1

Is Basque morphologically ergative?:

Western Basque vs. Eastern Basque

Gontzal Aldai

University of the Basque Country 1

Gontzal Aldai

Department of Linguistics and Basque Studies

University of the Basque Country

Cl. Nieves Cano ,33

01006, Vitoria-Gasteiz, Alava, Spain

gontzal.alday@ehu.es

1 I would like to thank the editor of Studies in Language, Bernard Comrie, and two

anonymous referees for their helpful comments on earlier versions of this paper. The

work reported in this article was made possible in part by a post-doctoral grant from the

Department of Education, Universities and Research of the Basque Government. I am

indebted to Nicolás A. G. for his invaluable help with the maps.

2

ABSTRACT

Although Basque’s case-marking pattern has conventionally been regarded as ergative, I

argue in this article that it is not straightforward to determine whether this pattern is

actually ergative or rather semantically-based. Several factors make this one a complex

issue. Perhaps the most important factor among these, and one which has gone largely

unnoticed in the literature, is the fact that there are clear dialectal differences regarding

the behavior of lexically-simple unergative verbs. Another complication comes from

establishing the specific contrast underlying the putative semantic split among Basque

intransitives. With these questions in mind, I propose that the case-marking system of

Western Basque should be considered semantically aligned instead of ergative. This

system is based on a contrast patientive / non-patientive. Conversely, Eastern and

Central Basque may be considered to have an ergative case-marking system, but one

which differs in some respects from that of the most typical ergative languages and

which is not too different from a semantic system.

3

1. Introduction

Basque has traditionally been considered to be a morphologically ergative language.

Ideed, Basque is often cited as an example of a language with thoroughgoing ergative

morphology (cf. Trask 2002:282; Hualde & Ortiz de Urbina eds. 2003:v). Yet, while it

is clear that Basque case-marking follows almost systematically a dominant pattern that

is not accusative, it is not obvious that this pattern is actually ergative. In truth, it is not

easy to determine whether the case-marking pattern of Basque is ergative (i.e. valence-

based) or rather semantically-based (i.e. split-intransitive). In order to distinguish

between these two types of case-marking, one needs to look at the behavior of

unergative predicates. However, in the Basque case there are at least three factors which

make this issue very difficult and which have posed difficulties for scholars working on

the subject.

First, we need to know what exactly unergative predicates are in Basque. In

other words, if we are to propose or reject a semantically-based system for Basque, we

need to understand the specific kind of semantic contrast that the system may reflect.

Traditionally, the contrast that has been looked at opposes agentive predicates vs.

patientive predicates. This is so because it has been considered that the semantic roles of

agent and patient exclude each other. However, in Aldai (2007), I proposed that the

agent / patient opposition in intransitives is rather a two-fold one. That is, a given

intransitive predicate can be classified as agentive vs. non-agentive and at the same time

as patientive vs. non-patientive. Thus, agentive non-patientive intransitive predicates are

those corresponding to such verbs as work, dance, run, etc.; intransitive verbs such as

die, fall, etc. represent non-agentive patientive predicates; and most importantly, the

verbs go, come, arrive, stand up, etc. typically yield agentive patientive predicates,

4

whereas verbs such as shine, beep, ring, etc. correspond to non-agentive non-patientive

predicates.

Second, many unergative verbs in Basque, probably the majority of them, are

compound verbs which consist of a nominal plus the light verb egin ‘to do’. Some of the

studies on Basque unergativity (notably Levin 1983) have focused on examining egin

compounds. Yet, what is crucial for determining the type of case-marking system of

Basque is not compound unergatives, but rather lexically-simple unergatives.

Unfortunately lexically-simple unergatives are not too frequent. Also, a fair number of

them are borrowings. Furthermore, it may be idiosyncratic of a given dialect whether a

particular unergative verb has a compound structure or is lexically simple.

Third, and perhaps most importantly, there are major dialectal differences

regarding the behavior of lexically-simple unergative verbs. To put it briefly, Western

Basque marks the subject of lexically-simple unergatives in the ERG case, and Eastern

Basque uses the ABS case in most of these instances, while Central Basque occupies a

transitional position which is closer to that of Eastern Basque.2

2 For the phenomenon at issue, I assume the following rough division of modern Basque

dialects: Western (Biscayan and Guipuzcoan), Central (Navarrese and Labourdin), and

Eastern (Lower Navarrese and Souletin). Note that the dialects correspond (albeit only

approximately) to political territories: i.e. Biscay, Guipuzcoa, Navarre, Labourd, Lower

Navarre and Soule. The Basque spoken in the territory of Alava, which was very similar

to Biscayan, has been practically lost in historical times. (Generalized attestations of

Basque start in the 16th century.) The Navarrese dialect has also suffered major

recession in this period. (Politically, Alava, Biscay, Guipuzcoa and Navarre now belong

to Spain, while Labourd, Lower Navarre and Soule belong to France.) See Figure 1.

5

FIGURE 1 (MAP OF THE BASQUE COUNTRY TO GO HERE)

For the above reasons, I argue in the present article that the issue under

discussion is better viewed from a dialectal (and historical) perspective. It seems that the

case-marking type of Basque is undergoing a reanalysis from ergative to semantically-

based. This reanalysis started in the Western area, and seems to be progressing towards

the East. Thus, I propose that the case-marking system of Western Basque should be

considered semantically aligned instead of ergative. More precisely, I argue that

Western Basque has a semantic case-marking system based on the contrast patientive

vs. non-patientive. On the other hand, Eastern and Central Basque may be considered to

have an ergative case-marking system, but, as we will see, one which differs in several

respects from that of the most typical ergative languages and which is in fact not too

dissimilar from a semantic pattern.3

2. The case-marking system of Basque

In this section, I introduce the bare facts regarding Basque case-marking. I give

prototypical examples for each of the canonical transitive clause and the two types of

canonical intransitive clauses of Basque, namely unaccusative and unergative clauses.

As will be discussed in Section 4, (Western) Basque unergatives are those intransitive

verbs that have a non-patientive subject. That is, patientivity, and not agentivity, is the

3 The research carried out in the present article should be complemented with a dialectal

and historical study of “indirect transitive constructions”, i.e. those two-place predicates

with an oblique (usually dative) object. As it turns out, indirect transitive verbs provide

historical evidence for an evolution towards a semantic alignment pattern in Western

Basque. (See Aldai 2007, and footnotes 6 and 11 below.)

6

factor that distinguishes between unaccusatives and unergatives in (Western) Basque.

Hence, agentive patientive predicates such as those pertaining to directed-motion verbs

like go, come, or arrive are unaccusatives and thus have ABS subjects. On the other

hand, non-patientive predicates, either agentive or non-agentive, are unergatives.

2.1. Transitive-clause marking

(1a) Peru-k sagarr-a-Ø jan du

Peter-ERG apple-DET-ABS eaten has

‘Peter has eaten the apple’

2.2. Marking of unaccusatives (patientive intransitives)

(1b) Non-agentive:

Peru erori da

Peter.ABS fallen is

‘Peter has fallen’

(1c) Agentive:

Peru etxe-ra joan da

Peter.ABS house-ALL gone is

‘Peter has gone home’

7

2.3. Marking of unergatives (non-patientive intransitives)

Egin-compounds:

(1d) Agentive:

Peru-k dantza-Ø egin du

Peter-ERG dance-ABS done has

‘Peter has danced’

(1e) Non-agentive:

Peru-k lo-Ø egin du

Peter-ERG sleep-ABS done has

‘Peter has slept’

Lexically-simple verbs:

(1f) Peru dantzatu da Eastern

Peter.ABS danced is

‘Peter has danced’

(1g) Peru-k dantzatu du Western

Peter-ERG danced has

‘Peter has danced’

Notice that auxiliary selection and subject case-marking are always mutually dependent

in Basque. That is, an ABS subject always triggers so-called intransitive auxiliaries,

such as izan ‘to be’ (3sg da), and conversely the intransitive auxiliaries always trigger

an ABS subject. On the other hand, ERG subjects always correspond to so-called

transitive auxiliaries like edun ‘to have’ (3sg du), and transitive auxiliaries always

correspond to ERG subjects.

8

2.4. Outline of the article

The above examples provide a rough approximation to the complex case-marking

pattern of Basque. It can be seen from those schematic data that there are mainly three

questions which need to be clarified. The present article is devoted to clarifying these

issues. In Section 4, I examine in detail the specific type of semantic contrast that occurs

between unaccusatives and unergatives in (Western)4 Basque (illustrated in examples

(1b)-(1c) vs. (1d)-(1g) above). In Section 6, I discuss the question of compound

unergatives vs. lexically-simple unergatives (examples (1d)-(1e) vs. (1f)-(1g) above).

From Section 7 onwards, I examine dialectal variation regarding lexically-simple

unergative verbs (examples (1f) vs. (1g) above). In Section 11, finally, I present a

number of conclusions arising from this research. I also address in this final section the

following question: To what extent is the ergative pattern of Eastern and Central Basque

significantly different from the semantic pattern of Western Basque? Before discussing

these issues, however, I offer in the next section a review of the literature on the topic.

3. Literature review

The proposal outlined above, which I will develop in the present article, is novel in

several respects. Maintaining that there is a split among Basque intransitive predicates is

4 Note that this semantic contrast affects both compound unergatives and lexically-

simple unergatives. Since the marking of compound unergatives is common to all

dialects, the unaccusative / unergative contrast is manifest (one way or another) in both

Western and Eastern Basque. Yet, it is only in Western Basque that the unaccusative /

unergative semantic contrast governs all instances of case-marking in intransitive

predicates. (See discussion in Section 6.2.)

9

not new, however. Various scholars, starting with Lafitte (1962:189-190) (see Section

9.1.1) and Lafon (1975), have already called attention to the fact that “a sizeable

number of verbs are syntactically intransitive (they cannot take direct objects) but none

the less require full transitive morphology (ergative subjects, transitive auxiliaries,

transitive agreement patterns)” (Trask 1997:229). After Lafitte (1962), practically all of

the ensuing studies of Basque transitivity acknowledge these data. Among them we can

cite Levin (1983:290-ff), Levin (1989), Ortiz de Urbina (1989:1-61), Salaburu (1992),

Oyharçabal (1992), Trask (1997:229-230), Trask (2002), and many others. However,

rather surprisingly, very few of these proposals have clearly stated that the data may

challenge the traditional view that Basque is morphologically ergative. Nevertheless,

there are two main positions on this issue, whose most significant representatives may

be Levin (1983, 1989) and Trask (1997, 2002), respectively (see Sections 4 and 6 for

more discussion on these two proposals).

Crucially, none of the above scholars has systematically differentiated between

Western dialects and Eastern dialects. (Also, the diachronic perspective has mainly been

ignored.) Furthermore, and most unexpectedly, even if some among these scholars

focus on Western-Basque data, none of them has unambiguously proposed that the case-

marking system of Western Basque is semantically aligned (e.g. “active”, “split

intransitive”, etc). Trask overtly rules out a semantic analysis (see Section 4). In turn,

Levin, Ortiz de Urbina and Oyharçabal get much closer to that possibility, but also

finish up excluding it (the last two, rather contradictorily).

The split-intransitivity facts of Basque are hard to match with a cross-

linguistically recognized alignment pattern of case-marking. As I have already outlined,

the Basque data are rather complicated. Now, with such an intricate picture to analyze,

practically the only models available which departed from the typical accusative and

10

ergative patterns were those coming from Georgian and the Native American languages

labeled “active”. Since Basque also diverges from these active languages in clear ways,

it appeared easier to end up assigning Basque to either one of the two long-established

case-marking patterns (Ortiz de Urbina, Oyharçabal), or not providing a characterization

of Basque’s case-marking altogether (Levin). Moreover, on the generativist side, many

of the case-marking typologies that the Basque formalists had as models were very

restricted. In some cases, there were only two types of case-marking systems available:

accusative and ergative. In addition, syntax and morphology were often entangled in a

holistic model. The unaccusative hypothesis did improve these classifications. Yet,

because of the too theory-internal nature of the generativist proposals and because of the

small number of languages examined, the unaccusative hypothesis did not resolve the

Basque issue, as it remained unclear for scholars working on Basque what a language

such as Italian could say for case-marking typologies.

Within the typological tradition too, there were scholars who overtly denied the

need for distinguishing case-marking systems other than accusative and ergative. Thus,

Dixon (1994:77) claims: “despite scholarly opinions, it is a clear fact that split-S

systems do involve a mixture of ergative and accusative patterns”. Similarly, Trask

(1979:387) argues: “in my opinion, it still remains to be established that the active

system is any more than an idiosyncratic development of the more usual ergative

system”.

Additionally, the tradition for considering Basque as ergative seems to have

been strong enough to dissuade some scholars from proposing a differing analysis. In

fact, the Royal Academy of Basque (see Section 10.2.5) still favors an ergative system

by recommending the use of ABS subjects for lexically-simple unergatives, (which

clearly goes against actual usage in Western Basque). In conclusion, all of the above

11

factors combined have made the analysis of the Basque case-marking system a very

difficult task.

4. The semantics of the unaccusative / unergative opposition in Basque

The terms unaccusative / unergative, besides being rather unfortunate, do not reflect a

unique semantic contrast. Put differently, split-intransitive systems may be based on

different semantic distinctions (cf. Mithun 1991). Thus, although traditionally these

systems have been called “active” in the typological literature, active-stative systems

only constitute a subset of all semantically-based systems. Other semantic contrasts of

this type involve agentive and patientive predicates. Yet, even within these systems,

there are more distinctions to make.

Levin (1983:290-ff) was the first to point out that there is a semantic split

conditioning the case-marking of intransitive predicates in Basque. She proposed that

the ERG / ABS split of Basque intransitives was by and large based on an agentive /

patientive distinction. This would give rise to two semantically homogeneous classes of

intransitive verbs. Thus, according to Levin, Basque unaccusatives would correspond to

patientive predicates, and Basque unergatives to agentive predicates. Now, while

Levin’s proposal is on the right track, I have already mentioned that there are three

types of problems with it. These are the issues of dialectal variation (see Sections 8, 9

and 10), of lexically-simple unergative verbs (see Section 6), and of the exact semantic

distinction underlying the unaccusative / unergative split of Basque (see below).

In what follows I attempt to provide a more accurate characterization of the

Basque unaccusative / unergative contrast by taking into account the two-fold agent /

patient opposition in intransitives I put forward in Section 1. Recall that I propose that,

in the case of Basque (and probably in other Western European split-intransitive

12

systems), the semantic factor which conditions the intransitivity split is close to the

contrast patientive vs. non-patientive.

Trask (2002: 269-272) argued against Levin’s hypothesis that Basque has a

semantically-based case-marking system. Thus, Trask (2002:271) claims “Basque

possesses a sizeable number of intransitive verbs which unquestionably take agent

subjects but which are morphologically strictly intransitive in all varieties, such as joan

‘go’, etorri and jin ‘come’, sartu ‘enter’, igan ‘go up’5, jaitsi ‘go down’, jaiki ‘get up’,

etzan ‘lie down’, zutitu ‘stand up’, belaunikatu ‘kneel’, jantzi ‘get dressed’, erantzi ‘get

undressed’, ezkondu ‘get married’, and many others”. However, Trask’s objections are

seriously undermined by the fact that he (apparently following Levin) opposes

agentivity vs. patientivity. The verbs Trask mentions are indeed all agentive. Yet, the

pertinent factor is whether they are patientive or not. And, as I will try to clarify below,

these verbs are clearly patientive. This is the reason why they have an ABS subject, i.e.

the reason for their being “morphologically strictly intransitive”.

The above verbs —and most particularly some of them such as jantzi ‘get

dressed’, erantzi ‘get undressed’— are actually similar in meaning to (transitive)

reflexives (although syntactically these are intransitive verbs). In these predicates, as in

reflexives, the agent of the action expressed by the verb and its undergoer refer to the

same entity. Therefore, in the verbs at issue, the subject is both agent and patient. (See

more discussion in Section 4.3.)

4.1. The data to account for

5 Actually, the verb igon ‘go up’ takes ERG subjects in the Biscayan dialect (see

Section 10.2.7).

13

Let us start by making more explicit the data which the semantic distinction we need to

capture ought to account for. Regarding lexically-compound verbs, egin compounds

have to correlate with unergatives. More precisely, unaccusatives should not be egin

compounds, whereas unergatives may be egin compounds (or may be lexically-simple).

Recall that egin compounds always have ERG subjects. In turn, regarding lexically-

simple verbs, unaccusatives should have ABS subjects in all dialects of Basque,

whereas unergatives should have ERG subjects in Western Basque (especially in

Biscayan) and ABS subjects in Eastern Basque (especially in Souletin).

Another characteristic of Basque split intransitivity is that it is generally

insensitive to the actual context (e.g. by the verb modifiers). In other words, Basque

unergatives and unaccusatives are largely lexicalized, so that their classification

corresponds to their prototypical use. This implies that learning whether an intransitive

verb of Basque is unaccusative or unergative, while mostly based on a semantic

contrast, can in specific cases simply have to be memorized. This can be a reason for the

conservation of idiosyncrasies. Additionally, the age and the token frequency of a given

verb may be factors in the contrast at hand. Thus, since the change in progress evolves

towards more ERG subjects in unergatives, old high-frequency verbs favor the

conservation of ABS subjects, while new verbs, particularly many recent borrowings,

favor the use of ERG subjects. (See more discussion in Sections 8, 9 and 10; especially

in 9.1.4 and 10.2.6.)

4.2. Defining patientivity

In order better to understand what the patientive / non-patientive contrast really means,

we should attempt to define patientivity in a more rigorous manner. I believe the

question is better addressed from a theory based on prototypes, such as Dowty’s (1991)

14

cluster-concept of proto-patient. Such a theory can better account for cross-linguistically

similar but partly different systems, and also for the minor variation within and across

dialects or varieties. The most relevant contributing properties for the patient proto-role

in intransitive subjects are: a) change of state or position, b) telicity, and c) affectedness.

Thus, 1) the most prototypically patientive of intransitive subjects are those

appearing in telic changes of state (e.g. break, open, die) or of position (e.g. stand up, sit

down, fall), including those telic changes that have a reflexive-like meaning (e.g. get

dressed, get undressed) and telic verbs of directed motion (e.g. arrive, come or even go,

especially with a telic modifier); and 2) less prototypically patientive intransitive

subjects contain those appearing in atelic (or not inherently telic) but affected

predicates: either stative, such as verbs of existence (e.g. live, stay, exist) and verbs of

spatial configuration (e.g. stand, sit, lie), or dynamic, such as atelic verbs of directed

motion (e.g. ascend, descend), atelic verbs of change of state (e.g. dim, widen, harden),

and atelic verbs that have a reflexive-like meaning (e.g. wash oneself, comb oneself).

By contrast, the most relevant properties of the prototypical non-patientive

intransitive subject oppose those of the prototypical patientive subject; namely: a) non-

change, i.e. process or activity, b) atelicity, and c) non-affectedness. Thus, 1) the most

prototypically non-patientive of intransitive subjects are those with non-affected

subjects appearing in atelic activities (e.g. work, dance, run, swim) or in processes of

emission (e.g. shout, bark, snore), including inanimate emissions (e.g. shine, glow,

sound) and weather processes (e.g. rain, snow); and 2) less prototypically non-

patientive intransitive subjects include those of atelic predicates which cannot be readily

conceptualized as affected or non-affected, either animate (e.g. those human activities or

bodily processes that may take a reflexive-like reading: sleep, rest, smile, defecate) or

15

inanimate (e.g. roll, vibrate), and also weather changes (e.g. dawn, get dark) and telic

emissions (e.g. burst, explode).

4.3. Understanding intransitive affectedness

I have used the term “affectedness” as one of the contributing properties of patientivity.

Yet, the concept of affectedness in intransitive predicates is difficult to grasp. By

comparing verbs of directed motion (e.g. come, arrive) with verbs of manner of motion

(e.g. run, swim) we may get closer to the idea of affectedness. Both of these groups of

verbs are typically agentive and, furthermore, volitional. However, only verbs in the

former group are inherently telic and have an affected subject. That is, the subject of

directed-motion verbs is typically the agent of the action expressed by the verb, but also

its undergoer. The patientive nature of the subjects of directed-motion verbs is better

observed in resultative and perfect constructions (e.g. he has arrived), which are typical

of these verbs. Conversely, in verbs of manner of motion, the subject is the agent of the

action, but not its undergoer. The non-patientive nature of the subjects of manner-of-

motion verbs is better observed in durative progressive constructions (e.g. he is

running), which are typical of these verbs.

Similarly, intransitive changes of state or of position (even agentive ones)

typically involve an affected (patientive) subject, whereas processes of emission (even

non-agentive ones) typically imply a non-affected (non-patientive) subject. In other

words, the subject of intransitive changes is the patient of the action conveyed by the

verb (even if it can also be its agent), while the subject of intransitive emissions is not

the patient (it would be something like the “sender” or “emitter”) of the action

expressed by the verb (even if it may not be an agent).

16

Patientivity of the subject of a given intransitive verb can be subject to variation

depending on its degree of affectedness. For example, the verb to work seems to provide

a clear-cut example of an agentive non-patientive predicate. However, if work is

understood in the sense of strive, struggle, take pains, work up a sweat, work flat out,

put one’s all into something, exert oneself, push oneself, then a patientive (sort of

reflexive) reading may also appear. The same can be said of other (in principle) non-

patientive predicates, such as enjoy, defecate, rest, sleep, etc.

4.4. Some difficult but revealing examples of Basque

Since the concept of patientivity just outlined is rather subtle and based on prototypes,

there are, of course, intransitive verbs which are hard to ascribe to one of the two above

classes. Just to mention a few, jump may be viewed as a patientive directed-motion

verb, if one focuses on the change of position of the subject; yet, it may be

conceptualized as non-patientive, if focusing on the activity of jumping itself. The latter

reading would be favored in an iterative situation with multiple jumps. In Basque, jauzi

egin, salto egin and saltatu ‘jump’ (the last two, Romance borrowings) are unergatives.

In turn, changes of state or of position that are not inherently telic may be partly

understood as non-patientive activities just as they can be conceptualized as patientive

changes. The same can be said of stative verbs of existence and stative verbs of spatial

configuration. In Basque, all these predicates are essentially unaccusatives. Yet, for

instance, the verbs aldatu ‘change’, eboluzionatu ‘evolve’, or existitu ‘exist’ are

occasionally heard with ERG subjects in Western Basque. Note that the last two are

recent loans. Their being new in the language can partly explain their more innovative

behavior towards a broader use of the ERG marker. Also, the compound verb gora egin

‘ascend, rise (e.g. price, temperature)’ is unergative.

17

The last verb I will examine now is boil. Although boil may seem a

straightforward change of state applying to a liquid, notice that, from a cognitive

perspective (unlike a physical perspective), there is not a clear-cut end-point delimiting

that change. Rather, what is cognitively noticeable is an activity occurring in the liquid

(after the boiling point has been reached). Thus, boil may be conceptualized as a non-

patientive activity instead of a patientive change. The Basque verb irakin / irakitu ‘boil’

is unergative (as in water-ERG boils).

Further research is needed in order to define patientivity more rigorously.

Particularly, entailment tests could be proposed which may provide a more quantifiable

characterization of this concept. I only hope to have established in the preceding

discussion the basic lines on which the patientive / non-patientive contrast lies,

especially in the case of Basque.

4.5. Predictions for Basque intransitive verbs

According to the above discussion, we can make the following predictions for Basque

intransitive verbs. The following sub-classes of intransitive verbs will have (in

principle) an unaccusative behavior: a) telic changes (including verbs of appearance and

disappearance); b) directed-motion verbs (e.g. of arrival and departure); c) stative verbs

of existence, position, spatial configuration, etc; and d) atelic changes. Exceptions

should more readily come from verbs in the last two classes (but see also the hypothesis

on directed-motion verbs of “emission” in Section 10.2.7).

In turn, the following sub-classes of intransitive verbs will have (in principle) an

unergative behavior: a) animate —particularly human— activities (including

intransitive speech verbs, intransitive meal verbs, etc); b) verbs of manner of motion

(undirected); c) animate emission verbs (even non-volitional); d) inanimate emission

18

verbs; e) weather processes; and f) verbs of extended temporal duration. Exceptions

should mainly be expected when these verbs take a reflexive-like reading.

I have already mentioned some in-between cases for which we can predict

variable behavior: a) verbs such as roll, vibrate, float, etc; b) weather changes; and c)

telic emissions. To these we can add some intransitive modal verbs (such as start and

finish), and directed-motion verbs of “emission”, which in the Biscayan dialect

constitute an awkward exception to the above generalization.

4.6. Testing the predictions

The remainder of this article will be devoted to testing in detail the predictions I have

just put forward. As we will see, by and large, the predictions will be borne out, even

though there are a few exceptions which are harder to account for in the light of the

patientive / non-patientive opposition proposed above. Regarding lexically-compound

intransitives, only a handful of egin compounds are among those predicted to be

unaccusatives. The rest are unergatives, as predicted. (Egin compounds are tested in

Section 6.) Regarding lexically-simple intransitives, very few verbs which are predicted

to be unaccusative have ERG subjects. Moreover, these exceptions come from the

Biscayan dialect (including the extinct Old Alavese), which is the most innovative and

“liberal” in the use of the ERG marker. In Sections 8, 9 and 10, I will focus on the

dialectal patterning of those lexically-simple intransitive verbs which are predicted to be

unergative. Before examining these questions, though, I need to say more on verbs

taking more than one case-marking type.

19

5. Verbs taking more than one case-marking array

In Basque, as in other languages, some lexical verbs can take more than one type of

case-marking. In this section, I make a brief digression and introduce some of these

verbs, in the hope that their study can shed light on the issue of ergative vs. semantic

alignment.

5.1. Causative alternation

A fair number of Basque verbs can be used as both intransitive and transitive, showing

an alternation whereby the transitive counterpart is the causative of the intransitive

counterpart. Typically, in this kind of alternation both the subject of the intransitive verb

and the object of the transitive verb have the semantic role of patient. The case-marking

pattern of these verbs is the same in all dialects. This is exemplified in (2) below.

(2a) Mikel etxe-an sartu da

Michael.ABS house-LOC entered is

‘Michael has got into the house’

(2b) Mikel hospitale-an sartu dute [haiek]

Michael.ABS hospital-LOC entered they.have [they.ERG]

‘They have put Michael in the hospital’

As illustrated in (2a), the subject of the intransitive verb, besides playing the role of

patient, may also be an agent. In other (perhaps more typical) cases, however, the

subject of the intransitive verb is a patient non-agent, as in (3a) below.

20

(3a) leiho-a apurtu da

window-DET.ABS broken is

‘The window has broken’

(3b) leiho-a apurtu dute

window-DET.ABS broken they.have

‘They have broken the window’

Other verbs that display this kind of alternation are ireki and zabaldu ‘open’, itxi and

hetsi ‘close’, berotu ‘get hot / heat’, erre ‘burn’, hil ‘die / kill’, etc. Among the

intransitive agentive ones, we can cite atera ‘go out / take out’, itzuli ‘return: come back

/ give back’, and a number of verbs that take a reflexive-like reading when they are

intransitive, e.g. garbitu ‘wash’, orraztu ‘comb’, etc. (Some verbs of this type can be

intransitive and transitive only in specific dialects, e.g. joan ‘go / take’, ibili ‘move /

use, manipulate’. Otherwise, they need to take the causative prefix era- for the transitive

reading: eroan / eraman ‘take’, erabili ‘use, manipulate’.)

Note that, if a lexical verb that is typically transitive (i.e. a verb which typically

has a direct object, an ERG subject and a transitive auxiliary) takes a referential ABS

subject and an intransitive auxiliary, then the most common reading by far (practically

the only possible reading) is the anti-causative reading just presented, whereby the

subject is a patient. For instance, if transitive kendu ‘remove, take off’ takes a referential

ABS subject, then it conveys the intransitive patientive reading ‘get away from, move

off’.

21

5.2. Intransitive / transitive alternation within non-patientive verbs

Other verbs of Basque can also function as intransitive and transitive, but the alternation

they show is not based on causation. Rather, the subject of these verbs always functions

as an (agent) non-patient in both the transitive and the intransitive counterparts. The

difference between transitive and intransitive readings, therefore, lies in the presence /

absence of a patientive direct object. Within these verbs, a further distinction can be

made.

5.2.1. Prototypically transitive

Some of the verbs showing this type of alternation are prototypically transitive. They

only take a derived intransitive reading when they are used to express an activity that

has a covert indefinite direct object. Basque transitive verbs which can take this derived

intransitive reading include verbs of ingesting (jan ‘eat’, edan ‘drink’, and “meal

verbs”— the latter are transitive only in the Spanish-Basque dialects: see Section 8.1.2),

and verbs of occupation (ikasi ‘study’, irakurri ‘read’, idatzi ‘write’, josi ‘sew’, erein

‘sow’, etc). These verbs take ERG subjects for both readings in all dialects of Basque.

This is illustrated in (4) below.

(4a) gizon-a-k sagarr-a jan du

man-DET-ERG apple-DET.ABS eaten has

‘The man has eaten the / an apple’

(4b) gizon-a-k jan du

man-DET-ERG eaten has

‘The man has eaten’

22

5.2.2. Prototypically intransitive

Other (agentive) non-patientive verbs of Basque may occasionally appear in a transitive

construction in some special uses, yet they are prototypically intransitive. Verbs of this

type include mintzatu ‘speak’, dantzatu ‘dance’, jolastu ‘play’, and a few others;

relevant examples are verbs of manner of motion with a pseudo-direct object denoting

the extent of the path being covered —run one mile—, or verbs of human activities with

a pseudo-direct object referring to the period of time over which the activity takes place

—fast two days. These verbs, whose case-marking pattern shows dialectal variation,

will be at the focus of our attention in the ensuing discussion. In Eastern and Central

Basque, these verbs display ABS / ERG subject alternations of the type illustrated in (5)

below.

(5a) [ni] euskara-z mintza-tzen naiz

[I.ABS] Basque-INST speak-ing I.am

‘I speak in Basque’

(5b) [ni-k] euskara mintza-tzen dut

[I-ERG] Basque.ABS speak-ing I.have

‘I speak Basque’

5.2.3. Transitive vs. intransitive

A few (agentive) non-patientive verbs of Basque may be transitive and intransitive, but

the presence / absence of a direct object entails a significant variation of meaning

between the two readings. Typical verbs of this type include unergatives with inanimate

subjects which are derived from transitive verbs by means of a metaphorical

interpretation. For example, the inanimate intransitives eman and jo ‘blow (wind), heat

23

(sun)’ come from the transitives eman ‘give’ and jo ‘hit’, respectively; the inanimate

intransitive hartu ‘take root (plant)’ comes from the transitive hartu ‘take, catch’. As

with the verbs illustrated in (4) above, intransitive verbs such as eman, jo or hartu also

take ERG subjects in all dialects of Basque.

5.3. Patientive / non-patientive alternation within intransitives

Finally, there are in Basque a few verbs which present another kind of alternation

regarding the semantic role of their subject, while always being intransitive. That is,

these verbs can have both patientive and non-patientive subjects without taking a direct

object. The patientive reading in this alternation often expresses the ingression to the

process conveyed by the non-patientive reading. This alternation includes verbs such as

pausatu ‘land, come to rest, calm down / pause, make a pause, rest’, airatu ‘take off /

fly’, argitu ‘lighten, clear up, dawn / shine, glow’, luzatu ‘get longer / be late’, etc.6

These verbs are in principle unaccusatives when they have patientive subjects,

and unergatives with non-patientive subjects. Yet, the existence of a doublet may have

implications for their case marking, leading either towards one common patterning or

conversely towards differentiation of the two readings.

6. Unergative egin compounds vs. unergative simple verbs

We have seen in Section 4 that unergative verbs in Basque can be defined as those

intransitive verbs which have a non-patient subject. Now, among Basque unergatives,

many (probably most) are construed as compound verbs consisting of the combination

6 Note also the alternation present in verbs with an indirect complement, such as heldu

and lotu ‘stick (patientive, unaccusative) / grasp, seize, hold on to (non-patientive,

unergative)’. (See note 3.)

24

of a lexical nominal item (usually in the ABS case with no article) and the light verb

egin ‘to do’. As illustrated in (1d)-(1e) above, the subject of these compound verbs is

always in the ERG case. This is common to all dialects of Basque. Geographically, egin

compounds are especially frequent in Western Basque, but they are actually present in

all dialects.

Not all compound unergatives in Basque are formed with egin. Since

unergatives are most typically employed in the progressive aspect, other light verbs

which frequently serve as the basis for Basque unergatives are those used to convey

progressive meaning. Thus, constructions such as those consisting of a nominal item in

the locative or adverbial cases plus the light verbs egon ‘be at’, ibili ‘be doing

something’, ari izan / aritu ‘be busy doing something’ also serve to express unergative

predicates; e.g. dantza-n ari da ‘s/he is dancing’ (lit. ‘s/he is busy in the dance’),

dantza-n ari izan da ‘s/he has been dancing’ (lit. ‘s/he has been busy in the dance’).

Note that, unlike in the case of egin compounds (see below), the light verbs used in

these constructions are themselves patientive intransitives. Therefore, their subject goes

into the ABS case (just as the subject of the transitive verb egin goes into the ERG, even

when an egin compound is patientive: see Section 6.1 below).

Another light verb which also serves as host for unergatives is erauntsi ‘be

(busy) doing something’. This intransitive light verb is very similar in meaning to the

ones just presented (e.g. arrantzu-n erauntsi du ‘s/he has been fishing’), yet erauntsi

exhibits the particularity of having ERG subjects. In fact, erauntsi is among the few

intransitive verbs having ERG subjects in the Souletin dialect (see discussion in Section

8.1.3). I should also mention that eman ‘give’ is the light verb of a number of

unergative-like “indirect transitive constructions” with a dative complement, which, due

to reasons of space, cannot be studied in this article (see note 3). Other light verbs

25

appearing in Basque unergatives of emission are the transitives bota ‘throw’ and atera

‘throw out’, as in txistua bota ‘spit’ and zarata atera ‘make a noise’. (The inherent

meaning of bota and atera is clearly appropriate as the basis for a verb of emission.) In

what follows, however, I will focus on the most typical and common compound source

of unergatives in Basque, namely egin compounds, disregarding other types of

compound unergatives.

Egin compounds look like originally transitive constructions that are undergoing

a process of lexicalization by object incorporation. This process does not appear to have

reached an advanced stage, though, since for the most part the indefinite nominal item

of egin compounds is not syntactically frozen and can behave in negative and

interrogative sentences like other direct-object NPs. In any event, in terms of their

source of grammaticalization, (most) egin compounds should be regarded as transitive.7

Two independent questions arise regarding egin compounds. First, is there any

semantic motivation for egin compounds? Second, do egin compounds provide

7 The exceptions to the transitive source of egin compounds come from a handful of

cases where the nominal item is not in the ABS case but in an oblique case, such as the

instrumental (hegaz egin ‘fly’, from hego ‘wing’), an adverbial (korrika egin or lasterka

egin ‘run’, from korri and laster ‘fast’), or the locative (dantzan egin ‘to dance’, from

dantza ‘dance’). Still, the subjects of these verbs appear in the ERG case. Note,

however, that this type of egin compound is only present in Western Basque. Thus,

hegaz egin is exclusive to the Biscayan dialect (and hegan egin to Guipuzcoan), korrika

egin is confined to Guipuzcoan, lasterka egin only shows up in Navarrese (together

with laster egin), and dantzan egin is restricted to Biscayan and Guipuzcoan (alongside

dantza egin). These data constitute further evidence in favor of the main thesis in this

article; namely, the different type of alignment in Western and Eastern dialects.

26

evidence for a split-intransitivity (semantically-based) analysis of Basque case-

marking?

6.1. Semantic motivation for egin compounds

Regarding the first question above, the answer is clearly yes: on the whole, egin

compounds are semantically motivated. Although there is not a one-to-one

correspondence between egin compounds and unergatives, nearly all egin compounds

are unergatives (and a majority of unergatives are egin compounds). This can hardly be

due to pure chance. What is crucial for the hypothesis of a semantic motivation for egin

compounds is that very few (predicted) unaccusatives belong in this group. Practically

the only putative unaccusatives which are egin compounds are the following: leher egin

‘explode’, alde egin and ospa egin ‘leave’, ihes egin ‘flee’, topo egin ‘run into’, and

some atelic verbs of directed motion, e.g. gora / behera egin ‘ascend, rise / descend,

drop (e.g. price, temperature)’, atzera / aurrera egin ‘progress, improve, go forward /

recede, go backward’.

The verb explode is actually a telic emission, as we have seen in Section 4.2.

Thus, variable behavior was predicted in this case (Section 4.5): it can be

conceptualized as a patientive change or as a non-patientive emission. As for the next

three unaccusative egin compounds just cited (alde egin, ospa egin and ihes egin), all

three belong to the exceptional sub-class of directed-motion verbs of “emission” which

will be examined in Section 10.2.7. Atelic verbs of directed motion have already been

discussed in Sections 4.2 and 4.4: recall that they are not prototypically patientive. On

the whole, therefore, the correlation between egin compounds and unergatives is very

high.8

8 Topo egin ‘run into’ is perhaps the most uncomfortable exception.

27

Moreover, the motivation for grammaticalizing unergative intransitive verbs in

the same fashion as transitive verbs is consistent with the similarities between these two

groups of verbs. Recall that unergatives in Basque have non-patientive subjects, just as

transitive verbs do. Furthermore, a non-patientive intransitive subject tends to imply

some kind of dummy cognate object, i.e. a dummy patient. This increases the

similarities with transitive predicates. It is precisely this sort of dummy cognate object

that is grammaticalized as the lexical item of Basque unergatives. The cognate object of

the intransitive verb occupies the formal position of the direct object of transitive

clauses, while a light verb plays the role of the main verb. Therefore, Trask’s (2002:

271-272) position against a semantic motivation for Basque unergatives, and

particularly for egin compounds, is not supported.

Actually, we can narrow down the semantic membership of egin compounds

even further. Leaving aside weather verbs (which are metaphorically animate), egin

compounds are by and large not only unergative verbs but unergatives that typically

have animate subjects, either volitional or non-volitional. Specifically, most egin

compounds belong to one of the following semantic sub-classes: verbs of volitional

human activities (e.g. lan egin ‘work’, jolas egin ‘play’, borroka egin ‘fight’, ehiza egin

‘hunt’, barau egin ‘fast’, dantza(n) egin ‘dance’), verbs of manner of motion (e.g.

korrika egin and laster(ka) egin ‘run’, hegaz egin ‘fly’, igeri egin ‘swim’), speech verbs

(e.g. hitz egin and berba egin ‘speak’, solas egin ‘chat’, ele egin ‘converse’), animate

sound emission verbs (e.g. oihu egin ‘shout’, garrasi egin ‘scream’, korrok egin ‘burp’,

aharrausi egin ‘yawn’, zurrunga egin ‘snore’, zaunka egin ‘bark’, miau egin ‘meow’,

irrintzi egin ‘neigh’), other animate emission verbs (e.g. eztul egin ‘cough’, doministiku

egin and usin egin ‘sneeze’, tu egin ‘spit’, putz egin ‘blow’), verbs of non-volitional

human activities (e.g. lo egin ‘sleep’, amets(etan) egin ‘dream’), verbs of human

28

reactions (e.g. barre egin ‘laugh’, irri egin ‘smile’, negar egin ‘cry’), verbs of bodily

processes (e.g. kaka egin ‘defecate’, pixa egin ‘urinate’).9

Although lexical patterns are hardly as systematic as grammatical patterns, what

the lexicalization of unergatives as egin compounds points out (even with its

exceptions) is that unergative verbs are differentiated in the lexicon of Basque because

of their special semantics. These semantic features make them closer to transitives in

several respects.

6.2. Egin compounds, simple unergatives, and the case-marking system of Basque

However, the lexical differentiation of many Basque unergatives does not necessarily

imply that Basque has a semantically-based case-marking system. As we will see for

Eastern Basque, it may still have a (somewhat special) ergative system. Note that

unergative egin compounds are grammaticalized as transitive verbs. Thus, it may be that

the reason for the ERG subject in egin-compound predicates is simply the very fact that

they have a transitive structure (cf. Trask 2002: 271). Egin compounds, therefore, are

neutral as to whether Basque case-marking is ergative or semantically-based. Learning

9 I have already mentioned that it may be idiosyncratic of a given dialect whether a

particular unergative verb is compound or lexically simple. However, by and large, only

verbs in the first three semantic classes just mentioned may be lexically simple as well

as compound; namely, verbs of volitional human activities, verbs of manner of motion,

and speech verbs. Additionally, we need to mention that intransitive meal verbs only

have a lexically-simple structure. Note, then, that the least volitional of animate

unergatives (i.e. animate emission verbs—except oihukatu ‘shout’—, verbs of non-

volitional human activities, verbs of human reactions, and verbs of bodily processes)

practically always have a compound structure in all dialects.

29

of egin compounds may be aided by their particular lexical-semantics, but learning of

their case-marking pattern may just be based on the model of transitive clauses. Hence,

studies of Basque unergativity such as Levin (1983), which are so extensively based on

egin compounds, are not entirely complete.

In order to determine definitely whether Basque has in the main a semantically-

based case-marking system or a valence-based (ergative) case-marking system, we need

to look at lexically-simple unergative verbs. This is so because lexically-simple

unergatives are full-fledged intransitive verbs, and thus their case-marking pattern is

clearly independent from that of transitives.10 If simple unergative verbs still take an

ERG subject, then Basque is certainly split-intransitive. Conversely, if lexically-simple

unergatives have an ABS subject, Basque is rather ergative.

Unfortunately, there are several problems with this approach. First, there are few

lexically-simple unergatives in Basque, especially in the Western dialects. Second,

among these few simple unergatives, a number of them typically have inanimate

subjects. Third, many simple unergatives are recent loans. In other words, if one wants

to examine in Western and Eastern Basque the behavior of the most prototypical

lexically-simple unergatives (i.e. agentive animate unergatives that are not recent loans),

what one finds is that a majority of these may be lexically-simple in Central and Eastern

dialects but have a compound structure in Western Basque. Thus, it is difficult to

directly compare lexically-simple unergatives in Eastern and Western dialects.

However, one can always test whether in Western Basque lexically-simple

unergatives are better accepted with ERG subjects or with ABS subjects (see Section

10.2.4). Also, a few lexically-simple animate unergatives can be found in the Western

10 Bear in mind, nonetheless, that some lexical verbs may be both transitive and

intransitive. For discussion on this issue, see Sections 5.2 above and 11.4 below.

30

dialects if one searches carefully for them. All in all, with all these problems to deal

with, the fact is that within prototypical lexically-simple unergatives there are major

differences between Western Basque and Eastern Basque regarding the subject’s case

(ERG or ABS). I address this dialectal distribution below.

7. Lexically-simple unergative verbs

In the following sections, I offer an approximate picture of the dialectal behavior of

lexically-simple (i.e. non-compound) unergative verbs in Basque. I will focus on those

unergatives that are typically agentive, volitional and animate, i.e. the most prototypical

of unergative verbs. The description is only tentative at this point. It is not the purpose

of this paper to provide a detailed dialectological study of the behavior of these verbs.

The semantic sub-classes that are pertinent for this study are presented below,

together with a list of the verbs that belong in each of the sub-classes.

(6) Agentive lexically-simple unergatives of Basque

a) verbs of manner of motion: korritu / lasterkatu ‘run’, hegatu / airatu ‘fly’,

igerikatu ‘swim’, paseatu ‘walk’, saltatu / jauzi ‘jump’

b) verbs of agentive human activities: dantzatu ‘dance’, jostatu / jolastu ‘play,

enjoy (children)’, jokatu ‘play, practice (games and sports)’, borrokatu /

gudukatu ‘fight’, iharduki / aharratu ‘quarrel, argue’, bidaiatu / biajatu ‘travel’,

arrantzatu ‘fish’, laboratu / trabailatu ‘work’, erreusitu ‘succeed’, erauntsi ‘be

(busy) doing something’

c) verbs of less-agentive human activities (including verbs of extended temporal

duration): barautu ‘fast’, deskantsatu / pausatu ‘rest’, beilatu ‘watch, stay

31

awake’, erreinatu ‘reign’, sufritu / padezitu ‘suffer’, iraun ‘last, endure’, tardatu

/ luzatu / berandu ‘be late’

d) speech verbs: mintzatu ‘speak’, and the nearly synonymous solastatu, elekatu,

hizketatu, xarduki

e) meal verbs: bazkaldu ‘have lunch’, afaldu ‘have dinner’, gosaldu ‘have

breakfast’ 11

Recall from note 9 that the least agentive of animate unergatives (i.e. animate emission

verbs, verbs of non-volitional human activities, verbs of human reactions, and verbs of

bodily processes) have a compound structure in all dialects. Hence, these sub-classes are

not relevant for the present inquiry. Recall also that most of the verbs in the list in (6)

that are not recent loans are usually egin compounds in Western varieties, and may also

have a compound structure in Central and Eastern varieties. Meal verbs (plus era- / ira-

prefixed verbs; see Section 8.1.3) are the only non-borrowed unergatives that are always

lexically-simple in all dialects.

The other unergative verbs we should examine are those which typically have

inanimate subjects. As it happens, inanimate unergatives are almost systematically

lexically-simple. Unfortunately, inanimate unergatives are rather scarce and have not

been studied separately and in detail in surveys of Basque unergativity, let alone Eastern

Basque unergativity. A list of Basque inanimate unergatives is given below.

11 Other verbs such as begiratu and behatu ‘look at’, jarraitu / jarraiki ‘follow, pursue’,

jazarri ‘revolt’, ihardoki ‘resist’, or oldartu ‘attack’ resemble unergatives in that they

are agentive non-patientive intransitive verbs. However, these verbs prototypically form

two-place predicates with indirect objects, as I discuss in Aldai (2007). (See note 3.)

32

(7) Inanimate lexically-simple unergatives of Basque

irakin, irakitu, herakitu ‘boil’, argitu / distiratu / dirdiratu / izarniatu ‘shine,

sparkle, glitter, glow, flash’, buhatu / jo / eman ‘blow (wind)’, jo / eman ‘heat

(sun)’, hartu ‘catch, take root (plant)’, kurritu ‘run, flow (water)’

In the main, inanimate unergatives can be said to be (unexpectedly) exceptional in their

case-marking. Unlike other unergatives, inanimate unergatives take ERG subjects in all

dialects of Basque. However, it is not clear that their ERG subjects should necessarily

be due to their specific lexical semantics. Since there are few inanimate unergatives in

Basque, it may rather be that various specific reasons have conspired to give the

impression of a generalized behavior. Thus, irakin / irakitu / herakitu is a verb in era- /

ira- (see Section 8.1.3); jo, eman and hartu are metaphorical readings of transitive verbs

(see Section 5.2.3); buhatu and kurritu, which are not always inanimate, are

nevertheless relatively recent loans —the former from Bearnese (see Section 9.1.4);

argitu has two different intransitive readings: one patientive (ABS), one non-patientive

(ERG) (see Section 5.3); dirdiratu and izarniatu are not documented in Souletin. More

research is needed regarding inanimate unergatives. In any event, we should

acknowledge that (perhaps due to the transitive-based metaphorical nature of some

inanimate unergatives and weather processes) ERG subjects are readily associated with

this type of intransitive verbs, particularly with inanimate emissions.

In what follows, I provide a description across dialects of lexically-simple

unergative verbs of Basque, focusing on agentive lexically-simple unergatives (cf. (6)

above). The main sources of data will be the following: a) The Ethno-linguistic Atlas of

the Basque Country (Aranzadi 1983, 1990); b) The General Dictionary of Basque

(Michelena 1987-2005); and c) my own research.

33

The Ethno-linguistic Atlas of the Basque Country gives the corresponding

dialectal translations for the following items which are relevant for the present study:

‘he hunts, he hunted’ (Aranzadi 1983: 58), ‘he fishes, he fished’ (59), ‘he speaks, he

spoke’ (Aranzadi 1990: 307), ‘he runs, he ran’ (312), ‘he jumps, he jumped’ (313), ‘he

swims, he swam’ (315), ‘we shall travel, we traveled’ (372), ‘it flies, it flew’ (455).

(The first two verbs, i.e. hunt and fish, can readily be transitive in Basque; see Section

5.2.1. Hence, they are not as significant as the other six, on which I will focus.)

8. Eastern Basque

Eastern Basque, and especially the Souletin dialect, displays a case-marking pattern

which is close to what one may expect to find in an ergative language.12 The great

majority of intransitive verbs, not only unaccusative verbs but also (lexically-simple)

unergatives, take ABS subjects and intransitive auxiliaries.

8.1. Souletin

8.1.1. The Ethno-linguistic Atlas of the Basque Country

Souletin is the dialect where non-compound unergatives are most strongly represented

in the Atlas. Interviews were carried out in seven localities of the Soule13 (see Figure 2).

Although occasionally egin —and e(r)auntsi— compounds do appear, generally all of

the six agentive unergative verbs cited in Section 7 occur in the Atlas as lexically-

12 Scholars familiar with Western-Basque data, however, will find Eastern and Central-

Basque data to be rather surprising.

13 Namely: Alçay (Altzai), Domezain (Domintxiñe), Esquiule (Eskiula), Etcharry

(Etxarri), Larrau (Larrañe), Sainte-Engrâce (Santa Garazi), and Viodos (Bildoze).

34

simple verbs. Thus, there are a fair number of examples that are relevant for the present

study.

FIGURE 2 (MAP OF SOULE & LOWER NAVARRE TO GO HERE)

- ‘Speak’: Two different lexically-simple verbs are elicited in the Atlas to translate ‘to

speak’ in Souletin: mintzatü and elhestatü. Both of these verbs always take ABS

subjects, without exception.

- ‘Run’: The verb lasterkatü is elicited in most of the Souletin interviews in the Atlas. It

always occurs with ABS subjects, with no exceptions.

- ‘Jump’: In all of the Souletin interviews, the lexically-simple verb jauzi is used for

‘jump’. It is employed with ABS subjects.

- ‘Swim’: The lexically-simple verb ige(r)iskatü appears most of the time in Souletin for

‘swim’. It is always used with ABS subjects, with no exceptions.

- ‘Travel’: The verb bidaiatü is the most frequent translation of ‘travel’. It is always

employed with ABS subjects. The Romance loan paseatü also occurs with ABS

subjects (in Esquiule —located actually in Bearn, to the east of Mauléon—, and

Etcharry).

- ‘Fly’: The lexically-simple verb hegaltatü is nearly always used to translate ‘fly’ in the

Atlas. It is always elicited with ABS subjects. The verb airatu shows up in Domezain

(on the boundary with the Lower Navarre); it too takes ABS subjects.

- ‘Hunt’ and ‘fish’: Additionally, the verbs ihizkatü and arrantzükatü are employed in

Larrau to translate ‘he hunted’ and ‘he fished’, respectively. The latter takes an ABS

subject, but the former takes an ERG subject.

35

8.1.2. My own research

- Meal verbs:

Since there is no information in the Ethno-linguistic Atlas regarding meal verbs, in a

small survey I carried out in a few localities of Lower Navarre —Saint-Étienne-de-

Baïgorry (Baigorri) and Saint-Jean-Pied-de-Port (Donibane-Garazi)—, and Soule —

Mauléon (Maule), Tardets (Atharratze) and Sunharette (Zunharreta), which is located

very close to Tardets—,14 I tested the kind of subject that meal verbs take in each of

these localities. Recall from Section 5.2.1 that meal verbs are intransitive in the French-

Basque dialects; they cannot take a direct object in the ABS case (as in the Spanish-

Basque dialects).

Unlike in the Lower Navarre (see below), I only found one pattern of usage with

meal verbs (bazkaltü ‘have lunch’, aihaltü ‘have dinner’) in the Soule. All speakers I

interviewed used ABS subjects with meal verbs.

- Other lexically-simple unergatives:

Besides the verbs cited above, all other regularly-used unergative verbs in the list of (6),

i.e. verbs such as dantzatü ‘dance,’ txosta(ka)tü ‘play’, borrokatü ‘fight’, aharratü

‘quarrel, argue’ always have ABS subjects in Souletin.

8.1.3. Exceptions

Souletin exceptions to the pattern of having ABS subjects for lexically-simple

unergative verbs mainly come from a handful of verbs that can commonly have a

transitive reading (cf. ihizkatü ‘hunt’ above, and Section 5.2), or from verbs marked

with the causative prefix era- / ira-. Among intransitive / transitive non-patientive

14 I am very grateful to Raquel Iturrino Aranzamendi for her help with this survey.

36

verbs, we can cite argitü ‘shine, glow / illuminate’,15 and the recent loans buhatü ‘blow

/ stoke’, and erreüsitü ‘succeed / achieve’. (Note that argitü and buhatü frequently have

inanimate subjects. Another inanimate unergative with ERG subjects in Souletin is

disti(r)atü ‘shine’. See Section 7.) Among era- / ira- verbs, we have e(r)auntsi ‘be

(busy) doing something’, erakitü / herakitü ‘boil’, and irain / irein ‘last, endure’.

Intransitive verbs in era- / ira- look as if they might be ancient causatives (thus

originally transitive verbs) which have undergone some kind of semantic change (cf.

Lafon 1975:337, Trask 2002:271). Therefore, the ERG subject of these verbs may be

explained by proposing the conservation of an ancient patterning. (Yet the 18th century

Souletin writer Eguiateguy employs ABS subjects for irein, and the 17th century

Souletin writer Tartas uses ABS subjects for herakitü.)

Other putative exceptional unergatives having ERG subjects in Souletin could

be (some) recent loans. The behavior of recent loans in Souletin should be examined in

more detail in further research.

8.2. Lower Navarrese

8.2.1. The Ethno-linguistic Atlas of the Basque Country

Unlike Souletin, Lower Navarrese does not show too many lexically-simple unergatives

in the Atlas. Interviews were carried out in six localities of the Lower Navarre16 plus in

15 As discussed in Section 5.3, argitu can also have a third reading, an unaccusative one;

namely ‘brighten, clear up, dawn’. As predicted, this unaccusative reading triggers ABS

subjects (particularly when it does not refer to a weather change), while the transitive

and unergative readings have ERG subjects.

16 Namely: Amendeuix (Amenduze), Bidarray (Bidarrai), Irissarry (Irisarri), Saint-

Martin-d’Arberoue (Donamartiri), Uhart-Cize (Uharte-Garazi), and Urepel (Urepele).

37

Valcarlos (Luzaide), which politically belongs to the Spanish Navarre but linguistically

is rather part of the French (i.e. Lower) Navarre (see Figure 2). Out of the six agentive

unergative verbs cited in Section 7, only three occur in the Atlas as lexically-simple.

(The rest are egin compounds or ari izan compounds.)

- ‘Speak’: Two verbs, mintzatu and elekatu, are elicited in the Atlas to translate ‘to

speak’ in Lower Navarrese (the latter in Valcarlos). Both of them always take ABS

subjects.

- ‘Travel’: The verb piaiatu / bidaiatu is the most frequent translation of ‘travel’ in

Lower Navarrese. It always takes ABS subjects when it is lexically-simple. The

Romance loan kurritu also occurs with ABS subjects (in Irissarry).

- ‘Fly’: Two lexically-simple verbs occur in Lower Navarrese to translate ‘fly’. The

verb hegaldatu / hegaldakatu / hegalatu shows up in Uhart-Cize (next to Saint-Jean-

Pied-de-Port), Irissarry, and Valcarlos. The verb airatu is employed in Urepel and

Saint-Martin-d’Arberoue. Both of these verbs are always elicited with ABS subjects.

8.2.2. My own research

In the survey I carried out in the Saint-Jean-Pied-de-Port area, I found that practically

all speakers consistently used ABS subjects for mintzatu ‘speak’, jostatu ‘play, enjoy’,

borrokatu ‘fight’, and (for those who employed a lexically-simple verb for ‘dance’17)

17 Many speakers of Lower Navarrese employed a compound verb (either jantza egin or,

more commonly, jantzan ari izan) for the intransitive reading of ‘to dance’, which is the

most typical one. On the other hand, these speakers used the lexically-simple verb

jantzatu for the transitive reading exemplified in mutilak baso-jantza jantzatu du ‘the

boy-ERG has danced the glass-dance (a specific type of dance)’.

38

jantzatu ‘dance’. However, there was a patent discrepancy concerning intransitive meal

verbs (to the extent that informants argued among themselves as to whether meal verbs

triggered ABS or ERG subjects). Therefore, it seems that an isogloss dividing ABS and

ERG subjects for meal verbs passes through the Saint-Jean-Pied-de-Port area. (Bear in

mind, nevertheless, that Saint-Jean-Pied-de-Port is a meeting point for the people of all

the Lower Navarre.) To the east of this isogloss (i.e. in eastern Lower Navarrese), meal

verbs take ABS subjects; to the west, in turn, (i.e. in western Lower Navarrese) ERG

subjects are used for meal verbs. As a matter of fact, all speakers I interviewed in the

Baïgorry area employed ERG subjects for meal verbs (while using ABS subjects for

mintzatu / solastatu, jostatu and borrokatu).

The only exceptions to the grammar(s) I have just described concerned a number

of teenagers I interviewed in the Saint-Jean-Pied-de-Port area. These adolescents gave

ERG subjects not only for meal verbs, but also for jantzatu and (with hesitation) also for

mintzo / mintzatu. They, however, still gave ABS subjects for jostatu. (We will see an

identical system elicited from children in Sare, Labourd.)

8.3. Conclusion: Eastern Basque

Souletin and eastern Lower Navarrese are varieties of Basque where ABS subjects are

almost consistently used in lexically-simple unergative verbs. There are certainly very

few exceptions to this generalization. Furthermore, some of these exceptions can be

accounted for by proposing influence from transitive readings or conservation of

erstwhile transitive patterns. In fact, even intransitive meal verbs systematically display

ABS subjects. Thus, these data constitute evidence for considering the case-marking

system of Eastern Basque as ergative, i.e. valence-based. (See Section 11 for more

discussion on these issues.)

39

9. Central Basque

Central Basque has a case-marking pattern which is difficult to label. For the majority

of speakers, the most prototypical unergative verbs take ABS subjects, as in Eastern

Basque. Yet, besides ancient transitives and era- / ira- verbs, meal verbs and other less-

common unergatives, especially some recent loans and inanimate unergatives, may take

ERG subjects (following a lexically-based pattern). Additionally, a minority of speakers

use ERG subjects for practically all unergatives.

9.1. Labourdin

9.1.1. Lafitte (1962)

Lafitte (1962: 189-190) presents a list of 28 verbs (which he calls “deponents”) that are

used intransitively but have transitive morphology in the Literary (Navarro)-Labourdin

dialect (19th-20th centuries). Most of the verbs Lafitte gives, plus a few others, appear

here in the lists of unergative verbs of (6) and (7) above. However, out of the sub-

classes of unergatives I consider in (6) and (7), it is mostly meal verbs (afaldu ‘have

dinner’, askaldu ‘have lunch’, bazkaldu ‘have lunch’, gosaldu ‘have breakfast’), verbs

of less-agentive human activities, including verbs of extended temporal duration

(beilatu ‘watch’, berandu / luzatu ‘be late’, iraun ‘last’), and inanimate unergatives

(argitu ‘shine’, dirdiratu ‘glow’, distiratu ‘glitter’, eman ‘blow (wind)’, hartu ‘take root

(plant)’, irakitu ‘boil’, izarniatu ‘flash’) that Lafitte cites as taking ERG subjects in

Labourdin.18

18 Lafitte gives also the verb ihardoki ‘resist’, which takes an indirect object in modern

Labourdin (see notes 3 and 11). This verb used to also mean ‘argue’, and in some

40

Thus, we have two very different groups of unergative verbs taking ERG

subjects in Labourdin. On the one hand, we have meal verbs, which conceivably may

have some transitive-like nuance, particularly if influenced by jan ‘eat’ (a prototypical

transitive verb). On the other hand, and perhaps quite unexpectedly, we have verbs that

are not very agentive, including inanimate unergatives. (For some discussion of

inanimate unergatives, see Section 7 above.) Additionally, we should also consider

recent loans, as we will see in Section 9.1.4 below.

Therefore, the sub-classes of unergatives not cited by Lafitte are: verbs of

manner of motion (except the loan kurritu ‘run’), verbs of agentive human activities

(except the loans laboratu ‘work’ and erreusitu ‘succeed’), and speech verbs. These

three sub-classes of unergatives take for the most part ABS subjects in Labourdin (as

we will see in Sections 9.1.2 and 9.1.3). We may say that these three sub-classes include

the most prototypical of Basque unergatives. Also, the unergative verbs which take

ABS subjects in Labourdin are typically rather ancient in the language.

9.1.2. The Ethno-linguistic Atlas of the Basque Country

In general, there are few non-compound unergatives elicited in the Atlas for Labourdin.

Still, the verbs for ‘speak’, ‘travel’ and ‘fly’ are most of the time elicited as lexically-

simple, and some lexically-simple unergatives are also used to translate ‘jump’, ‘swim’

and ‘run’. Interviews were made in seven localities of Labourd.19 In addition, the

varieties of Basque the variant xarduki conveys the meaning ‘speak, chat’. (See note

24.)

19 Namely: Arbonne (Arbona), Briscous (Beraskoitze), Ciboure (Ziburu), Hasparren

(Hazparne), Itsassou (Itsasu), Sare (Sara), and Urcuit (Urketa).

41

(Spanish) Navarrese town of Zugarramurdi is also considered to belong to the

Labourdin dialect. (See Figure 3.)

FIGURE 3 (MAP OF LABOURD AND NAVARRE TO GO HERE)

- ‘Speak’: In the Atlas, mintzatu and mintzo izan appear with ABS subjects in

Labourdin, except in the coastal town of Ciboure/Ziburu (next to St.-Jean-de-Luz)

which uses both types of subjects (ERG in the perfect, and ABS in the imperfective).

- ‘Travel’: The verb piaiatu is always elicited with ABS subjects in Labourdin (Briscous

—east of Bayonne/Baiona—, Hasparren, and Itsassou —between Hasparren and Sare).

Yet, in Ciboure again, the variant bidaiatu is used with ERG subjects. The French

borrowing promenatu, in turn, takes ABS subjects in Arbonne (between the St-Jean-de-

Luz—Ciboure area and the Bayonne-Biarritz area).

- ‘Fly’: The verb airatu always has ABS subjects in the Labourdin examples in the

Atlas (Arbonne, Briscous, Sare). It, however, takes both types of subjects in

Zugarramurdi (ABS in perfect aspect, ERG in imperfective aspect). Note that airatu can

mean not only ‘fly’ but also ‘start to fly, take off’. Also, the Romance loan bolatu

presents both kinds of subjects in Urcuit (east of Bayonne, and very close to Briscous):

ABS subject in the imperfective, ERG subject in the perfect.

- ‘Jump’: In Labourdin, the old loan saltatu is elicited twice with ERG subjects (the

above-cited Ciboure and Zugarramurdi), and once with ABS subject (Sare). (Otherwise,

the egin compound jauzi egin is employed for ‘jump’.)

- ‘Swim’: The verb igerikatu is given with ABS subjects in Sare, but with ERG subjects

in nearby Zugarramurdi. (Otherwise, the compound verb igerika ari izan is employed

for ‘swim’.)

42

- ‘Run’: The verb lasterkatu has ABS subjects in the only example where it is elicited as

a non-compound verb (Itsassou). (Otherwise, Labourdin informants employ compound

verbs such as laster egin, lasterka joan, or lasterka ibili.) The Romance loan kurritu, in

turn, takes ERG subjects, once again, in Zugarramurdi.

- ‘Fish’: In addition, the verb for ‘fish’ was translated as lexically-simple in Ciboure and

Zugarramurdi; namely as arrantzatu. While in the Soule (Larrau) arrantzükatü was

elicited with an ABS subject (see Section 8.1.1), in Ciboure and Zugarramurdi

arrantzatu takes ERG subjects.

9.1.3. My own research

- Meal verbs:

Although Lafitte (1962:190) mentions that meal verbs can readily take intransitive

auxiliaries and ABS subjects in Literary Navarro-Labourdin (19th-20th centuries),

according to my observations (and pending further research), meal verbs always have

ERG subjects in modern Labourdin and modern western Lower Navarrese. Therefore,

an attested change has taken place regarding the case-marking pattern of these verbs,

since it is certain that at least in Old Labourdin (16th-17th centuries) meal verbs used to

only take ABS subjects.

- Other lexically-simple unergatives:

Practically all speakers of Labourdin I interviewed —in Biriatou (close to Hendaye /

Hendaia, on the Spanish-French coastal border), Ascain (between Ciboure and Sare),

and Sare— used ABS subjects for mintzatu ‘speak’, jostatu ‘play, enjoy’, borrokatu

‘fight’, and dantzatu ‘dance’, although ‘to dance’ was seldom freely elicited as a non-

compound verb. As for the verb jokatu, all of the informants employed it with ERG

subjects. Apparently, jokatu is better known in Labourdin with the meaning ‘bet’ than

43

with the meaning ‘play’. In any case, the intransitive use of jokatu ‘play’ with ABS

subjects, which is attested in Old Labourdin, has been lost in modern Labourdin.

In turn, some non-adult informants I interviewed in Sare/Sara gave the following

responses for ‘speak’, ‘dance’ and ‘play’: ERG subjects for mintzatu and dantzatu; ABS

subjects for jostatu (though one child actually gave an ERG subject in his first

attempt20). (Recall that we found the same pattern with adolescents in the Lower

Navarre. It may, therefore, be worth pursuing, in further research, the hypothesis that a

significant grammar change may be going on among the new generations of speakers of

Labourdin and western Lower Navarrese.)

9.1.4. Recent loans: historical data

Recent loans provide evidence of the most innovative kind of alignment pattern in force

in a given stage of a language. In this section I show that the use in Labourdin of ERG

subjects with most unergative verbs that are recent loans goes back to at least the 16th

century, when the first Basque texts are documented. This evidence is easy to find in the

writings of the 16th century Labourdin author Leiçarrague (New Testament translation

in 1571), from Briscous, near Bayonne/Baiona, who was very given to employing raw

borrowings from Latin or Romance. Thus, Leiçarrague uses, among others, the

following unergative loans with ERG subjects: adulteratu ‘commit adultery’, beilatu

20 This child’s mother, who gave ABS subjects for all three of these verbs, made the

comment that her son might be influenced by the Unified Basque he learned at school.

However, note that the Unified-Basque norm proposed by the Royal Academy gives

ABS subjects for mintzatu, dantzatu and jostatu (see Section 10.2.5). Perhaps the

mother’s comment may be relevant if related to Western-Basque influence at school.

44

‘watch, stay awake’, erreinatu ‘reign’, laboratu ‘work, farm, cultivate’, predikatu

‘preach’, presumitu ‘boast’, sufritu ‘suffer’.

One may wonder whether Leiçarrague (and all authors thereafter) actually

employs ERG subjects for every intransitive verb that is a recent loan. The data,

however, do not support this hypothesis. In principle, recent borrowings that are

unaccusatives take ABS subjects, as predicted: e.g. aphointatu ‘get reconciled to’,

batheiatu ‘get baptized’, erresuzitatu ‘resurrect’, erretiratu ‘retire, withdraw, move

away, leave’ (cf. French se retirer), habitatu ‘live’. As a matter of fact, even some

unergative loans may take ABS subjects in Leiçarrague’s work: e.g. desperjuratu

‘commit perjury’, disputatu ‘argue’, trabailatu ‘work, take pains’ 21.

It may be unnecessary to ask ourselves why some unergatives that are recent

loans take ERG subjects in Old Labourdin and some others take ABS subjects. There

seems to be some lexically-based randomness in this process. Nevertheless, it appears

crucial to take into account that recent loans may be partly changing the semantic

opposition between the two classes of Basque intransitives. That is, recent intransitive-

verb loans of Basque from Romance (French, Spanish, etc.) often reflect the same type

of split-intransitive behavior that was present in the lending language. More precisely,

there is a clear tendency towards an identification of ABS-subject intransitives in

Basque with Romance se-reflexives. Note, for instance, that, although disputatu ‘argue’,

promenatu ‘walk’, and trufatu ‘mock at’ are in principle non-patientive intransitives, all

three correspond to French se-reflexives: cf. French se disputer, se promener, and se

21 Note that trabailatu and laboratu are not exactly synonyms. The former takes often

the reflexive-like reading ‘take pains’, whereas the latter may be transitive, as in

‘cultivate one’s land’.

45

truffer, respectively. This may be a reason for their ABS subject in Labourdin. (See the

discussion of loans from Spanish se-reflexives in Section 10.2.6).

All in all, since there seem to be few unergatives in French (or Spanish) that are

se-reflexives, i.e. since Romance se-reflexives are mainly unaccusatives, the main

generalization to be captured is that unergatives which are recent loans are by and large

more innovative, i.e. they take more ERG subjects, than older better-accepted

borrowings, such as dantzatu, jokatu, jostatu, saltatu, solas(ta)tu, etc, not to mention old

high-frequency assimilated unergatives, such as mintzatu, borrokatu, etc.

9.2. Navarrese

9.2.1. The Ethno-linguistic Atlas of the Basque Country

The verbs for ‘speak’, ‘fly’ and ‘jump’ are generally elicited as lexically-simple in

Navarrese. There are also in the Atlas some non-compound unergatives corresponding

to ‘swim’, ‘travel’ and ‘run’. Moreover, interviews were made in no less than seventeen

localities of Navarre22 (not to mention Zugarramurdi and Valcarlos, which were already

included in the Labourdin and Lower Navarrese dialects, respectively). Additionally,

the Atlas has information about the Eastern Navarrese area, in the Spanish Pyrenees.

(Note that the Basque variety of this area became practically extinct in the second half

of the 20th century.) Therefore, we have for Navarrese a good number of examples that

are relevant for our study. (See Figure 3.)

22 Namely: Alli, Echarri-Aranaz (Etxarri-Aranatz), Garzaron (Gartzaron), Goizueta,

Inza (Intza), Irañeta, Leiza (Leitza), Lecaroz (Lekaroz), Lesaca (Lesaka), Lizarraga,

Saldias, Sumbilla (Sunbilla), Urdiain, Urrizola, Usi, Xuarbe, Ziganda, and Eugui (Eugi).

46

- ‘Speak’: The most common verb employed to translate ‘speak’ in Navarrese (as well

as in Labourdin and Lower Navarrese) is mintzatu. In Navarrese, as in the other two

dialects, most of the time when mintzatu is elicited, it is used with ABS subjects. There

is, however, one exception in the Atlas; namely in Lesaca (close to the Guipuzcoan

border), where mintzatu takes ERG subjects. (Recall that there is in the Atlas an ERG

subject with mintzatu in the Labourdin locality of Ciboure, and that I obtained ERG

subjects with mintzatu from children in Sare. Both Ciboure and Sare are not too far

from Lesaca.)

- ‘Fly’: The verb hegatu is used with ABS subjects in all of the many (ten) Navarrese

localities where it is elicited (as well as in Ataun —south-east of Beasain— and

Oyarzun, in eastern Guipuzcoa; see Figure 4 below). The verb airatu takes ABS

subjects in Lecaroz, near Elizondo (but remember the ERG subject in nearby

Zugarramurdi). The Romance loan bolatu, finally, is employed with ABS subjects in

Lizarraga —south of Echarri-Aranaz— and Urrizola —east of Echarri-Aranaz—, but

with ERG subjects in nearby Irañeta (as well as in Urcuit in Labourd).

- ‘Jump’: There are a fair number of ERG subjects occurring with the old loan saltatu in

Navarrese (Lizarraga, Saldias —east of Leiza—, Sumbilla —south of Lesaca—, Urdiain

—west of Echarri-Aranaz—, Zugarramurdi), but the number of ABS subjects is greater

(to which we could add those elicited in Hernani and Renteria, near San Sebastian /

Donostia, in eastern Guipuzcoa).

- ‘Swim’: We find only a few instances of lexically-simple verbs used to convey this

meaning. The verb i(n)gerikatu shows an ABS subject in Irañeta —east of Echarri-

Aranaz— (but recall the ERG subject in Zugarramurdi). A similar verb with ERG

subject is given in Urrizola (and perhaps in Echarri-Aranaz).

47

- ‘Travel’: The recent loan (from Spanish) biajatu has ERG subjects in Saldias and

Ziganda —north of Pamplona/Iruña. (Otherwise, Navarrese informants employ

compound verbs for ‘travel’.)

- ‘Run’: The loan korritu is given with ERG subjects in Urdiain (and kurritu in

Zugarramurdi). (Otherwise, the compound verb laster(ka) egin is used.)

9.2.2. Eastern Navarrese

As mentioned above, the Ethno-linguistic Atlas of the Basque Country provides

information about the nearly extinct Eastern Navarrese varieties. Since these varieties

were located towards the east of the Basque territory, partly in contact with Souletin,

and since they were generally rather conservative, it may be interesting to see whether

unergative verbs take here ABS or ERG subjects. I thus present below data from six

localities of Eastern Navarrese.23 (All are located close to the Pyrenean border, from

Eugui to the east; see Figures 1 and 3.) As will be seen, it does seem that ABS subjects

were more common with some unergatives towards the east of this area (Esparza, in the

Salazar valley; Uztarroz, in the Roncal valley). However, since these varieties suffered

great influence from Romance, unergative loans taking ERG subjects are also frequent

here. In the main, therefore, there is no substantial difference between Navarrese and

Eastern Navarrese regarding the issue under consideration.

- ‘Speak’: There is a great variety of lexically-simple verbs conveying ‘speak’ in this

area. The verb mintzo / mintzatu is obtained in the neighboring localities of Mezquiriz

and Burguete. In Mezquiriz, both ABS and ERG subjects are used, and in Burguete one

23 Namely, ordered from west to east: Mezquiriz (Mezkiritz), Villanueva de Arce

(Iraberri), Burguete (Auritz), Abaurrea Alta (Abaurregaina), Esparza, and Uztarroz

(Uztarroze).

48

informant employs ABS subjects while a different informant gives ERG subjects.

Another verb meaning ‘speak’ is elekatu. It is employed also in Burguete and in

Abaurrea (between Burguete and Esparza), in both cases with ERG subjects. (Recall

however that elekatu had ABS subjects in nearby Valcarlos.) A third verb for ‘speak’ in

Eastern Navarrese is xarduki. It is used in Mezquiriz, Villanueva de Arce (south of

Burguete) and Esparza. This verb always appears with ERG subjects; however its

transitive origin makes it less relevant for the issue at hand.24

- ‘Jump’: The old loan saltatu shows up in Villanueva de Arce and in Uztarroz (in the

far east). It is used with ABS subjects. In Uztarroz too, the verb jauzi takes also ABS

subjects. A similar verb (jauzkatu) takes ERG subjects in Mezquiriz. The Spanish loan

brinkatu is elicited with ERG subjects in Abaurrea, but with ABS subjects in Villanueva

de Arce and Esparza.

- ‘Fly’: The Romance loan bolatu takes ERG subjects in Villanueva de Arce and

Abaurrea, but ABS subjects in Mezquiriz and in the eastern-most localities of Esparza

and Uztarroz.

- ‘Swim’: The recent loan (from Spanish) nadatu is used with ERG subjects in Uztarroz.

(Otherwise compound unergatives are employed in this area.)

- ‘Travel’: The recent loan biajatu is elicited in Abaurrea and Uztarroz, in both cases

with ERG subjects. (Compound unergatives are used in all other cases.)

24 The verbs xarduki and ihardoki have their origin in a transitive compound verb ihar

eduki (Michelena 1987: 333). This would explain their ERG subjects. Yet, note that

currently xarduki / ihardoki are totally lexicalized, involving bound fossilized

morphology (but an ERG subject).

49

9.2.3. My own research

In the survey I carried out in the valley of Baztan (Arizcun and Errazu), whose main

town is Elizondo, I found that in Navarrese (as in Labourdin), afaldu ‘have dinner’,

bazkaldu ‘have lunch’ and gosaldu ‘have breakfast’ always take ERG subjects.

It can also be concluded from my research, with the necessary reservations that

are inevitable in a small survey, that some speakers of Navarrese tend consistently to

use ABS subjects with the majority of prototypical lexically-simple unergative verbs

(excluding recent loans), while other speakers tend consistently to use ERG subjects.

(Note also that, in the Ethno-linguistic Atlas cited above, the two informants from

Abaurrea and Zugarramurdi —and the one from Ciboure in Labourd— consistently

used ERG subjects in all of the various lexically-simple unergatives they gave.)

Thus, some speakers of Baztan used ABS subjects with dantzatu ‘dance’, saltatu

‘jump’, jostatu ‘play (children)’, solastu (and mintzatu) ‘speak’, and airatu ‘fly’. Yet,

these same speakers gave ERG subjects with jokatu ‘play games and sports’ (which is

mostly transitive), and, as already mentioned, with meal verbs such as afaldu. Note that

the grammar I have just described corresponds to the main grammar I found in Labourd.

Conversely, other adult speakers of Baztan, fewer but younger speakers, seemed

to have a different grammar. They used ERG subjects with dantzatu, saltatu, bidaiatu

‘travel’, and, of course, with jokatu and meal verbs (afaldu). As for jostatu ‘play, enjoy’

and especially solastu ‘speak’, there was variation among these speakers and even

within individual speakers. One of these speakers, for instance, gave ABS subjects for

solastu when prompted to do so, but then he used ERG subjects when informally

speaking with the interviewer. (He also mentioned his speech was partly influenced by

what he called Unified Basque, learned at school; in fact, the influence in this case

comes rather from the Guipuzcoan dialect; see note 20.) Another speaker gave ERG

50

subjects for solastu, but with some hesitation. Conversely, she offered ABS subjects for

its synonym mintzatu (a verb that was apparently rare in her own speech). Note that this

second grammar I found in Baztan corresponds roughly to the grammar I elicited from

children in Labourd.

9.3. Conclusion: Central Basque

From the Central-Basque data presented above, we can conclude that the Central

dialects (Navarrese and Labourdin) occupy a sort of transitional position between the

semantic alignment of Western Basque (see below) and the more ergative system of

Eastern Basque.

Perhaps the clearest difference between Central Basque and Eastern Basque

involves meal verbs (afaldu, bazkaldu, gosaldu). In Eastern Basque (Souletin and

eastern Lower Navarrese), meal verbs have ABS subjects. However, in Central Basque,

meal verbs take ERG subjects, as in Western Basque.25 Moreover, we have seen that a

change has occurred in historical Labourdin, whereby meal verbs have taken up ERG

subjects.

Another noticeable difference between Central Basque and Eastern Basque is

that, in general, one finds more ERG subjects for lexically-simple unergatives in the

former. Also, within Central Basque, one finds on the whole more non-compound

unergatives with ERG subjects in Navarrese than in Labourdin. Even within Labourdin,

more ERG subjects are used with unergatives on the coast than in the interior. Hence,

the general conclusion is that, in non-compound unergatives, ERG subjects are more

25 Other verbs that take ABS subjects in Eastern Basque but ERG subjects in Central

Basque are jokatu ‘play, bet’, and the loan kurritu ‘run, walk, travel’ (cf. Section 8.2.1

for the latter).

51

frequent towards the (south)-west, and ABS subjects more frequent towards the (north)-

east.

Yet, there seems to be a more interesting generalization that can be made

regarding the use of ABS or ERG subjects with lexically-simple unergatives in Central

Basque. It appears that, by and large, there are two grammars available to speakers, so

that a given speaker will tend to consistently use the majority of simple unergatives

(especially the most prototypical of them, excluding recent loans) either with ABS

subjects or conversely with ERG subjects. Notice that the former grammar (ABS) is

more valence-based, whereas the latter (ERG) is more semantically-based. Exceptions

to these consistent grammars may be due to historical reasons or borrowing, or to

lexical extension. These exceptions, arguably, need to be learned by memory. In any

event, a comprehensive study of the sociolinguistic variation that non-compound

unergatives present is a matter for another work.

As already mentioned, the semantically-based grammar, with ERG subjects in

lexically-simple unergatives, is innovative in Basque. Its extension progresses

historically from Western Basque towards the east. This is the reason why more

speakers of Navarrese employ the innovative semantic system. In Labourdin (and

western Lower Navarrese), in turn, only (some) non-adult speakers have assumed a

semantically-based grammar (regardless of meal verbs). This is, nevertheless, an issue

which needs further empirical research.

Linked with the direction of the historical change just put forward is the issue of

the behavior of recent loans from Romance. The age of a given loan-verb in the

language is a factor here, since recent borrowings display a more innovative use while

older verbs have a more fixed case-marking pattern. Thus, recent loans such as biajatu

‘travel’ or nadatu ‘swim’, especially common in Eastern Navarrese (or in authors such

52

as Leiçarrague), are more open to the innovative use of the ERG marker in lexically-

simple unergatives.

10. Western Basque

Western Basque, particularly the Biscayan dialect, presents a case-marking pattern that

is quite straightforwardly semantically-based, and not ergative. Unergative verbs, not

only egin compounds but also lexically-simple unergatives, take ERG subjects (with

extremely few exceptions, if any; see Section 10.2.6 below). Furthermore, in the

Biscayan dialect some specific unaccusatives take also ERG subjects, apparently as a

further extension of the semantic reanalysis. The only problem for this analysis is that

there are few lexically-simple unergatives in Western Basque.

10.1. Guipuzcoan

10.1.1. The Ethno-linguistic Atlas of the Basque Country

There are not many non-compound unergatives elicited in the Guipuzcoan interviews in

the Atlas. These lexically-simple unergatives correspond mostly to the verbs for ‘jump’,

‘run’, and ‘fly’. Interviews were completed in fifteen localities of the Guipuzcoan

dialect26 (see Figure 4).

FIGURE 4 (MAP OF ALAVA, BISCAY & GUIPUZCOA TO GO HERE)

26 Namely: Alegria de Oria (Alegia), Asteasu, Ataun, Azpeitia, Elgoibar, Fuenterrabia

(Hondarribia), Guetaria (Getaria), Hernani, Legazpia (Legazpi), Motrico (Mutriku),

Oyarzun (Oiartzun), Renteria (Errenteria), San Sebastian (Donostia), Vidania (Bidania),

Zegama.

53

- ‘Jump’: The translation for ‘jump’ is given as lexically-simple, namely as the verb

saltatu, in nine localities of Guipuzcoa. Three of them (Hernani, Renteria —both close

to San Sebastian—, and Vidania, near Tolosa) use ABS subjects. The other six (Alegria

—also close to Tolosa—, Ataun —near Beasain—, Azpeitia, Hondarribia, Oyarzun and

San Sebastian) employ ERG subjects.

- ‘Run’: The lexically-simple loan verb korritu is used for ‘run’ in six localities of

Guipuzcoa (Ataun, Hernani, Hondarribia, Oyarzun, Renteria and Zegama). All six

employ ERG subjects, without exception.

- ‘Fly’: The verb ‘fly’ is given as lexically-simple, namely as hegatu, only in three

localities of Guipuzcoa (Ataun, Guetaria, and Oyarzun). Moreover, it is only used in the

perfect aspect, as a translation to ‘it has flown’ (otherwise, i.e. in imperfective aspect

and generally in the other localities, a compound verb is employed). The examples of

hegatu always take ABS subjects. But note that hegatu can also have the unaccusative

reading ‘start to fly, take off’, particularly when it is used in the perfect aspect.

- ‘Travel’: The recent loan biajatu is used in three localities of Guipuzcoa (Elgoibar,

Motrico —both in the west end of Guipuzcoa—, and Hernani). It always takes ERG

subjects.

- ‘Speak’: The verb ‘speak’ is never elicited as lexically-simple in Guipuzcoan. All of

the informants use the compound verb hitz egin.

- ‘Swim’: There are no lexically-simple examples of ‘swim’ in the Guipuzcoan Atlas.

All informants employ the compound verb igeri(an) egin.

- ‘Fish’: Additionally, we can mention that the verb for ‘fish’ is elicited as lexically-

simple in Elgoibar: the loan peskadu. As expected, it takes an ERG subject.

10.1.2. Historical data: The General Dictionary of Basque

54

- Meal verbs: All historical attestations of meal verbs (afaldu, bazkaldu) in the

Guipuzcoan dialect, starting in the 18th century, have ERG subjects.

- Dantzatu ‘dance’: The lexically-simple verb dantzatu is attested in a 17th-century text

from Tolosa. It is used with ABS subjects. The 18th-century author Mendiburu, from

Oyarzun, also employs ABS subjects for dantzatu. In the 19th-century Guipuzcoan

attestations, however, dantzatu is employed with both ABS and ERG subjects.

Lardizabal, from Zaldivia (near Beasain), and Soroa, from San Sebastian, use both types

of subjects; Petriarena (“Xenpelar”), from Renteria, employs ERG subjects.

- Jokatu ‘play, practice’: Guipuzcoan 19th- and 20th-century authors use the verb jokatu

with both ABS and ERG subjects; e.g. Aguirre, from Regil (between Azpeitia and

Tolosa), employs ABS subjects; Soroa, from San Sebastian, employs ERG subjects.

- Jostatu, jolastu ‘play, enjoy’: Old authors (e.g. Mendiburu, 18th century) use ABS

subjects for jostatu. Conversely, modern authors’ usage may vary; e.g. Etxaniz, from

Azcoitia (west of Azpeitia), employs ERG subjects with jostatu. (The distribution of the

verb jolastu is similar.)

- Deskantsatu ‘rest’: 18th- and 19th-century writers from Guipuzcoa employ mostly ABS

subjects with the verb deskantsatu (e.g. Irazusta, from Hernialde, near Tolosa;

Echagaray, from San Sebastian; Iturriaga, from Hernani, near San Sebastian; Alcain

“Udarregui”, from Aya, near Guetaria). However, an anonymous 19th-century text,

apparently from Azcoitia (west Guipuzcoa), uses ERG subjects with deskantsatu.

- Burrukatu ‘fight’: Guipuzcoan writers of the 19th and 20th centuries use ABS subjects

with the verb burrukatu: e.g. Beovide, from Azpeitia; Ormaechea (“Orixe”), from Oreja

(close to Tolosa); Aguirre, from Regil.

55

10.1.3. My own research

According to my observations, the following tentative picture obtains regarding the

usage of lexically-simple unergatives in current Guipuzcoan. Meal verbs (bazkaldu,

afaldu) always take ERG subjects in all varieties of modern Guipuzcoan. The same

occurs with other unergatives such as jokatu ‘play’, korritu ‘run’ and the recent loan

biajatu ‘travel’. As for the other major prototypical unergatives that are commonly used

as lexically-simple verbs (dantzatu ‘dance’, burrukatu ‘fight’, jolastu ‘play, enjoy’,

saltatu ‘jump’), there is variation regarding their subjects’ case. In eastern Guipuzcoan,

there are speakers who, quite consistently, use the majority of these verbs with ABS

subjects. This is especially true of the northeast area (Renteria, Oyarzun, Hondarribia),

which has traditionally been included in the Navarrese dialect. Other speakers of eastern

Guipuzcoan, however, employ ERG subjects for the majority of the verbs just

mentioned.

In western Guipuzcoan, in turn, all speakers use ERG subjects with the verbs

mentioned above. The only lexically-simple unergatives that can exceptionally be used

with ABS subjects in western Guipuzcoa are verbs such as hegatu ‘fly’ and deskantsatu

‘rest’, which can be both unergatives and unaccusatives. In the latter reading, these

verbs mean ‘start to fly, take off’ and ‘come to rest, calm down’, respectively.

Nevertheless, further research is needed to establish in detail the ABS / ERG isoglosses

corresponding to each of the unergative verbs under discussion.

10.2. Biscayan

10.2.1. The Ethno-linguistic Atlas of the Basque Country

Very few lexically-simple unergatives show up in the Biscayan interviews in the Atlas.

In fact, non-compound examples only occur with the verbs for ‘run’ and ‘jump’, (and

56

with ‘fish’). Interviews were carried out in twelve localities of Biscay,27 plus in three

localities of western Guipuzcoa (Anzuola/Antzuola, Mondragón/Arrasate, and Oñate /

Oñati) and one of Alava (Aramayona/Aramaio), where the Biscayan dialect is spoken

(see Figure 4).

- ‘Run’: The lexically-simple loan verb korridu is employed in five localities of Biscay

(Bermeo, Guernica, Munguia, Ochandiano and Zamudio, near Bilbao). It always takes

ERG subjects. (Otherwise, egin compounds are elicited.)

- ‘Jump’: The loan saltadu is elicited in four localities of Biscay (Bermeo, Guernica,

Munguia and Zamudio). It always takes ERG subjects. (The compound unergative salto

egin is used in all other cases.)

- ‘Swim’: Egin compounds are always elicited for ‘swim’ (mainly, uger egin).

- ‘Fly’: All informants employ egin compounds for ‘fly’ (mostly, hegaz egin).

- ‘Speak’: The verb ‘speak’ is never elicited as lexically-simple in Biscayan. All

informants use the compound verb berba egin.

- ‘Fish’: The verb for ‘fish’ is lexically-simple (arrantzatu) in Mondragón, near Oñate;

and also, in the form of the loan peskadu, in Amorebieta (west of Durango), Zamudio

(near Bilbao) and Zeberio (west of Zeanuri). Both verbs are always used with ERG

subjects.

27 Namely: Amorebieta, Bermeo, Elorrio, Guernica (Gernika), Marquina (Markina),

Munguia (Mungia), Ondarroa, Orozco (Orozko), Ochandiano (Otxandio), Zamudio,

Zeanuri, and Zeberio.

57

10.2.2. Historical data: The General Dictionary of Basque

- Meal verbs: Meal verbs are always employed with ERG subjects by Biscayan writers

(attestations start in the 18th century).

- Dantzadu ‘dance’: The lexically-simple verb dantzadu is not very common among

Biscayan authors. Still, at least the 18th-century writer Astarloa, from Durango, and the

19th-century writer Arrese-Beitia, from Ochandiano, do employ this verb. Both of them

use it with ERG subjects.

- Jokatu ‘play, practice’: The 17th-century author Capanaga, from Mañaria (near

Durango), employs ABS subjects with jokatu. Modern authors (e.g. Aguirre, from

Ondarroa; Bustinza “Kirikiño”, also from Mañaria) employ ERG subjects.

- Jostatu, jolastu ‘play, enjoy’: The verbs jostatu and jolastu are not well documented in

historical Biscayan. Nevertheless, modern Biscayan authors who use these verbs

employ ERG subjects; e.g. Gandiaga, from Mendata (near Guernica), uses ERG

subjects with jostatu; Enbeita, from Mugica (also near Guernica), and Echeita, from

Mundaca (near Bermeo), use ERG subjects with jolastu.

- Burrukatu ‘fight’: The verb burrukatu is not commonly attested in historical Biscayan.

The 18th-century author Zavala, who lived in Guipuzcoa, employs ABS subjects with

burrukatu. Currently, ERG subjects are used with this verb (see Section 10.2.4 below).

10.2.3. Extinct dialect from Alava

Since in the territory of Biscay there are very few non-compound unergatives, it is

interesting to look at the extinct dialect of Alava, located in the southwest (see Figures 1

and 4), for it basically had Biscayan-like grammar but also some Eastern lexical

influence (mostly from Pamplona). As it turns out, Old Alavese not only looks like

58

Biscayan regarding the presence of ERG subjects in unergative verbs, but may even be

more innovative in this respect.

Betolaza (16th century), from Betolaza (see Figure 4), uses ERG subjects not

only with the verbs padezidu ‘suffer’, which takes ERG subjects in most dialects, and

igo ‘ascend’, which takes ERG subjects in Biscayan, but also with the unaccusative loan

erresuzitadu ‘resurrect’. This use of the ERG marker is difficult to explain on the basis

of a patientive / non-patientive contrast (see nonetheless the discussion on directed-

motion verbs of “emission” in Section 10.2.7 below). Recall from Section 9.1.4 that the

16th-century Labourdin writer Leiçarrague employs ABS subjects with erresuzitatu. So

do, for instance, Beriayn, from Beriain, and Elizalde, from Muzqui (Muzki), both of

them 17th-century speakers of southern Navarrese (see Figure 3).

Perhaps more importantly, Gamiz (18th century), from Sabando (eastern Alava:

see Figure 4), employs ERG subjects with deskantsatu ‘rest’, hegatu ‘fly’, and mintzatu

‘speak’, verbs which generally have ABS subjects in more Eastern dialects.

10.2.4. My own research

Lexically-simple unergatives are not frequent in the Biscayan dialect. However, I have

observed in my small survey that, even when they do not ordinarily employ these verbs,

Biscayan speakers do have intuitions about some of them. That is, verbs such as jolastu

‘play’, burrukatu ‘fight’, dantzadu ‘dance’, korridu ‘run’, saltadu ‘jump’, oihukatu

‘shout’, deskantsadu ‘rest’, and biajadu ‘travel’ are not totally unfamiliar to Biscayan

speakers and may be tested among them. Psycholinguistic testing would be more

appropriate for a partly artificial situation like this. Nevertheless, I have found (without

carrying out any sophisticated experimental design) that ERG subjects are much more

readily accepted with the above verbs than ABS subjects. More precisely, while

59

lexically-simple unergatives with ERG subjects may sound unusual, ABS subjects are

totally rejected by speakers (unless some patientive reading may be found). In other

words, Biscayan speakers cannot accept ABS subjects with unergative verbs (in an

unergative reading).

This is the reason why the Biscayan speaker finds the proposal made by the

Royal Academy of Basque for Unified Basque (see below) to be very bizarre.

10.2.5. Unified Basque

The Royal Academy of Basque (cf. Euskaltzaindia 2000) has made the following

proposals for the use of unergative verbs in Unified Basque: afaldu ‘have dinner’ and

bazkaldu ‘have lunch’ (when intransitive) should take ERG subjects in the Spanish-

Basque dialects and ABS subjects in the French-Basque dialects; barautu ‘fast’ (when

intransitive) should take ABS subject; bidaiatu ‘travel’ should always take an ERG

subject; borrokatu ‘fight’ always with an ABS subject; dantzatu ‘dance’ (when

intransitive) always with an ABS subject (an ERG subject is explicitly forbidden);

deskantsatu ‘rest’ (when intransitive) always with an ABS subject; hegatu ‘fly’ always

with an ABS subject; jokatu ‘play’ (when intransitive) either with an ABS subject or

with an ERG subject; jolastu and jostatu ‘play, enjoy’ always with an ABS subject;

korritu ‘run’ always with an ERG subject; mintzatu and solastatu ‘speak, talk’ always

with an ABS subject; ; paseatu ‘walk’ (when intransitive) always with an ABS subject;

saltatu ‘jump’ always with an ABS subject.

It is fairly clear that, on the whole, the Academy has favored a rather ergative

(i.e. valence-based) norm, which is conservative and closer to the current usage in the

Eastern dialects. This norm favors the use of ABS subjects in lexically-simple

unergative verbs (especially in prototypical unergatives), since they are after all

60

intransitive verbs. The only exceptions to this claim in the above list are korritu and

(quite surprisingly; cf. Sections 8.1.1, 8.2.1 and 9.1.2) bidaiatu / piaiatu.

The Academy’s position in favor of an ergative norm is understandable, not only

because of conservatism and because it follows the major written tradition, but also

because lexically-simple unergatives are more frequent towards the East. In other

words, the Academy has proposed to use each lexically-simple unergative verb with the

kind of subject it takes in the area where it is most common. However, the Academy has

not taken into consideration the fact that there are two major case-marking grammars in

Basque: one Eastern, another Western. Thus, as expected from prescriptive rules of this

type, the speaker of Biscayan, which possesses a semantically-based (patientive / non-

patientive) case-marking grammar, intuitively perceives this valence-based norm as

odd.

10.2.6. Putative exceptions

Before concluding the discussion on Biscayan, I address the question whether there is

any exception at all in this dialect whereby an unergative verb takes ABS subjects. The

answer is that there is no clear exception to the semantically-based pattern, although we

will see some verbs taking ABS subjects in Biscayan which might be classified as

unergatives (albeit non-prototypical unergatives).

The verb mintzatu ‘talk, speak’ has been mentioned as an exception (“the only

counter-example I am aware of”, Levin 1983:307, footnote 22) to the behavior of

unergative verbs. Indeed, mintzatu is usually employed in written Basque with an ABS

subject. However, although Levin focused on Western-Basque data, she did not take

into account the dialectal issue. The verb mintzatu is not indigenous to the Western

dialects. If a speaker of Biscayan or Guipuzcoan uses this verb in writing, s/he is using a

61

loan from more Eastern dialects. Furthermore, we have seen in Section 10.2.3 that the

18th-century Alavese author Gamiz employed ERG subjects with mintzatu.

Intransitive verbs such as dibertitu ‘enjoy oneself’, mendekatu ‘take revenge’,

saiatu ‘attempt’, ahalegindu ‘take pains’, ahaztu ‘forget’, gogoratu / gomutatu /

akordadu ‘remember’, ezkondu ‘get married’ may look like exceptions to the dialectal

division I am proposing, for they do have ABS subjects in all dialects, including

Biscayan. However, note that most of these verbs are borrowings from Romance and

that the meaning of dibertitu, mendekatu, saiatu, ahalegindu, ahaztu, gogoratu and

ezkondu is close to that of the Spanish se-reflexives divertirse, vengarse, probarse,

esforzarse, olvidarse, acordarse and casarse, respectively. (Saiatu and ahalegindu are

nearly synonyms.) As pointed out in Section 9.1.4, it seems Romance borrowings have

had an influence on the semantic opposition between the two classes of Basque

intransitives towards an identification of some ABS-subject predicates (i.e. some

unaccusatives) with Romance se-reflexives (also unaccusatives, generally). Therefore,

the ABS subject of the above verbs is not exceptional from that perspective.

The verb ibili ‘walk, move (undirected)’ is quite idiosyncratic in Basque.

Although the primary meaning of this frequently-used verb seems to involve an

unergative activity, in truth its multiple readings are so unpredictable that it can also

mean something like ‘be busy doing something’ or just ‘be (around)’. At any rate, its

ABS subject in Biscayan (as in all dialects) is not straightforward from a semantic

perspective, and its high frequency of occurrence might be invoked to explain its odd

behavior, via conservatism (ibili belongs to the handful of verbs that have synthetic

forms). The closest equivalents of ‘walk, stroll’ in the Western dialects are paseatu or

buelta bat eman. As predicted, the lexically-simple unergative verb paseatu has ERG

subjects in Biscayan.

62

The modal verb hasi ‘begin’ is perhaps the verb that looks closest to an

exception to the general behavior of unergatives in Biscayan. Hasi can be transitive, but

its typical use is intransitive, usually with a clausal modifier. When it is intransitively

used, hasi takes ABS subjects in Biscayan (and elsewhere). Note that its antonym

amaitu ‘finish’ takes ERG subjects even when it is intransitive. If hasi is a non-

patientive verb (it is certainly not a clear-cut case of non-patientivity), then it would

constitute an exception to the generalization presented above for Biscayan. Note that

hasi is also a high-frequency verb.

10.2.7. Unaccusative directed-motion verbs of “emission”

I have already mentioned that there are a handful of lexically-simple verbs which

presumably should be classified as unaccusative, but which nonetheless take ERG

subjects in Biscayan, unlike in other dialects. Besides recent loans such as existitu

‘exist’ and eboluzionatu ‘evolve’, already discussed in Section 4.4 above, the most

important verbs in this list are urten ‘go out, exit’ and igon ‘go up, ascend’.

The verb urten takes ERG subjects not only in Biscayan but also in western

Guipuzcoan, where it has the form irten / erten. The Ethno-linguistic Atlas of the

Basque Country gives ERG subjects for urten in all localities of the Biscayan dialect

(including those in Guipuzcoa and Alava), plus in at least Azpeitia, Legazpia —between

Oñate and Zegama—, and Zegama, of the Guipuzcoan dialect. The verb igon, in turn,

only takes ERG subjects in Biscayan, although the Atlas gives also an ERG subject for

Legazpia (igo). To these two verbs, we could add the verb erresuzitadu ‘resurrect’,

which we found with ERG subjects in Old Alavese (see Section 10.2.3).

It is not easy to explain these data. The simplest hypothesis would be to propose

that each of these three verbs takes ERG subjects independently, for some idiosyncratic

63

reason. (I have already mentioned in Section 4.2 that ‘ascend’ can be atelic in many

cases.) However, I will tentatively propose a hypothesis that will try to account globally

for the exceptionality of the three verbs just mentioned, and also for that of a few egin

compounds which are (in principle) unaccusatives too (see Section 6.1). The hypothesis

will assume that the use of ERG subjects in Biscayan and adjacent varieties for the

verbs urten, igon and erresuzitadu is due to a further, more innovative, extension of the

semantic reanalysis that the ERG marker has undergone in Western Basque. That is, the

reanalysis of the ERG marker has reached in these varieties not only pure unergatives

but also a few unaccusatives that for some reason resemble unergatives. In what follows

I attempt to explain what reason that may be.

If we could find a semantic relationship between ‘go out’, ‘go up’ and

‘resurrect’, we could propose that the extension of the ERG marker has occurred in a

specific semantic sub-class of unaccusatives. Finding a semantic link between these

three verbs is not straightforward, though. If (speculatively) ‘up’ could be cognitively

associated with ‘out from a reference point’ (vs. ‘down’ associated with ‘towards a

reference point’), one could propose that the three verbs above can roughly be classified

as directed-motion verbs of “emission”. The hypothesis, then, would only need to argue

that directed-motion verbs of “emission” are semantically closer to unergatives than

other unaccusatives, such as verbs of arrival. In fact, we have already seen in Section

4.2 that processes of emission are among the most prototypical of non-patientive

intransitives.

A small piece of evidence in favor of this hypothesis comes from the following

unaccusative verbs which, unexpectedly, are egin compounds: alde egin / ospa egin

‘leave’, ihes egin ‘flee’. Note that, according to the hypothesis just presented, these

three verbs may also be classified as directed-motion verbs of “emission”. If this is

64

correct, all six verbs discussed above, even though basically unaccusatives, are closer to

unergatives because they can be conceptualized as “emission” verbs.

In conclusion, if we accept that the six verbs at hand have some cognitive

resemblance to unergatives, all we need to assume further is that the lexically-simple

verbs in this sub-class (urten, igon, erresuzitadu) have taken on an ERG subject as a

further extension of the reanalysis of the ERG marker. Of course, this would only

happen in Western varieties of Basque. This is the reason why verbs such as atera /

jal(g)i / elkhi ‘go out’, which are not employed in Biscayan, always have ABS subjects.

10.3. Conclusion: Western Basque

The dialectal data on lexically-simple unergatives presented in the preceding section

corroborates the conclusion that by and large Western Basque, especially Biscayan, has

a semantic case-marking system, which is based on the contrast patientive / non-

patientive. This system has emerged as the reanalysis of an erstwhile ergative pattern.

Thus, currently, ABS subjects correspond to patientive predicates (unaccusatives), and

ERG subjects to non-patientive predicates (either to lexically-simple or compound

unergatives, or, of course, to transitive verbs).

Biscayan and Guipuzcoan have very few exceptions to this patientive / non-

patientive pattern. In eastern Guipuzcoan, the exceptions are a handful of unergative

verbs that may (still) take ABS subjects. In Biscayan, by contrast, the main exceptions

pertain to some unaccusative predicates which may take ERG subjects. Some of these

exceptions (e.g. existitu, eboluzionatu) are easier to explain, as they correspond to recent

loans that are not prototypically patientive. Other exceptions, such as urten ‘go out’ and

igon ‘go up’ are more recalcitrant.

65

In any event, it is fairly clear that Biscayan is the most innovative dialect in that

the ERG marker has extended its use to all unergative subjects, and it may be

undergoing further extension to non-prototypical patientive contexts. The prototype-

based approach assumed in this work, therefore, is able to handle not only typical cases

but also the variation created by non-prototypical ones. As in other dialects, the age of a

given verb is also a factor here. Recent borrowings may display a more innovative

behavior, while old high-frequency verbs may be more conservative.

To conclude it may be worth recalling that there are few lexically-simple

unergatives in Western Basque. The majority of unergatives present a compound

structure with the light verb egin. Actually, it might be the case that the massive

presence of egin compounds in Western-Basque unergatives might have helped trigger

the use of ERG subjects for lexically-simple unergatives too. The scarcity of

(indigenous) lexically-simple unergatives might have been a factor for the more

innovative behavior of the ERG marker. Conversely, in Eastern Basque, where more

(indigenous) lexically-simple unergatives are present, there might have been a pressure

towards maintaining the ergative pattern.

11. Conclusion: the case-marking system of Basque

11.1. The present paper

In this article, I have shown that defining the case-marking pattern of Basque is quite a

complex task. All dialects of Basque make, one way or another, a semantic distinction

within intransitive predicates. The first issue that needs to be addressed is the specific

nature of this semantic distinction. I have argued in Section 4 that the semantic contrast

between Basque intransitives is best captured by a prototype-based approach, and that it

can be defined as opposing patientive intransitives (unaccusatives) vs. non-patientive

66

intransitives (unergatives). Although more research is needed towards a precise

definition of patientivity, I have presented above its basic lines.

A majority of Basque unergatives are lexically-compound, many of which are

hosted by the light verb egin ‘do’. Egin compounds take ERG subjects in all dialects of

Basque; yet it seems that the reason for having ERG subjects is their transitive-like

structure. Lexically-simple unergatives, therefore, constitute the main object of study

for determining the Basque case-marking pattern. In Sections 8, 9 and 10, I have shown

that the patterning of lexically-simple unergatives varies a great deal from Eastern

Basque to Western Basque. Roughly, lexically-simple unergatives take ABS subjects in

Eastern Basque, while they take ERG subjects in Western Basque.

It is fairly safe to assume that a reanalysis has occurred in Western Basque,

whereby the ERG marker has extended its applicability and taken up the function of

marking non-patientive subjects, not only transitive but also intransitive. This reanalysis

is extending towards the Eastern area. We can conclude, then, that Western Basque has

a semantic case-marking pattern, which is based on the contrast patientive / non-

patientive. Eastern and Central Basque, on the other hand, have in the main an ergative

case-marking system, although particularly in Central Basque there are a fair number of

exceptions where lexically-simple unergatives take ERG subjects.

11.2. Prospects

The dialectal data on lexically-simple unergatives presented in the present paper

constitute a very interesting object of study for further research. Moreover, the data are

worth examining from different angles. First, a detailed sociolinguistic study can be

carried out in order to establish better the variation that occurs between Basque dialects

regarding the case-marking of specific lexically-simple unergatives. Secondly, on a

67

more theoretical level, the data can serve as the basis for testing models of lexical

diffusion which may account for the way the gradual reanalysis of the Basque system

progresses. We already know that the Basque unaccusative / unergative contrast is

governed, especially in the Central transitional area, not only by a semantic distinction

but also by secondary factors related to conservative vs. innovative behavior. Thus,

thirdly, a psycholinguistic (synchronic) model is needed to explain how Basque

speakers learn and handle exceptions to the semantic distinction that governs the bulk of

the unaccusative / unergative contrast. This model should leave room for the non-

discreteness that may be caused: a) by verbs that are neither prototypically patientive

nor prototypically non-patientive, and b) by the innovative patterning of recent loans

and the conservative patterning of old high-frequency verbs.

11.3. Eastern Basque’s ergative marking vs. Western Basque’s semantic marking

In the light of the dialectal variation that the Basque data display, the question arises as

to whether the ergative pattern of Eastern and Central Basque is significantly different

from the semantic pattern of Western Basque. This is a complicated question.

I have argued above that by and large two main grammars can be defined for

Basque case-marking. One of these grammars marks the majority of prototypical

lexically-simple unergatives with ABS subjects; the other grammar marks the majority

of prototypical lexically-simple unergatives with ERG subjects. Speakers of Eastern

Basque and most speakers of Central Basque use the former grammar. Speakers of

Western Basque use the latter. Thus, I argue that there is a noticeable difference

between Eastern and Western Basque case-marking. Furthermore, this difference can

intuitively be perceived by speakers of Basque. Therefore, while we can still label

68

Eastern-Basque grammar as ergative (valence-based), we have to define Western-

Basque grammar as semantically-based.

However, even within the most consistent valence-based grammar of Eastern

Basque, we need to take into account that the following intransitive clauses have ERG

subjects: a) unergative egin compounds, b) derived intransitive (unergative) readings of

prototypically transitive verbs (see Sections 5.2.1 and 5.2.3 above, and Section 11.4

below), c) erstwhile-transitive unergatives (see Section 8.1.3, and note 24), d) some

unergative loans, e) one or two inanimate unergatives. This leaves us with only a few

lexically-simple unergatives on the basis of which to determine whether Eastern Basque

is ergative or semantically-based. It is true that the most prototypical unergatives of

Eastern Basque are included within these few lexically-simple verbs, and that

practically all of them take ABS subjects. But it is also true that there are not many

more lexically-simple unergatives with ABS subjects in Eastern Basque than the ones I

have discussed throughout this paper. This is certainly a very small number of

unergative verbs taking ABS subjects.

Note, additionally, that in Central Basque there are even more exceptions to a

valence-based case-marking grammar. These exceptions make the grammar of Central

Basque quantitatively different from that of Eastern Basque. Thus, I would argue that

the differences in case-marking between Eastern and Central Basque have to do only

with the assignment of specific lexical verbs. Conversely, I would argue that the

differences between Central and Western Basque are more significant.

Yet, a legitimate question arises as to whether a small amount of unergative

verbs suffices to measure when a divergence in case-marking is only due to lexical

assignment differences and when it is due to differing grammars. (As a matter of fact, if

mutual unintelligibility exists between Eastern and Western Basque, it is not due to their

69

different case-marking patterns as much as it is due to their differences at the

phonological, morpho-phonological and lexical levels.)

In conclusion, I propose that the case-marking system of Eastern Basque can be

considered ergative, but in some respects it lies not too far from a semantically-based

system. Another way of looking at this issue is by proposing that the Eastern-Basque

case-marking pattern is not “typically ergative”. I address this question below.

11.4. Towards a typology of ergative languages

One of the characteristics that are typical of ergative languages is the presence of

antipassives. Even if the definition of “antipassive” is not totally established, I will refer

by this term to the use of ABS subjects in “derived” transitive clauses with mass, non-

specific (covert), non-referential, incorporated, or indefinite objects. As it turns out, we

do not find in Basque, not even in Eastern Basque, genuine antipassives. Although we

have seen in Section 5.2.2 that there exist in Eastern and Central Basque a small number

of constructions which may resemble antipassives (cf. (5) above), in truth these

correspond to some canonically intransitive verbs that may be used (in specific cases) as

transitive, by taking a somewhat peculiar direct object NP in the ABS case.

In the most typical ergative languages (Australian, Polynesian, Eskimo, etc),

antipassives apply to canonical transitive verbs, and transform prototypical transitive

sentences of the type ERG-ABS into derived intransitive sentences of the types ABS-Ø

or ABS-OBL. For instance, a verb such as eat would change its ERG subject in a

prototypical transitive clause, such as The man ate the apple, into an ABS subject in

sentences with non-specific object omission, as in The man ate, or with a non-referential

object as in The man ate apples. As we have seen in Section 5.2.1, nothing like this

occurs in Basque. All verbs that are considered to be canonically transitive always have

70

ERG subjects when they have an agentive non-patientive subject, regardless of the kind

of object they may have, or even if they do not have any overt object. This happens in

all dialects of Modern Basque, and indeed also in Old Basque. Note that egin

compounds are after all transitive-like constructions with a non-referential incorporated

object, and yet they take ERG subjects.

The conclusion, therefore, is that the case-marking system of Eastern Basque is

quite different from the systems of the typical ergative languages. In Aldai (2007), I

dubbed the Eastern-Basque-type system “loose ergative”, along the lines of Harris

(1985:125). The main difference between a loose ergative system and the strict ergative

systems of typical ergative languages is that the latter display ERG / ABS antipassive

alternations according to the individuation (referentiality, definiteness) of direct objects,

whereas the former has a fixed ERG marking regardless of the degree of individuation

of objects. Besides this, or precisely as an extension of it, the system of Eastern Basque

presents also ERG subjects in egin compounds and in a few lexically-simple

unergatives. In other words, the ERG marker is very generalized and liberally used not

only in Western Basque but even in Eastern Basque.

Some scholars (cf. Trask 1997, 2002) have emphasized the fact that Basque has

“thoroughgoing ergative morphology” and fails to “exhibit any kind of split” (2002:

282). While this statement is descriptively correct, it seems to imply that Basque is

“very ergative”. However, the following paradox should be noted here. The massive

generalization of ergative morphology is in fact an indication of “loose ergativity”.

Strict ergative systems are those that use a very restricted concept of transitivity. The

ERG marker is only employed in selected contexts, such as transitive clauses with

individuated objects. Conversely, the generalization of the ERG marker to low-

transitivity contexts, such as transitive clauses with non-individuated or omitted objects,

71

leads to loose ergative systems. From here, a further extension of the ERG marker to

intransitive clauses may lead towards a “more semantically-based” pattern.

Thus, I propose that the case-marking pattern of Eastern Basque can still be

considered ergative, but also that it should be considered “loose ergative”, or “more

semantically-based” than the pattern of the typical ergative languages.

References

Aldai, Gontzal. 2007. “From ergative case marking to semantic case marking: the case

of historical Basque”. In Mark Donohue and Søren Wichmann (eds.), The

typology of semantic alignment. Oxford: Oxford University Press. 197-218.

Aranzadi Zientzia Elkartea. 1983. Euskalerriko Atlas Etno-linguistikoa (EAEL), Lehen

zatia [Ethno-linguistic Atlas of the Basque Country, 1st part]. San Sebastian:

Aranzadi Zientzia Elkartea.

Aranzadi Zientzia Elkartea. 1990. Euskalerriko Atlas Etno-linguistikoa (EAEL),

Bigarren zatia [Ethno-linguistic Atlas of the Basque Country, 2nd part]. San

Sebastian: Aranzadi Zientzia Elkartea.

Dixon, R.M.W. 1994. Ergativity. Cambridge: Cambridge University Press.

Dowty, David. 1991. “Thematic proto-roles and argument selection”. Language 67:

547-619.

Euskaltzaindia. 2000. Hiztegi batua [Unified lexicon]. Bilbao: Euskaltzaindia = Royal

Academy of the Basque Language.

Harris, Alice. 1985. Diachronic syntax: The Kartvelian case. (Syntax and Semantics,

vol. 18.). Orlando: Academic Press.

Hualde, José Ignacio & Jon Ortiz de Urbina (eds.). 2003. A grammar of Basque. Berlin /

New York: Mouton de Gruyter.

72

Lafitte, Pierre. 1962. Grammaire basque: navarro-labourdin littéraire. 2nd edition (1st

edition 1944). Reprinted by Ikas / Elkar, Bayonne / San Sebastian, 1979.

Lafon, René. 1975. “Indices personnels n’exprimant rien de déterminé dans les verbes

basques”. In Mélanges linguistiques offerts á Emile Benveniste. Paris: Société

Linguistique de Paris. 331-337.

Levin, Beth. 1983. On the nature of ergativity. PhD dissertation, MIT.

Levin, Beth. 1989. “The Basque verbal inventory and configurationality”. In Laszlo

Maracz & Pieter Muysken (eds.), Configurationality: The typology of

asymmetries. Dordrecht: Foris Publications. 39-62.

Michelena, Luis. 1987. Palabras y textos. Bilbao: University of the Basque Country.

Michelena, Luis. 1987-2005. Diccionario General Vasco – Orotariko Euskal Hiztegia

[General Dictionary of Basque]. 16 volumes. Bilbao: Euskaltzaindia = Royal

Academy of the Basque Language.

Mithun, Marianne. 1991. “Active/Agentive case marking and its motivations”.

Language 67: 510-546.

Ortiz de Urbina, Jon. 1989. Parameters in the grammar of Basque. Dordrecht: Foris

Publications.

Oyharçabal, Bernard. 1992. “Structural case and inherent case marking: Ergaccusativity

in Basque”. In Joseba A. Lakarra & Jon Ortiz de Urbina (eds.), Syntactic theory

and Basque syntax. San Sebastian: Diputación Foral de Guipuzcoa. 309-342.

Salaburu, Pello. 1992. “Euskara, hizkuntza ergatiboa ote da?” [= Is Basque an ergative

language?]. In Luis Villasanteri omenaldia. Iker 6. Bilbao: Euskaltzaindia =

Royal Academy of the Basque Language. 417-433.

Trask, R.L. 1979. “On the origins of ergativity”. In Frans Plank (ed.), Ergativity:

Towards a theory of grammatical relations. London: Academic Press.

73

Trask, R.L. 1997. The history of Basque. London / New York: Routledge.

Trask, R.L. 2002. “Ergativity and accusativity in Basque”. In Kristin Davidse &

Béatrice Lamiroy (eds.), The nominative & accusative and their counterparts.

Amsterdam: John Benjamins. 265-284.