case and meaning in uyghur nominalized...

Case and Meaning in Uyghur Nominalized Clauses

Alya Asarina

March 18, 2010

1 Introduction

We will consider the relationship between case-marking and the scope and definiteness of

noun phrases in Uyghur. The focus will be on nominalized embedding constructions:

(1) q1z-(n1N)girl-(gen)

kil-iS-icome-ISH-3

mohimimportant

‘It’s important for a girl to come.’


kil-iS-icome-ISH-3

kireknecessary

‘It’s necessary for a girl to come.’

There are major differences between:

• embedding by modal and non-modal adjectives1

• genitive-marked and unmarked subject options

Under non-modal adjective (example (1)):

• The subject remains inside the -ish phrase.

• An unmarked subject must scope below the adjective, while a genitive-marked

subject can take high or low scope.

Under modal adjective (example (2)):

• The subject raises out of the -ish phrase.

• An unmarked subject can take high or low scope, while a genitive-marked subject

must scope above the adjective.

Proposal for modal vs. non-modal adjectives:

(3) a. The -ish phrase that combines with non-modal adjectives is a DP.

1Non-modal adjective is used here simply as a term to distinguish the adjectives discussed in this sectionfrom those presented in section 5.

1

Alya AsarinaCase and Meaning in Uyghur Nominalized Clauses

Generals DefenseMarch 18, 2010

b. The -ish phrase that combines with modal adjectives is an NP/GenP.

c. T in Uyghur has an EPP property that must be satisfied by a DP.

Consequently:

(4) a. When the embedding predicate is a non-modal adjective, the -ish phrase raises

to spec, TP to satisfy EPP.

b. When the embedding predicate is a modal adjective, the subject of the -ish

phrase raises to spec, TP to satisfy EPP.

The optional genitive-marking on the subject is derived from the optional presence of the

functional head Gen, which:

• assigns overt genitive case

• has an EPP feature

Scope facts are deduced base on structural position together with the following principle:

Restriction on reconstruction in Uyghur:

(5) Reconstruction of overtly case-marked noun phrases is prohibited.

The account assumes Diesing (1992).

2 Background

In this section:

• theoretical background (Diesing (1992))

• data on Uyghur direct objects

• analysis of Uyghur direct object data, including key components of later proposals:

– optionally present case-assigning functional head

– prohibition against reconstruction for case-marked noun phrases

2.1 Proposal of Diesing (1992)

• There is an existential closure operator at the VP/vP level that binds all free variables

in its scope.

Two options for indefinites:

• binding by existential closure

• generalized quantifier, Quantifier Raising (QR)

2



(6) A girl came.

(7) a. Existential Closure: b. QR:

∃ vP

(a) girl came

a girl1 ∃ vP

1 came

2.2 Uyghur Direct Objects: Data

Uyghur direct objects can be:

• accusative-marked & high

• unmarked & low

Uyghur direct objects:2

(8) MehemmetMehemmet

nan-*(ni)bread-*(acc)

jaXSiwell

jaX-t-ibake-past-3

‘Mehemmet baked the bread well.’


jaXSiwell

nan-(*ni)bread-(*acc)

jaX-t-ibake-past-3

‘Mehemmet baked bread well.’

Accusative-marked objects:

• are syntactically high

• scope above the verb

• can be definite

Unmarked objects:

• are syntactically low

• scope below the verb

• cannot be definite

Uyghur DO’s – case-marking and interpretation:

(10) menI

doXtordoctor

izi-wat-i-menlook.for-prog-impf-1sg

‘I’m looking for a doctor.’ [de dicto]

(11) menI

doXtor-nidoctor-acc


‘I’m looking for a (specific) doctor.’ [de re]

2Data from MIT Spring 2009 graduate field methods (24.942) class notes.

3



Uyghur definite DO – accusative marking obligatory:3

(12) AntonAnton

Laura-*(ni)Laura-*(acc)

kor-d-isee-past-3

‘Anton saw Laura.’

(13) quSbird

buthis

dan-*(ni)worm-*(acc)

kor-d-isee-past-3

‘The bird saw this worm.’

2.3 Analysis of Turkish and Uyghur Direct Objects

It has been commonly proposed (Diesing (1992), Kelepir (2001), Cagri (2005)) that Turkish

unmarked objects are inside VP, while accusative-marked direct objects are outside of VP.

Concrete proposal for Uyghur:

vCASE:

• assigns overt accusative


• is higher than v θ

• projected optionally

v θ:

• assigns subject theta-role

• is lower than vCASE

• v θP is the scope of existential closure

(14) a. vCASE: b. No vCASE:

vCASEP

DO-acc vCASE′

v θP

. . . tDO . . .

vCASE

v θP

. . .DO-∅ . . .

No existential closure for accusative-marked object:

(15) menI

doXtor-nidoctor-acc


3Data from MIT Fall 2009 undergraduate field methods (24.910) class notes.

4



‘I’m looking for a (specific) doctor.’ [de re]

* ‘I’m looking for a doctor.’ [de dicto]

Existential closure possible for subject:

(16) ihtdog

sirt-taoutside-loc

qawi-wat-i-dubark-prog-impf-3

‘Dogs are barking outside.’

What is the relevant difference between objects and subjects?

• The scope of existential closure is v θ.

• Accusative-marked objects cannot reconstruct into base position from spec, vCASEP.

• Unmarked subjects can reconstruct into spec, v θ from spec, TP.



(18) Accusative-marked object:

vCASEP

DO-acc

DO > V

vCASE′

v θP

. . . tDO . . .

vCASE

(19) Subject:

TP

Subj

QP indefinite

T′

v θP

tSubj

∃ bound indef.

v θ′

. . .

T

Definites are not permitted under the scope of existential closure.

5



(20) Case-Marking and Interpretation (interim summary):

definite interpretation of indef.

environment case-marked? allowed? high? low?

direct objectyes ✓ ✓ ✗

no ✗ ✗ ✓

3 Non-Modal Proposition-Taking Adjectives

In this section:

• -ish phrases, and their similarities to possessed DP’s

• analysis of non-modal adjective embedding of -ish phrases, including the relationship

between case and interpretation

3.1 The Nominal Nature of -ish Phrases

A non-modal adjective like mohim (‘important’) can embed:

• a regular DP

• an -ish phrase

Possessed DP:

(21) min-*(1N)I-*(gen)

kitav-imbook-1sg

mohim/eXmijetlikimportant/useful

‘My book is important/useful.’

Possessed -ish phrase:

(22) min-*(1N)I-*(gen)

oqu-S-imread-ISH-1sg


‘My reading is important/useful.’

Proposal for -ish:

(23) -ish is a nominalizing suffix.

-ish phrases:

• bear possessor agreement (see above)

• bear case morphology (see below)

The subject of a directly embedded proposition in Uyghur can be accusative-marked or

unmarked. (Shklovsky and Sudo (2009))

• unmarked → shifted interpretation for pronouns

6



• accusative-marked → non-shifted interpretation for pronouns

Verbal predicate – unmarked subject, shifted:

(24) AhmetAhmet

[min-1N[I-gen

q1z-imgirl-1sg

ket-t-i]leave-past-3]

didisaid

‘Ahmet said that his daughter left.’ [shifted]

* ‘Ahmet said that my daughter left.’ [non-shifted]

Verbal predicate – accusative-marked subject, non-shifted:

(25) AhmetAhmet

[min-1N[I-gen

q1z-im-nigirl-1sg-acc


didisaid

* ‘Ahmet said that his daughter left.’ [shifted]

‘Ahmet said that my daughter left.’ [non-shifted]

The same pattern is found with an -ish phrase embedded under a non-modal adjective.

Non-modal adjective – unmarked -ish phrase, shifted:

(26) Otkur

Otkur

[min-1NI-gen


mohim]important

didisaid

‘Otkur said that his studying is important.’ [shifted]

* ‘Otkur said that my studying is important.’ [non-shifted]

Non-modal adjective – accusative-marked -ish phrase, non-shifted:

(27) Otkur

Otkur

[min-1NI-gen

oqu-S-im-niread-ISH-1sg-acc

mohim]important

didisaid

* ‘Otkur said that his studying is important.’ [shifted]

‘Otkur said that my studying is important.’ [non-shifted]

Conclusion: An -ish phrase embedded by a non-modal adjective behaves just like a pos-

sessed DP.

Proposed below: This type of -ish phrase can satisfy EPP on T.

3.2 Case-Marking and Interpretation of -ish Phrase Subjects

The subject of an -ish phrase under a non-modal adjective can be genitive-marked (seen

above) or unmarked.

Genitive-marked subject:

• definite or indefinite

• scopes above or below adjective

7



Unmarked subject:

• indefinite only

• scopes below adjective

Key points:

• genitive-marked subject raises out of the scope of existential closure to satisfy EPP on

Gen, and can QR even higher

• unmarked subject remains in the scope of existential closure

Genitive-marked -ish phrase subject – high or low:

(28) q1z-n1Ngirl-gen

kil-iS-icome-ISH-3

mohimimportant


[∃ > important]

[important > ∃]

Unmarked -ish phrase subject – low only:

(29) q1zgirl

kil-iS-icome-ISH-3

mohimimportant


*[∃ > important]

[important > ∃]

Definite -ish phrase subject – genitive-marked only:

(30) Ajgul-*(nuN)Ajgul-*(gen)

kil-iS-icome-ISH-3

mohimimportant

‘It’s important for Ajgul to come.’

Suppose that an -ish phrase optionally contains the functional projection Gen, which:

• assigns genitive case to the subject/possessor


Then there are two possible -ish phrase structures.

8



Genitive -ish phrase subject (repeated from example (28)):


kil-iS-icome-ISH-3

mohimimportant


(32) Non-modal adjective (with Gen):

∃ > important TP

DP

GenP

DP

q1z-n1N

girl-gen

important > ∃

Gen′

NP

v θP

t v θ′

kil

come

-iS-i

-ISH-3

Gen

D

T′

AP/PredP

t mohim

t important

T

9



Unmarked -ish phrase subject (repeated from example (29)):

(33) q1zgirl

kil-iS-icome-ISH-3

mohimimportant


(34) Non-modal adjective structure (without Gen):

TP

DP

NP

v θP

DP

q1z

girl

v θ′

kil

come

-iS-i

-ISH-3

D

T′

AP/PredP

t mohim

t important

T

Structure without Gen:

• -ish phrase subject does not raise out of existential closure → definites prohibited

• interpretation in situ → subject < adjective

• no QR possible

Ordering with respect to adverbials confirms that an unmarked subject is lower than a

genitive-marked subject. Adverb placement:

• above or below genitive-marked subject

• above unmarked subject

Unmarked subject is lower than genitive subject:

(35) (æte)(tomorrow)

Ajgul-n1N/q1z-n1NAjgul-gen/girl-gen

(æte)(tomorrow)

kil-iS-icome-ISH-3

mohimimportant

‘It’s important for Ajgul/[a girl] to come tomorrow.’

(36) (æte)(tomorrow)

q1zgirl

(??æte)(??tomorrow)

kil-iS-icome-ISH-3

mohimimportant

‘It’s important for a girl to come tomorrow.’

10







no ✗ ✗ ✓

-ish subj. w/ non-modal adj.yes ✓ ✓ ✓

no ✗ ✗ ✓

4 Verbal Embedding of -ish Phrases

In this section:

• a verbal predicate can take an -ish phrase object

• the nominalized phrase is not a full TP, so there is no EPP feature on T in it

• the case-interpretation pattern is the same as when the embedding predicate is a non-

modal adjective

-ish phrases can also be embedded under verbal predicates.

Verbal embedding:

(38) Otkur

Otkur

[Ajgul-n1N[Ajgul-gen

kel-gen-lik-i-ni]come-RAN-LIQ-3-acc]

didisaid

‘Otkur said that Ajgul came.’

New suffixes:

-ran: Subordinate clause marker. Expresses perfective meaning when no other aspectual

morphology is present (phonologically [gen], [Kan]).

-liq : Allomorph of -ish that is used when another suffix intervenes between the verbal root

and the nominalizer.

The nominalized phrase under a verb contains an AspP, but not a TP, so there is no

EPP-bearing T.

Verbal embedding with aspect marking:


[min-1N[I-gen

oqu-wat-Kan-liq-im-ni]read-prog-RAN-LIQ-1sg-acc]

didisaid

‘Mehemmet said that I am reading right now.’

No verbal embedding with tense marking:

(40) *MehemmetMehemmet

[min-1N[I-gen

oqu-[d/di/dim]-Kan-liq-im-ni]read-[past/past-3/past-1sg]-RAN-LIQ-1sg-acc]

didisaid

Subject case and scope properties are the same as with embedding by non-modal adjectives:

11





• scopes above or below verb

Unmarked subject:

• indefinite only

• scopes below verb

Key points:

• genitive-marked subject raises out of the scope of existential closure to satisfy EPP on

Gen, and can QR even higher

• unmarked subject remains in the scope of existential closure

Genitive-marked -ish phrase subject – high or low:

(41) Otkur

Otkur

[q1z-n1N[girl-gen


didisaid

‘Otkur said that a girl came.’

[∃ > say]

[say > ∃]

Unmarked -ish phrase subject – low only:

(42) Otkur

Otkur

[q1z[girl


didisaid

‘Otkur said that a girl came.’

*[∃ > say]

[say > ∃]

Definite subject – genitive obligatory:

(43) Otkur

Otkur

[Ajgul-?*(n1N)[Ajgul-?*(gen)


didisaid

‘Otkur said that Ajgul came.’

Just as with non-modal adjectives, the -ish phrase optionally contains Gen.

12



(44) Verbal embedding (with Gen):

∃ > say TP

Otkur

Otkur

T′

VP

DP

GenP

DP

q1z-n1N

girl-gen

say > ∃

Gen′

NP

AspP

v θP

t v θ′

kel

come

-gen

-RAN

-liq-i-ni

-LIQ-3-acc

Gen

D

didi

said

past

Structure with Gen:

• -ish phrase subject raises out of existential closure → definites allowed

• interpretation in raised position → subject < verb

• QR to a higher position → subject > verb

13



(45) Verbal embedding (without Gen):

TP

Otkur

Otkur

T′

VP

DP

NP

AspP

v θP

DP

q1z

girl

v θ′

kel

come

-gen

-RAN

-liq-i-ni

-LIQ-3-acc

D

didi

said

past

Structure without Gen:

• -ish phrase subject does not raise out of existential closure → definites prohibited

• interpretation in situ → subject < verb

• no QR possible

Ordering with respect to locatives confirms that an unmarked subject is lower than a genitive-

marked subject. Locative placement:



Unmarked subject is lower than genitive subject:


[(sorun-Ka)[party-dat

q1z-n1Ngirl-gen

(sorun-Ka)(party-dat)


didisaid

‘Mehemmet said that a girl came to the party.’


[(sorun-Ka)[(party-dat)

q1zgirl

(??sorun-Ka)(??party-dat)


didisaid

‘Mehemmet said that a girl came to the party.’

14







no ✗ ✗ ✓


no ✗ ✗ ✓

-ish subj. w/ verbyes ✓ ✓ ✓

no ✗ ✗ ✓

5 Modal Adjectives

In this section:

• differences between modal adjectives and non-modal adjectives

• proposal that the subject of the -ish phrase raises out of it

• account of why the subject of the -ish phrase raises out of it

• data supporting the claim that the -ish phrase subject is high

• data supporting the claim that the -ish phrase subject started low

• analysis of the relationship between -ish phrase subject case and scope

5.1 Modal Adjectives vs. Non-Modal Adjectives

Some types of modality in Uyghur are expressed by adjectives.

kirek: deontic/epistemic necessity

lazim: deontic/epistemic necessity

mumkin: epistemic possibility

We will see:

• evidence that the subject of the -ish phrase raises out of it when the embedding pred-

icate is a modal adjective

• a proposed account of why this is

Modal adjectives have different -ish phrase subject case properties from the adjectives pre-

viously discussed.



• prohibited with mumkin (‘possible’)

15



Umarked subject:


Modal necessity adjectives:

(49) men-(1N)I-(gen)

kitapbook


kirek/lazimnecessary

‘I have to read a book.’

Modal possibility adjective:

(50) men-(*1N)I-(*gen)

kitapbook


mumkinpossible

‘I might read a book.’

Non-modal adjectives:

(51) min-*(1N)I-*(gen)

kitapbook



‘My reading a book is important/useful.’

Modal adjectives also behave differently under direct embedding. Recall: the subject of a

directly embedded proposition in Uyghur can be accusative-marked or unmarked. (Shklovsky

and Sudo (2009))

• unmarked → shifted interpretation for pronouns

• accusative-marked → non-shifted interpretation for pronouns

Verbal predicate – nominative subject, shifted (repeated from (24)):

(52) AhmetAhmet

[min-1N[I-gen

q1z-imgirl-1sg


didisaid

‘Ahmet said that his daughter left.’ [shifted]

* ‘Ahmet said that my daughter left.’ [non-shifted]

Verbal predicate – accusative subject, non-shifted (repeated from (25)):

(53) AhmetAhmet

[min-1N[I-gen

q1z-im-nigirl-1sg-acc


didisaid

* ‘Ahmet said that his daughter left.’ [shifted]

‘Ahmet said that my daughter left.’ [non-shifted]

Shklovsky and Sudo (2009) propose:

• unmarked subjects are low

• accusative-marked subjects are high

Non-modal adjectives: -ish phrase raises and gets accusative

Modal adjectives: -ish phrase subject raises and gets accusative

16



Non-modal adjective – accusative -ish phrase, non-shifted (repeated from (27)):

(54) Otkur

Otkur

[min-1NI-gen

oqu-S-im-niread-ISH-1sg-acc

mohim]important

didisaid

* ‘Otkur said that his studying is important.’ [shifted]

‘Otkur said that my studying is important.’ [non-shifted]

The subject of the -ish phrase with an non-modal adjective cannot be accusative marked.

Non-modal adjective – no accusative marking on subject of -ish phrase:


[Ajgul-ni[Ajgul-acc

oqu-S-istudy-ISH-3

mohim]important]

didisaid

Modal adjective – nominative subject, shifted:

(56) Otkur

Otkur

[men[I

oqu-S-imstudy-ISH-1sg

kirek]necessary]

didisaid

‘Otkur said that his studying is necessary.’ [shifted]

* ‘Otkur said that my studying is necessary.’ [non-shifted]

Modal adjective – accusative subject, non-shifted:

(57) Otkur

Otkur

miniI-acc

oqu-S-[i/1N]read-[3/2sg]

kireknecessary

didisaid

* ‘Otkur said that his studying is necessary.’ [shifted]

‘Otkur said that my studying is necessary.’ [non-shifted]

Modal adjective – no accusative marking on -ish phrase:


[Ajgul-[∅/n1N/ni][Ajgul-[nom/gen/acc]

oqu-S-i-niread-ISH-3-acc

kirek]necessary]

didisaid

This suggests that with modal adjectives, the subject of the -ish phrase is high:

• outside of -ish phrase

• in main clause subject position

Proposal for modal (vs. non-modal) adjectives (following Trinh (2009)):

(59) a. The -ish phrase that combines with non-modal adjectives is a DP.

b. The -ish phrase that combines with modal adjectives is an NP/GenP.

c. T in Uyghur has an EPP property that must be satisfied by a DP.

Consequently:

(60) a. When the embedding predicate is a non-modal adjective, the -ish phrase raises

to spec, TP to satisfy EPP.

17



b. When the embedding predicate is a modal adjective, the subject of the -ish

phrase raises to spec, TP to satisfy EPP.

Modal adjective:


kitapbook


kireknecessary

‘I have to read a book.’

(62) Modal adjective – genitive-marked subject of -ish phrase satisfies EPP of

T:

TP

DP

men-1N

I-gen

T′

AP/PredP

GenP

t Gen′

NP

v θP

t v θ′

kitap oqu

book read

-iS-im

-ISH-1sg

Gen

kirek

necessary

T

18



(63) Modal adjective – unmarked subject of -ish phrase satisfies EPP of T:

TP

DP

men

I

T′

AP/PredP

NP

v θP

t v θ′

kitap oqu

book read

-iS-im

-ISH-1sg

kirek

necessary

T

Compare the modal and non-modal adjective structures.

Adjectival embedding:


kil-iS-icome-ISH-3

mohim/kirekimportant/necessary

‘It’s important/necessary for a girl to come.’

(65) Non-modal adjective (abbreviated) – -ish phrase satisfies EPP of T:

TP

DP

q1z-(n1N) kil-iS-i

girl-(gen) come-ISH-3

T′

AP/PredP

t mohim

t important

T

19



(66) Modal adjective (abbreviated) – subject of -ish phrase satisfies EPP of

T:

TP

DP

q1z-(n1N)

girl-gen

T′

AP/PredP

NP/GenP

t kil-iS-im

t come-ISH-1sg

kirek

necessary

T

Non-modal adjectives: -ish phrase satisfies EPP of T

Modal adjectives: -ish phrase subject satisfies EPP of T

5.2 Evidence for High Surface Position of -ish Phrase Subject

Data showing that the subject of the -ish phrase is outside the -ish phrase at some level in

modal adjective constructions:

• direct embedding (seen above)

• placement of bolsa

• placement of mu

5.2.1 Direct Embedding

As seen above, in embedded modal adjective constructions the subject of the -ish phrase can

take accusative case.

Modal adjective – accusative subject, non-shifted:

(67) Otkur

Otkur

meniI-acc

oqu-S-[i/1N]read-[3/2sg]

kireknecessary

didisaid

* ‘Otkur said that his studying is necessary.’ [shifted]

‘Otkur said that my studying is necessary.’ [non-shifted]

In contrast, accusative case on the subject of the -ish phrase is not possible for non-modal

adjectives.

20



Non-modal adjective – accusative subject, non-shifted:


[Ajgul-ni[Ajgul-acc

oqu-S-istudy-ISH-3

mohim]important]

didisaid

The subject of the -ish phrase with modal adjectives thus:

• acts as the subject of the embedded proposition

• is high in the structure, above the “monster” projection (Shklovsky and Sudo (2009))

5.2.2 Bolsa

Bolsa (morphologically bol (‘be’) + -sa (conditional suffix)) is a topic marker. With modal

adjectives, it can appear directly after the subject to the -ish phrase, but not after the entire

-ish phrase.4

Modal adjectives – bolsa after -ish phrase subject:


bolsatop



‘My reading is necessary.’

(70) menI

bolsatop


mumkin.possible

‘My reading is possible.’

Modal adjectives – no bolsa after -ish phrase:

(71) *men-(1N)I-(gen)


bolsatop

kirek/lazim/mumkin.necessary/necessary/possible

Thus the -ish phrase is not a constituent for the purposes of bolsa placement with modal

adjectives. Non-modal adjectives display the opposite pattern.

Non-modal adjectives – bolsa after -ish phrase:

(72) min-1NI-gen


bol-satop

mohim/eXmijetlik/qimmetimportant/useful/expensive

‘My reading is important/useful/expensive.’

Non-modal adjectives – no bolsa after -ish phrase subject:

(73) *min-1NI-gen

bol-satop


mohim/eXmijetlik/qimmetimportant/useful/expensive

Example (72) shows that an -ish phrase can be a topic.

4The data in this subsection for kirek (‘necessary’) and qimmet (‘expensive’) is also found in Trinh (2009).

21



5.2.3 Mu

-Mu (‘also’, ‘even’) can affix to a focused element:5:

(74) Otkur

Otkur

eqilliq,smart

veand

John-muJ.-MU

eqilliq.smart

‘Otkur is smart, and J. is also smart.’

It can also affix to a phrase containing the focused element:

(75) min-1NI-gen

q1z-imdaughter-1sg

eqilliq.smart.

Otkur-n1N

Otkur-gen

q1z-i-mudaughter-3-MU

eqilliq.smart.

‘My daughter is smart. Otkur’s daughter is also smart.’

-Mu cannot appear on an element that does not contain (or is not contained in) the target

of focus:

(76) Otkur

Otkur

kitap-nibook-acc

oqu-d-i,read-past-3,

veand

Otkur.-(*mu)

Otkur-(*MU)

Xet-niletter-acc

oqu-d-i.write-past-3

‘Otkur read a book, and Otkur wrote a letter.’

-Mu can appear on an entire DP when the possessor is focused, as in example (75). Similarly,

with non-modal adjectives mu can appear on the -ish phrase when subject of the -ish phrase

is focused.

Non-modal adjectives – focused subject, -mu on -ish phrase:

(77) Otkur-nuN

Otkur-gen

ket-iS-ileave-ISH-3

mohim/eXmijetlik/jaXSi.important/useful/good.

Ajgul-n1N-(mu)Ajgul-gen-(MU)

ket-iS-i-(mu)leave-ISH-3-(MU)

mohim/eXmijetlik/jaXSi.important/useful/good

‘Otkur’s leaving is important/useful/good. Ajgul’s leaving is also important/useful/good.’

If the subject of the -ish phrase is obligatorily outside of the -ish phrase in modal construc-

tions, we expect -mu not to appear on the entire -ish phrase when the subject is focused.

That is precisely what we find.

Modal adjectives – focused subject, *-mu on -ish phrase:

(78) Otkur-nuN

Otkur-gen

ket-iS-ileave-ISH-3

kirek/lazim.necessary.

Ajgul-n1N-(mu)Ajgul-gen-(MU)

ket-iS-i-(*mu)leave-ISH-3-(*MU)


‘Otkur’s leaving is necessary. Ajgul’s leaving is also necessary.’

5Examples (74) through (76) are from Hartman (2009).

22



(79) Otkur

Otkur

ket-iS-ileave-ISH-3-

kirek/lazim/mumkin.nec./nec./poss..

Ajgul-(mu)Ajgul-(MU)

ket-iS-i-(*mu)leave-ISH-3-(*MU)

kirek/lazim/mumkin.necessary/nec./possible

‘Otkur’s leaving is necessary/possible. Ajgul’s leaving is also necessary/possible.’

Note also that when the object of an -ish phrase under a modal adjective is focused, -mu

can appear on the -ish phrase.

Modal adjective – focused object, -mu on -ish phrase:

(80) Otkur-nuN

Otkur-gen

kitapbook

oqu-S-iread-ISH-3

kirek.necessary.

veand

u-n1Nhe-gen

Xet-(mu)letter-(MU)

oqu-S-i-(mu)read-ISH-3-(MU)

kirek.necessary.

‘Otkur has to read a book. And he also has to read a letter.’

(81) Otkur

Otkur

kitapbook

oqu-S-iread-ISH-3

kirek.necessary.

veand

uhe

Xet-(mu)letter-(MU)

oqu-S-i-(mu)read-ISH-3-(MU)

kirek.necessary.

‘Otkur has to read a book. And he also has to read a letter.’

5.3 Evidence for Low Underlying Position of -ish Phrase Subject

Having seen that the semantic subject of the -ish phrase is outside the -ish phase, we must

now ask whether this is a raising or a control construction:

• Does the subject of the -ish phrase raise? or

• Is the subject of the -ish phrase a PRO?

Evidence from Negative Concord Item (NCI) licensing shows that raising is possible; evidence

from agreement suggests that control is not available. As shown in (82) and (83), the NCI

hitSkim (‘nobody’) requires negation in order to be licensed.

(82) hitSkimn-body

oqu-mi-d-iread-neg-past-3

‘Nobody read.’

(83) *hitSkimn-body

oqu-d-iread-past-3

As (84) shows, an NCI subject of an -ish clause with kirek can be licensed by negation in

the embedded predicate, regardless of case on the NCI:

23



Modal adjective – NCI licensed by negation on -ish phrase:

(84) hitSkim-(n1N)n-body-(gen)

ket-mas-liq-ileave-neg-LIQ-3

kireknecessary

‘It’s necessary that nobody leave’

Note that negating an -ish phrase results in a -liq phrase:6

(85) Otkur-nuN

Otkur-gen

oqu-mas-liq-iread-neg-LIQ-3

mohim/kirekimportant/necessary

‘It’s important/necessary for Otkur not to read.’

NCI licensing with modal adjectives is in contrast to control predicates, for which only main

clause negation licenses hitSkim.

Control construction – NCI licensed by main-clause negation:

(86) AjgulAjgul

kitapbook

oqu-S-qaread-ISH-dat

tiriS-mi-d-itry-neg-past-3

‘Ajgul didn’t try to read a book.’

(87) hitSkimn-body

kitapbook

oqu-S-qaread-ISH-dat

tiriS-mi-d-itry-neg-past-3

‘Nobody tried to read a book.’

Control construction – NCI not licensed by negation on -ish phrase:

(88) AjgulAjgul

kitapbook

oqu-mas-liq-qaread-neg-LIQ-dat

tiriS-t-itry-past-3

‘Ajgul tried not to read a book.’

(89) *hitSkimn-body

kitapbook

oqu-mas-liq-qaread-neg-LIQ-dat

tiriS-t-itry-past-3

This shows that a raising structure with modal adjectives is available. Agreement properties

of the -ish phrase suggest that a control structure is not available. In particular, agreement

on the -ish phrase under a modal adjective is required with 1st and 2nd person subjects.

Modal adjective – agreement on -ish phrase required:


ket-mas-lik-*(im)leave-neg-LIQ-*(1sg)

kireknecessary

‘It’s necessary for me not to leave.’

(91) menI

ket-iS-*(im)leave-ISH-*(1sg)

kirek/mumkinnecessary/possible

‘It’s necessary/possible for me to leave.’

6-ish and -liq are allomorphs, with -ish being selected iff the stem is a bare verbal root. (Thomas (2009))

24



By contrast, an -ish phrase under a control predicate does not show any agreement marking.

Control construction – agreement on -ish phrase prohibited:

(92) menI

kitapbook

oqu-S-(*im)-niread-ISH-(*1sg)-acc

ojli-wat-i-menwant-prog-imp-1sg

‘I want to read a book.’

(93) menI

kitapbook

oqu-S-(*im)-qaread-ISH-(*1sg)-dat

tiriS-t-imtry-past-1sg

‘I tried to read a book.’

Thus in modal adjective constructions:

• NCI’s are licensed by negation on -ish phrase: raising is possible

• agreement with 1st/2nd person subjects is required on -ish phrase: control is not

possible

5.4 Subject Case and Interpretation

The subject of the -ish phrase with modal adjectives has the following properties.



• scopes above adjective

Umarked subject:



Key points:

• both genitive-marked and unmarked subjects raise to spec, TP

• unmarked subjects can reconstruct, but genitive-marked ones cannot

Genitive -ish phrase subject – high only:


kil-iS-icome-ISH-3

kireknecessary

‘A girl has to come.’

[∃ > necessary]

*[necessary > ∃]

25



Unmarked -ish phrase subject – high or low:

(95) q1zgirl

kil-iS-icome-ISH-3

kireknecessary


[∃ > necessary]

[necessary > ∃]



Genitive-marked subjects:

• outside the scope of existential closure → can be definite

• cannot reconstruct → must scope high

Unmarked subjects:

• outside the scope of existential closure → can be definite

• can reconstruct → may scope high or low

26



Genitive -ish phrase subject (repeated from (94)):


kil-iS-icome-ISH-3

kireknecessary


(98) Modal adjective – genitive-marked subject structure:

TP

DP

q1z-n1N

girl-gen

∃ > necessary

T′

AP/PredP

GenP

t Gen′

NP

v θP

t v θ′

kil

come

-iS-i

-ISH-3

Gen

kirek

necessary

T

27



Unmarked -ish phrase subject (repeated from (95)):

(99) q1zgirl

kil-iS-icome-ISH-3

kireknecessary


(100) Modal adjective – unmarked subject structure:

TP

DP

q1z

girl

∃ > necessary

T′

AP/PredP

NP

v θP

t

necessary > ∃

v θ′

kil

come

-iS-i

-ISH-3

kirek

necessary

T





no ✗ ✗ ✓


no ✗ ✗ ✓


no ✗ ✗ ✓

-ish subj. w/ modal adj.yes ✓ ✓ ✗

no ✓ ✓ ✓

28



6 Test Case: Double Embedding

In this section:

• -ish phrase subjects under embedded modal adjectives behave like -ish phrase subjects

subjects under unembedded non-modal adjectives and verbs

• because nominalized embedding contains something smaller than a TP, there is no

EPP on T, and the observed pattern is predicted

What happens when we have two layers of a modal adjective structure is nominalized and

embedded under a verb?

• genitive marking on -ish phrase subject

• no genitive marking on -ish phrase

Embedded modal adjective:

(102) Otkur

Otkur


kil-iS-icome-ISH-3

kirek-liq-i-ni]necessary-LIQ-3-acc]

didisaid

‘Otkur said that Ajgul has to come.’

No case marking on -ish phrase:

(103) ??Otkur

Otkur


kil-iS-i-n1Ncome-ISH-3-gen


didisaid

(104) ??Otkur

Otkur

[q1z[girl

kil-iS-i-n1Ncome-ISH-3-gen


didisaid




Umarked subject:

• indefinite only

• scopes below adjective

Key points:

• genitive-marked subjects can get case from the higher or lower -ish phrase, which

corresponds to high and low readings

• since there is no T, unmarked subjects remain in situ, inside the scope of existential

closure

29



Double embedding – definite subject genitive:

(105) Otkur

Otkur

[Ajgul-*(n1N)[Ajgul-*(gen)

kil-iS-icome-ISH-3


didisaid

‘Otkur said that Ajgul has to come.’

Genitive subject – high or low:

(106) Otkur

Otkur

[q1z-n1N[girl-gen

kil-iS-icome-ISH-3


didisaid

‘Otkur said that a girl has to come.’

[∃ > necessary]

[necessary > ∃]

Unmarked subject – low only:

(107) Otkur

Otkur

[q1z[girl

kil-iS-icome-ISH-3


didisaid


* [∃ > necessary]

[necessary > ∃]

The case-marking and interpretation pattern observed here is:

• like embedding under non-modal adjectives and verbs

• different from embedding under modal adjectives

Nevertheless, this pattern is predicted!

Two possible sourced for genitive case:

• Gen in lower -ish phrase (under necessary)

• Gen in higher -ish phrase (under say)

There is no TP in the nominalized structure, and therefore no EPP on T. Thus, when both

Gen’s are absent, the subject is:

• unmarked

• low

Unmarked subject (repeated from (107)):

(108) Otkur

Otkur

[q1z[girl

kil-iS-icome-ISH-3


didi.said


30



(109) Structure for double embedding with unmarked subject (no Gen):

TP

DP

Otkur

T′

VP

DP

NP

AP/PredP

NP

v θP

DP

q1z

girl

v θ′

kel

come

-iS-i

-ISH-3

kirek

necessary

-liq-i-ni

-liq-3-acc

D

didi

said

past

The unmarked subject is under the scope of existential closure. Consequently, it:

• cannot be definite

• must scope below the modal

Ordering with respect to adverbials confirms that an unmarked subject is lwoer than a

genitive-marked subject. Adverb placement:



Unmarked subject lower than genitive-marked subject:

(110) Otkur

Otkur

(æte)(tomorrow)

Ajgul-n1N/q1z-n1NAjgul-gen/girl-gen

(æte)(tomorrow)

kil-iS-icome-ISH-3

kirek-liq-i-ninecessary-LIQ-3-acc

didisaid

31



‘Otkur said that Ajgul/[a/the] girl has to come tomorrow.’

(111) Otkur

Otkur

(æte)(tomorrow)

q1zgirl

(*æte)(*tomorrow)

kil-iS-icome-ISH-3

kirek-liq-i-ninecessary-LIQ-3-acc

didisaid

‘Otkur said that a girl has to come tomorrow.’

Three possibilities for genitive-marking of -ish phrase subject:

Gen in low -ish phrase? Gen in high -ish phrase?

yes no

no yes

yes yes

The third option does not yield any additional interpretations, and so will not be illustrated.

• low Gen → subject < modal

• high Gen → subject > modal

• high Gen, low Gen: subject outside of existential closure → definites allowed

32



Genitive subject (repeated from (106)):

(112) Otkur

Otkur

[q1z-n1Ngirl-gen

kil-iS-icome-ISH-3

kirek-liq-i-ni]necessary-LIQ-3-acc

didi.said


(113) Structure for double embedding with genitive subject (low Gen):

TP

DP

Otkur

T′

VP

DP

NP

AP/PredP

GenP

DP

q1z-n1N

girl-gen

Gen′

NP

v θP

t v θ′

kil

come

-iS-i

-ISH-3

Gen

kirek

necessary

-liq-i-ni

-LIQ-3-acc

D

didi

said

past

33



(114) Structure for double embedding with genitive subject (high Gen):

TP

DP

Otkur

T′

VP

DP

GenP

DP

q1z-n1N

girl-gen

Gen′

NP

AP/PredP

NP

v θP

t v θ′

kil

come

-iS-i

-ISH-3

kirek

necessary

-liq-i-ni

-LIQ-3-acc

Gen

D

didi

said

past

34



(115) Case-Marking and Interpretation (summary):




no ✗ ✗ ✓


no ✗ ✗ ✓


no ✗ ✗ ✓


no ✓ ✓ ✓

-ish subj. w/ modal adj. under verbyes ✓ ✓ ✓

no ✗ ✗ ✓

7 Conclusion

7.1 Comments on Case

Questions about Uyghur case (set aside so far):

• Do unmarked noun phrases have case?

• If so, what kind of case, and where does it come from?

• If not, how are they licensed?

• Why is their case distinguished from overt case for the purposes of reconstruction?

• Shuld null nominative case be distinguished from other null cases?

At least three possible views:

1. Uyghur has several morphologically null cases (nominative, accusative, genitive). (the

boring view)

2. Noun phrases do not require case. (Shklovsky and Sudo (2010)) [best fit for above

data]

3. Some noun phrases are bare NP’s, and bare NP’s do not require case. (Cagri (2005),

Ozturk (2005))

The restiction on reconstruction proposed above will be helpful in this discussion.



35



7.1.1 The Boring View

Proposal:

• accusative and genitive have null allomorphs

• null accusative and genitive are assigned in situ

Problems:

• large variety of null cases (nominative, accusative, genitive)

• no connection between a case being morphologically null and being assigned in situ

• restriction on reconstruction looks very strange

Verdict: This is not a good theory of Uyghur case.

7.1.2 Shklovsky and Sudo (2010)

Proposal:

• noun phrases do not require case

• all unmarked noun phrases are caseless

• some noun phrases bear an abstract feature that is targeted by an accusative-assigning

head with EPP

• alternative formulation: vCASE and Gen can target any DP, but are present optionally

Nice consequence:

• restriction on reconstruction looks very reasonable – only caseless noun phrases can

reconstruct

Verdict: This looks pretty good.

7.1.3 Cagri (2005)

This proposal was originally made for Turkish; we will consider how well it could be applied

to Uyghur.

Proposal:

• some noun phrases are DP’s, and some are NP’s

• DP’s require case

• NP’s do not receive case

• nominative (which is morphologically null) is a real case

• NP’s cannot be definite; they take low scope

• only DP’s can satisfy EPP (on T)

What works:

Genitive -ish phrase subjects:

• full DP

36




• outside scope of existential closure

Unmarked -ish phrase subjects with non-modal adjective embedding:

• NP

• indefinite

• under existential closure

What doesn’t work: unmarked -ish phrase subjects under modal adjectives

• subject satisfies EPP on T → must be a DP

• subject a DP → cannot reconstruct under existential closure (false!)

(117) Modal adjective – unmarked subject structure:

TP

DP

q1z

girl

∃ > necessary

T′

AP/PredP

NP

v θP

t

necessary > ∃

v θ′

kil

come

-iS-i

-ISH-3

kirek

necessary

T

Verdict: The proposal does not fit some of the Uyghur data.

7.2 Summary

There is a variety of environments in Uyghur where presence or absence of case-marking

affects the interpretation of a DP.

37



(118) Case-Marking and Interpretation:




no ✗ ✗ ✓


no ✗ ✗ ✓


no ✗ ✗ ✓


no ✓ ✓ ✓

-ish subj. w/ modal adj. under verbyes ✓ ✓ ✓

no ✗ ✗ ✓

All the data in the table above has been accounted for. Proposals:

Case assignment and EPP: vCASE and Gen are optionally projected. They assign overt

case an bear an EPP feature. This accounts for the null/accusative and null/genitive

alternations.

-ish phrases: -ish phrases are nominal. Non-modal adjectives and verbs embed full DP

-ish phrases, while modal adjectives embed impoverished ones. The EPP feature of T

is not satisfied by these impoverished -ish phrases.

Restriction on reconstruction: Reconstruction of overtly case-marked noun phrases is

prohibited.

References

Cagri, Ilhan. 2005. Minimality and Turkish relative clauses. Doctoral Dissertation, Univer-

sity of Maryland.

Diesing, Molly. 1992. Indefinites . MIT press.

Hartman, Jeremy. 2009. On the additive suffix -mu in Uyghur. 24.942 squib, Spring 2009.

Kelepir, Meltem. 2001. Topics in Turkish syntax: Clausal structure and scope. Doctoral

Dissertation, MIT.

Ozturk, Balkiz. 2005. Case, referentiality, and phrase structure. John Benjamins Publishing

Company.

38



Shklovsky, Kirill, and Yasutada Sudo. 2009. Shifted indexicals in Uyghur. In The 40th

annual meeting of the North East Linguistic Society (NELS40).

Shklovsky, Kirill, and Yasutada Sudo. 2010. No case licensing: Evidence from Uyghur.

Handout from Penn Linguistics Colloquium 34.

Thomas, Guillaume. 2009. Remarks on three Uyghur suffixes. 24.942 squib, Spring 2009.

Trinh, Tue. 2009. Some notes on Uighur modal sentences. 24.942 squib, Spring 2009.

39

case and meaning in uyghur nominalized...

Documents