Subject-object asymmetries in
Zazaki Argument Ellipsis:
A problem for the anti-agreement theory
RYAN WALTER SMITH
UNIVERSITY OF ARIZONA
1 ST N O R T H A M E R I C A N C O N F E R E N C E O N
IRANIAN LINGUISTICS
04/29/2017
Introduction
 Many languages permit argument ellipsis (AE).
 An argument may be null, permitting sloppy or
quantificational interpretations not attested with overt
pronouns.
 In some languages (Persian), sloppy/quantificational
readings are not possible in subject position.

One explanation of this is that the absence of agreement in a
language permits AE in the first place (Saito 2007), and that
subject/object asymmetries arise due to subject-verb
agreement (Sener & Takahashi 2010; Sato & Karimi 2016).
Goal of the talk
 I provide evidence from Zazaki (Northwestern
Iranian) against Saito’s anti-agreement theory of
AE.

Due to the split-ergative agreement pattern of Zazaki, the antiagreement theory predicts that AE should be possible with
subjects in the past tense, and with objects in the present.

This is not the case: Zazaki permits object AE, and disallows
subject AE, regardless of which argument is agreed with.
Overview
1.
2.
3.
4.
5.
Introduce argument ellipsis and subject/object
asymmetries therein.
Discuss the anti-agreement theory and its merits.
Introduce Zazaki AE and the problem it poses for
the anti-agreement idea.
Discuss problems for other approaches
Conclude with some possible alternatives.
Argument Ellipsis
 Many languages exhibit argument ellipsis (AE):
arguments may be left unpronounced, leading to the
possibility of the types of strict/sloppy ambiguities
observed in cases of ellipsis.

These contrast with the use of an overt pronoun, which, in
most circumstances only permit strict readings.

Extensively studied in the context of East Asian languages like
Mandarin (Huang 1991) and Japanese (Hoji 1998; Tomioka
2003; akahashi 2008;).
Strict/Sloppy ambiguities in Japanese
(1) Taroo-wa jibun-no sensei-o mita
Taroo-Top self-Gen room-Acc saw
‘Taro saw his teacher’
(2) Hanako-mo __ mita
Hanako-also saw
Hanako also saw (her own or Taro’s teacher)’
(3) Hanako-mo kare-o mita
Hanako-also he-Acc
saw
‘Hanako also saw him’ (only Taro’s teacher)
Strict/Sloppy cont.
 These are attested in the subject position in Japanese as well.
(4) Taroo-wa jibun-no teian -ga
saiyoo sareru
to
omotteiru
Taroo-Top self-Gen proposal-Nom accept do-pass-prs C think-prs
‘Taro thinks his proposal will be accepted’
(5) Hanako-mo __ saiyoo sareru
to omotteiru
Hanako-also
accept do-pass-prs C think-prs
‘Hanako also thinks (Taro’s/her own proposal) will be accepted’
(6) Hanako-mo sore-ga saiyoo sareru
to omotteiru
Hanako-also that-Nom accept do-pass-prs C think-prs
‘Hanako also thinks it will be accepted’ (only Taro’s proposal)
Sloppy readings of null objects in Persian
 Persian also permits argument ellipsis, showing the same
sloppy/strict ambiguity (Sato & Karimi 2016).
(7) Bahâr mo’allem-esh-o dust dâr-e
Bahâr teacher-3.Sg-RÂ friend have-3.Sg
‘Bahar likes her teacher’
(8) Mohsen ham _ dust dâr-e
Mohsen also friend have-3.Sg
‘Mohsen also likes (his own/Bahâr’s teacher)’
(9) Mohsen ham un-o
dust dâr-e
Mohsen also 3.Sg-RÂ friend have-3.Sg
‘Mohsen also likes him/her’ (only Bahâr’s teacher)
The subject/object asymmetry in Persian AE
 Unlike Japanese, however, sloppy readings are not generally
possible in subject position. A similar restriction is found in
Turkish (Sener & Takahashi 2010).
(10) Bahâr goft ke dust-hâ-sh miyân
Bahar said that friend-Pl-3.Sg come-3.Pl
‘Bahar said that her friends are coming’
(11) Mohsen goft ke __ ne-miyâ-n
Mohsen said that Neg-come-3.Pl
‘Mohsen said they’re not coming’ (only Bahâr’s friends)
(12) Mohsen goft ke unâ ne-miyâ-n
Mohsen said that 3.Pl Neg-come-3.Pl
‘Mohsen said they’re not coming’ (only Bahâr’s friends)
Interim Recap and a Question
 Japanese and Persian both permit AE.
 They differ in that while Japanese permits AE in
subject and object position, Persian (as well as
Turkish) does not permit it in subject position
 Why should this be?
Enter the anti-agreement theory
 Saito (2007) proposes the anti-agreement theory of AE:
(13) The anti-agreement theory of argument ellipsis
A language will allow argument ellipsis if it lacks
phi-agreement with its arguments.

Languages like Japanese and Korean lack agreement, and thus allow
AE.

Languages like English possess subject-verb agreement (as well as
agreement between v and the object to assign accusative Case), and
thus lack AE entirely.
Anti-agreement: a success for Persian?
 Languages like Persian (and Turkish) possess subject-
verb agreement.
 As such, the anti-agreement theory correctly predicts
that AE will be unavailable in the subject position of
these languages, while permitting it in object position.

As such, Sener & Takahashi (2010) and Sato & Karimi (2016) have
argued that the anti-agreement theory receives support from Turkish
and Persian, respectively
 But what if we expand the empirical domain a bit?
Enter Zazaki
 Zazaki is a Northwestern Iranian language, spoken primarily in
eastern Turkey.
 Zazaki permits AE: null objects are ambiguous between a strict and
sloppy reading.
(14) Muhsin malım-ē
Muhsin teacher-ez.3.sg.m
‘Muhsin sees his teacher’
xo
self
vēnen-o
see-3.sg.m
(15) Rıza ki vēnen-o
(16) Rıza ki ey
vēnen-o
Rıza also see-3.sg.m
Rıza also 3.sg.m.obl see-3.sg.m
‘Rıza also sees’ (strict/sloppy)
‘Rıza also sees him’ (strict only)
Zazaki: Like Persian after all?
 Like Persian and Turkish, Zazaki exhibits a subject/object asymmetry:
subjects do not permit sloppy readings.
(17)Muhsın-i
vat ke dost-ē
xo oda ken-o
pak
Muhsin-obl.sg.m said that friend-ez.sg.m self room.f do-3.sg.m clean.m
‘Muhsin said that his friend cleans the room’
(18) Rıza-y
vat ke banyo-y
k-en-o
pak
Rıza-obl.sg.m said that bathroom-obl.sg.m do.3.sg.m
clean
‘Rıza said that cleans the bathroom’
(only Muhsin’s friend)
(19) Rıza-y
vat ke o
banyo-y
ken-o
pak
Rıza-obl.sg.m said that 3.sg.m.dir bathroom-obl.sg.m do-3.sg.m clean
‘Rıza said that he cleans the bathroom’ (only Muhsin’s friend)
 At first glance, this seems to offer further support for the anti-agreement
theory.
Zazaki split-ergativity and a prediction
 Zazaki is split-ergative.
 In the present, the verb agrees with the subject, but in the past,
it agrees with the object.
 The anti-agreement theory makes a prediction here:
in the past tense, the subject/object asymmetry
should be the reverse of the one in the present tense.

We should expect the object to fail to possess sloppy readings
in the past, while the subject should permit such sloppy
readings.
A failed prediction
 This is not borne out! Null objects continue to possess
sloppy interpretations in the past tense, while null
subjects are unambiguous.
(20) Muhsin-i
dost-ē
xo di-y
Muhsin-obl.sg.m friend-ez.3.pl self saw-3.pl
‘Muhsin saw his friends yesterday’
(21) Rıza-y
ki di-y
Rıza-obl.sg.m also saw-3.pl
‘Rıza also saw (his own or Muhsin’s friends)’
A failed prediction, cont.
 Subjects are unambiguous despite lack of agreement
(22) Muhsin-i
vat ke dost-ē
xo oda kerd-e
pak-e
Muhsin-obl.sg.m said that friend-ez.sg.m self room.f did-3.sg.f. clean-sg.f
‘Muhsin said that his friend cleaned the room’
(23) Rıza-y
Rıza-obl.sg.m
vat ke banyo
kerd
said that bathroom did.3.sg.m
pak
clean.sg.m
‘Rıza said that cleans the bathroom’ (only Muhsin’s friend)
 The anti-agreement theory thus makes an incorrect prediction
about the distribution of AE in Zazaki!
Alternative approaches?
 Saito’s approach is not the only theory of AE out
there.
 Another idea has been presented by Ohtaki (2012):
that AE is only possible in languages with nonfusional case morphology.


Japanese and Korean are highly agglutinative (watashi-ga ‘Inom’), and therefore allow AE.
English case morphology is fusional (I, my, me), so no AE!
More problems
 Ohtaki’s approach isn’t fine-grained enough: ceteris
paribus, it predicts that a language will either have
AE or it won’t.

We shouldn’t expect to find languages where AE is possible in
some positions but not in others.
 Even putting that aside, the theory makes the
incorrect prediction that Zazaki shouldn’t have AE,
since Zazaki case morphology is fusional.

Case, number, and gender are fused in pronouns (ez ‘1.sg.dir’,
mı ‘1.sg.obl’), and the same features are expressed in a single
morpheme on nouns.
Where to go from here?
 There are a few possible directions to explore for AE
in these languages.
1. Sato’s (2015) idea that subject/object asymmetries
arise from definiteness/specificity restrictions on
the subject in some languages.
2. A null pronoun approach (Hoji 1998; Tomioka
2003): sloppy interpretations of null
subject/objects arise because null pronouns are
property anaphora.

Connected to the fact that many (all?) of these languages have
bare number-neutral nominals.
Conclusion and future research
 I have argued against Saito’s anti-agreement theory
of AE on the basis of evidence from Zazaki.
 I have also argued against Ohtaki’s non-fusional case
morphology approach on the way.
 Future research will look more closely at AE in
Zazaki (as well as Japanese and Persian),
manipulating factors like topicality and contextual
licensing to gain a deeper understanding of the
phenomenon
THANK YOU!!
TO HEIDI HARLEY, ROBERT HENDERSON,
SIMIN KARIMI, RYOICHIRO KOBAYASHI, AND
YOSUKE SATO FOR DISCUSSION. TO ROYA
KABIRI, MOHSEN MAHDAVI MAZDEH, RANA
NABORS FOR PERSIAN JUDGMENTS. TO MY
CONSULTANT MESUT ASMEN KESKIN FOR HIS
ZAZAKI JUDGMENTS. AND TO AUDIENCES AT
THE UNIVERSITY OF ARIZONA AND AT NACIL1
FOR LISTENING!
References
Huang, C. T. J. (1991). Remarks on the status of the null object. Principles and parameters in
comparative grammar, 56-76.
Otani, K., & Whitman, J. (1991). V-raising and VP-ellipsis. Linguistic Inquiry, 345-358.
Saito, M. (2007). Notes on East Asian Argument Ellipsis. Language Research 43(2), 203-227.
Sato, Y. (2015). Argument ellipsis in Javanese and voice agreement. Studia Linguistica, 69(1), 5885.
Sato, Y., & Karimi, S. (2016). Subject-object asymmetries in Persian Argument Ellipsis and the
Anti-agreement Theory. Glossa: a journal of general linguistics 1(1).
Sener, S., & Takahashi, D. (2010). Ellipsis of arguments in Turkish and Japanese. Nanzan
Linguistics 6, 79-99.
Takahashi, D. (2008). Quantificational null objects and argument ellipsis. Linguistic Inquiry,
39(2), 307-326.
Tomioka, S. (2003). The semantics of Japanese null pronouns and its cross-linguistic implications.
The interfaces: Deriving and interpreting omitted structures, 61, 321.
Appendix: Quantificational Readings of AE
 Takahashi (2008) points out that null arguments that
refer to quantified NPs are ambiguous between referring
to the same set of entities (the “E-type” reading) and a
reading where it refers to a different set of objects (the
“quantificational reading”).
(24) Taro-wa san-nin-no sensei-o
mi-ta
Taro-Top three-cl-gen teacher-acc see-pst
‘Taro saw three teachers’
(25) Hanako-mo __ mi-ta
Hanako-also
see-pst
‘Hanako also saw (the same three or different teachers)’
Quantificational readings in Persian AE
 Persian also permits this ambiguity.
(26) Mohsen se-tâ
mo’allem-o da’vat kard
Mohsen three-cl teacher-râ invite did
‘Mohsen invited three teachers.’
(27) Royâ ham __ da’vat kard
Royâ also
invite did
‘Royâ also invited (the same or different set of
teachers’
No quantificational readings in subject position
 As you might expect, Japanese permits
quantificational readings in subject position, but
Persian does not.
(28) Mohsen goft ke se-tâ dâneshju ingilisi mi-xun-an
Mohsen said that three-cl student english imp-read-3.pl
‘Mohsen said that three students are studying English’
(29) Rahâ goft ke farânse mi-xun-an
Rahâ said that French imp-read-3.pl
‘Rahâ said that they are studying French’ (E-type only)
Zazaki Quantificational AE in objects
 Zazaki permits quantificational readings of null objects in both the
present and the past.
(30) Muhsın hirē malım-an
dawet
k-en-o.
Muhsin three teacher-obl.pl invitation do-pres.ind-3.sg.m.
‘Muhsin will invite three teachers’
(31) Rıza ki dawet
k-en-o
Rıza also invitation do-pres.ind-3.sg.m
‘Rıza will also invite’ (Quant / E-type)
(32) Muhsin-i
hirē malım-i
dawet
kerd-i
Muhsin-obl.sg.m. three teacher-dir.pl invitation did-3.pl.
‘Muhsin invited three teachers’
(33) Rıza-y
ki dawet kerd-i
Rıza-obl.sg.m also invitation did-3.pl
‘Rıza also invited’ (Quant/E-type)
No quantificational AE for Zazaki subjects
 No quantificational readings in the subject position!
(34) Muhsin-i
vat ke hirē ṭeleb-an
İngılızki wend
Muhsin-obl.sg.m said that three student-obl.pl English read.3.sg
‘Muhsin said that three students studied English’
(35) Rıza-y
vat ke Fransızki wend
Rıza-obl.sg.m said that French
read.3.sg
‘Rıza said that studied French’ (E-type only)
(36) Muhsin-i
vat ke hirē ṭelebe-y
İngılızki
Muhsin-obl.sg.m said that three student-dir.pl English
‘Muhsin said that three students study English’
(37) Rıza-y
vat ke Fransızki wanen-ē
Rıza-obl.sg.m said that French read-3.pl
‘Rıza said that study French’ (E-type only)
wanen-ē
read-3.pl