Not all verbs that license non-structural case can equally passivize with a pro- moted, agreeing subject. The examples in (25)–(28) show oblique passivization of verbs that license dative and instrumental on the internal argument. For some verbs, only the present passive is possible, as in (26b), while for others, neither is acceptable, as shown in (27). Finally, (28) shows that both forms of the participle are acceptable.
(25) a. Jon-as pirmininkav-o posėdži-ui.
John-nom.sg chair-pst.3 meeting-dat.sg
‘John chaired the meeting.’
b. Posėd-is buvo Jon-o pirmininkau-jam-as.
meeting-nom.sg aux.pst.3 John-gen.sg chair-prs.pp-nom.sg.m
‘The meeting was (being) chaired by John.’
c. ??Posėd-is buvo Jon-o pirmininkau-t-as.
meeting-nom.sg aux.pst.3 John-gen.sg chair-pst.pp-nom.sg.m
‘The meeting was chaired by John.’
(26) a. Vaik-as padė-jo motin-ai.
child-nom.sg help-pst.3 mother-dat.sg
‘The child helped the mother.’
b. Motin-a būdavo/buvo vaik-o paded-am-a.
mother-nom.sg aux.hab.3/aux.pst.3 child-gen.sg help-prs.
pp-nom.sg.f
‘The mother usually was/was being helped by the child.’
c. *Motin-a buvo vaik-o padė-t-a.
*mother-nom.sg aux.pst.3 child-gen.sg help-pst.pp-nom.sg.f Intended: ‘The mother was helped by the child.’
(27) a. Lietuv-a prekiau-ja gintar-u.
Lithuania-nom.sg trade-prs.3 amber-ins.sg
‘Lithuania trades (in) amber.’
b. *Gintar-as Lietuv-os prekiau-jam-as.
amber-nom.sg Lithuania-gen.sg trade-prs.pp-nom.sg.m
‘Amber is traded by Lithuania.’
(28) a. Alkohol-is (pa)-kenk-ė kepen-ims.
alcohol-nom.sg (prf)-harm-pst.3 liver-dat.pl
‘Alcohol harmed liver.’
b. Kepen-ys yra/buvo pa-kenk-t-os/kenki-am-os
liver-nom.pl aux.prs.3/pst.3 prf-harm-pst.pp-nom.pl.f/harm- prs.pp-nom.pl.f
alkoholi-o.
alcohol-gen.sg
‘The liver was harmed by alcohol.’
Additionally, the animacy of the argument can affect ability of a verb to passivize (Kristina Lenartaitė, p.c.). Note the unacceptability of the passive for the animate argument in (29), as compared with the inanimate argument in (28).
(29) a. Alkohol-is (pa)-kenk-ė Jon-ui.
alcohol-nom.sg (prf)-harm-pst.3 John-dat.sg
‘Alcohol harmed (the) liver.’
b. *Jon-as yra/buvo pa-kenk-t-as/kenki-am-as
Jonas-nom.sg aux.prs.3/pst.3 prf-harm-pst.pp-nom.sg.m/harm- prs.pp-nom.m.sg
alkoholi-o.
alcohol-gen.sg
Intended: ‘John was harmed by alcohol.’
A further issue for oblique passivization is that not only direct internal argu- ments (which i hear take to include patients and themes) can become subjects under passivization, as discussed by Lenartaitė (2009: 74–75). Noun phrases marked with other theta roles can be appear as agreeing nominative subjects of passive participles, including locations, shown in (2), repeated hear as (30); ins- truments, shown in (31); and means, shown in (32). These noun phrases, while not direct internal arguments of the verbs, may not be adjuncts. Details of such a distinction are beyond the scope of this chapter.
(30) a. Žvėr-ys gyven-a urv-uose.
animal-nom.pl live-prs.3 cave-loc.pl
‘Animals inhabit caves.’
b. Urv-ai yra žvėri-ų gyven-am-i.
cave-nom.pl aux.prs.3 animal-gen.pl live-prs.pp-nom.pl.m
‘Caves are inhabited by animals.’ (From Ambrazas 2006: 322)
(31) a. Tu valg-ai šit-uo šaukšt-u.
you(sg):nom eat:prs.2sg this-ins.sg.m spoon-ins.sg
‘You are eating with this spoon.’
b. Š-is šaukšt-as tavo valg-om-as.
this-nom.sg.m spoon-nom.sg your(sg) eat-prs.pp-nom.sg.m Literally: ‘This spoon is being eaten by you.’ (Jablonskis [1922] 1997:
132, in Lenartaitė 2009: 75)
(32) a. Žmon-ės šit-uo keli-u dabar ne-be-važiuo-ja.
people-nom.pl this-ins.m.sg road-ins.sg now neg-cnt-drive- prs.3
‘People don’t drive this way anymore.’
b. Šit-as keli-as dabar (žmoni-ų) this-nom.sg.m road-nom.sg now (people-gen.pl) ne-be-važiuo-jam-as.
neg-cnt-drive-prs.pp-nom.sg.m
‘The road isn’t driven by people anymore.’ (Jablonskis [1922] 1997: 132, in Lenartaitė 2009: 75)
In addition to the locative and instrumental noun phrases shown above, even an internal argument inside a prepositional phrases can become nominative sub- jects of passives, as in (33), somewhat like the English pseudo-passive This bed has been slept in.
(33) a. J-is atsak-ė į klausim-ą.
he-nom.sg answer-pst.3 in question-acc.sg
‘He answered the question.’
b. Klausim-as j-o yra/buvo
question-nom.sg he-gen.sg aux.prs.3/aux.pst.3 atsaky-t-as.
answer-pst.pp-nom.sg.m
‘The question was answered by him.’
All of these examples of non-canonical passives show that passivization is not a useful test for either structural vs. lexical case, or (internal) argumenthood.
This could have to do with the fact that passivization involves two changes to the syntactic structure: The external argument is demoted, and (generally), the
internal argument (or some NP, in the case of, e.g., English pseudo-passives) is promoted. For Lithuanian, it appears that the first change is key, while the second is less important. This is evidenced by the existence of (non-agreeing) passive participles from one-place predicates, such as miegoti ‘to sleep’ (see Wiemer 2006 for more on such impersonal passives). Thus, the fact that many different types of non-external arguments/participants can become the agreeing subject of a passive is not entirely unjustifiable. However, there are differences in acceptabi- lity between verbs and between the present and the past passive.
At the very least, this is evidence that the traditional distinction between structural and inherent case is not adequate enough to capture the different ( morphological) case patterns and voice alternations in Lithuanian. In the next subsection, I will attempt to show how a finer distinction of non-structural case can be useful in interpreting the above data, relying on previous analyses in which arguments marked with different cases can occupy different structural positions.