Applicative Voice: Geminate and 'Give'-Causatives in Sason Arabic 1

Akkuş – geminate causatives in SA Tu+ 5; February 8-9, 2020

Applicative Voice: Geminate and ‘give’-causatives in b. oretman ki tı-qarri kemal lala kitab. Sason Arabic teacher be.3f 3f-read.caus Kemal this.m book ‘The teacher is making Kemal read this book.’ (Yakut 2013:33a) Faruk Akkuş1 University of Pennsylvania c. oretman ki tı-qarri lala kitab mışa kemal. [email protected] teacher be.3f 3f-read.caus this.m book to Kemal ‘The teacher is making Kemal read this book.’ (Yakut 2013:33b) 1 Introduction • In contrast, in ‘give’ causatives, the causee is introduced only as a PP. Sason Arabic (SA) has four strategies of expressing causatives: (i) ablaut, (ii) ‘make’ causatives, (iii) gemination, and (iv) ‘give’ causatives. (4) ‘give’ causatives (GiveC) • Ablaut applies to unaccusative verbs. a. ado dolab-ad-en mışa tamirci addil gave.3pl shelf-pl-their to repairman fix.inf (1) ablaut ‘They had the repairman fix their shelves.’ a. l¯ake tal-e b. tel-tu l¯ake (Lit: They gave their shelves to the repairman to fixing) stain came.out-3f came.out.caus-1sg stain b. ımm-a mışa fatma şi adıd-u addil ‘The stain came out.’ ‘I got the stain out.’ mother-her to Fatma food gave-it.m fix.inf ‘Her mother had Fatma cook the food.’ • Causatives embedded under ‘make’ (MC), (2), involve (Lit: The food, her mother gave it to Fatma to fixing) (Erguvanlı-Taylan 2017:221:30) – overt embedded theme – This strategy is a result of contact with Kurdish (Akkuş 2017; Akkuş and – infinitival verb Benmamoun 2018; Erguvanlı-Taylan 2017). – null embedded agent, interpreted as indefinite. • Today’s focus is on (3) and (4). (2) ‘make’ causatives (MC) Proposal a.a Ga sa nazf haydan. village.lord made.3m clean.inf wall • Gemination exhibits an active-passive alterna- (5) VP ‘The village lord made (someone) clean the wall.’ (Akkuş accepted:1) tion, whereas the GiveC manifests a passive structure. V VoicePappl b.a Ga sa nazf haydan mı nes-ma tawwil. • Both embed a distinct VoiceP, which assigns a village.lord made.3m clean.inf wall by person-a tall DP VoiceP 0 causee T-role as opposed to the canonical Initiator appl ‘The village lord had the wall cleaned by someone tall.’ role of VoiceP. VoicePappl VP • Gemination allows the causee to be expressed either as a DP or a PP headed by • This VoiceP can be called applicative VoiceP (à T Causee la Legate 2014). (mı)şa ‘for, to’, as in (3). V DP (3) gemination Roadmap a. kemal ku i-qri lala kitab. kemal be.3m 3m-read.ipfv this.m book §2 Active-passive in geminates and the GiveC ‘Kemal is reading this book.’ §3 Applicative VoiceP §4 Structures for geminates and the GiveC 1Many thanks to Julie Anne Legate, David Embick and Martin Salzmann for invaluable com- ments and discussions. Thanks also to thank Abbas Benmamoun, Hamid Ouali, Usama Soltan, Matt §5 Conclusions Barros, and the audience at NACAL 47 for feedback. Usual disclaimers apply. §6 Appendix Scan for PDF

1 Akkuş – geminate causatives in SA Tu+ 5; February 8-9, 2020

2 Active-passive alternation (8) a. ams adi-tu dolab-ad-i mışa tamirci addil yesterday gave.3pl shelf-pl-my to repairman fix.inf A variety of diagnostics demonstrate that geminates exhibit an active-passive alter- ‘Yesterday, I had my shelves fixed by the repairmam.’ nation (as in MC; see Akkuş 2019) and that the GiveC behaves as passives. b. ams dolab-ad-i ın-ado mışa tamirci addil (mı-nni) yesterday shelf-pl-my pass-gave.3pl to repairman fix.inf by-me • An initial clue with regard to the structure of geminates comes from passivization ‘Yesterday I made the repairman fix my shelves.’ asymmetries. (‘Yesterday, my shelves were made fixed by the repairman by me’) – With the DP causee, (6b), it is the DP causee that raises to become the grammatical subject, (6c). c. *ams (mışa) tamirci ın-ada dolab-ad-i addil (mı-nni) yesterday (to) repairman pass-gave.3m shelf-pl-my fix.inf by-me – Raising the theme leads to ungrammaticality, (6d). ‘I made the repairman fix my shelves yesterday.’ (Intended: ‘Yesterday, the repairman was made fix my shelves by me’) (6) a. leyla qar-e alu kitabad Leyla read.pst-3f these.m books • This contrast, however, does not necessarily indicate active-passive alternation ‘Leyla read these books.’ (cf. double-object vs dative-shift in English). b. qarri-tu leyla alu kitabad The evidence for the active-passive alternation and the adjunct status of the read.caus-1sg Leyla these.m books PP comes from (i) the interpretation in the absence of the causee, (ii) sluicing, (iii) ‘I made Leyla read these books.’ nonpassivizable idioms, and (iv) secondary predicates (see Appendix). c. leyla ın-qarr-e [_ alu kitabad ] (mı oratman) Leyla pass-read.caus-3f [ these.m books ] (by teacher) 2.1 The interpretation of the null causee ‘Leyla was made to read these books by the teacher.’ • The causee is optional. d. *alu kitabad ın-qarr-o [leyla _] (mı oratman) • The null causee is interpreted as existential (like a missing ‘by’-phrase) rather these.m books pass-read.caus-3pl [Leyla ] (by teacher) than pronominal (like a pro-dropped argument), (9). Intended: ‘The books were made (by the teacher) to be read by Leyla. (9) leyla qarr-e alu kitabad Leyla read.caus-3f these.m books – When the causee is a PP, (7a), the theme argument ends up as the grammatical subject, as such shows verbal agreement, (7b). YES: ‘Leyla made someone read these books.’ NO: ‘Leyla made him/her/them read the books.’ (7) a. qarri-tu alu kitabad mışa leyla • The interpretation of the null causee as existential also explains the grammati- read.caus-1sg these.m books to Leyla cality of (10a) only in the absence of a DP causee. ‘I made Leyla read these books.’ – The absence of a DP causee indicates that it is not projected, which in turn b. alu kitabad ın-qarr-o [_ mışa leyla ] (mı oratman) allows the theme argument to be raised. these.m books pass-read.caus-3pl [ to Leyla ] (by teacher) – The raising of the theme is possible regardless of whether a PP causee is ‘These books were made (by the teacher) to be read by Leyla.’2 realized or not, (10b), in line with the adjuncthood status of the PP.3

(10) a. alu kitabad ın-qarr-o [ (*Leyla) _] (mı oratman) • The GiveC patterns as like the geminates with a PP causee, thus (8). these.m books pass-read.caus-3pl [ (*Leyla) ] (by teacher) ‘The books were made (by the teacher) to be read by Leyla.’

2See e.g. Camilleri et al. 2014 for the same restriction in ditransitives in other Arabic varieties 3The adjuncthood status of the PP is also supported by clefting, in that similar to Turkish and such as Egyptian Arabic, Hijazi Arabic and Maltese. Egyptian Arabic, only arguments can be clefted. Expectedly, the PP causee may not be.

2 Akkuş – geminate causatives in SA Tu+ 5; February 8-9, 2020

b. alu kitabad ın-qarr-o [ _ (mışa leyla) ] (mı oratman) # (Someone) opened the door with a key. these.m books pass-read.caus-3pl [ (to Leyla) ] (by teacher) # The door was opened with a key. ‘The books were made (by the teacher) to be read (by Leyla).’ b. fada babe wara mıfta, #hama mı-arafe ande • The same interpretation is observed in the GiveC, as such the absence of the PP opened.3m door with key, but neg-knew-3f who causee leads to an existential reading, (11). ‘(He) opened the door with a key, #but she didn’t know who.’

(11) a. ams dolab-ad-i ın-ado mışa tamirci addil mı-nni • Given that the null causee is interpreted as an existential (cf. sect 2.1), the yesterday shelf-pl-my pass-gave.3pl to repairman fix.inf by-me following arguments also follow from an active-passive alternation, and not two ‘Yesterday my shelves were made by me to be fixed by the repairman.’ different argument structures.4 b. cf. ams dolab-ad-i ın-ado addil mı-nni • With a DP causee, the embedded structure behaves like a canonical active for yesterday shelf-pl-my pass-gave.3pl fix.inf by-me sluicing, (15), such that the remnant cannot be headed by a preposition. ‘Yesterday my shelves were made by me to be fixed (by someone /#by him).’ (15) leyla qarr-e nes-ma alu kitabad, hama m-o-re (*mışa) Leyla read. - person-a these. books, but know to 2.2 Sluicing caus 3f m neg-1sg- ande • While VP ellipsis may allow voice mismatching, sluicing does not (Merchant who 2013); also true in SA. ‘Leyla made someone read these books, but I don’t know who.’

(12) VP ellipsis • With a PP causee, the embedded clause behaves as passive for sluicing, (16). a. kemal kul çax i-xsel potad ta bad ma kınno. Kemal every time 3m-wash clothes if yet not are (16) a. leyla xassal-e alu potad, hama m-o-re *(mışa) ‘Kemal washes the clothes every time if they are not already.’ Leyla wash.caus-3f these.m clothes, but neg-1sg-know to ande b. ala bilgisayar itix in-fıde mı ande le irıllu. who this.m computer can pass-open by who that wants ‘Leyla had these clothes washed, but I don’t know by who.’ ‘This computer can be turned on by anyone who wants to.’ b. leyla qarr-e alu kitabad mışa nes-ma, hama (13) sluicing Leyla read.caus-3f these.m books to person-a, but m-o-re *(mışa) ande a. kitab ın-qara, hama m-o-re *(mı) ande neg-1sg-know to who book pass-read.3m but neg-1sg-know *(by) who ‘Leyla had these books read by someone, but I don’t know by who.’ ‘The book was read, but I don’t know *(by) who.’ • Expectedly, in the GiveC, the embedded clause behaves as passive. b. sadqe boş samaq ın-qafal-o, hama mı-arafe *(mı) ande believed.3f many ﬁsh pass-caught-3pl, but neg-knew *(by) who (17) leyla ad-e alu kitabad (mışa nes-ma) qaru, hama ‘She believes many ﬁsh to have been caught, but she didn’t know *(by) Leyla gave-3f these.m books (to person-a) read.inf, but who.’ m-o-re *(mışa) ande • An implicit agent can license sluicing, (13a), but a null pronoun (pro-dropped neg-1sg-know to who argument) cannot, (14b). ‘Leyla had these books read by someone, but I don’t know by who.’

(14) a. fada babe wara mıfta. Sluicing demonstrates that geminates exhibit an active-passive alternation, whereas opened.3m door with key the GiveC behaves as passive for sluicing. ‘(He) opened the door with a key.’ ← requires established topic 4Thanks also to a Tu+ 5 reviewer for pointing this out.

3 Akkuş – geminate causatives in SA Tu+ 5; February 8-9, 2020

2.3 Nonpassivizable idioms (22) a. ımm-u harraG-e Leyla ro le Kemal mother-his burned.caus-3f Leyla heart of Kemal SA has a class of nonpassivizable idioms, as in (18). These idioms are another test for the active-passive alternation (cf. Kayne 1975; Folli and Harley 2007). ‘His mother made Leyla break Kemal’s heart.’ b. ımm-u harraG-e ro le Kemal (mışa Leyla) (18) a. kemal qaraf faGz le şeytan mother-his burned.caus-3f heart of Kemal to Leyla Kemal broke.3m leg of devil ‘His mother had Kemal’s heart be broken (by Leyla).’ ‘Kemal finally got lucky.’ (lit. broke the devil’s leg) (23) ımm-u ad-e ro le Kemal (mışa Leyla) harx b. faGz le şeytan ın-qaraf mı kemal mother-his gave-3f heart of Kemal to Leyla burn.inf leg of devil -broke. by Kemal pass 3m ‘His mother had Kemal’s heart be broken (by Leyla).’ ‘The devil’s leg was broken by Kemal.’ ‘*Kemal finally got lucky.’ Interim Summary • These idioms may occur in geminates only in the case of a DP causee, (19a), but • Geminates show an active-passive-like alternation, and the GiveC a passive not a PP causee, (19b). structure. • The DP causee in geminates is an argument, (19) a. nihayet qarrıf-tu kemal faGz le şeytan • The PP causee in both geminates and the GiveC is an adjunct like a ‘by’- finally broke.caus-1sg Kemal leg of devil phrase. ‘I finally made Kemal get lucky.’ (lit. broke the devil’s leg) b. nihayet qarrıf-tu faGz le şeytan (mışa kemal) 3 Applicative VoiceP finally broke.caus-1sg leg of devil to Kemal ‘I finally had the devil’s leg broken by Kemal.’ Although the embedded event in geminates and the GiveC does contain a second NOT: Kemal finally got lucky. VoiceP, it is not a canonical VoiceP, but an applicative VoiceP. • These idioms are also not possible in the GiveC, (20). • This applicative VoiceP assigns a different T-role (causee versus initiator);

(20) adi-tu faGz le şeytan (mışa kemal) qarf – (i) instrument phrases, (ii) agent-oriented adverbs, or (iii) agent-oriented gave.caus-1sg leg of devil to Kemal break.inf comitatives cannot be associated with the embedded causee. ‘I finally had the devil’s leg broken by Kemal.’ – (iv) the causee is introduced with a different preposition than canonical NOT: Kemal finally got lucky. agents are introduced with. • Idioms of this sort contrast with passivizable idioms, (21). 3.1 Instrument phrases (21) a. kemal hatarax ro-i • Instrumentals are diagnostics for an external argument layer (i.e. Voice) (Bruen- Kemal burned.3m heart-my ing 2013; Alexiadou et al. 2015, also Fillmore 1968). ‘Kemal broke my heart.’ Lit: ‘Kemal burned my heart.’ (24) a. bina ın-faşş-e mı işçiyad wara çakuçad apartment pass.pfv-demolished-3f by employees with hammers b. ro-i ın-hatarax mı kemal ‘The apartment was demolished by the employees with hammers.’ heart-my pass-burned.3m by Kemal ‘My heart was broken by Kemal.’ b. *bina ın-qalab-e mı rua wara çakuçad apartment nact-fell.over-3f by itself with hammers • Unlike non-passivizable idioms, which require a DP causee, such idioms impose ‘The apartment fell over by itself with hammers.’ no restriction; (22) for geminates and (23) for the GiveC. • They are also grammatical in the MC, and can modify the embedded agent, (25).

4 Akkuş – geminate causatives in SA Tu+ 5; February 8-9, 2020

(25) a. si-to aGet şurvan wara ibre b. ım-mu ade lalu potad mışa kemal xassil bı sabır made-2pl sew.inf pants with needles mother-his gave.3f these clothes to Kemal wash.inf with patience ‘You had someone [sew the pants with needles].’ ‘His mother made Kemal wash these clothes patiently.’ YES: His mother was patient. b. kemal sa buaG sir glımboz-e wara sope. NOT: Kemal was patient. Kemal made.3m paint do.inf turtle-f with stick ‘Kemal, with the stick, had [someone paint the turtle].’ ‘Kemal had [someone paint the turtle with the stick].’ 3.3 Choice of the preposition • Instrumentals pick out only the causer in both the geminates, (26), and the The causee diﬀers from the canonical VoiceP Initiator in terms of the preposition GiveC, (26b). heading the PP adjunct. • The PP adjunct in both short passives and the MC are headed by the preposition (26) a. ım-mu xassle hasan potad wara furça gbir-e mı ‘by, from’, (29). mother-his washed.caus.3f Hasan clothes with brush big-f ‘His mother made Hasan wash the clothes with a big brush.’ (29) a. ala cam mı kemal ın-qaraf bı-l-qasti. YES: His mother used the brush [to force Hasan to do washing possibly this glass by Kemal pass-broke.3m with-the-intention with another instrument]. ‘This glass was broken by Kemal deliberately.’ NOT: Hasan used the brush. b. ım-mu ade lalu potad mışa hasan xassil wara furça gbir-e b. kemal sa xassil potad mı mara-ma pir-e. mother-his gave.3f these clothes to Hasan wash with brush big-f kemal made.3m wash.inf clothes by woman-a old-f ‘His mother made Hasan wash the clothes with a big brush.’ ‘Kemal had the clothes washed by some old woman.’ YES: His mother used the brush ... • The PP adjunct causee in both geminates, (30a), and the GiveC, (30b), is headed NOT: Hasan used the brush. by preposition mışa ‘to, for’.

3.2 Agent-oriented adverbs (30) a. oretman ki tı-qarri lala kitab mışa kemal. teacher be.3f 3f-read.caus this.m book to Kemal Agent-oriented adverbs in SA provide another testing ground wrt the T-role the ex- ‘The teacher is making Kemal read this book.’ ternal argument of the embedded event bears (Ernst 2001; Matsuoka 2013, i.a.). b. ams adi-tu dolab-ad-i mışa tamirci addil • In the MC, these adverbs can modify the action of the embedded agent, (27). yesterday gave.3pl shelf-pl-my to repairman fix.inf (27) bolum tı-si mez sınavad le qabul wara diqqat. ‘Yesterday, I had my shelves fixed by the repairmam.’ department 3f-make look.inf tests of acceptance with care Summary ‘The department makes (someone) [check acceptance tests carefully].’ • Geminates show an active-passive-like alternation, and the GiveC a passive structure. • Agent-oriented adverbs cannot be associated with the causee in either geminates, (28a), or the GiveC, (28b); they exclusively target the causer.5 • This embedded VoiceP assigns a different T-role (causee versus initiator) than the canonical VoiceP. (28) a. oratman ki tı-qarri kemal lala kitab bı sabır teacher be.3f 3f-read.caus Kemal this.m book with patience 4 Structures for geminates and the GiveC ‘The teacher is making Kemal read this book patiently.’ YES: The teacher is patient. • Legate’s (2014) analysis of passive: NOT: Kemal is patient. – The passive is a variant of a functional head that introduces a DP in its 5Agent-oriented commitatives behave identically. See Appendix. specifier,

5 Akkuş – geminate causatives in SA Tu+ 5; February 8-9, 2020

– This configuration could be common to both VoiceP and ApplP (see also Anagnostopoulou 2003, Alexiadou et al. 2006, Schäfer 2012 for the sugges- b. VoiceP tion that an applicative head introduces the non-canonical external arguments, i.e. oblique causers. cf. Pylkkänen 2008). DP Voice’ • One prediction of this analysis is that an active-passive-like alternation should be possible on Appl. ... Voice – True in Acehnese. (Initiator) ... • Similar to its Voice counterpart, this passivization does not necessarily end up VoiceapplP with a morphological reflex (e.g. Harley 2017b; Pitteroff 2014, 2015).

∃+Voiceappl VP (31) Active ApplP θ a. qarri-tu leyla alu kitabad V DP read.caus-1sg Leyla these.m books ‘I made Leyla read these books.’ – In the second semantic denotation, Applpass leaves the initiator position b. VoiceP open, i.e. λe.λx.causee(x,e), to be accessed and saturated by the ‘to’- phrase. DP Voice’ (33) Passive ApplP with ‘to’-phrase ... Voice a. qarri-tu alu kitabad mışa leyla (Initiator) read.caus-1sg these.m books to Leyla ... VoiceapplP ‘I made Leyla read these books.’

0 DP Voiceappl b. VoiceP

Voiceappl VP DP Voice’ θ ... V DP Voice (Initiator) • Voicepass has two associated semantic denotations (see also Bruening 2013; Alex- 6 ... iadou et al. 2015; Legate 2014; Legate and Akkuş 2017). VoiceapplP – In the ﬁrst one, which does not combine with a ‘to’-phrase, the initiator is Voice P PP existentially bound on the Applpass head. appl

(32) Passive ApplP without ‘to’-phrase Voiceappl VP a. qarri-tu alu kitabad θ read.caus-1sg these.m books V DP ‘I had these books read.’ • The causee may be generated as a DP in Spec,ApplP, and must become the 6Semantically, the passive needs to allow the external θ-role to be satisﬁed by the ‘by’-phrase grammatical subject when passivized (see e.g. Rackowski and Richards(2005); (‘to’-phrase in the CG and geminates), when present, and to otherwise be interpreted existentially Legate(2014) for high ApplP in Austronesian languages). (e.g. Bach 1980; Keenan 1985; Williams 1987; Parsons 1990; Bruening 2013; Alexiadou et al. 2015; Reed 2018). • It receives a causee T-role from the Appl head, as in (31b).

6 Akkuş – geminate causatives in SA Tu+ 5; February 8-9, 2020

• Alternatively, the causee may be introduced like the initiator in the canonical References passive:

– in a PP adjunct, in which the P assigns a causee T-role to its DP complement, Aissen, Judith, and Jorge Hankamer. 1980. Lexical extension and grammatical trans- this causee being tied semantically to the causee T-role introduced by Appl, formations. In Proceedings of the Sixth Annual Meeting of the Berkeley Linguistics as in (32b). Society, 238–249. – without a PP adjunct, the causee is interpreted existentially on the Appl Akkuş, Faruk. 2017. Peripheral Arabic Dialects. In The Routledge Handbook of Arabic head. Linguistics, ed. Elabbas Benmamoun and Reem Bassiouney, 454–471. Routledge. Akkuş, Faruk. 2019. Evidence from Sason Arabic for A-movement¯ feeding licensing relations. Revised & resubmitted to Linguistic Inquiry. 4.1 SA as a high applicative language Akkuş, Faruk. accepted. Variable embedded agent in Sason Arabic. Journal of Lin- • Haddad(2014) shows that non-argumental attitude datives in Lebanese Arabic guistics. may occur with unergative predicates (see also Al-Zahre and Boneh(2010, 2016) Akkuş, Faruk, and Elabbas Benmamoun. 2018. Syntactic outcomes of contact in for the same argument in Syrian Arabic and Hebrew). Sason Arabic. In Arabic in contact, ed. Stefano Manfredi, Mauro Tosco, and Giorgio Banti, volume 6, 37–52. John Benjamins. (34) a. Na:dya ştaGalit-la: şi nisQ se:Qa Al-Zahre, Nisrine, and Nora Boneh. 2010. Coreferential dative constructions in syrian Nadia worked-her.dat some half hour arabic and modern hebrew. Brill’s Journal of Afroasiatic Languages and Linguistics ‘Nadia worked [her] for about a half hour.’ (Haddad 2014:66) 2:248–282. Al-Zahre, Nisrine, and Nora Boneh. 2016. Pronominal non-core datives in syrian b. Pibn Na:dya harab-la: min l-madrase marra te:nye arabic. Brill’s Journal of Afroasiatic Languages and Linguistics 8:3–36. son Nadia escaped-her.dat from the-school second time Alexiadou, Artemis, Elena Anagnostopoulou, and Florian Schäfer. 2006. The prop- ‘Nadia’s son ran away [her] from school again.’ (Haddad 2014:88) erties of anticausatives crosslinguistically. In Phases of interpretation, ed. Mara Frascarelli, volume 91, 187–211. Mouton de Gruyter. • These datives merge as high applicatives in Lebanese Arabic (Haddad 2014). Alexiadou, Artemis, Elena Anagnostopoulou, and Florian Schäfer. 2015. External • SA also has such affected datives, (35), thus SA is also independently a high arguments in transitivity alternations: A layering approach, volume 55. Oxford applicative language. University Press. Anagnostopoulou, Elena. 2003. The syntax of ditransitives: Evidence from clitics. (35) a. kemali şarab-lui/k mayn Berlin/New York: Mouton de Gruyter. Kemal drank-him water Bach, Emmon W. 1980. In defense of passive. Linguistics and Philosophy 3:297–341. ‘Kemal drank [him] water.’ Bruening, Benjamin. 2013. By-phrases in passives and nominals. Syntax 16:1–41. Bruening, Benjamin, and Thuan Tran. 2015. The nature of the passive, with an b. leyla boş faqaz-ı-nni ams analysis of vietnamese. Lingua 165:133–172. Leyla much ran-3f-me yesterday Camilleri, Maris, Shaimaa ElSadek, and Louisa Sadler. 2014. A cross dialectal view ‘Leyla ran [me] a lot yesterday.’ of the arabic dative alternation. Acta Linguistica Hungarica 61:3–44. Deal, Amy Rose. 2009. The origin and content of expletives: Evidence from “selection”. Syntax 12:285–323. 5 Conclusions Erguvanlı-Taylan, Eser. 2017. Language Contact in Anatolia: the case of Sason Arabic. In Endangered languages of the caucasus and beyond, ed. Ramazan Korkmaz and • Geminate causatives in SA manifest an active-passive alternation, whereas ‘give’ Doğan Gürkan, 209–225. Brill. causatives exhibit a passive configuration. Ernst, Thomas. 2001. The syntax of adjuncts, volume 96. Cambridge University Press. • Both embed a second VoiceP, however this VoiceP exhibits distinct behavior Fillmore, Charles. 1968. The case for Case. In Universals in linguistic theory, ed. from the canonical, agentive VoiceP, which warrants identifying it as a distinct E. Bach and R.T. Harms. Holt, Rinehart and Winston. category. Folli, Raffaella, and Heidi Harley. 2007. Causation, obligation, and argument struc- • The causee in both constructions is generated in applicative Voice. ture: On the nature of little v. Linguistic Inquiry 38:197–238.

7 Akkuş – geminate causatives in SA Tu+ 5; February 8-9, 2020

Haddad, Youssef A. 2014. Attitude datives in lebanese arabic and the interplay of tax. Doctoral Dissertation, University of Pennsylvania. syntax and pragmatics. Lingua 145:65–103. Thráinsson, Höskuldur. 2007. The syntax of icelandic. Cambridge University Press. Harley, Heidi. 2017a. The “bundling” hypothesis and the disparate functions of lit- Williams, Edwin. 1987. Implicity arguments, the binding theory, and control. Natural tle v. In The verbal domain, ed. Roberta D’Alessandro, Irene Franco, and Ángel Language & Linguistic Theory 5:151–180. J. Gallego, 3–28. Oxford University Press. Yakut, Ayşe B. 2013. Syntax of Sason Arabic: A Descriptive Re- Harley, Heidi. 2017b. ‘Passive’ as agreement: Suppressing subjects in Hiaki. University view. Manuscript, Boğaziçi University URL https://cpb-us-w2. of Arizona, manuscript. wpmucdn.com/web.sas.upenn.edu/dist/4/121/files/2016/05/ Holisky, Dee Ann. 1987. The case of the intransitive subject in Tsova-Tush (Batsbi). Yakut-2013-Syntax-of-Sason-Arabic-2id3xfy.pdf. Lingua 71:103–132. Jung, Hyun-Kyoung. 2014. On applicatives and causatives. Doctoral Dissertation, University of Arizona. Kayne, Richard. 1975. French syntax. the transformational cycle. Cambridge, MA: MIT Press. Keenan, Edward L. 1985. Passive in the world’s languages. In Language typology and syntactic description, volume 1: Clause structure, ed. Timothy Shopen, 243–281. Cambridge University Press. Key, Gregory. 2013. The morphosyntax of the turkish causative construction. Doctoral Dissertation, University of Arizona. Legate, Julie Anne. 2014. Voice and v: Lessons from acehnese. MIT Press. Legate, Julie Anne, and Faruk Akkuş. 2017. Turkish passive impersonals. Paper presented at the 48th meeting of the North East Linguistic Society, University of Iceland, Iceland. Levin, Beth, and Malka Rappaport Hovav. 1995. Unaccusativity: At the syntax-lexical semantics interface. Cambridge, MA: MIT Press. Matsuoka, Mikinari. 2013. On the notion of subject for subject-oriented adverbs. Language 89:586–618. Merchant, Jason. 2013. Voice and ellipsis. Linguistic Inquiry 44:77–108. Munro, Pamela, and Lynn Gordon. 1982. Syntactic relations in western muskogean: A typological perspective. Language 81–115. Parsons, Terence. 1990. Events in the Semantics of English, volume 5. Cambridge, Ma: MIT Press. Pitteroﬀ, Marcel. 2014. Non-canonical ‘sich lassen’ middles. Doctoral Dissertation, University of Stuttgart. Pitteroﬀ, Marcel. 2015. Non-canonical middles: a study of personal let-middles in German. The Journal of Comparative Germanic Linguistics 18:1–64. Pylkkänen, Liina. 2008. Introducing arguments. Cambridge, MA: MIT Press. Rackowski, Andrea, and Norvin Richards. 2005. Phase edge and extraction: A tagalog case study. Linguistic Inquiry 36:565–599. Reed, Lisa A. 2018. Against control by implicit passive agents. In Romance Languages and Linguistic Theory 14: Selected papers from the 46th Linguistic Symposium on Romance Languages (LSRL), volume 14, 279–292. John Benjamins. Schäfer, Florian. 2012. Two types of external argument licensing: the case of causers. Studia Linguistica 66:128–180. Sigurðsson, Einar F. 2017. Deriving Case, Agreement and Voice Phenomena in Syn-

8 Akkuş – geminate causatives in SA Tu+ 5; February 8-9, 2020

6 Appendix (38) a. kemali kar-a xanni (sarxoşi ). Kemal wrote-3m song (drunk) • Sluicing ‘Kemal composed the song drunk.’ Diﬀerent interpretations in the GiveC depending on whether sluicing targets the main b. xanni ın-kara (??sarxoş) (mı nes-ma). clause or the embedded clause, (36). song.m pass-wrote.m (??drunk) by someone – In (36a), the remnant mı ande “by who” indicates that the sluice can only ‘The song was composed drunk by someone.’ target the matrix clause, an impersonal passive, not the caused event “build”. • Secondary predicates are not licensed with the GiveC, (39), but are compatible – In (36b), the remnant mışa ande “to who” indicates that it can only target with geminates only when the causee is a DP, (40). the caused event “build” in the complement of “give”. – In either interpretation, leaving out the preposition on the remnant results (39) GiveC: Depictives Impossible in ungrammaticality, (36c). a. nanai mı-na-di daq zıGar-na ek (sarxoşi/??k ) . we neg-1pl-give beat.inf children-our (drunk) (36) a. ın-ada beyt mışa nes-ma addil, hama m-ore mı ande ‘We don’t let anyone beat our children drunk.’ pass-gave house to person-a build.inf but neg-know.1sg by who ‘It was made someone build the house, but I don’t know by who’ b. beaqıl ye dar hamıl haşiş (??bitkin). YES: who made somebody build the house unwise cop.3sg give carry.inf grass (??tired) NO: who built the house ‘It would be unwise to make someone carry the grass tired.’

b. ın-ada beyt mışa nes-ma addil, hama m-ore mışa (40) Geminates pass-gave house to person-a build.inf but neg-know.1sg to a. Depictives Possible with DP causee ande who nanai qarri-na kemalk kitab-na (sarxoşi/k ). we read. Kemal book-our (drunk) ‘It was made someone build the house, but I don’t know by who’ caus-1pl YES: who built the house ‘We made Kemal read our book drunk.’ NO: who made somebody build the house b. Depictives Impossible with null causee c. ın-ada beyt mışa nes-ma addil, hama m-ore *(mışa / haşiş nana hammıl-na-u e (sarxoş ). pass-gave house to person-a build.inf but neg-know.1sg *(to / i k i/??k grass we carried.caus-1pl-3m (drunk) mı) ande by) who ‘The grass, we made someone carry it drunk.’

• Secondary Predicate Licensing c. Depictives Impossible with PP causee nana hammıl-na haşiş mışa işçiyad (sarxoş ). Depictives require projection in SA: not allowed in passives even when the agent is i i/??k we carried.caus-1pl grass to workers (drunk) realized as a PP. k ‘We made the workers carry the grass drunk.’ (37) a. nes-ma amal araba (sarxoş ) i i • Agent-oriented comitatives person-a drove car drunk They tend to pattern with instrument phrases and agent-oriented adverbs in picking ‘Someonei drove the car (drunki).’ out an external argument layer (Bruening 2013; Alexiadou et al. 2015). b. araba ın-amal-e (??sarxoş) car.f pass-drove-f (??drunk) (41) a. bina ın-faşş-e wara sırraG ﬁ-ya ‘The car was driven (??drunk).’ apartment pass-demolish-3f with burglar in-it.f ‘The apartment was demolished with the burglar inside.’

9 Akkuş – geminate causatives in SA Tu+ 5; February 8-9, 2020

(the burglar was helping with the demolishing from inside) • It is hard to ﬁnd an import of this in NOM-ACC languages, yet ergative languages are useful. b. bina ın-qalab-e wara sırraG ﬁ-ya In Tsova-Tush, where a single intransitive verb can appear with either nominative apartment nact-fall over-3f with burglar in-it.f or ergative marking. ‘The apartment fell over with the burglar inside.’ (the burglar was inside when the building fell over) • Holisky(1987:105) points out that “the nominative and the ergative can be used with one and the same intransitive verb, depending on the responsibility or ac- • The comitative reading is also available in the MC, (42). tiveness of the subject”7

(42) kemal sa hamıl mase wara hasan (45) a. (as) vuiž-n-as Kemal made carry.inf table with Hasan 1sg.erg fell.aor-1sg.erg ‘Kemal made someone carry the table with Hasan.’ ‘I fell down, on purpose.’ (Hasan helped carry the table) b. so vož-en-sO • In the case of geminates and the GiveC, however, the comitative reading is not 1sg.nom fell.aor-1sg.nom available with the causee, but only with the matrix causer. ‘I fell down, by accident.’ (Tsova-Tush; Holisky 1987:105)

(43) a. leyla hammıl-e kemal mase wara hasan Creek (Muskogean) uses three types of agreement, I, II, and III ∼ Agent, Patient, Leyla carried.caus-3f Kemal table with Hasan Dative. Depending on the thematic roles/volitionality, fluid verbs get marked either I or II (Munro and Gordon 1982). ‘Leyla made Kemal carry the table with Hasan.’ YES: Leyla and Hasan made Kemal carry the table. (46) a. transitive NO: Leyla made Kemal and Hasan carry the table ca-ìakpaal-íck-is b. leyla ade mase mışa kemal hamıl wara hasan 1sII-turn.lg-2sI-ind Leyla gave table to Kemal carry.inf with Hasan ‘You are turning me over.’ ‘Leyla made Kemal carry the table with Hasan.’ a. unergative YES: Leyla and Hasan made Kemal carry the table. iimoomiit nocaay-ey-s. NO: Kemal and Hasan carried the table on.purpose yawn-1sI-ind • On T-role, applicatives and causatives ‘I am yawning on purpose.’ A Tu+ 5 reviewer questions the connection between the T-role in the embedded structure and the type of VoiceP, citing Bruening and Tran(2015). b. unaccusative moyhceys ca-nocaay-is. (44) a. The sentient, talking door opened wide deliberately. not.meaning.to 1sII-yawn-ind b. The sentient, talking door slammed shut deliberately. ‘Not meaning to, I am yawning.’ c. The Iceman froze solid deliberately. Secondly, even for English, the speakers I have consulted all prefer to have ‘itself’ in the examples in (44), pointing to an unergative structure. As such, d. The robot broke open deliberately. (Bruening and Tran 2015:40) • Bruening and Tran(2015) cite these examples to argue that adverbs such as (47) a. *Dora shouted hoarse. ‘deliberately’ do not require an agent. b. Dora shouted herself hoarse. • However, I believe it is not as simple or innocuous an assumption as they suggest. 7In Tsova-Tush, 3rd person is always ergative, and only for 1st/2nd person, this alternation with the adverb is attested. Transitive clauses pattern like 3rd person. This property indicates that this • First, it could indeed be that clauses with ‘deliberately’ (or other adverbs ex- can’t be because we are dealing with different langauge types; it is the same language using this pressing volitionality) have a different structure. morphology.

10 Akkuş – geminate causatives in SA Tu+ 5; February 8-9, 2020

(48) a. The river froze solid. (Levin and Rappaport Hovav 1995:39) b. kelp bıl-qasti faqaz. dog with-purpose ran.3m b. *The river froze itself solid. ‘The dog ran deliberately.’ – This might mean that even for speakers who accept (44), they might have a null reflexive. Icelandic provides another testing ground for the role of agentivity: in general, the Impersonal Passive is grammatical with unergatives only, not unaccusatives. Moreover, clauses with strictly unaccusative syntax are infelicitious with such adverbs. • However, various examples of unaccusatives are found. As suggested in Thráins- son(2007:268), “semantic features like agentivity or volition may play a role in • Deal(2009) shows that there-insertion rules out transitives, unergatives and in- licensing the impersonal passive in Icelandic” (see also Sigurðsson 2017:366). choatives. – Only a subset of unaccusatives are possible with there-insertion in (49). (53) a. þaD var fariD snemma af staD. expl was gone early from place (49) There appeared a shadowy figure in the doorway. ‘People left early.’ likewise: accumulate, coexist, emerge, hover, ... (Deal 2009:1a) b. Enn er barist og dáiD fyrir föDurlandiD. • Speakers find a contrast regarding the availability of ‘deliberately’ in clauses with still is fought and died for the.fatherland and without there. ‘People are still fighting and dying for their fatherland.’ (Sigurðsson 2017:(249)) (50) a. A shadowy figure deliberately appeared in the doorway.8 • Sigurðsson(2017) argues that such configurations involve a weak object position, b. *There appeared deliberately a shadowy figure in the doorway. as such they also require a reflexive (thus patterning like unergatives again). • This contrast indicates that in truly unaccusative structures, certain adverbs are • Moreover, they involve additional agent role, and an agent position. indeed disallowed. • Moreover, the connection between T-role and the phrase hosting the causee has Furthermore, in SA, there is a sharp contrast between the availability of adverbs already been argued for in the context of other languages. such as ‘deliberately’ and the predicate type. • Key(2013) and Harley(2017a) argue that in Turkish the productive causatives – Unaccusative verbs are not compatible with agentive adverbs even when are headed by a dedicated CausP (or an ApplP with a null applicative head, e.g. coerced. Jung 2014). (51) a. # nahar bıl-qasti tala ala sari. • This argument has been used to explain why agent-oriented adverbs cannot pick sun with-purpose appeared.3m this morning the causee, (54). ‘The sun appeared deliberately this morning.’ (54) anne çocuğ-a kitab-ı isteksizce oku-t-tu. b. # şelç/baGling bıl-qasti hedi zab. mother child-dat book-acc reluctantly read-caus-pst snow/Iceman with-purpose slow melted.3m ‘The mother reluctantly made [the child read the book].’ ‘The snow / Iceman melted slowly deliberately.’ NOT: ‘The mother made [the baby read the book reluctantly].’ – On the other hand, unergatives are licit with such adverbs. • This could also explain why the causative of a passive is impossible.

(52) a. baGling bıl-qasti zay. (55) a. *Kadın et-i kasap tarafından kes-il-dir-di. Iceman with-purpose laughed.3m woman meat-acc butcher by cut-pass-caus-pst ‘Iceman laughed deliberately.’ ‘The woman had the meat be cut by the butcher.’ (Aissen and Hankamer 1980:239)

8The adverb is placed before the verb to avoid the confound Deal calls ‘outside verbals’.