In light of new data brought to my attention by Kristina Lenartaitė (p.c.), it appears that there is no clear semantic distinction between the verbs that allow oblique passivization, either present or past, such as atstovauti ‘to represent’, ieškoti ‘to search for’, kenkti ‘to harm’, and those that do not allow oblique passi- vization, such as prekiauti ‘to trade in’, contra Anderson (2009). However, there is a subtle difference in the acceptability of the past passive. In (26), the past passive participle padėtas ‘helped:pst.pass’ is completely ungrammatical, while the past passive participles for other verbs are not. In Anderson (2009), I argued that the difference could be based on the fact that some verbs have a semantic motivation for the particular case that is used, while others have a purely lexical case. Dative case is highly associated with recipients and bene/malefactives in Lithuanian, and other languages (cf. Cuervo 2003 for datives in Spanish). Instrumental is asso- ciated with the means of performing an action (e.g., write with a pen). Thus, the infelicity of the (past) passive with the verbs padėti ‘to help’ and prekauti ‘to trade in’ could be due to the fact that these verbs have inherent case, predictable from the theta role, on the internal argument, rather than either structural or lexical case. Recall that Woolford (2006) proposes a distinction between inherent and lexical on the basis of passivization. While her claim that lexical case-governing verbs cannot undergo passivization is refuted by the data above, there still may be something to her distinction between two types of non-structural case.
Additionally, I claim that the examples with non-internal arguments, as shown in (31)–(33), have semantic case, given the fact that the morphological
case in these examples contributes to the overall meaning of the sentence. Note that these are also shown with the present, rather than past, passive participles.
I suggest in Anderson (2013a,b) that the past passive, being more resultative in nature than the present passive, only allows affected arguments to be promo- ted, and that only lexical case can occur on noun phrases in the internal argu- ment position. This position is in the scope of the past passive participle. Noun phrases that receive inherent or semantic case (both of which involve theta roles), however, occur in a different position, where the present passive participle can have scope. As shown above, the present passive participle is acceptable for more verbs, and the past passive participle is not entirely grammatical for the verbs that seem to lack this semantic connection with the case licensed on the internal argument, e.g., the judgment on (26c).
Thus, the arguments whose case marking is related to their theta role cannot become subjects of past passives as easily as they can in present passives. This would require the non-accusative internal arguments that can be promoted in the past passive to be considered patients. Thus, my earlier claims may still hold true: Verbs that license a case on the internal argument (or adjunct) that is based on the theta role the verb assigns are not patients. However, not all patients are marked with struc- tural case; accusative may be overridden by a strong lexical requirement of the verb.
We can see how this plays out with two verbs that license instrumental on the internal argument, but do not both allow the past passive participle: prekauti ‘to trade in’ and tikėti ‘to believe in’. The item that is traded can be construed as the means of performing the action of trading, while the thing or person that is believed in is not the means of believing. I conclude that the instrumental case that occurs on the argument with prekauti is an instance of semantic case, while tikėti has a strong lexical requirement for instrumental case, which is not based on the meaning of instrumental in other contexts. However, the lexical semantics of this verb may pre- clude it from having the right structure for forming the past passive participle, as not all accusative-assigning verbs form this participle either. The type of event described by the verb, i.e., the state described by tikėti, also plays a role in determining the ability of a verb to passivize. This conclusion is still tentative, as there must be esta- blished a clear diagnostic for distinguishing between the “statal” and the “actional”
passive for Lithuanian, as well as a better understanding of the lexical semantics of more verbs that appear to license purely lexical case.
4 Case in purpose clauses
Another instance of non-canonical case marking is found in the behavior of internal arguments in infinitival purpose clauses, discussed in great detail by
Franks and Lavine (2006) and Arkadiev (2014). Unlike these works, which present analyses for the case and word order, I will focus on the facts relevant to case typology. In these purpose clauses, the structural accusative case that is usually licensed by the verb is apparently overridden by dative or genitive in these non- finite subordinate clauses. Additionally, there is a strong preference for object- verb word order with dative and a weak preference for the same word order with genitive (rather than the default7 verb-object word order). Genitive case occurs when the main-clause verb is a motion verb; dative occurs elsewhere, as shown in (34)–(35).
(34) Genitive (from Ambrazas 2006: 557)
a. Išsiunt-ė sūn-ų daktar-o pakvies-ti.
send-pst.3 son-acc.sg doctor-gen.sg invite-inf
‘He sent his son to get the doctor.’
b. Išvažiav-o keli-o taisy-ti.
arrive-pst.3 road-gen.sg fix-inf
‘They came to fix the road.’
(35) Dative (from Ambrazas 2006: 557) a. Iššov-ė žmon-ėms pagąsdin-ti.
fire-pst.3 people-dat.pl scare-inf
‘He fired to scare people.’
b. Pastat-ė daržin-ę šien-ui sukrau-ti.
build-pst.3 hayloft-acc.sg hay-dat.sg keep-inf
‘They built a hayloft to keep hay.’
The genitive and dative case in such purpose clauses is related to the overall interpretation of the sentence, like semantic case found on nominal adverbial expressions. Additionally, it generally occurs with infinitives of verbs that license accusative on their direct objects. The examples below show the ungrammaticality of the dative and genitive in a purpose clause with the verb rūpintis ‘to take care of’, which requires an instrumental argument. The dative or genitive is not possible;
as seen elsewhere with “case conflicts” (cf. Babby 1994), lexical case cannot be
“overridden”. The word order is also important here: OV is possible under certain discourse conditions, such as shift in functional sentence perspective.
7 Lithuanian, as other languages with rich morphological case marking, has relatively free word order, which is largely determined by information structure. The constituents of a sentence occur in the order subject, verb, object in neutral circumstances. Scrambling is permitted for expres- sing differences in functional sentence perspective (Ambrazas 2006: 691–692).
(36) a. Mes pastat-ėme ligonin-ę [rūpintis vaik-ais].
we:nom build-pst.1.pl hospital-acc.sg take.care:inf child-ins.pl
‘We built a hospital to take care of children.’
b. Mes pastatėme ligoninę [#vaik-ais rūpintis].
children-ins.pl take.care:inf
c. Mes pastatėme ligoninę [*vaik-ams rūpintis].
children-dat.pl take.care:inf
(Franks & Lavine 2006: 250) (37) a. Atėj-o [rūpintis draug-u].
arrive-pst.3 take.care:inf friend-ins.sg
‘He came to take care of his friend.’
b. Atėjo [#draug-u rūpintis].
friend-ins.sg take.care:inf c. Atėjo [*draug-o rūpintis].
friend-gen.sg take.care:inf (Franks & Lavine 2006: 255) As noted in Section 3.1, genitive internal arguments seem to be somewhere between structural and non-structural. Both lexically required and quantificatio- nal genitive can be overridden by dative in purpose infinitivals, as demonstrated by Arkadiev (2014).
(38) a. Jie nor-i [išveng-ti kar-o].
they:nom.pl want-prs.3 avoid-inf war-gen.sg
‘They want to avoid war.’
b. %Jie dė-jo pastang-as [kar-ui išveng-ti].
they:nom.pl put-pst.3 efforts-acc.pl war-dat.sg avoid-inf c. %Jie dė-jo pastang-as [išveng-ti kar-ui].
they:nom.pl put:pst.3 efforts:acc.pl avoid:inf war:dat.sg
‘They made efforts to avoid war.’ (Arkadiev 2014: 61–62)
Dative can also occur on the subject8 of the infinitival clause if it is interpre- ted as benefitting from the action described by the main verb (Ambrazas 2006:
557–558).
8 Infinitives may have the subject expressed as a dative argument under other circumstances as well, such as non-finite modal sentences, so I will focus on the object case in these constructions.
(39) a. Pastūm-ė kėd-ę sveči-ui atsisės-ti.
move-pst.3 chair-acc.sg guest-dat.sg sit.down-inf
‘He moved the chair for the guest to sit down.’
b. Iškas-ė griov-į vandeni-ui nutekė-ti.
dig-pst.3 ditch-acc.sg water-dat.sg flow-inf
‘They dug a ditch for the water to flow away.’
The genitive and dative of purpose are not limited to subordinate clauses but can also occur without an infinitive in the lower clause.
(40) Išsiunt-ė sūn-ų daktar-o (pakvies-ti).
sent-pst.3 son-acc.sg doctor-gen.sg invite-inf
‘(He) sent his son {for the doctor / to get the doctor}.’ (Ambrazas 2006: 557)
(41) Parvež-ėm lent-ų nam-ui (apmuš-ti).
bring-pst.1.pl board-gen.pl house-dat.sg cover-inf ‘We brought some boards {for the house/to cover the house}.’ (Ambrazas
2006: 557)
However, the infinitive is not always optional.9 Compare (35a) with (42).
(42) Iššov-ė žmon-ėms *(pagąsdin-ti)
fire-pst.3 people-dat.pl frighten-inf
‘He fired for people.’ (Franks & Lavine 2006: 271)
The dative without the infinitive, as in (41) and (42), is only possible if the action in the main clause is done for the benefit of the people. This seems to indicate the there is a strong association with the dative and the meaning of “for X purpose”.
Dative case-marked noun phrase can be added in many contexts to indicate a purpose or goal.
(43) Nusipirk-au kiaušini-ų pyrag-ui.
buy-pst.1.sg egg-gen.pl pie-dat.sg
‘I bought some eggs for the pie’
(44) Įvyk-o vakar-as, skirt-as rašytoj-ui.
take.place-pst.3 evening-nom.sg dedicated-nom.sg.m writer-dat.sg
‘There was an evening dedicated to the writer’
9 The order of the asterisk and parenthesis indicates that the exclusion of the infinitive is ungrammatical.
There is a semantic correlation between a benefactive or recipient theta role and dative case in Lithuanian. For dative noun phrases like those in (43), it appears that the dative noun phrase is an adjunct modifying the first noun:
The eggs are for the pie. In (44), the dative is an argument of the adjective skirtas
‘meant for, dedicated to’. The actual status of such datives as adjuncts or an opti- onal argument is unclear, but does not affect my claims. Likewise, the genitive has a similar meaning: The noun phrase is always interpreted as the goal of the motion in the main clause, even without an infinitive in the lower clause. The default interpretation is “to get”. Examples of the genitive without an infinitive are shown in (45) and (46), which were overheard outside a cornerstore in Vilnius.
(45) a. Išėj-o pien-o.
go.out-pst.3 milk-gen.sg
‘He went out for milk.’
b. Išsiunt-ė sūn-ų daktar-o.
sent-pst.3 son-acc.sg doctor-gen.sg
‘He sent his son for the doctor.’ (Ambrazas 2006: 557)
(46) – Kur eini? – Al-aus.
where go:prs.2.sg beer-gen.sg
‘Where are you going?’ ‘For beer.’
The goal of the present work is not to account for the assignment or licen- sing of case; Franks and Lavine (2006) and Arkadiev (2014) offer two potential analyses. Rather, my goal here is to highlight the problem presented by the case patterns in such constructions in Lithuanian. The main issue for the dative and genitive case in purpose clauses is that the semantics of such clauses allow a case other than (the expected) accusative to be licensed on the direct object. Further evidence that these should be considered semantic case, rather than structural, or inherent, is that the morphological case can affect the semantic interpretation of the sentence. This is also supported by the fact that the case marking is possi- ble without the verb present, particularly with dative. While there are many inte- resting facts regarding the distribution, summarized in Arkadiev (2014), issues of analysis are beyond the scope of the current work.
The evidence presented by Arkadiev (2014) regarding “case conflicts” in these constructions, e.g., the (in)ability of the dative to override another case, also shows that semantic case “takes precedence” (Richardson 2008: 44) over structural case. While this particular semantic case pattern may be on its way out (Arkadiev 2014 claims that accusative is acceptable, or required for some speakers), the next section features a different semantic case that has various
applications in Lithuanian: instrumental, alternating with accusative in a variety of verb classes.
5 Instrumental alternations
The final non-canonical case pattern is the alternation of accusative and instru- mental with certain verbs. There are four semantic classes of verbs that license either accusative or instrumental on the internal argument (Ambrazas 2006: 512–
513): verbs of throwing, verbs of moving a body part, verbs of making sound, and verbs of dressing/wearing clothing, shown in (47a–d), respectively.
(47) a. Berniuk-ai mėt-ė akmen-is/akmen-imis į lang-ą.
boy-nom.pl throw-pst.3 stone-acc.pl/ins/pl in window-acc.sg
‘The boys threw stones at the window.’
b. On-a trauk-ė peči-us/peči-ais.
Ona-nom.sg shrug-pst.3 shoulder-acc.pl/ins.pl
‘Ona shrugged her shoulders.’
c. Apsauginink-as žvang-in-o rakt-us/rakt-ais.
guard-nom.sg jingle-caus-pst.3 key-acc.pl/ins.pl
‘The guard jingled the keys.’
d. Moter-is avėsi/avėjo bat-us/bat-ais.
woman-nom.sg put.on:pst.3/wear:pst.3 shoe-acc.pl/ins.pl
‘The woman put on/wore shoes.’
For each of the verbal classes, there is a slight difference in meaning associa- ted with the difference in case. Generally, this difference is related to whether the internal argument is interpreted as an (affected) patient or a means of performing the action (Šukys 2006, Anderson 2011, 2013b, c). As argued by Anderson (2011), the accusative is used with these verbs when the internal argument has more fea- tures of a prototypical patient in the sense of Dowty (1991). When the internal argument is not a proto-patient, instrumental is licensed. This also corresponds with the claims of Lenartaitė (2010: 204–205) that verbs of throwing and verbs of dressing not only involve a case alternation, but a diathetic (argument structure) alternation: Only the accusative internal object is an argument of the verb, while the instrumental is more of an adjunct.
Because of these differences in meaning and argument structure, based on the morphological case assigned to the internal argument, I propose that the ins- trumental is an instance of semantic case, rather than lexical case. Furthermore,
the accusative case also plays a role in the interpretation of these sentences, accounting for why this structural case is not always replaced by the semantic case, as is the case with infinitival purpose clauses, discussed in Section 4. First, I will review the facts for each of the verbal classes (Sections 5.1–5.4) and then further discuss the implications of this alternation for case theory (Section 5.5).