Dative ( first ) complements in Basque

This article examines dative complements of unergative verbs in Basque, i.e., dative arguments of morphologically “transitive” verbs, which, unlike ditransitives, do not co-occur with a canonical object complement. We will claim that such arguments fall under two different types, each of which involves a different type of non-structural licensing of the dative case. The presence of two different types of dative case in these constructions is correlated with the two different types of complement case alternations which many of these predicates exhibit, so that alternation patterns will provide us with clues to identify different sources for the dative marking. In particular, we will examine datives alternating with absolutives (i.e., with the regular object structural case in an ergative language) and datives alternating with postpositional phrases. We will first examine an approach to the former which relies on current proposals that identify a low applicative head as case licenser. Such approach, while accounting for the dative case, raises a number of issues with respect to the absolutive variant. As for datives alternating with postpositional phrases, we claim that they are lexically licensed by the lower verbal head V.

identity of their morphological patterns of case marking and agreement with those of ditransitive configurations.The following transitive sentence can be used to introduce the basics of Basque case-related marking: (1) Nik Joni etxera dirua eraman d-io-t I.ERG Jon.DAT home money(ABS) carry PRS-3DAT-1ERG 'I have carried the money home to Jon' The verbal auxiliary diot contains a first person ergative -t suffix for the subject, a zero third person absolutive mark for the singular object and an -iomarker probably containing a third person singular dative mark -o-plus a predative flag -i-.Like many other languages (French and Spanish, for example), Basque shows strict person constraints on the types of possible ditransitive configurations like these; in particular, other than a few historically attested counterexamples, ditransitives are only possible with a third person object (as in I sold it to you) but not with any other person (as in You sold me to him).This means that, morphologically, all ditransitives are third person direct object transitives.Now consider a verb configuration like that of (2): (2) Nik neure asmoari eutsi d-io-t I.ERG my plan.DAT hold PRS-3DAT-1ERG 'I've held on to my plan' There are only two arguments associated with the predicate eutsi 'hold on to', the ergative subject nik 'I' and the dative complement.Notice that, as indicated in the morpheme-to-morpheme discussion of the ditransitive morphology of (1) above, the third person absolutive marker is zero.A consequence of this fact is that the auxiliary pattern in (2) is identical to that of (1), even though we may want to say that the ditransitive auxiliary pattern has a zero object mark while the bivalent unergative in (2) has no object mark.For completeness sake, unergative predicates with a dative argument clearly differ from unaccusative bivalent predicates like the following (3): (3) Ni lanari lotu n-a-tza-io I.ABS work.DAT buckle-down 1ABS.ROOT-3DAT 'I've buckled down to work' In nominative-accusative languages like Spanish or English, where there is no auxiliary selection based on transitivity, there is no drastic morphological difference between the unaccusative and the unergative bivalent predicate: they are both subject/indirect object configurations.But in an ergative language like Basque with very clear difference between "transitive" and "intransitive" marking and auxiliary selection, the difference is quite evident.However, in that case, the bivalent unergative (2) is formally identical to the ditransitive (1), so it is little wonder that these configurations did not attract the attention of Basque grammarians, who lumped them together with regular ditransitive constructions with two object arguments.Bivalent unergative structures are found with verbs like ekin 'to engage in', eragin 'to push', heldu 'to hold on to something', begiratu 'to look at', itxaron 'wait for', etc. (see Fernández & Ortiz de Urbina (2010) for a detailed description of these verbs).Abstracting away from the ergative morphology and from the case or postpositional nature of Basque dative marking, such configurations might be akin to predicates of the type of hablar 'talk to' in Spanish or talk to, and listen to in English.We will identify different sources for the dative marking of such complements on the basis of different alternation patterns they present.While not all bivalent unergatives participate in alternations involving the case of their complement, most do, although not necessarily in the same dialect.In as much as possible, we identify which variants are dialectal and which co-occur within the same dialect, the most interesting case.In section 2 we begin with a set of dative--taking unergatives whose members largely coincide with single complement dative-taking predicates on other languages (see Blume 1998 for an early analysis based on verbs with comparable meanings from seven languages).We will show that complements to these predicates typically display dative/absolutive case alternations in Basque and explore some of the issues arising from an inherent case analysis for the dative alternant.In section 3 we present a different set of bivalent unergatives, making up a fairly coherent class of 'aspectual' predicates.These are cross-linguistically less prominent than the first set and display dative/postpositional phrase alternations.Conflation analyses for datives are difficult to motivate for this class, and we suggest that lexical case as in Woolford (2006) may account for these datives.

Dative/absolutive alternations: inherent dative
Many bivalent unergative structures alternate with standard transitive configurations, so that what looks like the same argument appears sometimes marked dative and sometimes looks like a typical object, marked absolutive.Predicates like abisatu 'warn ', begiratu 'look', bultzatu 'push', deitu 'call', entzun 'hear', eskertu 'thank' itxaron 'wait', lagundu 'help' or ukitu 'touch', many of them also displaying dative first complements in other languages (Blume 1998), can be placed in this group.All of them show dialectal variation (absolutive vs. dative) in their complement case, and in some cases both markings can co-occur in the same dialect for what looks like the same type of complement of a given verb.Most are agentive verbs, as expected for unergatives, but object marking does not seem to be sensitive to the presence of agentive vs. non-agentive subjects.Rather, factors such as telicity have been claimed to correlate with the dative/absolutive marking of the complement, at least in the case of some predicates.We turn to these cases first in section 2.1, addressing in section 2.2 the question of actual case checking configurations that might be involved for each case.We first outline a possible analysis of the dative as an inherent case licensed by a low applicative head, using conflation as an ancillary hypothesis.Finally, in section 3 we propose that datives alternating with postpositional phrases actually involve lexical case licensing by the verbal head V.

The distribution of dative and absolutive complements
Following Hopper and Thompson's (1980) account of transitivity as a clause-level phenomenon, Blume (1998) identifies dative objects as indicating low transitivity departing from a standard agent/patient argument configuration.According to her analysis, such objects occur in predicates indicating complex events where "each participant in the complex event is independently active in at least one of the subevents" (Blume 1998: 254).This is most clear in the case of predicates like lagundu 'help', and, less clearly, itxaron 'wait', deitu 'call' and the like, on the assumption that the participant helped, waited upon or called is independently involved in another subevent.In the case of contact verbs like bultzatu 'push' or ukitu 'touch', one may claim that the 'pushed' participant is forced to participate in a moving event independent of the pushing, but it is difficult to extend this analysis to the object of ukitu 'touch'.Svenonius (2002) also relies on subevents to account for dative nominals in Icelandic, but these relate to tense, aspect and Aktionsart.In fact, Etxepare (2003) claims that at least some Basque predicates convey different aspectual nuances associated with case choice.In this section we will examine these purported aspectual differences associated with dative or absolutive objects, trying to see if any single aspectual type can be associated with either case.As we will show, this is not an easy task, making it difficult to extend the aspect-based analysis to non-alternating predicates.This suggests that the source of unexpected dative case for the first complement of bivalent unergatives may lie somewhere else.This will be discussed in section 2.2.Let us first try and see whether a uniform aspectual account of the dative complement of this first set of predicates can be obtained.
According to Etxepare (2003), in a verb like bultzatu 'push' the dative signals in some dialects a non-affected reading of the object akin to that found with the objects of conative constructions in English.The allative phrase introduces a telic delimitation on the predicate which is compatible with the absolutive but not with the dative: (4) a. Mikel bazterrera bultzatu du Mikel(ABS) corner-to push '(S/he) has pushed Mikel' It is not clear at present how widespread this aspectual difference might be, both lexically and dialectally.Western dialects which require dative marking for the object admit the allative in (b).In any event, where it exists, the distinction seems far from systematic, and in this section we will concentrate for the most part on the previous predicate and on the following two alternating verbs also described in Etxepare (2003) 'Some unknown people have helped/accompanied us' With deitu 'call' in ( 5), speakers of central varieties like Etxepare find an interpretation with the object marked absolutive whereby Xabier is called to go to somebody else's office, whereas the dative marked configuration only means that they have called him at his office.Western speakers, on the other hand, use the dative alternant in (5b) with both meanings.Etxepare (2003) also claims that lagundu 'help' in ( 6) is interpreted only as 'help' with the absolutive object (6a), while it is ambiguous between a 'help' and an 'accompany' interpretation in (6b) with a dative object (here reflected through the agreeing inflection).Again, in non-monitored western varieties only the dative marked object is used, also with both interpretations.
It is not easy to find a single thread unifying the semantic factors involved in the case alternation of even this small set of verbs.Telicity and causation come immediately to mind but, as we will see below, neither can be extended to subsume the nuances found in all three predicates.With bultzatu 'push' and deitu 'call', the allative (bazterrera 'to the corner' and bulegora 'to the office') is interpreted as the resulting location, although in the case of bulegora deitu 'call to the office' with absolutive object, it does not have the delimitating effect of measuring out the event, or at least the delimitation is not on the calling event, but on some subevent whose syntactic correlate (if any) is unclear.Even so, it is unlikely that this approach can be extended to lagundu: although there has been no in-depth study of its dialectal distribution, even if it were the case that the 'accompany' interpretation was found with delimitating expressions ('accompany home'), it would not fit the previous pattern, since this interpretation is precisely found with the dative marking only, rather than with the absolutive as in the previous two verbs.The causative-like interpretation, the 'alternating' one for this predicate, is not apparently affected by telicity nuances.Alternatively, rather than the aspectual make-up of these predicates, one could capitalize on the presence of a causative layer of meaning.Help verbs typically take as complements actions carried out by the helped participant, and their syntax often mirrors that of more standard causative predicates.But, again, there is no correlation between the presence of a causative semantic/syntactic layer and any particular case: the basic causative-like meaning of lagundu 'help' co-occurs with both absolutive and dative, and the same would hold true of bultzatu 'push'.Positing for bultzatu 'push' and deitu 'call' a causative component different from that of other activity verbs not showing this alternation seems quite contrived.Bultzatu 'push' is a regular agentive verb not participating in the inchoative/causative alternation, and while deitu 'call' in 'call to the office' can be paraphrased as 'make go' it is far from clear that the causative element is part of the lexical meaning of the verb.The conclusion we draw is that even within the dialect described in Etxepare (2003) it is perhaps not viable to provide unified general accounts for this alternation on the basis of telicity or causative layers.
If we examine these items from a broader perspective, we also encounter different patterns converging into the present distribution.A look at the information on dialectal and diachronic distribution of a verb like deitu 'call' contained in the Orotariko Euskal Hiztegia (OEH, Basque Comprehensive Dictionary) shows that while the dative complement was and is pervasive in all dialects, an innovation in the central Gipuzkoan area introduced the absolutive variant; in that area the alternation can be found in the same speakers.In fact, OEH provides one attestation of the two uses in the same sentence by the same author: (7) Beregana askotan deitu gaitu eta orain ere deitzen him-toward often call 1ABS-ROOT-(3ERG) and now also call.IMPF digu

PRS-1DAT-(3ERG)
'He often calls us to him and is also now calling us' (Gco I 409) Notice that the two variants fit well with the nuance mentioned by Etxepare (2003), since the monotransitive pattern occurs with an allative phrase, while the bivalent unergative occurs with a more bare complement (in both cases pro is retrieved via the inflection/clitic).On the other hand, the verb lagundu 'help' belongs to a group of predicates which in pre-18th century texts occur mostly as bivalent unaccusative, i.e., with absolutive subject and dative complement (see Mounole 2009Mounole , 2011)); from that time, it shows a north/south (or east/west) dialectal split, with absolutive complements in the north and dative complements in western areas, with occasional overlappings.Finally bultzatu 'push' shows yet a different pattern, since it has been attested mostly from the 19th century in the central area; an important aspect that we will capitalize on later in this article is the fact that there exists a noun+light verb construction where the noun dei 'call' and the noun bultza 'push' combine with light egin 'do'.In fact, quite a few of the examples with dative marking of the verb bultzatu 'push' gathered in the OEH dictionary show either the noun itself with light egin 'do' or a form identical in shape to the noun which the OEH (V: 619--20) identifies as verbal (a 'participle'), and where no missing light verb seems to be understood.The following two examples contrast an instance of a regular imperfective participial bultzatzen (in (8a), found in the same sentence and with the same marking as imperfective laguntzen) and an attestation of the 'defective' form bultza (8b): (8) a. Nik biotz guztiz laguntzen deutsat, / bultzatzen deutsat I.ERG heart all.INSTR help.IMPF PRS.3DAT.1ERGpush.IMPF aidera.air-to 'I help him with all my heart, push him to the air' (Gand Elorri 88) b.Ori aditu zuanean, bultza zion Txirritak kaxuelari that hear AUX.when push PST.3DAT.(3ERG)Txirrita.ERG pot.DAT 'When Txirrita heard that, he pushed the pot' (Salav 46) We will return to these data below in the context of the discussion on conflation analyses and dative marking.
For all three verbs examined above and also for others traditionally taking dative complements, there has been an increase in the use of absolutive marking beginning from the 20th century, as in (9a).This may be attributed to a complex interaction between traditional dialects, the contemporary standard dialect, and Spanish.Thus, a verb like begiratu in the meaning 'look at' takes dative or allative complements in all historical dialects (9b): (9) a. Liburua itxi eta azala begiratu dut berriro book.ABS close and cover look-at PRS.ROOT.1ERGagain 'I closed the book and looked at the cover again' ( Gauza txikien liburua, P. Aristi, 90) b.Gereziondoari luzaz begiratu nion cherry-tree.DAT long look-at 1ERG.3DAT.PST 'I looked at the cherry tree for a while' ( Anaiaren azken hitzak., D. Landart, 189) However, absolutive complements as in (9a), only occasionally found in western dialects before last century, are easily encountered in contemporary standard texts with inanimate objects: ibaia, mapa, mundua begiratu dut 'I've looked at the river, map, world'.Similarly, with a verb like itxaron 'wait for', mostly found in southern dialects, OEH (IX: 693) notes that "the complement appears in the dative if it is animate; otherwise it can be absolutive or dative", and it is not uncommon to find advise to use dative with animates and absolutive with inanimates: zuri, nori, Claudineri itxaron 'wait for you, for whom, for Claudine' with dative as opposed to txanda, eguna, soldata itxaron 'wait for the turn, for the day, for the salary'.This may be connected with the existence of differential object marking (DOM) patterns in Spanish, which marks with dative morphology animate direct objects.Hypercorrection to avoid what is perceived as a Spanish-influenced DOM-like dative marking may have contributed to the increase in absolutive marking mentioned before.Remember that unergative bivalent patterns with dative complements are only found with a small set of verbs; most bivalent predicates taking transitive morphology display the monotransitive ergative/absolutive configuration.Animate dative objects with these verbs are occasionally found in colloquial registers (not necessarily in 'new' speakers), but are also stigmatized in prescriptive contexts (including schools).Notice also that unergative bivalent configurations are not recognized in traditional grammatical descriptions of Basque and are often lumped together with ditransitive patterns with identical morphology.As a result, dative complements of the verbs under discussion here, without accompanying absolutive arguments, become "suspicious", while absolutive complements are not.This may be one of the factors accounting for the spread of absolutive patterns coinciding with the spread of Basque literacy and the wider reach of prescriptive grammar from the second half of the 20th century, as a result of the development of a standard dialect and the incorporation of Basque grammar into the school curriculum.Let us turn now to a discussion of the case marking configurations themselves.

Case checking configurations
In the context of the preceding sociolinguistic sketch, it should come as no surprise that traditional prescriptivist Basque grammar actually converges with contemporary approaches in the analysis of bivalent unergative predicates as essentially surface variants of ditransitive patterns.A common explanation for the dative marking of the object hari 'to him' in hari begiratu diot 'I've looked at him' is that there is an unexpressed (first) object complement 'something', so that the dative falls under the more common "second" complement type.This intuition is supported by syntactic tests which show that these datives pattern with indirect objects rather than with direct objects.For instance, unlike direct objects, these seemingly "first" datives cannot license secondary predicates (see McFadden 2004 for German, Pylkkänen 2008 for English and Fernández & Ortiz de Urbina 2009 for Basque).The 'first-is-second' approach is congruent with Torrego's (2010) analysis of animate objects of agentive verbs taking dative morphology in Spanish, which builds upon the conflation analysis developed by Hale & Keyser (1993, 2002).At least in some cases, this approach even has some diachronic plausibility in the case of Basque.We will return to NP+egin 'do' construction below, but, as Mounole (2009Mounole ( , 2011) ) indicates, some of the verbs taking this configuration are historically N-V compounds, where the noun would be the first argument of the verb.Thus, itxaron 'wait for' derives from hitz edun 'word hold', and iguriki (also meaning 'wait for' and which up to the 18th century took dative complements) derives from egun eduki 'day hold'.These data are of course suggestive but not demonstrative; the fact that the latter verb evolved into an absolutive taking monotransitive pattern from the 18th century on indicates that, independently of the diachronic origin, the synchronic structure need not coincide with the original one.
Torrego (2010) revamps Torrego's (1998) analysis of agentive verbs taking dative-like objects in Spanish.A sentence like (10) would contain a structure as in ( 11): (10) Jon contrató a su hermano The dative-like animate object marked with the preposition a would in fact be a fairly canonical second complement, bearing an inherent dative case associated with the applicative head and bearing a goal/beneficiary role with respect to the first complement of the applicative head.This analysis can be easily applied to the Basque dative complements under discussion, making them also canonical second complements.This can derive support from the NP+egin 'do' construction mentioned above.There is a sizable group of Basque verbs, mostly agentives corresponding to unergative in many other languages, which are expressed by this periphrasis in Basque: hitz egin 'word do, speak', irri egin 'laugh make, laugh', alde egin 'part make, leave', so egin 'gaze make, watch', dei egin 'call make, call', etc. (Ortiz de Urbina 1989, Uribe-Etxebarria 1993, Laka 1993, Bobaljik 1993, Fernández 1997, Etxepare 2003, among many others).The subject bears ergative case and the noun is not a fully independent syntactic constituent, even failing to be marked with a determiner, a very uncommon situation in Basque.There is, then, both deitu 'call' and dei egin 'call', the first one with the alternation described above and the second one with a dative complement (if taking a complement at all).Following Torrego (2010) and in the spirit of Laka (1993), the light verb v DO may or may not be lexicalized as egin 'do' (or, alternatively, it may take a VP headed by light egin 'do' or not; the latter would in turn take the low applicative as a complement).In any case, if egin is not present, conflation of the verbal and nominal heads would produce deitu 'call'.This verb will occur with a dative complement if the noun dei had been associated with a goal via the applicative head, or without any complement, as an unergative predicate if no applicative head has been selected.This straightforward analysis would receive some support from the irregular morphology mentioned above in connection with bultzatu 'push'.As we indicated in section 2.1, a large number of the dative complement examples of this predicate are expressed not exactly by this verb but by means of the bultza egin locution.However, there are also some 'intermediate' forms where bultza, although lacking verbal morphology, has been claimed to be acting not as a noun but as a verb (as in (8b) above).These cases are actually expected under the conflation analysis, especially for a verb like this attested mostly from the 19th century.
The "low applicative cum conflation" analysis accounts for the unergative/monotransitive alternation of these verbs as well as for the dative case of the goal argument.It would then be a good account of bivalent unergatives, capturing the traditional suggestion that the dative is not actually a "first" but a "second" complement.It faces, however, some problems, some predicate-specific and some more general.We devote the rest of this section to an admittedly very tentative discussion of some ramifications of these.Beginning with the dative, notice that part of the appeal of positing low applicative heads as in Pylkkänen (2008) lies in the possibility of establishing thematic relations between the basic noun and the applied noun.In Pylkkänen's proposal, low applicative heads denote a transfer of possession relation between the direct object and the applied argument, which is usually the goal or the source of the transfer.Verbs like deitu fit well into this thematic characterization on a broader interpretation in which there is transfer of a message to a goal.Even cases like bultzatu, eragin 'push, move' can be construed as involving the transfer of a 'push' to a person or object.
In other dative taking verbs, though, whether alternating or not, this is more complicated.Thus, begiratu 'watch', originally a loanword from Romance vigilare 'invigilate', has been reanalyzed so its root is the allative phrase 'to the eye' converted into a verb.If we pay attention to this morphology (and this is congruent with a conflation analysis, especially if this 'reanalysis' reflects the speakers' perception in the borrowing language), the thematic makeup of this predicate would involve an initiator transferring a theme argument to a location which is in an inalienable possession relation with respect to another argument.We may even consider the possessor a beneficiary of the transfer, but in any event we would need a massive type of conflation to incorporate the goal, theme and verbal heads.Similarly, the dative associated with itxaron 'wait for', diachronically a compound meaning 'hold the word', can easily be analyzed, in the original meaning, as involving a goal or beneficiary to whom something is transferred.From a synchronic perspective, though, this is a completely opaque verb where assuming a transfer relationship seems unwarranted.It is not impossible to think of 'give x a y' paraphrases for many of these verbs, but this is too broad a characterization of the bivalent unergative class: many monotransitive predicates also admit those paraphrases, so the question arises as to why dative objects occur with a small set of verbs only.Characterizing the class as that involving a transfer relationship between a theme and a goal/beneficiary, on the other hand, is too narrow an analysis and leaves out members of the class, as we have just seen.This questions the validity of the approach at least for all the verbs sharing a first dative argument.
The more general issue associated with this approach, though, emerges when we consider the absolutive/dative alternation we are examining here and the way the analysis can relate both alternants.More specifically, while the analysis accounts for the dative case in a fairly homogeneous case, it is now the absolutive variant that requires some attention, since this analysis often forces us to posit quite different structures and derivations for variants which are intuitively very close.This is not a problem per se, but an analysis of the small sample of cases examined in this paper shows again that this alternation actually hides quite different situations and, perhaps, case checking mechanisms and structures.
Deitu 'call' again exemplifies this: if deitu is the conflated form of dei (egin) 'call do', dei-is the morphological reflex of the theme argument, and the (apparent) dative argument would correspond to a second, goal participant.But then it is not clear what dei-is in the use of this verb with absolutive object.It is either a different verb, or the absolutive complement is not in fact the object of a regular monotransitive verb but, rather, a dative--shifted absolutive like students in teach the students.We will return to this possibility after discussing the problem of a verb like erantzun 'answer'.This verb usually takes dative complements, but 'answering a question' can occur with dative (12a) or, less commonly, absolutive (12b) marking: (12) a. galdera-ri erantzun-go d-io-t question-DAT answer-FUT PRS-3DAT-1ERG 'I will answer the question' b. galdera erantzun-go d-u-t question(ABS) answer-FUT prs(3ERG)-3DAT-1ERG 'I will answer the question' The applicative approach accounts well for the dative argument 'introduced' over dativeless bases (unergative or perhaps transitive in our cases) and (12a) would fall squarely under the type of data it is meant to explain, since it would be akin to 'give/do an answer to the question'.In the absolutive object case (12b), though, what was the applied argument in (12a) appears as a standard internal (non-conflated) argument.Thematically, the (b) example is relatively unexpected, since the question is not the theme with respect to the answer.Rather, this looks again parallel to examples like teach the students, that is, to dative-shifted goals occurring without the theme object.This is not a problem in and of itself were it not for the fact that the dative argument in (12a) is quite generally identified as dative-shifted: most approaches take the dative marker in Basque to be equivalent to the dative case of English, rather than to the prepositional to phrase (see Elordieta 2001 in particular).If this is correct, galdera 'answer' in (12b), just like the absolutive object of deitu 'call' mentioned above, is the internal argument of the verb, so we need to postulate quite different lexical verbs for the two alternants.
In other cases, though, there seems to be hardly any basis to claim that the alternating nominal is linked to different thematic roles.Lagundu 'help' is one such example.Recall that this predicate has two meanings, 'help' and 'accompany', and that although the second one is associated with a dative complement, the first one can alternate between a dative and an absolutive one (see (6a-b) above), presumably with the same semantic content.The low applicative analysis forces us to consider the dative alternant an applied argument associated with a conflated object, but the predicate lagundu would still exist as a lexical item heading its VP layer and with a regular object complement checking its case with it (Chomsky 2008).Other differences noted above following Etxepare (2003) present similar problems: while bultzatu 'push' with a dative complement can be accounted for by the conflation analysis, the absolutive variant requires a different explanation.We might want to retain the conflation analysis for this predicate if the case of the absolutive variant does not correspond to the internal argument, for instance if in push Mikel to the corner the verb takes a different type of complement, say, for the sake of the argument, a small clause rather than a nominal element.But then, the licensing mechanism for the absolutive case would not coincide with the regular inherent case licensed by V.As (13) shows, the case of the subject of a complement small clause is licensed by the verbal head (lexical case as in Woolford 2006).The verb eritzi 'consider' would license the dative case of horri 'to that': 'I consider that interesting' We might expect that if the small clause in push [Mikel to the corner] is introduced as a specifier of the applicative, its subject would occur with inherent dative case, licensed by the applicative head just like the simple nominal dative specifier, rather than with the absolutive.The situation might be similar to the case licensing alternation in bark at someone vs. bark someone awake, but, in any event, the licenser for the absolutive nominal seems elusive in this approach.
Summarizing, the type of analysis we have termed "conflation cum applicative", although quite enlightening for dative complements, makes the analysis of the absolutive variants perhaps considerably more complex than one would possibly expect and desire.The line of analysis is appropriate, more so than just claiming that predicates taking dative objects, after all a fairly small class, are just the result of lexical marking/listing.The fact that the (small) set of predicates of this type often shows similar properties cross--linguistically (Blume 1998) indicates that lexical marking cannot be involved here.The semantic/aspectual analysis and the syntactic mechanisms in the opposite line of analysis, though, are not, at least to our mind, completely satisfactory alternatives.Lexical marking, on the other hand, may be more appropriate for the type of datives we introduce in the following section.

Dative vs. PP alternations: lexical dative
We have introduced a different type of dative case in the preceding section: lexical dative, "idiosyncratic nonstructural case, lexically selected by particular verbs, licensed by V inside the VP proper at vP structure" (Woolford 2006: 126).In fact, if the conflation analysis captures one traditional intuition on dative marking for "first" complements (namely, that there is a hidden absolutive complement in the structure), lexical case can help formalize an alternative intuition about at least some of these datives: that they are licensed by the verb as a marked, "quirky" departure from the more basic absolutive case licensing.We will not explore this possibility for the types of predicates discussed in section 2. Instead, we will briefly present a set of bivalent unergative predicates whose dative complements are not, as far as we know, cross-linguistically associated with "unexpected" dative case single complements.Verbs like eragon and ekin 'engage in' belong to this type.One indication that their dative complements differ from those in the previous section is that the alternations patterns they participate in are also drastically different.Thus, the datives of verbs like eragon and ekin 'engage in' alternate with postpositional phrases.We illustrate with the latter predicate.Example (14a) shows this verb with an activity dative complement, while it displays a postpositional inessive complement in ( 14b While postpositional alternants for these verbs are mostly dialectal, the fact is they exist while, on the other hand, no dative/absolutive alternation has been attested in any dialect.The two predicates mentioned above have an aspectual meaning, and they often occur with nominalized complements bearing the same case markings as in the previous examples (dative or inessive).The dative complement pattern in (14a) also differs from the bivalent unergatives in the previous section in that there is no transfer reading available, and in fact no unexpressed object can be discerned that would make a conflation analysis plausible.Moreover, in terms of complement case, these verbs pattern with another set of aspectual predicates such as ari 'be engaged in', lotu 'begin', jarraiki, jo 'continue', or hasi 'begin'.The latter set occurs mostly in bivalent unaccusative configurations in which the complement may appear in the dative or with the same range of (usually postpositional) markers found with ekin, eragon 'be engaged in'.The number of event participants in the two sets is identical: the subject of the unaccusative verb and the event denoting complement (typically a clause or an eventive noun).These are also the two participants we find in ekin and eragon, and no extra argument can be discerned even though these unergatives also look morphologically "transitive".There is then no motivation for any low (or high) applicative head introducing the dative nominal associated with any more or less hidden first object.Diachronically, the dative case of these verbs seems to be related to well known grammaticalization patterns for different prepositional markings associated with aspectual verbs (Bybee et al. 1994).
We would like to claim that this dative is a lexical case licensed by V, as in Woolford (2006).This is in effect a diacritic marking which, in essence, treats this marking as "exceptional".There are few bivalent unergative predicates of this type, as expected for this type of case.The lexical connection with the verbal head may account for the occasional occurrence of examples in which the dative marking has been "fossilized" in the agreement and is dissociated from the complement, which appears in a prepositional form: The dummy dative inflection corresponds to no dative participant: the aspectual verb takes a complement event in the inessive.These are isolated examples, but they point at a verb-centered licensing of this type of dative in general.
If this approach is on the right track, then, Basque predicates taking dative "first" complements would in fact exemplify two different types of non--structural case as in Woolford (2006): inherent case, perhaps licensed by an argument-introducing applicative head; and lexical case, licensed by V.The different structural conditions under which case is checked in the two types of datives would correlate with the existence of two very different types of case alternations associated with datives in Basque.