In Lithuanian, accusative case on internal arguments is replaced with genitive case when the verb is negated, as in (15). This is referred to as the genitive of negation and is obligatory in Lithuanian. The genitive of negation does not hold for non-accusative licensing verbs, as in (16).
6 In this chapter, I will not discuss non-agreeing passives, which can be formed from almost any verb, including zero- and one-place predicates, but with a different semantic function, namely evi- dential mood. For more on these, see Wiemer (2006), Lavine (2006), inter alia.
(15) a. J-is mėgsta al-ų.
he-nom.sg like:prs.3 beer-acc.sg
‘He likes beer.’
b. J-is ne-mėgsta al-aus/*al-ų.
he-nom.sg neg-like:prs.3 beer-gen.sg/*acc.sg
‘He doesn’t like beer.’
(16) a. J-is atstovau-ja darbinink-ams.
he-nom.sg represent-prs.3 worker-dat.pl
‘He represented the workers.’
b. J-is ne-atstovau-ja darbinink-ams/*darbinink-ų.
he-nom.sg neg-represent-prs.3 workers-dat.pl/*gen.pl
‘He didn’t represent the workers.’
A strict distinction between structural accusative and other cases licensed by the verb can be established on the basis of genitive of negation. The only possible exception is with genitive case licensed by the verb: It is impossible to test if the case marking is due to the negation or the verb, but this does not affect my analysis greatly.
It should also be noted that the genitive of negation can also, but need not, apply to accusative adverbs of time, as discussed by Holvoet and Judžentis (2004).
(17) a. J-is musų firm-oje išdirb-o vien-us met-us.
he-nom.sg our firm-loc.sg worked-pst.3 one-acc.pl.m year-acc.pl
‘He worked a year in our firm.’
b. J-is musų firm-oje ne-išdirb-o vien-ų he-nom.sg our firm-loc.sg neg-worked-pst.3 one-gen.pl
met-ų /?vien-us met-us.
year-gen.pl /?one-acc.pl.m year-acc.pl
‘He didn’t work a year in our firm.’ (Holvoet & Judžentis 2004: 71) Certain accusative adverbials seem to function like accusative direct objects, highlighting the complex situation of transitivity in Lithuanian. Similarly, certain unaccusative verbs also allow genitive case marking on the subject.
(18) Pas mus atvažiav-o sveči-ų.
at we:acc arrive-pst.3 guests-gen.pl
‘Some guests arrived at our place.’ (Holvoet & Judžentis 2004: 64)
A second instance of structural case behaving consistently differently from non-structural case is the case preservation of internal arguments with deverbal
nouns. For verbs that license accusative case, the object is expressed in the geni- tive case under nominalization of the verb. The genitive case-marked noun pre- cedes the noun it is associated with, as in other instances in Lithuanian (e.g., genitive of possession, quality), as in (19).
(19) *laišk-ą/laišk-o rašymas *letter-acc.sg/gen.sg writing
‘the writing of letters.’
This instance of genitive case, like the genitive of negation, appears to be structu- ral: It is not associated with any semantic relationship, but is due to the structural relation between the noun and the nominalized verb.
In nominalizations from verbs that license genitive case, it is unclear if the object occurs in genitive due to the verb or the nominalization. The word order is the same as for accusative verbs, perhaps indicating that the case is structural as for accusative verbs.
(20) a. tams-os baimė dark-gen.sg fear
‘fear of the dark’
b. *baimė tams-os
fear dark-gen.sg
Nominalizations from verbs that license an oblique case (other than genitive) differ in two ways from their accusative counterparts. First, they retain the mor- phological case on the object, and second, the word order is reversed: The object follows the nominalized verb, as in (19).
(21) a. vadovavimas darbinink-ams/*darbininkų representing worker-dat.pl/*gen.pl
‘the representing of the workers’
b. tikėjimas ateit-imi belief future-ins.sg
‘belief in the future’
Babby (1994) identifies instances in which two possible cases being licensed as “case conflicts”. According to him, lexical case can override structural case in such conflicts but not vice versa, accounting for the difference in case for numeri- cal expressions in Russian. This account can be extended to the Lithuanian geni- tive of negation: While it may be an obligatory structural case, licensed by the negation, it cannot override a non-structural case. Similarly, the genitive under nominalization only applies to structural case objects. The word order difference
creates a problem, however, because it does not appear that the accusative objects and oblique objects are in the same structural position relative to the deverbal noun they are associated with. One possibility is that all objects are initially in the same position, but the genitive case marking activates movement to the specifier position of the noun phrase, yielding the object-verb word order. Another analy- sis, suggested by Peter Arkadiev (p.c.) is that the deverbal noun does not contain the v head that would normally license Accusative case, triggering the movement to SpecNP for genitive case assignment.
The facts of genitive under negation and under nominalization show that there is a difference in syntactic behavior between verbs that license accusative objects and those that license an oblique case. Because it is genitive case in such constructions that is the defining feature of a structural-case licensing verb, it is unclear how to categorize verbs that license genitive case.
One piece of evidence that genitive on internal arguments is strutural case comes from the fact that this case can be overridden with the distributive prepo- sition po (Axel Holvoet, p.c.).
(22) a. suvalg-ė obuol-į.
eat-pst.3 apple-acc.sg
‘(S)he/They ate an apple.’
b. suvalg-ė po obuol-į.
eat-pst.3 prep apple-acc.sg
‘They ate an apple each.’
(23) a. Ar nor-ite obuoli-o?
q want-prs.2.pl apple-gen.sg
‘Do you want an apple?’
b. Ar nor-ite po obuol-į?
q want-prs.2.pl prep apple-gen.sg
‘Do you want an apple each?’
(24) a. Atstovav-o darbinink-ui.
represent-pst.3 worker-dat.sg
‘(S)he/they represented the worker.’
b. *Atstovav-o po darbinink-ą.
represent-pst.3 prep worker-dat.sg
‘They represented a worker each.’
The examples in (22) and (23) show that accusative and genitive internal argu- ments have the same behavior when in the scope of po ‘each’, while (24) shows
that the lexical dative case cannot be overridden (cf. Babby 1994: 643 for similar data in Russian). Thus, it is unclear if the passivization of verbs that require geni- tive case marked internal arguments should be considered oblique passivization or not.