Let us start from an exposition of the facts. The Latvian debitive has an A argument (the experiencer argument of the model predicate and agent of the embedded pre- dicate) in the dative. The B argument is supposed to be in the nominative if it is a noun or a third-person pronoun. It is in the accusative if it is a first- or second-per- son pronoun or a reflexive pronoun (Endzelīns 1951: 972). Compare (27) with (3):
(27) Pirms tu ko sak-i, man ir
before 2sg.nom anything.acc say-prs.2sg 1sg.dat be.prs.3 tevi jā-brīdina, ka vis-s, ko tu 2sg.acc deb-warn that all-nom.sg.m what.acc 2sg.nom teik-si, var tikt lieto-t-s pret tevi.
say-fut.2sg may.prs.3 become.inf use-ppp-nom.sg.m against 2sg.acc
‘Before you say anything I must warn you that anything you say can be used against you.’
http://zagarins.net/jg/jg182/JG182_Bekmans.htm (accessed July 2012) That the use of the accusative for first- and second-person pronouns is an effect of the animacy hierarchy (the accusative being reserved for NPs at the upper end of the hierarchy) was already recognized by Timberlake (1974). Timberlake com- pares the Latvian debitive construction to the Fennic nominative object construc- tion, which shows a similar animacy constraint, as shown in (28) and (29):
(28) Finnish
Sinu-n tọytyy luke-a tọmọ kirja.
2sg-gen be.necessary.prs.3sg read-inf dem book.nom
‘You must read this book.’
(29) Finnish
Sinu-n tọytyy kutsu-a minu-t.
2sg-gen be.necessary.prs.3sg invite-inf 1sg-acc
‘You must invite me.’
Although the parallel is valid in principle, it is only to a certain extent: In the Fennic construction, there is a separate embedded infinitival clause with gram- matical relations largely intact apart from the use of a phonologically empty PRO subject. The complement of the infinitive is therefore clearly an object. This cannot be said of the Latvian debitive construction, which does not contain a syntactically distinguishable embedded clause. That the B-argument of a debitive construction is an object may therefore not be taken for granted.
In the dialects, the situation is different from what was stated for Standard Latvian: In many Latvian dialects, the object of a verb in the debitive is always in the accusative (Endzelīns 1951: 971). This situation is also spreading in colloquial Standard Latvian, although this tendency is opposed by prescriptive grammarians.
Historically speaking, the nominative is the original case, as the debitive construction has arisen from an existential construction or its expanded form, a possessive construction of the type mihi est, in which the subject was modified by an infinitival relative purpose clause (cf. Holvoet 1998 for a historical outline).
In Old Latvian, the original meaning was retained:
(30) Old Latvian
Man ir Barriba jaehd, 1sg.dat be.prs3 food.nom.sg deb.eat ko juhs ne sinnat
that.acc 2pl.nom neg know.prs.2pl
‘I have meat to eat that ye know not of.’ (Glück’s New Testament, John 4.32) (Ich habe eine Speise zu essen, von der ihr nicht wisset)
The modern meaning would be “I have to eat food”. The original subject has undoubtedly lost part of the properties associated with subjecthood, the ques- tion is, however, to establish exactly how many.
Let us first look at coding properties. The alternation of nominative and accusative as cases encoding the second-ranking argument points to the obvious conclusion that we are not dealing with a canonical subject here. This impression is reinforced by the facts concerning agreement. The debitive comprises a form of the auxiliary būt ‘be’. In the compound tenses of this auxiliary, the participle should show agreement with the nominative if it is a subject. This can be illustrated by the copular construction:
(31) Sien-as kādreiz ir bij-uš-as balt-as.
wall-nom.pl once be.prs.3 be-ppa-nom.pl.f white-nom.pl.f
‘The walls have once been white.’
In the debitive construction, however, this agreement is usually absent, and the participle is in the default agreement form, the masculine singular:
(32) Šaj-os gad-os ir bij-is
those-loc.pl year-loc.pl be.prs.3 be-ppa.nom.sg.m
jā-lasa dažād-as ziņ-as,
deb-read various-nom.pl.f news.item-nom.pl tai skaitā arī šokējošas, kad pašai diktorei trīcējušas kājas no uztraukuma.
http://www.diena.lv/sabiedriba/pasi-mazakie-648149 (accessed July 2012) ‘In those years one has had to read various news items, among them shocking ones, which made the newsreader’s legs shake with excitement.’
But constructions with agreeing auxiliaries also occur. The Academy Grammar (Bergmane et al. 1959: 618) cites an example with agreement alongside an analo- gous construction without agreement. Both examples are in the so-called oblique mood, an evidential construction based on the use of a participle instead of a finite form. This participle is capable, in principle, of agreeing with the nomina- tive NP in number and gender. In (33), there is no agreement, while (34) displays agreement with a nominative plural NP:
(33) Jau lemeš-u uzasināšan-ai ogl-es
pcle ploughshare-gen.pl sharpening-dat.sg coals-nom.pl bij-is jā-aizņemas no kalēj-a.
be-ppa.nom.sg.m deb-borrow from blacksmith-gen.sg ‘The coals needed for the sharpening of the ploughshares had [reportedly]
to be borrowed from the blacksmith.’ (E. Birznieks-Upītis)
(34) Lin-i bij-uš-i jā-kaltē, un vec-ais
flax-nom.pl be-ppa-nom.pl.m deb-dry and old-nom.sg.m.def
tēv-s tos labi prat-is izraudzīt.
father-nom.sg 3.acc.pl.m well know.how-ppa.nom.sg.m select.inf ‘The flax [reportedly] had to be dried, and the old father knew very well how to select it.’ (J. Akuraters)
The Grammar notes, however, that this agreement is rare (Bergmane et al.
1959: 617). We may probably rely on the authors of the Academy Grammar when they state that constructions with agreement are less frequent, but they can certainly be found even in the present-day language. Word order probably has a certain influence on the presence of absence of agreement, cf. the following example with preverbal B argument:
(35) Es gan spriežu no malas, jo man
1sg.nom ptcl judge.prs.1sg from aside because 1sg.dat
par laimi šād-a izvēl-e nav
fortunately such-nom.sg.f choice-nom.sg be.prs.3.neg bij-usi jā-izdara.
be-ppa.nom.sg.f deb-make
‘True, I can judge only from aside, as I have never been compelled to make such a choice.’
http://www.calis.lv/forums/tema/18191705-piespiedu-karta/ (accessed July 2012)
Behavior properties yield no decisive evidence. The datival NP is capable of controlling reflexivization and does so regularly.3
(36) Latvian
Tā ir pasaul-es lielāk-ā
dem-nom.sg.f be.prs.3 world-gen.sg greatest-nom.sg.f.def nelaime, ka vecāk-iem jā-redz sav-i
misfortune-nom.sg that parent-dat.pl deb-see poss.refl-nom.pl.m
bērn-i aizej-am.
child-nom.pl go.away-cvb
http://www.calis.lv/forums/tema/15206129-engelisu-maminas/51/
(accessed July 2012)
3 In the construction with the debitive, reflexive pronouns proper can be controlled only by the dative NP.
The nominative NP seems to be unable to control reflexivization if the datival A argument precedes the B argument:
(37) Bet siev-ām vīr-i aizvien jā-redz
but wife-dat.pl husband-nom.pl always deb-see ar vis-iem viņ-u veikal-iem
with all-dat.pl 3-gen.pl business-dat.pl (Pāvils Rozītis) ‘But wives should always view their husbands in the context of their (i.e., their husbands’) business occupations.’
As in the case of patikt, however, topicalization and fronting render the nomina- tive B argument capable of controlling reflexivization:
(38) Atgriežoties pie biskvīta, kad tas izcepies,
tas jā-ņem ārā, jā-atbrīvo no sav-a
it.nom.sg.m deb-take out deb-loosen from poss.refl-gen.sg.m follij-a žodziņ-a un cepam-ā papīr-a, foil-gen enclosure-gen and baking-gen.sg.m.def paper-gen lai atdziest.
so.that get.cold.prs.3
‘To get back to the sponge cake, when it’s ready, it should be taken out, loosened from its foil enclosure and from the baking paper to let it get cold.’
http://spoki.tvnet.lv/receptes/Aromatiska-zemenu-kuka/544597 (accessed July 2012)
When topicalized and occurring sentence-initially, such a nominative NP is also treated as a pivot/controller in coordination, as in example (39). This shows, together with pairs of contrasting sentences like (4) and (6), how important topic- hood is for control of reflexivization:
(39) Vid-es, sociāl-ie un ekonomisk-ie
environment-gen social-nom.pl.m.def and economic-nom.pl.m.def jautājum-i ir savstarpēji cieši saistīt-i
issue-nom.pl be.prs.3 mutually closely connected-nom.pl.m un ỉ ir jā-risina vienoti.
and be.prs.3 deb-solve conjointly
‘Environmental, social and economic issues are closely interrelated and must be dealt with conjointly.’ (http://archive.politika.lv/temas/vide_un_
ilgtspeja/17419/) (accessed July 2012)
This points to the conclusion that the debitive, just as the construction with patikt, provides us with an example of recoverable subjecthood. This recovery of subjecthood, however, is not complete, as the nominative may not control reflexi- vization even when topicalized and sentence-initial, as shown by (33).
The apparently recoverable subject properties of the B argument explain the specific properties of the debitive with regard to voice. Although the debitive, as mentioned above, has been called a kind of passive, the full debitive construc- tion, comprising both A and B argument, is more similar to the active one in that the nominative-marked NP (the original object) is normally the second-ranking argument, the dative NP (the original subject) being least oblique and normally assuming the position of unmarked topic:
(40) Mums ar Kuzm-u jā-padarin-ot četr-i 1pl.dat with Kuzma-acc deb-make-obl four-nom.m baļķ-u vārt-i un jā-uzstat-ot tie
beam-gen.pl gate-nom.pl and deb-set.up-obl 3.nom.pl.m straum-es vid-ū.
stream-gen.sg mid-loc.sg
‘Kuzma and I [he said] would have to make four beam gates and to set them up amidst the stream.’ (Valentīns Jakobsons)
This example could be transformed into a non-debitive active construction without the slightest change in word order:
(41) Mēs ar Kuzm-u padarinājām četr-us 1pl.nom with Kuzma-acc make.pst.1pl four-acc.m baļķ-u vārt-us un uzstatījām tos beam-gen.pl gate-acc.pl and set.up.pst.1pl 3.acc.pl.m straum-es vid-ū.
stream-gen.sg mid-loc.sg
‘Kuzma and I made four beam gates and set them up amidst the stream.’
But the agent/modal experiencer may be backgrounded or generalized and, as a result, may not be represented in syntactic structure. In many languages, Latvian not excluded, a typical device enabling the suppression of the agent is the passive. Passive debitives, however, do not occur.4 For a structure like
4 Or rather, a debitive actional (dynamic) passive does not occur. What is possible is a debitive derived from a resultative passive (with the auxiliary būt rather than the regular passive auxiliary tikt), as in adresei jābūt apzīmētai (address.dat.sg deb-be indicate.ppp.dat.sg.f) ‘the address must be indicated’ (German die Adresse muss angegeben sein). The corresponding actional passive *adresei jātiek apzīmētai (German die Adresse muss angegeben werden) is not used;
instead, we have simply adrese jāapzīmē (address.nom.sg deb-indicate).
(42) Lin-i ir jā-kaltē.
flax-nom.pl be.prs.3 deb-dry
‘The flax must be dried.’
the natural English translation will be in the passive. In the Latvian construction with the debitive, however, the effect otherwise achieved by passivization is achieved merely by moving the nominative NP to clause-initial, topical position – the dative NP can be deleted without further morphosyntactic changes. The nominative can be compared to the nominative subject in the corresponding passive construction:
(43) Lin-i tiek kaltē-t-i.
flax-nom.pl become.prs.3 dry-ppp-nom.pl.m
‘The flax is being dried.’
The resemblance is, of course, only partial, because even when occurring in the typical subject position, the nominative in the debitive construction does not always control auxiliary agreement. Still, between the active-like debitive construction and that with recovery of subject properties by the B argument there is probably a diffe- rence in the assignment of grammatical relations. Does this mean that in the active- like debitive construction, it is actually the same as in the active non-debitive cons- truction? Could the nominative NP in debitive constructions like (3) be characterized as a nominative object that, to an increasing degree, is adjusting its case marking to its real syntactic function – that of direct object? I think not, despite appearances. It is true that the nominative NP represents the object of the “embedded” verb, but in view of the fact that in the debitive construction the argument structures of higher (modal) and embedded predicates are collapsed, we should not a priori assume identical grammatical relations for the object of the embedded verb and the second- ranking NP with the debitive. The debitive construction should be compared to other constructions with datival quasi-subjects, such as those with vajag, žēl, etc. These constructions are, as I have argued above, basically intransitive and their second- ranking (more oblique) argument is not a direct object but a demoted intransitive subject. That this interpretation applies to the debitive construction is suggested by the partial recoverability of subjecthood by the nominative NP. The behavioral properties of NPs with the debitive are shown in Table 3.
Tab. 3: Subject properties with the debitive
A B
Case marking DAT NOM/ACC
Control of agreement No Yes*
Control of reflexivization Yes Yes*
Pivot/controller in coordination No Yes*
6 Conclusion
When we look at the list of subject properties shown in Tables 1–3, certain regu- larities stand out clearly. In the left column, we find only control of reflexiviza- tion, a feature of that we cannot be sure whether it is really a subject property.
In the right column, subject properties correlate in an interesting way with case marking. Nominative marked B arguments show themselves capable of reco- vering subject properties when they are topicalized, even though they do not fundamentally differ from oblique marked B arguments when non-topicalized;
as soon as nominative marking is replaced with oblique marking, this recovera- bility of subject properties is lost. This shows that there is really no stable con- centration of subject properties in any argument as long as we have a nominative marked argument; when there is no nominative marked argument, the distribu- tion of subject properties (which I have described as “diffuse” in this chapter) is more stable, i.e., independent from topicalization and word order, but this does not go hand in hand with a stronger concentration of subject properties in A. The best way of accounting for the types discussed here is thus through the notion of diffuse subjecthood.
From a diachronic perspective, at least two of the three constructions discussed here had, at the outset, argument B as their subject. As we see most clearly in the case of the debitive, historical shifts lead to a loss of subject properties in B, but without a clear increase in subject properties in A. Claiming that the oblique (accusative) marking on B is a kind of object marking is pointless because an object presupposes a transitive subject. That is why I prefer to refer to the B arguments as demoted intran- sitive subjects. The three types discussed in this article are once more compared in Table 4, arranged in an order different from that in which they are discussed above:
It shows that those subject properties of B conditional on topicalization and clause- initial position are absent when B is encoded by an oblique case, but this does not go hand in hand with an increase of subject properties in A, and the overall picture is, in all cases, that of a construction with diffuse grammatical relations.
Tab. 4: Subject properties – overview table
Patikt Debitive Vajadzēt
A controls agreement No No No
B controls agreement Yes Yes* No
A controls reflexivization Yes Yes Yes B controls reflexivization Yes* Yes* No
A is pivot/controller No No No
B is pivot/controller Yes* Yes* No
The three types of constructions discussed in this article present a fundamen- tal unity: They display a least-oblique datival NP alongside an argument oscilla- ting in its marking between nominative and accusative (and genitive, a case that can encode both intransitive subjects and direct objects). These cases compete as means of encoding what is here characterized as a demoted intransitive subject in a specific configuration with a datival quasi-subject. This configuration is dia- chronically quite stable, but shows minor changes in case marking characterized here as “obliqueness adjustment”, i.e., a replacement of the pattern DAT-NOM with DAT-ACC not motivated by transitivity but rather by the tendency to bring the pattern of case marking in accordance with that of syntactic obliqueness.
Acknowledgments
My thanks are due to the co-editors of this volume, to an anonymous reviewer, and to Ilja Seržant, Nicole Nau, and Wayles E. Browne, for useful criticism and comments on earlier versions of this article. For its shortcomings, I remain solely responsible.
Abbreviations
acc accusative aux auxiliary cond conditional cvb converb dat dative deb debitive def definite dem demonstrative f feminine fut future gen genitive imp imperative inf infinitive loc locative m masculine neg negation nom nominative
obl oblique mood pl plural
poss possessive pronoun ppa past active participle ppp past passive participle prs present
pst past ptcl particle
refl reflexive pronoun rel relative pronoun sg singular
References
Aikhenvald, Alexandra Y., Robert M. W. Dixon & Masayuki Onishi (eds.) Non-canonical marking of subjects and objects. Amsterdam, Philadelphia: John Benjamins.
Allen, Cynthia. 1986. Reconsidering the history of like. Journal of Linguistics 22: 375–409.
Andrews, Avery D. 2001. Non-canonical A/S marking in Icelandic. In Alexandra Y. Aikhenvald, Robert M. W. Dixon, & Masayuki Onishi (eds.) Non-canonical marking of subjects and objects, 85–111. Amsterdam, Philadelphia: John Benjamins.
Barðdal, Jóhanna. 2001. The perplexity of Dat-Nom verbs in Icelandic. Nordic Journal of Linguistics 24: 47–70.
Barðdal, Jóhanna & Thórhallur Eythórsson. 2006. Control infinitives and case in Germanic:
‘Performance error’ or marginally acceptable constructions? In Leonid Kulikov, Andrej Malchukov, & Peter de Swart (eds.) Case, valency and transitivity, 147–177. Amsterdam, Philadelphia: John Benjamins.
Berg-Olsen, Sturla. 1999. A syntactic change in progress: The decline in the use of the non-prepositional genitive in Latvian, with a comparative view on Lithuanian. MA thesis, Oslo University.
Bergmane, Anna, Rūdolfs Grabis, M. Lepika & Evalds Sokols (eds.) 1959. Mūsdienu latviešu literārās valodas gramatika 1. Fonētika un morfoloģija [Grammar of Modern Standard Latvian 1. Phonetics and morphology]. Rīga: Latvijas PSR Zinātņu akadēmijas izdevniecība.
Bhaskararao, Peri & Karumuri V. Subbarao (eds.). 2004. Non-nominative subjects, Vols. 1–2.
Amsterdam, Philadelphia: John Benjamins.
Bielenstein, August. 1863/1864. Die lettische Sprache nach ihren Lauten und Formen erklọrend und vergleichend dargestellt, Bd. I–II. Berlin: Ferdinand Dümmler.
Blake, Barry J. 2001. Case. 2nd edition. Cambridge: Cambridge University Press.
Dalrymple, Mary & Irina Nikolaeva. 2011. Objects and information structure. Cambridge:
Cambridge University Press.
De Haan, Ferdinand. 2006. Typological approaches to modality. In William Frawley (ed.) The expression of modality, 27–69. Berlin, New York: Mouton de Gruyter.
Endzelīns, Jānis. 1951. Latviešu valodas gramatika [Grammar of Latvian]. Rīga: Latvijas valsts izdevniecība.
Haspelmath, Martin. 2001. Non-canonical marking of core arguments in European languages.
In Alexandra Y. Aikhenvald, Robert M. W. Dixon & Masayuki Onishi (eds.) Non-canonical marking of subjects and objects, 53–83. Amsterdam, Philadelphia: John Benjamins.
Holvoet, Axel. 1998. Notes on the rise and grammaticalization of the Latvian debitive.
Linguistica Baltica 7: 101–118.
Holvoet, Axel. 2013. Obliqueness, quasi-subjects and transitivity in Baltic and Slavonic. In Ilja Seržant & Leonid Kulikov (eds.). 2013. The diachronic typology of non-prototypical subjects, 257–282. Amsterdam, Philadelphia: John Benjamins.
Jackendoff, Ray. 1972. Semantic interpretation in generative grammar. Cambridge MA: MIT Press.
Keenan, Edward L. 1976. Towards a universal definition of ‘subject’. In Charles N. Li (ed.) Subject and topic, 303–333. New York: Academic Press.
Keenan, Edward L. & Bernard Comrie. 1977. Noun phrase accessibility and Universal Grammar.
Linguistic Inquiry 8(1): 63–91.
Malchukov, Andrej L. & Helen de Hoop. 2011. Tense, aspect, and mood based differential case marking. Lingua 121: 35–47.