishing a semantic definition which always rightly predicts whether a particular predicate will be unergative or unaccusative. It is important to clarify this aspect because it is of the utmost relevance in order to establish to what extent Basque differs from sae in the marking of grammatical relations, and thus to understand the developments undergone by its verbal morphology.
Most unergative verbs fall into one of these three groups (see Aldai, 2009a).3 Firstly, we have X egin compounds like lo egin ‘sleep’, korrika egin ‘run’, negar egin ‘cry, weep’. In these, egin ‘do’ functions as a light verb which, combined with another element, can express any kind of semantically intransitive action, but triggers ergative morphology on the np and on the verb. Secondly, we have three verbs, iraun ‘last’, irakin ‘boil’, irudi ‘seem, resemble’, which can be conjugated synthetically and hence belong to the autochthonous, non-borrowed lexicon of the language. Thirdly, we have some verbs which are unergative in western dialects but unaccusative in non-western ones, as one can see by comparing (1a) with (1b).
As Aldai highlights, western dialects are clearly innovative on this point, since it is fairly clear that they have developed a tendency to express some intransitive predicates through unergative constructions. Hence, Old Common Basque was closer to a pure ergative system, the morpho-syntactic marking of which was based primarily on valency.
4 The tam System of Basque
Basque verbal morphology is relatively complex (Trask, 1997: 102–109). Here I will just summarize, within that complexity, what applies to our discussion, leaving aside several parameters.
Most Basque vps – in the functional sense – are analytic and comprise a nonfinite verb which contains the verbal lexeme and an overt or covert morpheme expressing tam values, plus an auxiliary verb which contains more tam morphemes, again overt or covert and amalgamated or not, and also the argument markers for A, S, O and Rec (= Recipient). Using ekarri ‘bring’ as an example, the possible forms of the non-finite verb within a vp are: verb-stem (ekar), perfective participle (ekarr-i), imperfective participle (ekar-tzen), and prospective participle (ekarr-i-ko). Auxiliary verbs have distinct sets which make up quite a sophisticated system. In modern Standard Basque, the indicative mood has a present set and a past set; the subjunctive a present set and a past set; and the potential a present, a past, and a hypothetical set, whereas the imperative has no time.
In addition, there is a closed class of verbs which allow synthetic conjugation.4 These strong verbs belong to the nuclear lexicon of the language (‘be’, ‘have’, ‘go’, ‘come’, ‘walk’, ‘carry’, ‘bring’), and are not borrowed. In them, one single word gathers the verbal lexeme, the tense and mood morphemes, and the argument markers. However, in today’s Basque all of them have imperfective semantics. That means that, in order to express perfective and prospective or future predicates, one has to resort to analytic constructions even with these verbs. Thus, one can say (2a) to express the present imperfect, but one has to resort to (2b) and (2c) to express the perfect and the prospective.
Finally, for any finite verb, be it the auxiliary of an analytic construction or a synthetic one, voice and grammatical relations in general are expressed by the use of four different sets with four different valencies: monovalent intransitive (marker for S), bivalent intransitive (markers for S and Rec), bivalent transitive (markers for A and O), trivalent transitive (markers for A, O, and Rec). From the point of view of the contrast between Basque and sae, the sets with a Rec marker are not meaningful, as they arose as a result of an inner development with no clear match in neighboring languages, and they were not affected by the Latin-Romance influence either. Hence, in the following sections, we shall restrict our analysis to intransitive and transitive sets with no Rec. Bearing all this in mind, let us have a look at a minimal sketch of the tam system in Basque. This is represented in Table 1.
Note that, unlike in sae, the intransitive/transitive distinction is crucial for the layout of argument markers within finite verbal forms, because it splits all these into two well-defined types, with two different morphologies. In Table 2, I represent the template of a standard present form in sae, as opposed to the template of a standard present form in Basque.5
The mere comparison between n-a-tor ‘I come’ and da-kar-t ‘I bring’ should illustrate the drastic contrast in morphology which the intransitive/transitive opposition implies, as well as the fact that this is one of the features which diametrically opposes Basque to sae. Such a difference in morphological structure of finite verbal forms must have been of crucial importance for the structuring of Latin and Romance influence, as it blocked the borrowing of finite verbal forms. A further problem was that, unlike in noun inflection – and unlike the finite verbal morphology of sae –, the Basque verb conjugation has prefixes as argument markers, which must have been another hindrance to borrow finite verbs from Latin and Romance. The solution has turned out to be the development of a robust analytic system, which covers all possible tam values required by any verb and hence allows massive borrowing, and in which a non-finite form with a simple morphology hosts the borrowed verbal lexeme, whereas the complex morphology brought about by the intransitive/transitive distinction is expressed by an autochthonous finite form.
To some extent, this situation is comparable to that of German (see Drosdowski et al., 1984), in which verbs borrowed from Latin and Romance do not take ge- to form the past participle (e.g., er hat inspirier-t), cannot build the preterit or the Konjunktiv ii by apophonic alternation (er würde inspirieren, as opposed to er schwämme), etc. However, in German the entry of new verbs is not as traumatic as in Basque: they are hosted in the weak class, and build the (finite, synthetic, one-piece) present form like any other verb (er inspirier-t). In Basque, the intr./tr. opposition and the peculiar disposition of S/O/A markers makes the job too difficult for synthetic inflection. Hence, analytic conjugation is required not only to form e.g. irrealis conditional apodoses (the Basque correspondence of Konjunktiv ii, as in inspiratuko luke ‘(s)he would inspire’), but also the standard present form, as in inspira-tzen du ‘(s)he inspires’.
5 The Basque (Analytic) Present Perfect and Narrative Past
I will next show what the analytic perfect is like in historical Basque, how Basque may have replicated the analytic perfect construction of Late Latin, and how this has evolved within Basque from Late Antiquity to the present.
By means of the perfective participle and the indicative auxiliaries (present or past), one can form, in Basque, the present perfect and the narrative past, as shown in Table 3.
The perfective participles can be divided basically into three groups, according to their morphological formation: 1) those which are formed by adding -i to the verb-stem (etorr-i ‘come’, ekarr-i ‘bring’…); 2) those which are formed by adding -tu to the verb-stem (kanta-tu ‘sing’, dantza-tu ‘dance’, senti-tu ‘feel, notice’, har-tu ‘take’…); and 3) those in which the verb-stem and the perfective participle are identical (hil ‘die, kill’, erre ‘burn’, jarraiki ‘continue’…). Many verbs of the second group come from Latin or Romance through the past passive participle (Lat. laudatu(m) > lauda-tu, etc.) but the suffix -tu has apparently spread to several autochthonous verb-stems to form perfective participles (har-tu ‘take’, sar-tu ‘go in, take into’, sal-du ‘sell’…). By contrast, the suffix -i of the first group is an autochthonous one, present in autochthonous verbs which have a prefix (*)e- and sometimes synthetic inflection.6 Now, the fact that the participles of the first group (-tu participles of Latin/Romance origin) and those of the second group (autochthonous -i participles) have ended up in complementary distribution suggests that, when Basque replicated the analytic perfect of Latin, a set of autochthonous verbal elements was coupled with the past passive participle. This is doubly interesting, first because -tu is the closest thing Basque has to an inflectional morpheme borrowed from Latin/Romance (see Gardani, Arkadiev and Amiridze, 2015, for a cross-linguistic overview of borrowing of inflectional morphemes), secondly because both allomorphs imply a combination of two kinds of borrowing: of concrete phonological matter in the case of -tu, of a functional and semantic morphological pattern in the case of -i (mat-borrowing and pat-borrowing respectively, in the terminology coined by Sakel, 2007).
As for the analytic construction as a whole, it is reasonable to think that Basque acquired the construction [participle + ‘be / have’ present auxiliary] to express a perfect – resultative in origin – because of the influence of Late Latin and/or Romance. We might conceive that it came about as a calque, from a phase previous to any contact in which Basque had nothing similar. However, recent theory on contact-induced grammaticalization shows that, when a replica language (= R) replicates a pattern of a model language (= M), “the new structure Rx is in most cases not entirely new; rather, it is built on some structure (Ry) that already existed in the replica language, and what replication then achieves is that it transforms Ry into Rx” (Heine and Kuteva, 2005: 40–41). Since Old Common Basque, and even the phase previous to it, is supposed to have had verbs for ‘be’ and ‘have’, we might instead conceive that what Old Common Basque did was to turn a structure Ry (“a minor use pattern”, in Heine and Kuteva’s words) into Rx (“a major use pattern”). Ry would have been a resultative marked – i.e., not commonly used – construction of the type etorr-i n-aiz ‘I am come’ / ekarr-i du-t ‘I have it brought’, in pre-ocb (although the auxiliaries are forms of today’s Standard Basque).
As regards the chronology of the replication process, we should first consider how and in what chronology the perfect construction developed and became grammaticalized within Late Latin and Romance. Handbooks (see e.g., Lausberg, 1962/1988: 319–325) describe the rise of the periphrastic perfect as a long process which starts in Classical Latin itself – if not earlier –, as a periphrasis limited, in transitives, to a handful of verbs, the accomplishment of which can imply the possession expressed by habere (as in equitatum ex omni provincia coactum habeo; Caes. BGall. 1, 15, 1). This capacity was metaphorically transferred to any verb, and by Late Latin became the main construction to express anteriority and was grammaticalized as a resultative perfect. Its intransitive counterpart came from passive-intransitive expressions like res cognita est, which also indicated the resulting state and ended up in complementary distribution with the transitive perfect constructions. The resultative meaning and the consequent relevance for the present, though, lingered as long as the participle agreed in gender and number with the subject in intransitive constructions and with the object in transitive ones, from which moment on these could move on to show a past of various kinds. In Spanish, this agreement did not cease to exist until the 13th-15th centuries (see Penny, 1991/1998: 159).
Within this temporal range, the terminus post quem for the replication of the construction from M to R – i.e., from Latin/Romance to Basque – must be its grammaticalization as a perfect in M, with intransitive and transitive verbs in complementary distribution. At least regarding the transitive compound perfect, this grammaticalization is usually set in about the 4th-6th centuries (see Väänänen, 1968/1995: 229, with examples of Saint Augustine, Gregory of Tours, and Oribasius), although the scope of the process may have varied depending on many factors. Let us recall that this is roughly the period attributed to Old Common Basque.
Now, before I go on to describe the semantic and functional development of the resultative perfect in Basque once it had been replicated, I would like to stress that the important takeaway of this section is the following: once the periphrastic perfect had arisen in Late Latin or Common Romance, this became the only sector of the verbal morphology which was sensitive to the intransitive/transitive nature of the verb (as still occurs in French, and in Medieval Castilian), just as occurred in Old Common Basque throughout its whole verbal morphology, due to it ergative alignment. Moreover, its analytical structure, with the lexical component hosted in a non-finite verb and kept away from the morphological intricacies of the auxiliary, was exactly what Basque needed in order to overcome the morphotactic clash between its finite verb and that of Latin-Romance (see Table 2). In the replicating process from M to R, to use Sakel’s (2007) terms, the participle-to-participle would be: a mat-borrowing when the Late Latin past passive participle -tu was borrowed as such, as from (3c) to (3d); and a pat-borrowing when it was replicated by means of an autochthonous perfective participle, as from (3a) to (3b); the replication of the auxiliaries ‘be’ / ‘have’ would be also a pat-borrowing.7
The perfect construction, then, could be used as an entry point for Latin-Romance lexicon to get into Basque, which enabled the borrowing of loan verbs such as nekatu ‘get tired’ (< Lat. necatu), bekatu ‘sin’ (< Lat. peccatu), barkatu ‘forgive’ (< past passive participle of Lat. parco), laudatu ‘praise’ (< Lat. laudatu), begiratu ‘look at, observe; beware’ (< Lat. vigilatu), aditu ‘listen to; understand’ (< Lat. auditu), etc. This was a strong functional motivation for its replication.
The semantic development of the periphrastic perfect in Basque has run parallel to that of Romance languages, with the resultative meaning going along the pathway > perfect > near past > distant past, and reaching different phases of that progression depending on geography and period. By 16th century Basque (see Mounole, 2014: 34–41, 45–56, 295–312), the construction [perfective participle + izan ‘be’ / *edun ‘have’ present auxiliary] already indicates perfect in all dialects, and hence the way is paved to indicate past – whether near or distant –, a process which is under way in that century and will develop in the following two centuries. In Lazarraga (a 16th century western author), a hodiernal past can already be perceived when using the construction, as shown in (4a). In continental dialects, in turn, there are some cases in which the construction still preserves its original resultative meaning, objective or subjective with intransitive auxiliaries, possessive with transitive ones,8 as shown in (4b) by means of a sentence by Leizarraga (a 16th century eastern, continental author).
Despite this and some other conservative attestations in 16th and 17th century continental authors in which the perfect construction holds on to its original resultative meaning, in the long run the semantics of this construction has developed further in continental or French territory than in peninsular or Spanish territory. Nowadays, in peninsular dialects under the pressure of Castilian, the analytic construction [perfective participle + izan ‘be’ / *edun ‘have’ present auxiliary] entails hodiernal past, just as in (4a), or action/event/state within the time frame indicated by an adjunct included in the same clause, besides perfect itself. In continental dialects, the tendency is for the same construction to encroach on more distant pasts, thus adding the narrative past – in the sense of perfective action/event in the past – to the aforementioned functions (Lafitte, 1944/1979: 373).9
Mounole (2014: 296–302) has summarized the evolution of the perfect construction as outlined in (5).
Peninsular dialects would not have reached the last stage. This development of the construction [perfective participle + izan ‘be’ / *edun ‘have’ present auxiliary] is very similar to the one of neighboring languages, with the continental dialects matching the model of German and French – whether due to influence of the latter or not –, and with the peninsular dialects matching the Castilian model of hodiernal past (or past included within the time frame of an adjunct of the same clause).
A final issue to consider is the original function and the later development of the construction [perfective participle + izan ‘be’ / *edun ‘have’ past auxiliary], with past auxiliaries (see Table 3), also replicated by Old Common Basque. In Romance languages, and more generally in Standard Average European, this construction expresses the pluperfect, whether past in the past, or perfect in the past holding on to the original resultative meaning. As Aldai (1998) already envisaged, this must also have been the original function of the construction in Basque. In fact, the construction [perfective participle + izan ‘be’ / *edun ‘have’ past auxiliary] expresses pluperfect in all dialects of 16th century Basque. However, by the same period, and likewise in all dialects, the construction is also indicating narrative past, yet in this function it competes with: a) past synthetic forms, some of which can contextually express narrative past in 16th century Basque (e.g., 16th c. nekarren ‘I brought (aor.) ~ I brought (ipfv.)’ > 21st c. nekarren ‘I brought (ipfv.)’), and b) the analytic construction [verb-stem + *edin / *ezan past auxiliary] (see Section 8). In the 16th century, both competitors are on the wane in this function, and this is taken over by the construction [perfective participle + izan ‘be’ / *edun ‘have’ past auxiliary], a process which is practically accomplished in nearly all dialects by the 17th century.
Mounole (2014: 303–304) summarizes the evolution of the construction as outlined in (6).
It appears, then, that in Basque the construction [perfective participle + izan ‘be’ / *edun ‘have’ past auxiliary] has gone a step further than in Standard Average European.
6 The Basque (Analytic) Imperfective Present and Past
Let us now examine the construction [imperfective participle + izan ‘be’ / *edun ‘have’ present / past auxiliary], which is of later formation than the one considered in Section 5, and has no parallel in neighboring Romance languages. This is the procedure to express the imperfective present and past with most verbs, i.e., those with no synthetic conjugation (like dantzatu ‘dance’ and kantatu ‘sing’ in Table 4). In the few verbs which do have synthetic conjugation and hence allow a synthetic / analytic opposition (like etorri ‘come’ and ekarri ‘bring’ in Table 4), the analytic construction expresses some kind of habitual meaning (‘usually’, ‘every day’…).
In modern Standard Basque and in central dialects, the verbal noun -t(z)e and the imperfective participle -t(z)en are formed by adding -te(n) to sibilant-ending and -n-ending verb-stems (ikus-ten ‘seeing’, egi-ten ‘doing’, etc.), and -tze(n) to the rest of the verb-stems (etor-tzen ‘coming’, ibil-tzen ‘walking’, kanta-tzen ‘singing’, etc.). The suffix -te, which appears in all dialects in a varying number of contexts, was originally a derivational suffix implying ‘time of’ (as in euri-te ‘rainy season’). In the function under discussion, however, some competing forms ended up as allomorphs (see Urgell, 2006). In the west, in a number of contexts we have -eta and -keta, two suffixes which appear in toponyms, in the plural of the local cases of noun-declension, and as derivational suffixes implying action (lapurr-eta ‘theft’). Also in a varying number of contexts across dialects, we have -tze as well, which, whatever its exact origin, appears to contain the derivational suffix -(t)za, with several meanings (as in bizitza ‘life’).
Similarly to what occurs in neighboring languages, in modern Standard Basque the imperfective present and past cover habitual, gnomic, planned future notions, and pure present or past notions with stative verbs. In the cases of verbs which allow synthetic conjugation (recall that these are also imperfective), the analytic construction takes over the habitual meaning. This situation does not greatly differ from the one of 16th century Basque (see Mounole, 2014: 27–34).
As for the origin of this imperfective construction, Mounole (2008), refining Trask (1995), has hit upon an elegant explanation to account for it. Since the imperfective participle is an originally derivational suffix implying ‘time of’, or action, to which -n is attached, and since -n is the inessive marker (as in Bermeo-n ‘in Bermeo’), Trask proposed that egi-te-n ‘doing’ was, in origin, ‘on doing’. The origin of today’s imperfective construction, then, must have been very similar to that of the present continuous in English, the development of which is well attested since Old English. In modern forms, it would go: I am on doing of that > I am a-doing (of) that > I am doing that. Mounole has further accounted for the mismatch of auxiliaries in historical Basque. Originally, both joa-te-n da ‘(s)he is going (lit. (s)he is on going)’ and *egi-te-n da ‘(s)he is doing (lit. (s)he is on doing)’ would have been intransitive progressive periphrases. Later, the verbal noun in the inessive would have been reanalyzed as part of the vp, and, by analogy with the perfect periphrases (joan da ‘(s)he has gone’ vs. egin du ‘(s)he has done’), the intransitivity/transitivity of the auxiliary would have been determined by the lexical verb of the construction. This development is represented in (7).
All this means that, at some point in the Middle Ages (after Old Common Basque, hence in a dialectal period), Basque developed a progressive construction. The functional motivation for this may have been that the synthetic forms antecedent of the present forms (e.g., nator ‘I come’, dakart ‘I bring’, etc.), originally progressive themselves, were becoming present (a process which perhaps has to be understood within a drift of the language as a whole from aspect-prominence to tense-prominence). This brought about the need for a new (present) progressive construction, which by the 16th century had also become present itself (except with synthetic verbs, see Table 4).
The most interesting aspect of the process, however, is how this construction arose formally. Unlike the periphrastic perfect, the structure [imperfective participle + izan ‘be’ / *edun ‘have’ present / past auxiliary] did not come about as the result of any replication, but as an original construction with no parallels in neighboring Romance. It started off as a marked intransitive periphrasis expressing progressivity, but, as it gained in productivity and the need was felt to integrate any lexical possibility into the whole verbal system (sensitive to the intransitive / transitive opposition in Basque), the two-fold system of auxiliaries was extended from the construction [perfective participle + izan ‘be’ / *edun ‘have’ present / past auxiliary]. We can conclude, then, that it was not a replication, but the extention of a pattern acquired or at least reinforced in a previous replicating process.
7 The Basque (Analytic) Future
Basque has also had an analytic construction of the structure [prospective participle + izan ‘be’ / *edun ‘have’ present / past auxiliary], expressing future ever since the 16th century (if the auxiliary is past, future in the past, or even conditional perfect). Some examples are represented in Table 5.
The prospective participle is formed by adding -ko or -(r)en to the perfective participle. Roughly, in the west -ko is used, in the east -(r)en is used (-ren after vowels), and in a central area -ko is added to vowel-ending, -en to -n-ending perfective participles (ikusi-ko ‘see’, but erran-en ‘say’). This suggests that, like the construction [imperfective participle + izan ‘be’ / *edun ‘have’ auxiliary], it must have arisen at some point in the dialectal period, after Old Common Basque.
Nowadays, the analytic future of the structure [prospective participle + izan ‘be’ / *edun ‘have’ present auxiliary] is the only construction specialized in expressing this tense. However, in 16th century Basque, it competed with several other possibilities. Restricting our analysis to the future with present auxiliaries, some of these possibilities are (see Mounole, 2014: 59–70, for a full account): (a) present synthetic forms, some of which could contextually express future, (b) present synthetic forms to which the suffix -te/-ke is added, (c) present synthetic forms turned into a kind of contextual subjunctive by the suffixation of -(a)n and reanalyzed as future, (d) constructions of the type [verb-stem + *edin / *ezan present auxiliary + -te/-ke], (e) constructions of the type [present participle + izan ‘be’ / *edun ‘have’ present auxiliary + -te/-ke].
The suffix -te/-ke, which in later Basque has grammaticalized as a morpheme of potential mood (or, in eastern dialects, epistemic modality), could cover a wider scope of meanings in 16th century Basque, depending on the context and the type of morphological combination, such as future tense, potential mood or epistemic modality. To put it briefly, the synthetic du-ke and the analytic ukan-en du were in competition to express ‘(s)he will have’. Eventually, ukan-en du has prevailed in this function, whereas du-ke has been fenced into potential or epistemic functions.
Both forms of the prospective participle are identical with one of the two genitives Basque has: -ko is the local genitive (e.g., Bermeo-ko ‘of Bermeo’), -(r)en the possessive one (e.g., Peru-ren ‘Peru’s’). This opens up the possibility that the future constructions are originally periphrases, perhaps deontic, built upon the genitive, with western dialects having opted for the local genitive, eastern ones for the possessive genitive. This hypothesis has always been latent among analysts, yet nobody has substantiated it in the form of a theory, perhaps because the details would be much more difficult to explain than for the rise of the imperfective construction. To begin with, the genitive idea is not among the most frequent sources of futures (see Bybee et al., 1994: 251–253, where Basque is the only case), however logical we may find an evolution like ‘I am of going/doing’ > ‘I will go/do’.10
In fact, such an idea was already hinted at by Mitxelena (1981/1987: 48), who suggested that the Basque future construction might be a calque of the Romance deontic construction of the type he de ir / hacer ‘I have to go/do ~ I will go/do’. If so, we would have to postulate a process similar to the one of (7) in Section 6, but with the analogy going in the opposite direction. The original marked structure would have been the transitive periphrasis ‘have of’, with a deontic meaning, and later, because of the need to grammaticalize the structure, the intransitivity / transitivity of the auxiliary would be determined by the lexical verb, by analogy with the perfect and perhaps other sectors of the verbal morphology. This hypothetical process, again face to face with the perfect, is represented in (8).
If Mitxelena’s suggestion is right and we do have a calque of a Romance deontic structure, then Basque would have replicated a structure of a model language, but would have transformed it into something else (future tense). Using the formalism coined by Heine and Kuteva (2005), the process could be figured in the following manner: Mx > Rx / Ry > Ry. Apparently, the function of that structure Mx was not exactly the one required by R, but, unlike the future-expressing devices which medieval Basque must have had at the time of its grammaticalization (see the five ones catalogued as (a-e) above, regarding 16th century Basque), (i) Mx was an analytic construction, which allowed the lexical component to be kept aside from the categories expressed in the finite verb (a condition required for productive borrowing of verbs, as we saw in Section 5), and (ii) after an analogical process like the one figured in (8), the replication of Mx could take ‘be’ / ‘have’ as non-suffixed auxiliaries, thus coupling the periphrastic perfect (the innovative and productive pattern), from a formal point of view.
As Matras (2011: 288) puts it, when talking about speakers of an R language which needs a particular structure and is in contact with an M language, “[s]canning through the entire repertoire [of M], the speaker identifies a construction that would serve this particular task [of R] most effectively”. The Romance deontic periphrastic construction had the afore-mentioned formal advantages and was semantically close to the future (a primary need in a tense-prominent language), so it served the task of making the whole verbal system coherent, finally ousting all the competing structures to express the future (see (a-e) above), already in the historical period.
8 Non-indicative Moods
What we have said so far might lead one to believe that Basque acquired verbal analytic constructions from Latin/Romance. Yet, while I have defended that the perfect analytic construction arose due to contact with Latin or most probably Romance – as do most authors –, there is also strong evidence that, before borrowing the perfect construction, Basque already had another analytic construction, arisen and developed on internal grounds. Hence, note that, although the construction of this section is the last one to be described “in a synchronically logical order” (because in historical Basque it corresponds to marked moods), it is older than the ones of Sections 5–7, and must have existed in pre-ocb.
In addition to the tense/aspect values mentioned until now, which are all indicative, today’s Basque has a set of marked moods (potential, subjunctive, and imperative), which in analytic constructions are expressed morphologically (in synthetic verbs, in turn, contextually). These moods are formed through a distinct set of auxiliaries, built upon roots different from the ones of izan ‘be’ (intr.) and *edun ‘have’ (tr.), with no apparent correspondence in neighboring Romance languages, namely: *edin and *ezan, the lexical sources of which are perhaps ‘come’ or ‘become’, and ‘have-2’, respectively.11 Now, since – at least in modern Standard Basque – the potential is morphologically derived from the subjunctive, and the imperative is defective and formed by other procedures (although it shares the same verbal roots as the subjunctive and potential), for the sake of both simplicity and effectiveness in the diachronic account of facts, I shall restrict my analysis to subjunctive forms. The template of these in their analytic version is represented in Table 6.
The verb-stem is the bare form of the non-finite verb, to which different suffixes can be attached to build the different non-finite forms that we have seen (see Trask, 1995; 1997: 211–217). In many autochthonous verbs, the structure of the verb-stem comprises an (*)e- attached to a verbal root, typically of the structure cvc (e-tor ‘come’, e-kar ‘bring’), or of a combination of a primary structure like this plus a/some prefix(es) (e-ra-kar ‘attract’) and/or a/some suffix(es) (e-du-ki ‘have’). Most verbs for which at least a synthetic form is attested belong to this type.
In loanwords borrowed from the past passive participle of Latin, the verb-stem is formed by removing -tu: kanta-tu → kanta. This is apparently a back formation, generated by the analogy with the process ekarr-i (perfective participle) → ekar (verb-stem) in autochthonous verbs.
In today’s Basque, with the exception of some archaism maintained in far eastern dialects, by means of the construction [verb-stem + *edin / *ezan present or past auxiliary (+ -teke, -ke)], one expresses subjunctive notions in subordinated clauses, mainly final, completive and jussive. The auxiliary always bears the suffix -(a/e)n or -(a/e)la.
In 16th (and in fewer cases 17th) century Basque, however, the same construction is used in some few functions which are not subjunctive but indicative, the most prominent of which are, in the present, realis conditional protases (etor ba-da-di ‘if (s)he comes’, ekar ba-de-za ‘if (s)he brings’), and, in the past, the narrative past – in main clauses –, as we have seen in Section 5 (etor nendin ‘I came’, ekar nezan ‘I brought’). Both these functions are in retreat at that time, ousted by the synthetic or analytic forms of the present with the finite verb suffixed by ba- ‘if’, and by the thriving construction [perfective participle + izan / *edun past auxiliary], respectively. (9a) shows an example of the first case, and (9b) two of the second.12
The relegation of such constructions to a morphological subjunctive also entailed their syntactic relegation. Regarding the past, the process has already been described in Section 5. According to both Aldai (2009b) and Mounole (2014), the construction [perfective participle + izan ‘be’ / *edun ‘have’ past auxiliary] would have been first mainly associated to subordinate clauses. Later, it would have gone through the process outlined in (2) of Section 5, the last stage of which implied their prevalence in main clauses, thus limiting [verb-stem + *edin ‘come’ / *ezan ‘have-2’ past auxiliary] – without -te/-ke – to subordinate clauses with a finite verb suffixed by -(a/e)n or -(a/e)la.
To sum up, pre-Old Common Basque must have already had an analytic construction of the structure [verb-stem + *edin ‘come’ / *ezan ‘have-2’ present and past auxiliary], which has no parallel in Latin and must be an internal development. Along with synthetic verbs, it must have covered all tam functions allowed by the verbal system. When Old Common Basque replicated the Late Latin structure [perfective participle + ‘be’ / ‘have’ present or past auxiliary], and Medieval Basque created more analytic constructions by extending the auxiliaries ‘be’ / ‘have’ to them, the construction [verb-stem + *edin ‘come’ / *ezan ‘have-2’ present and past auxiliary] was progressively relegated to subordinating contexts and to the expression of the aorist, until in historical Basque it was grammaticalized as a subjunctive (and as further marked moods, by means of different derivations).
9 Chronological Summary
Let us briefly recapitulate, now in chronological order, the processes which have been assumed so far. Although some dates have been mentioned, particularly in Sections 2 and 5, in this summary we are interested in the structural development of events rather than in specific dates, which would, furthermore, pose problems of a complexity which cannot be tackled in this paper.
1) Before beginning to borrow any grammatical structure from Latin and/or Romance – not necessarily before borrowing any linguistic element nor of having any linguistic contact –, Basque verbal morphology had, most probably, at least two kinds of vps: synthetic forms and analytic forms of the structure [verb-stem + *edin / *ezan auxiliary].
It is tempting to identify this period with pre-ocb, i.e., with the phase previous to the Old Common Basque mentioned in Section 2 (largely 5th-6th centuries), as the period proposed by Mitxelena for ocb is roughly simultaneous with the grammaticalization of the resultative perfect in Late Latin.
It would be beyond the scope of this paper to go into the details regarding which forms and functions the existing synthetic and analytic vps had. I would like to simply point out, firstly, that the relative importance of synthetic forms must have been much larger in this period, and, secondly, that this rendered the borrowing of verbs impossible, due to the incompatibility of finite-verb templates, Basque being sensitive to the intransitive / transitive opposition.
2) For that reason, when in Late Latin or Common Romance a compound perfect was grammaticalized which consisted of a past passive participle of simple morphology and an auxiliary verb, this was most welcome in Old Common Basque and was felt as the right entry point for Romance lexicon, inasmuch as that sector of the verbal morphology was sensitive to the intransitive / transitive opposition also in the donor language. Once replicated, the meaning of this construction ran parallel to that of Romance, functioning as a resultative perfect until the last centuries of the Middle Ages, and becoming a past from then on. The resultative meaning is still present in some (Basque) eastern attestations of the 16th century.
The massive borrowing of past passive participles turned the suffix -tu into a productive suffix to form verbs. This was attached to loan verbs as well as to some autochthonous elements (har-tu, sar-tu, sal-du, etc.). Autochthonous verbs which had synthetic inflection, however, resorted to the autochthonous suffix -i to form the perfective participle, or perhaps some version of this already existed in the R language with some minor function and was simply coupled with the borrowed – actually mat-borrowed – -tu. In loan verbs ending in -atu or -itu, in turn, -tu was removed in order to form the verb-stem.
3) As loan verbs grew in number, but had no possibility of being conjugated synthetically, a pressure arose to extend the analytic model of the compound perfect and of the construction [verb-stem + *edin / ezan auxiliary] to express more tam notions. Thus, an intransitive progressive periphrasis of the structure [[verbal noun in the inessive] [izan auxiliary]] was turned into [imperfective participle + izan / *edun auxiliary] by analogy with the perfect and, in the course of time, grammaticalized into a present.
Likewise, another periphrasis formed by the perfective participle to which the local or possessive genitive was attached, plus a set of auxiliaries which are more difficult to guess – but certainly izan and/or *edun, i.e., the historically indicative ones –, gave rise to a future construction, perhaps deontic in origin and replicated from a Romance model, which in historical Basque would prevail as the expression of the future tense.
As for the relative chronology of the three analytic constructions (perfective, imperfective, future), the imperfective is later than the perfective a) because, according to the series of events accepted, the auxiliaries of the former became the intr./tr. ‘be’/ ‘have’ by analogy with those of the latter, and b) because the imperfective participle displays dialectal diversity, and hence must be later than ocb. The future construction is also later than the perfective construction a) because the prospective participle is formed upon the perfective participle, and hence the formation of the latter must chronologically precede it, and b) because the prospective participle displays dialectal diversity.
Less straightforward is it to establish the relative chronology between the imperfective construction and the future construction. However, although scholars have not overtly expressed a relative chronology on this point, after Urgell (2006) most agree that the large dialectal diversity of forms of the verbal noun argues in favor of a particularly late grammaticalization of the progressive construction [verbal noun + -n ‘on’ + izan ‘be’ present / past auxiliary], which turned into [imperfective participle + izan ‘be’ / *edun ‘have’ present / past auxiliary].
4) By a process which is under way at the beginning of the written tradition (16th century), the construction [verb-stem + *edin / *ezan auxiliary] – without -te/-ke – is fenced into subjunctive functions, suffixed by -(a/e)n or -(a/e)la and restricted to subordinate clauses, thus giving rise to a morphologized subjunctive (only in analytic constructions, as in synthetic verbs this continues to be contextual until the present day). In the present, the function of expressing realis conditional protases is taken over by present forms (synthetic or analytic, and also prefixed by ba- ‘if’), whereas, in the past, the expression of the narrative past is taken over by the construction [perfective participle + izan / *edun past auxiliary], whose original function – later still kept, in many varieties in competition with other forms – was that of the pluperfect.
The creation of a subjunctive, which basically occurred between the 15th and the 17th centuries, may have been motivated, or encouraged, by the influence of Romance. It has nothing to do, however, with the rise of a Basque written tradition, the translation of Spanish and French texts into Basque, etc., as one might think regarding the synchronicity of events (see Mounole, 2014 and Aldai, 2009b, for more details on this issue).
5) Finally, the prevailing of the construction [prospective participle + izan / *edun auxiliary] as the expression of future over synthetic and analytic verbal forms suffixed by -te/-ke brought about the specialization of these as potential or epistemic, and hence the creation of a morphologized potential mood. This phenomenon has no parallel in neighboring Romance languages nor in (at least western) sae, where potentiality and/or epistemic notions are expressed by modal (but independent) verbs.
10 Conclusion
In the preceding sections, it has been shown how Latin and Romance influence has been crucial for the configuration of the whole Basque verbal morphology, even though the only construction which Basque has replicated as such has been the perfect one. In the other constructions (progressive > present, and future), rather than a replication, what occurred was the extension of the pattern intr. aux. ‘be’ / tr. aux. ‘have’ to further periphrastic structures that arose in medieval Basque, in a process as a result of which they became analytic vps.
I believe that the key point to understand this process whereby analytic vps became the productive pattern is the need to borrow vocabulary, and particularly verbal vocabulary. While Basque was the only language in southwestern Europe which survived the Romanization process, from a cultural point of view the Basque-speaking population did not escape a thorough Romanization in many realms. The result was a massive borrowing of words at all levels.
Now, there was nothing to prevent nouns, adjectives or adverbs from entering Basque and taking part in this process, as these elements have no prefixes in today’s Basque nor could they have had them at the time of the first borrowings. Verbs, by contrast, were left out of the process for the reason put forward in Section 4: some of the argument markers were prefixed, and the morphological template was different in intransitive and transitive verbs, unlike in the donor language. In these circumstances, the perfect construction came up as a solution which could serve as a cornerstone to introduce verbal lexemes into the language, as it hosted these in a morphologically simple participle, while the messy morphology was seen to by an autochthonous auxiliary.
This pattern with intr. ‘be’ / tr. ‘have’ auxiliaries was later analogically transferred to other constructions in order to create more tam possibilities formed analytically, i.e., to an emerging intransitive progressive periphrasis, and to a periphrastic future.
I also believe that the preponderant role that the perfective participle has among non-finite forms in modern Basque may be a remnant of that stage at which the analytic perfect was the entry point for Latin/Romance verbal vocabulary to get into the language. It cannot be by chance that the neutral form of Basque verbs – as found in dictionaries or word-listing contexts – is the perfective participle, again unlike in neighboring languages.
References
Aldai, Gontzal 1998. A propósito del aoristo vasco. Fontes Linguae Vasconum 79: 377–386.
Aldai, Gontzal 2009a. Is Basque morphologically ergative? Western Basque vs. Eastern Basque. Studies in Language 33: 783–831.
Aldai, Gontzal 2009b. Aoristo perifrástico, perfectivo y pluscuamperfecto: Leizarraga vs. Lazarraga. In R., Etxepare and R., Gómez and J. A., Lakarra (eds.), Festschrift for Bernard Oyharçabal, 19–36 (= Anuario del Seminario Julio Urquijo43).
Bybee, Joan, Perkins, R. and Pagliuca, W. 1994. The evolution of grammar. Tense, aspect and modality in the languages of the world. Chicago: University of Chicago Press.
Drosdowski, Günther et al. 1984. Duden. Grammatik der deutschen Gegenwartssprache. Mannheim / Leipzig / Wien / Zürich: Dudenverlag.
Euskaltzaindia. 1987. Euskal gramatika. Lehen urratsak II. Bilbao: Euskaltzaindia.
Gardeni, Francesco, Arkadiev, Peter and Amiridze, Nino 2015. Borrowed morphology: an overview. In Francesco, Gardeni and Peter, Arkadiev and Nino, Amiridze (eds.), Borrowed Morphology, 1–23. Berlin, Boston, Munich: Mouton De Gruyter.
Heine, Bernd and Kuteva, Tania 2005. Language Contact and Grammatical Change. Cambridge University Press.
Lafitte, Pierre 1944/1979. Grammaire basque (Navarro-labourdin littéraire). San Sebastián: Elkarlanean.
Lausberg, Heinrich 1962/1988. Lingüística románica. II. Morfología. Madrid: Gredos.
Matras, Yaron 2011. Grammaticalization and language contact. In B., Heine and H., Narrog (eds.), The Oxford Handbook of Grammaticalization, 279–290. Oxford University Press.
Mitxelena, Koldo 1981/1987. Lengua común y dialectos vascos. In Palabras y textos, 35–55. Bilbao: University of the Basque Country.
Mounole, Céline 2008. Sintaxi diakronikoa eta aditz multzoaren garapena: inperfektibozko perifrasiaren sorreraz. Anuario del Seminario Julio Urquijo 51: 585–603.
Mounole, Céline 2014. Le verbe basque ancien: étude philologique et diachronique (= Anuario del Seminario Julio Urquijo 48). University of the Basque Country.
Nedjalkov, Vladimir P. and Jaxontov, Sergej J. 1988. The typology of resultative constructions. In Sergej J., Nedjalkov (ed.), Typology of Resultative Constructions, 3–62. Amsterdam / Philadelphia: John Benjamins.
Penny, Ralph 1991/1998. Gramática histórica del español. Barcelona: Ariel.
Pineda, Anna and Berro, Ane 2020. Hybrid intransitives in Basqe. Glossa: a journal of general linguistics 5(1): 22.
Sakel, Jeanette 2007. Types of loan: Matter and Pattern.In. Yaron, Matras and Jeanette, Sakel (eds.), Grammatical borrowing in cross-linguistic perspective, 15–29. Berlin / New York: Mouton De Gruyter.
Trask, R. L. 1995. On the History of Non-Finite Verb Forms in Basque. In José I., Hualde and Joseba A., Lakarra and R. L., Trask (eds.), Towards a History of the Basque Language, 207–234. Amsterdam / Philadelphia: John Benjamins.
Trask, R. L. 1997. The History of Basque. London / New York: Routledge.
Urgell, B. 2006. Para la historia del sustantivo verbal en vasco. In Joseba L., Lakarra and José I., Hualde (eds.), Studies in Basque and historical linguistics in memory of R.L. Trask, 921–948 (= Anuario del Seminario Julio Urquijo40).
Väänänen, Veikko 1968/1995. Introducción al latín vulgar. Madrid: Gredos.
This paper has been written within the framework of two research projects: (1) Monumenta Linguae Vasconum 6: avances en cronología de la historia y la prehistoria de la lengua vasca(mlv6) [= pid2020-118445gb-I00], led by Blanca Urgell and funded by the Ministry of Education and Science of the Spanish Government, and (2) Historia de la lengua vasca y lingüística histórico-comparada(hmlv-lhc) [= it1344-19], led by Joaquín Gorrochategui and funded by the Government of the Basque Autonomous Community. I would also like to thank Blanca Urgell and two anonymous reviewers for their helpful comments and suggestions. All errors remain mine.
Abbreviations employed in the paper: A = Agent; aor = Aorist; aux = Auxiliary verb; C = Century; G = Genitive; hab = Habitual; hyp = Hypothetical; intr = Intransitive; ipfv = Imperfective; ipv = Imperative; M = Model language; N = Noun; O = Object; ocb = Old Common Basque; pfv = Perfective; po = Postposition; pot = Potential; ppp = Past passive participle; prosp = Prospective; prs = Present; pst = Past; R = Replica language; Rec = Recipient; rel = Relative clause; S = Single argument of canonical intransitive verb; sae = Standard Average European; subj = Subjunctive; subord = Subordinator; tam =Tense-Aspect-Mood; temp = Temporal; tr = Transitive; vb = Verb(al); vp = Verbal Phrase (in the functional sense).
See also Pineda and Berro (2020), for a recent analysis which introduces some interesting elements which cannot be taken up here.
They are 68 in 15th-16th century Basque according to Mounole’s (2014: 15) count, 26 in today’s Basque according to Euskaltzaindia (1987), although the use of some of these is very rare.
Note also that I represent the present finite verb in Basque by means of a synthetic form. Were I to instead represent it by an auxiliary verb, the template would remain the same, except for the fact that, instead of a verbal lexeme, we would have a root devoid of any lexical meaning.
Most perfective participles of the third group are also autochthonous, but, since they have a heterogenous origin, for the sake of simplicity I will not consider them.
The auxiliaries of (3b) and (3d) are forms of modern Standard Basque, but the perfective participles can be traced back to Old Common Basque.
I am adopting Nedjalkov and Jaxontov’s (1988) classical terminology, also employed by Mounole (2014).
This extension has been traditionally explained through French influence, where e.g., il est allé has evolved to express distant past ‘he went’ (last week, last year, ten years ago). However, Mounole (2014: 296–302) has pointed to some complications for the postulation of such an influence, at least as the first and main factor. To begin with, the influence which Basque has received has been mainly that of Gascon, where the construction is not used to indicate narrative past.
Interestingly, Mounole (2014: 67–69) puts forward two cases, in the Alavese dialect of the 16th century, in which the future construction clearly expresses not a temporal but a deontic notion. One of them appears in a passage of Lazarraga’s bucolic novel, where we read: Çuec jaquin-go do-çu, arçay noble-ac…, which according to the context should translate as ‘You (pl.) are to know, noble shepherds…’ (in Spanish, ‘Habéis de saber…’). Perhaps in these attestations there underlies the original deontic meaning of the future construction, in a fossilized expression. Note that the Castilian future itself comes from a deontic, although in this case we have to start not from the structure he de ir, but from ir he > iré ‘I will go’ (Penny, 1991/1998: 193–198).
I call it ‘have-2’ because in origin it might be related to izan, which can mean both ‘be’ and ‘have’, but is a verb-stem different from *edun. It turns out, then, that the izan of intransitive indicative auxiliaries and the *ezan of transitive auxiliaries of non-indicative moods might be cognates.
The auxiliary roots are glossed as subjunctive, as in today’s Basque, although in these authors they do not yet function as such.