Available online at www.sciencedirect.com

ScienceDirect

Adjectival participles in Turkish

Aslı Gürer *

www.elsevier.com/locate/linguaLingua 149 (2014) 166--187

Bogaziçi University, Turkey

Received 30 May 2013; received in revised form 12 July 2014; accepted 16 July 2014Available online 19 August 2014

Abstract

Adjectival participles have been classified either as lexical participles or phrasal participles which are then further classified as targetand resultant state participles based on some syntactic and semantic diagnostics as elaborated in Kratzer (1994, 2000), Anagnos-topoulou (2003), Embick (2004) for German, Greek and English respectively. In this study we present the typology of adjectival participlesin Turkish and show that unlike German, Greek and English, Turkish bears distinctive morphology for the three types of participlesproposed in the literature. Based on diagnostics of event-oriented manner adverb modification, the present study argues that lexicaladjectival participles formed with --(I)k directly attach to the root without an eventive layer as is the case in German and Greek lexicaladjectival participles but phrasal adjectival participles with --(I)lI and --mIş morphology both have an eventive layer. A distinction is drawnfor phrasal adjectival participles based on compatibility with (i) by phrases, (ii) purpose clauses and (iii) agent oriented adverbials whichindicate an additional voice projection for resultant state participles with --mIş morphology. We propose that phrasal adjectival participleswith --(I)lI denote target state participles and AspTARGET does not directly attach with the root but selects for vP projection. Phrasaladjectival participle with --mIş denote resultant state participles and unlike German and similar to its Greek counterpart, AspRESULTANTselects for vP and VoiceP. Additionally, phrasal participles derived from atelic activity verbs become grammatical only with --mIşparticiples when accompanied by telic expressions with an end point. We argue that the semantics of the aspectual stativizers differ for--(I)lI and --mIş participles and hence we get different restrictions on formation. This study shows that Turkish provides clear evidence forthe tripartite classification of adjectival participles, and differences can be captured through different internal structures in syntax andsemantic differences between the aspectual heads.© 2014 Elsevier B.V. All rights reserved.

Keywords: Lexical adjectival participles; Phrasal adjectival participles; Argument structure

1. Introduction

Investigation of adjectival participles has been an intriguing issue as adjectival-stative participles have distinctiveproperties from underived adjectives and verbal-eventive participles with which they can share the same morphology as inEnglish. In German and Greek, adjectival passives and verbal passives bear distinct morphology as exemplified in (1a-b)for German. Adjectival passives are used with the auxiliary ‘sein’; verbal passives are used with the auxiliary ‘werden’.

Abbreviations: ABL, ablative; ACC, accusative; AGR, agreement; AOR, aorist; AUX, auxiliary; ASP, aspect; CAUS, causative; CM,compound marker; DAT, dative; GEN, genitive; FUT, future; INST, instrumental; LOC, locative; NEG, negation; NOM, nominative; NOML,nominalizer; PASS, passive; PAST, past tense; PERF, perfective; PL, plural; POSS, possessive; PROG, progressive; PRS, person; PTPL,participle; REF, reflexive; SG, singular; VERBL, verbalizer; VOICE, voice; QP, question particle.* Tel.: +90 5558729816.E-mail address: gurer.asli@gmail.com.

http://dx.doi.org/10.1016/j.lingua.2014.07.0130024-3841/© 2014 Elsevier B.V. All rights reserved.

(1) a. Das Theorem ist bewiesen. Adjectival passive

A. Gürer / Lingua 149 (2014) 166--187 167

1 The

-(I)k: -ık-mIş: -m

suffixe

, -ik, -uış, -mi

s used in the participial formation c

k, -ük --(I)lI: -ılı, -ili, -ulu, -üş, -muş, -müş --(I)l/-(I)n:-ıl,-il,-ül, -u

The theorem is proven.

b. Das Theorem ist bewiesen worden.

an surfac

lül, -ın,-in-,-

Verbal passive, perfect

The theorem is proven gotten The theorem has been proven. (Kratzer, 2000, 5)

For German adjectival participles, Kratzer (2000) makes a two way distinction as lexical and phrasal adjectival participles.Phrasal adjectival participles denote states resulting from prior events while lexical adjectival participles denotecharacteristic states. Phrasal adjectival participles are further analyzed as target state participles denoting reversiblestates and hence compatible with the adverbial immer noch ‘still’ as in (2a), and resultant state participles denotingirreversible states which are incompatible with the same adverbial as illustrated in (2b).

(2)

a. Die Geisslein sind immer noch versteckt.

e in the follo

un, -ün -n

The little goats

are still hidden. b. Das Theorem ist (* immer noch) bewiesen.

The theorem

is (*still) proven. (Kratzer, 2000, 2)

Anagnostopoulou (2003) makes a three way distinction for Greek adjectival participles as (i) lexical adjectival, (ii) targetstate and (iii) resultant state participles. Lexical adjectival participles surface with the suffix --tos while both types of phrasaladjectival participles surface with the suffix --menos. In Greek, both target state and resultant state participles have anevent implication. Additionally, implicit external argument implication is allowed with resultant state participles asillustrated in (3) below with the compatibility of --menos participle with a by-phrase (Alexiadou and Anagnostopoulou,2008).

(3)

Ta keftedakia ine tiganis-men-a apo tin Maria the meatballs are fried by the Mary ‘The meatballs are fried by Mary.’

Embick (2004) also makes a three-way distinction for English as (i) eventive passives, (ii) resultative and (iii) stativeparticiples.

(4)

The door was opened. a. Eventive passive

Someone opened the door.

b. Resultative

The door was in a state of having become open.

(5)

The door was open. c. Stative (Embick, 2004, 1)

Embick (2004) claims that resultative participles include a verbalizing head which statives lack. The presence of v,eventive layer, makes adverbial modification possible in resultatives.

Turkish adjectival participles can bear --(I)k, --(I)lI and --mIş morphology as illustrated in the following examples.1

(6)

a. aç-ık open-PTPL ‘open’ yırt-ık tear-PTPL ‘torn’

wing forms due to vowel harmony in Turkish:

A. Gürer / Lingua 149 (2014) 166--187168

2 Theproduct

(1) s

In this a

b.

re are

ive as

at-ılık

rticle I

kapa-lı

close-PTPL ‘closed’ sak-lı hide-PTPL ‘hidden’

c.

kapa-n-mış close-VOICE-PTPL ‘closed’ aç-ıl-mış open-VOICE-PTPL ‘opened’

In the present study, I argue that Turkish shows a three-way distinction for (i) lexical adjectival participles with --(I)k

morphology, (ii) target state participles with --(I)lI morphology and (iii) resultant state participles with --mIş morphology.Adjectival participles have different internal structures in syntax as illustrated in (7) below.

(7)

ASPLEXICAL ASPTARGET ASPRESULTATIVE

√ ASP vP ASP VoiceP ASPaç -ık -lı -mış

‘open’ √ vP v -ıl

kapa ‘close’ √ v

aç ‘open’

ASPLEXICAL is directly attached to the root. Phrasal adjectival participles have an eventive layer above the root whichmakes event-oriented manner adverb modification possible. ASPRESULTATIVE is further differentiated from ASPTARGET inthat it selects for a Voice projection which makes agent oriented adverb modification possible.

The next section presents the distinction between adjectival and verbal passives. Section 3 makes the firstclassification for Turkish participles as lexical and phrasal. Compatibility with adverbial modification and becomingcomplements of ‘make, become, build’ provide clear evidence for this classification. Lexical adjectival participles with --(I)kappear in prenominal position but they are incompatible with event-oriented manner adverb modification as is the casewith underived adjectives and additionally they can be the complements of ‘make, become, build’ which we take as anindication of lack of an eventive layer in line with Embick (2004). Phrasal adjectival participles with --(I)lI and --mIşmorphology can appear in prenominal position and they are licit with event oriented manner adverb modification indicatingthe presence of an eventive layer. Section 4 illustrates how phrasal adjectival participles differ based on the diagnostictests proposed for German and Greek. Phrasal adjectival participles with --(I)lI denote reversible states and they arecompatible with adverbial ‘still’ and ‘for-adverbials’ while phrasal adjectival participles with --mIş denote irreversible statesand they are incompatible with these adverbs. I further demonstrate that similar to Greek, Turkish resultant stateparticiples allow implicit external argument based on which I propose different internal structures for phrasal participles.Restrictions on participle formation are addressed in Section 5 which shed lights on the internal structure of participleswhich is the locus of Section 6. In line with Alexiadou and Anagnostopoulou (2008), I suggest that distinctive properties oflexical and phrasal adjectival participles can be captured through different internal structures for ASPLEXICAL, ASPTARGET

and ASPRESULTATIVE and different semantic compositions for the aspectual stativizer in each case.

2. Adjectival and verbal passives in Turkish

In Turkish, participles that function as adjectives can surface with -(I)k, --(I)lI, and -mIş morphology which are veryproductive in Turkish as illustrated in (6a-c) above.2

also some other forms such as --(I)lIk, - gIn, -mAdIK which form adjectival participles as exemplified in (1) which are not asthe set with -(I)k, --(I)lI, and -mIş morphology.

‘on sale’, yay-gın ‘widespread’, kır-ıl-madık ‘not shattered’

will concentrate on adjectival participles formed with --(I)lI, -(I)k and -mIş morphology.

A. Gürer / Lingua 149 (2014) 166--187 169

In the literature adjectival participles are disambiguated from verbal (perfect and passive) participles in that adjectivalparticiples can surface in prenominal position as is the case with pure adjectives and only adjectival participles cansurface as complements of ‘act, remain, seem, sound, look’ as proposed by Wasow (1977), Williams (1981), Bresnan(1982) cited in Anagnostopoulou (2003). In Turkish, participles with --(I)lI, -(I)k and -mIş morphology can appear inprenominal position as is the case with underived adjectives as presented in (8--10).

(8)

a. yeni kapı adjective ‘new door’

b.

kapa-lı kapı participle close-PTPL door ‘closed door’

(9)

a. eski bina adjective ‘old building’

b.

yık-ık bina participle demolish-PTPL building ‘demolished building’

(10)

a. güzel kutu adjective ‘nice box’

b.

aç-ıl-mış kutu participle open-VOICE-PTPL box ‘opened box’

Secondly, only adjectival participles can surface as complements of ‘seem, look, remain, sound etc.’ Similar to underivedadjectives (11-13a), Turkish adjectival participles with --(I)lI, -(I)k and -mIş morphology may appear as complements oflook, remain and seem as in (11-13b).

(11)

a. Kapı yeni görün-üyor. door new look-PROG ‘The door looks new’

b.

Kapı kapa-n-mış görün-üyor. door close-VOICE-PTPL look-PROG ‘The door looks closed’

(12)

a. Kutu temiz dur-uyor. box clean remain-PROG ‘The box remains clean’

b.

Kutu aç-ık dur-uyor. box open-PTPL remain-PROG ‘The box remains open’

(13)

a. Yumurta-lar taze görün-üyor. egg-PL fresh look-PROG ‘The eggs look fresh’

b.

Yumurta-lar diz-ili görün-üyor. egg-PL order-PTPL look-PROG ‘The eggs look ordered’

The property of surfacing in attributive position for adjectival participles to distinguish them from verbal-eventive participlesis based on the assumption that adjectives can appear in prenominal position but not verbal eventive participles. HoweverTurkish is a head final language and relative clauses also surface in prenominal position as in (14) and hence adjectivalparticiples in (11-13b) can be suggested to be reduced relative clauses.

(14)

iki sene önce al-dıg-ım ev two year before buy-NOML-1PRS AGR house ‘The house that I bought two years ago’

A. Gürer / Lingua 149 (2014) 166--187170

The other point is that the participle forms in (10b) and (11b) with -mIş surfaces with --(I)l/-(I)n morphology which is the formused with reflexives (15a), middles (15b), in some reciprocals (15c), and verbal-eventive passives (15d).

(15)

3 One ohaving th

(1)

(2)

As all theverbal ev

a.

*

*

f the ree same

MaryJohn

coordinentive

Çocuklar

viewers sugg category as

is unhappy

is fast and a

ated phrasespassives and

yıka-n-mış.

est that acceptability of co illustrated below:

and in the kitchen. good soccer player.

are semantically parallel i hence we present the re

children

wash-VOICE-PAST ‘Children have washed themselves.’

b.

Bu çeşit ev-ler çabuk sat-ıl-ır. this kind house-PL fast sell-VOICE-AOR Literally: ‘These kinds of houses are sold easily.’

(Taneri, 1993, 1, with my modifications)

c. Anne-m ve ben birbir-imiz-e

ordination depends on the coordina

n (16--17) this test on its own is not csults of other diagnostic tests in thi

sar-ıl-dı-k.

mother-1SG POSS and I each other-1PL POSS-DAT embrace-VOICE-PAST-PL ‘My mother and I embraced each other.’

(Taneri, 1993, 17, with my modifications)

d. Bu mektup-lar çocuk-lar tarafından yaz-ıl-mış.

this

letter-PL child-PL by write-VOICE-PAST ‘These letters were written by children.’

Now we will see how adjectival participles are distinguished from relative clauses which also appear in prenominalposition and verbal-eventive passives with which --mIş adjectival participle can be homophonous as in (15d).

Firstly, we will test whether coordination with pure adjectives yield grammaticality with participles which revealswhether they are of the same category or not (Aya-Meltzer 2010). As illustrated in (16) adjectival participles with --Ik, -(I)lIand --mIş marker can be coordinated with underived adjectives.

(16)

a. Müze aç-ık ve kalabalık mı? museum open-PTPL and crowded QP ‘Is the museum open and crowded?’

b.

Duvar-lar yaz-ılı ve pis wall-PL write-PTPL and dirty ‘The walls are written on and dirty.’

c.

Ev-i boya-n-mış ve yeni mi? house-POSS paint- VOICE-PTPL and new QP ‘Is her house painted and new?’

We apply the same test to verbal eventive passives. As adjectival participles with --mIş and verbal passives can behomophonous in order to make sure that the form is a verbal eventive passive; we will use a verbal passive with differentmorphological markers.

(17)

a. Sınıf-lar temiz ve hazırla-n-dı mı? class-PL clean and prepare-VOICE-PAST QP Intended reading: ‘Are the classes clean and prepared?’

b.

Film eglenceli ve izle-n-ecek mi? film amusing and watch-VOICE-FUT QP Intended reading: ‘Is the film amusing and will we watch it?’

As the ungrammaticality of (17a-b) indicates, verbal passives which are not homophonous with --mIş adjectival participlecannot be coordinated with underived adjectives.3

ted phrases being semantically parallel, not on them

onclusive for distinguishing adjectival participles froms section.

A. Gürer / Lingua 149 (2014) 166--187 171

Note that in alternative questions the distinction between verbal-eventive passives (18d) and adjectival participles(18a-c) is not observed and coordination with an underived adjective is licit with all forms.

(18)

a.

*

*

*

Ev-i

temiz mi yoksa kirle-n-miş mi? house-3SG POSS clean QP or dirty-VOICE-PTPL QP ‘Is her house clean or dirty?’

b.

Ev-i yeni mi yoksa yık-ık mı? house-3SG POSS new QP or demolish-PTPL QP ‘Is her house new or demolished?’

c.

Ev-i müsait mi yoksa kapa-lı mı? house-3SG POSS residential QP or close-PTPL QP ‘Is her house residential or closed?’

d.

Ev-i temiz mi yoksa hazırla-n-ma-dı mı? house-3SG POSS clean QP or prepare-VOICE-NEG-PAST QP ‘Is her house clean or (is it) not prepared?’

We suggest that the alternative propositions in the alternative questions form separate domains and it is a VP levelcoordination. Hence lack of a distinction between verbal passives and adjectival participles cannot be an argument fortaking the --mIş participles as verbal-eventive passive.

Additionally, as illustrated in (11--13), adjectival participles can surface as complements of ‘remain, seem, look’ butverbal complements are not compatible with these verbs as indicated below.

(19)

a. Kapı boya-n-ıyor görün-üyor. door paint-VOICE-PROG look-PROG Intended reading: ‘The door looks it is being painted’

b.

Sınıf hazırla-n-dı dur-uyor class prepare-VOICE-PAST remain-PROG Intended reading: ‘‘The class remains in a prepared condition.’

Finally, restrictions on participial formation with --mIş morphology, which is discussed in detail in Section 5, furtherindicates that the participles with --mIş form a distinctive class than the verbal eventive passives. Activity verbs withoutaccompanying telic expressions cannot form --mIş participles (20a-b), while it is possible to form verbal-eventive passiveswith the same group of verbs (21a-b).

(20)

a. koş-muş çocuk run-PTPL child Intended reading: ‘the child who has run’

b.

200 metre koş-muş çocuk 200 meter run-PTPL child ‘The child who has run 200 meters.’

(21)

a. Dün koş-ul-muş, bugün degil. yesterday run-VOICE-PAST today not Literally: ‘It was run yesterday not today.’

b.

Her sene koş-ul-ur, sadece bu sene degil. Every year run-VOICE-AOR only this year not Literally: ‘It is run every year not only this year.’

Based on the data given in (16--20) we suggest that although --mIş adjectival participle can be homophonous with verbal-eventive passives in Turkish; --mIş adjectival participles form a distinctive class.

Now we will see how adjectival participles differ from relative clauses which also surface in prenominal position inTurkish as in (14). It is already noted in Turkish that a lexical item can surface between the modifying clause and thehead noun of relative clause (Kornfit, 2001). Additionally, relative clauses can be embedded in a genitive-possessiveconstruction in which case the head noun agrees with the genitive noun phrase as illustrated in (22a). (22b) indicatesthat demonstrative pronoun and the genitive noun phrase can surface between the modifying clause and the headnoun.

(22) a. şu ben-im kimse-nin al-ma-dıg-ı ev-im

A. Gürer / Lingua 149 (2014) 166--187172

*

that

I-GEN nobody-GEN buy-NEG-NOML-3SG POSS house-1SG POSS ‘That house of mine which nobody buys’

b.

kimse-nin al-ma-dıg-ı şu ben-im ev-im nobody-GEN buy-NEG-NOML-3SG POSS that I-GEN house-1SG POSS ‘That house of mine which nobody buys’

However this is not possible with the participle forms with -(I)k, --(I)lI and -mIş morphology as illustrated in (23-25b). Notethat this is also out for underived adjectives as in (26b).

(23)

a. şu ben-im yık-ık ev-im that I-GEN demolish-PTPL house-1SG POSS ‘That house of mine which is demolished.’

b.

yık-ık şu ben-im ev-im *

(24)

a. şu ben-im kapa-lı kapı-m that I-GEN close-PTPL door-1SG POSS ‘That door of me which is closed’

b.

kapa-lı şu benim kapı-m *

(25)

a. şu ben-im yık-ıl-mş ev-im that I-GEN demolish-VOICE-PTPL house-1SG POSS ‘That house of mine which is demolished’

b.

yık-ıl-mş şu ben-im ev-im *

(26)

a. şu benim eski ev-im that I-GEN old house-1SG POSS ‘That house of mine which is old’

b.

eski şu benim ev-im

The ungrammatical structures in (23--26) become grammatical when we add an overt auxiliary ‘ol’ following the participle.

(27)

a. yık-ık ol-an şu ben-im ev-im demolish-PTPL be-NOML that I-GEN house-1SG POSS ‘That house of mine which is demolished’

b.

kapa-lı ol-an şu benim kapı-m close-PTPL be-NOML that I-GEN door-1SG POSS ‘That door of mine which is closed’

c.

yık-ıl-mş ol-an şu benim ev-im demolish-VOICE-PTPL be-NOML that I-GEN house-1SG POSS ‘That house of mine which is demolished’

d.

eski ol-an şu benim ev-im old be-NOML that I-GEN house-1SG POSS ‘That house of mine which is old’

If we assume that --(I)lI, -(I)k, -mIş participles are reduced relative clauses, we have to assume a similar analysis for underivedadjectives as illustrated in (27d). As --(I)lI, -(I)k, -mIş participles differ from relative clauses in a similar vein with underivedadjectives as in (27a-d), we suggest that they are not clause level structures and they cannot be grouped with relative clauses.

Based on these diagnostic tests, we suggest that participles that appear with the --(I)lI, -(I)k, -mIş suffixes are adjectivaland they differ from verbal-eventive passive participle and relative clauses. Section 3 reanalyzes adjectival participles andfurther classifies them as lexical and phrasal adjectival participles.

3. Lexical and phrasal adjectival participles in Turkish

Within adjectival participles a further distinction is made as lexical and phrasal adjectival participles (Kratzer, 2000;Anagnostopoulou, 2003; Embick, 2004). The aspectual stativizer projects over a verbal projection which denotes

A. Gürer / Lingua 149 (2014) 166--187 173

eventivity in phrasal adjectival participles and hence phrasal adjectival participles are compatible with event-orientedmanner adverb modification. With lexical adjectival participles, the aspectual head directly attaches to the root whichmakes manner adverb modification illicit. Although aspectual head is present in both participles which makes stativizationpossible, phrasal adjectival participles denote states resulting from prior events while lexical adjectival participles denotecharacteristic states without making reference to a prior event is the case with underived adjectives.

Now we will focus on adjectival participles with --(I)k, --(I)lI and --mIş morphology and see whether they aredifferentiated based on this diagnostic. As illustrated in (28--30) as is the case in Greek and German, in Turkish not all theforms that we analyzed as adjectival participles in Section 2 are compatible with event-oriented manner adverbmodification. The adjectival participles with --mIş and --(I)lI suffixes are compatible with adverb modification in attributiveand predicative positions as in (28--29).

(28)

a.

*

*

*

*

*

saç-ı

özensizce yap-ılı-ydı. hair-3SG POSS sloppily make-PTPL-PAST ‘Her hair was sloppily done’

a’.

özensizce yap-ılı saç sloppily make-PTPL hair ‘Sloppily done hair’

b.

metal koruyucu-lar sıkıca kapa-lı-ydı. --(I)lI adjectival participle metal saver-PL tightly close-PTPL-PAST ‘The metal savers were tightly closed’

b’.

sıkıca kapa-lı metal koruyucu-lar tightly close-PTPL metal saver-PL ‘Tightly closed metal savers’

(29)

a. kagıt-lar özensizce kes-il-miş-ti. paper-PL sloppily cut-VOICE-PTPL-PAST ‘The papers were sloppily cut’

a’.

özensizce kes-il-miş kagıt-lar sloppily cut-VOICE-PTPL paper-PL --mIş adjectival participle ‘Sloppily cut papers’

b.

musluk başı dikkatlice bük-ül-müş-tü. faucet knob carefully bend-VOICE-PTPL-PAST ‘The faucet knob was carefully bent’

b’.

dikkatlice bük-ül-müş musluk başı carefully bend-VOICE-PTPL faucet knob ‘Carefully bent faucet knob’

Adverbial modification yields ungrammaticality with participles bearing --(I)k morphology. Note that adverbial modificationis also not possible with underived, simple adjectives as indicated in (31).

(30)

a. kagıt-lar özensizce kes-ik-ti. paper-PL sloppily cut-PTPL-PAST Intended reading: ‘The papers were cut sloppily’

a’.

özensizce kes-ik kagıt-lar sloppily cut-PTPL paper-PL Intended reading: ‘Sloppily cut papers’ --(I)k adjectival participle

b.

musluk başı dikkatlice bük-ük-tü. faucet knob carefully bend-PTPL-PAST Intended reading: ‘The faucet knob was carefully bent’

b’.

dikkatlice bük-ük musluk başı carefully bend-PTPL faucet knob Intended reading: ‘Carefully bent faucet knob’

(31)

a. Saç-ı özensizce pis-ti. hair-POSS sloppily dirty-PAST Intended reading: ‘Her hair was sloppily dirty’

*

A. Gürer / Lingua 149 (2014) 166--187174

a’.

*

özensizce

pis saç sloppily dirty hair Intended reading: ‘Sloppily dirty hair’

Based on this distinction, we claim that the forms with --mIş and --(I)lI are phrasal adjectival participles while the ones with--(I)k are lexical adjectival participles.

Embick (2004) notes that an adverb, that is compatible with a lexical/stative participle, is also compatible with thephrasal participles but with an additional reading. In (32a) the door was open but (probably) it is no longer open, (32b) hasthe same interpretation but additionally it has the interpretation that the door is in an open state but the opening event hastaken place recently.

(32)

a. the recently open door b. the recently opened door (Embick, 2004, 5)

The same interpretational difference holds in Turkish. When the participle bears --(I)k suffix, which is a lexical adjectivalparticiple according to the first diagnostic test of adverbial modification, in (33a) the only interpretation is that the door wasopened recently but it is no longer open. When the participle bears --mIş suffix, it has an additional reading according towhich the door is still open but the opening event has taken place recently (33b).

(33)

a. kapı geçenlerde aç-ık-tı, şimdi degil. door recently open-PTPL-PAST now not ‘The door was open recently not now’

b.

kapı geçenlerde aç-ıl-mış-tı, şimdi degil. door recently open-VOICE-PTPL-PAST now not ‘The door was opened recently not now’

The following contexts clearly indicate that both interpretations are possible in (33b).

(34)

A: Gümrük kapı-sın-dan mal geç-ir-mek için customs gate-3SG POSS-ABL goods pass-CAUS-NOML for bugün-ü bekli-yor-du-m. Ama kapı-yı erken aç-mış-lar. today-ACC wait-PROG-PAST-AGR but gate-ACC early open-PAST-AGR ‘In order to pass the goods from the customs gate I was waiting for Monday. But they opened the door early.’

B:

Kapı geçenlerde aç-ıl-mış-tı, bugün degil. Gate recently open-VOICE-PTPL-PAST today not Birkaç gün daha aç-ık kal-acak Several day more open-PTPL remain-FUT ‘The door was opened recently, not today. It will remain open for several days.’

(35)

A: Gümrük kapı-sın-dan mal geç-ir-mek için bugün-ü customs gate-3SG POSS-ABL goods pass-CAUS-NOML for bekli-yor-du-m. Ama kapı geçenlerde aç-ıl-mış ve bugün wait-PROG-PAST-AGR but gate recently open-VOICE-PTPL-PAST and today artık aç-ma-yacak-lar-mış. no longer open-NEG-FUT-AGR-PAST ‘In order to pass the goods from the customs gate I was waiting for today but the door was opened recentlyand it wouldn’t be opened any longer.’

In (34), the context forced the interpretation in which the door was opened recently and additionally the door is still open. In(35), on the other hand, the interpretation is that the door was opened recently but the door is not open now. The adjectivalparticiple with --mIş is licit in both contexts.

Embick (2004) further points out that lexical adjectival participles denoting characteristic state can be complements ofverbs of creation such as ‘make, build’ and verbs of change of state such as ‘become’ but phrasal adjectival participles,denoting states resulting from prior events, cannot be complements of these verbs as illustrated in (36a-b) below forEnglish.

(36)

a. This door was built open. b. This door was built opened. (Embick, 2004, 6)

A. Gürer / Lingua 149 (2014) 166--187 175

If our analysis is on the right track, we expect only the adjectival participles with --(I)k to surface as complements of‘become, make’. As illustrated in (37--39), this prediction is fulfilled and the adjectival participles with --(I)k differs from theparticiples formed with --(I)lI and --mIş in that only the forms with --(I)k can be complements of ‘become, make’.

(37)

a.

*

*

*

*

Bu

bahçe kapı-sı aç-ık yap-ıl-mış. this garden door-CM open-PTPL make-VOICE-PAST ‘This garden door is made open’ --(I)k adjectival participle

b.

Bu cetvel eg-ik ol-muş. this ruler bend-PTPL become-PAST ‘This ruler became bent’

(38)

a. Bu bahçe kapısı aç-ıl-mış yap-ıl-mış. this garden door-CM open-VOICE-PTPL make-VOICE-PAST Intended reading: ‘This garden door is made opened’

b.

Bu cetvel eg-il-miş ol-muş. --mIş adjectival participle this ruler bend-VOICE-PTPL become-PAST Intended reading: ‘This ruler became bent’

(39)

a. Bu site-ler diz-ili ol-muş. this building-PL order-PTPL become-PAST Intended reading: ‘These buildings became ordered.’ --(I)lI adjectival participle

b.

Bu kitabe-ler yaz-ılı yap-ıl-mış. this inscription-PL write-PTPL make-VOICE-PAST Intended reading: ‘These inscriptions are made written.’

This test further indicates that adjectival participle with --(I)k denote states not resulting from prior events while adjectivalparticiples with --(I)lI and --mIş denote states making reference to a prior event.

To sum up, Turkish distinguishes lexical and phrasal adjectival participles based on (i) incompatibility of lexicaladjectival participles with event-oriented manner adverb modification, (ii) additional interpretation that we get with phrasaladjectival participles in cases where both types of adjectival participles are compatible with adverb modification, (iii)compatibility of lexical adjectival participles with verbs of creation and verbs of change of state as this group of participlesdenote characteristic state.

Based on these data we suggest the following internal structure for --(I)k lexical adjectival participles.

(40)

ASPLEXICAL

√ ASP

çiz -ik ‘scratch’

The aspectual stativizer is attached directly to the root. As illustrated in (33b) --(I)k participles allow some adverbialmodification but there is not a verbal projection to encode eventivity. Hence event oriented, manner adverbial modificationis not possible with --(I)k participles. Adjectival participles with --(I)k can surface as complements of verbs of creation asthey denote characteristic state.

The next section discusses the semantic and syntactic properties of phrasal adjectival participles with --(I)lI and --mIşmarkers.

4. Phrasal adjectival participles in Turkish

In the literature a further distinction is made for phrasal adjectival participles as target state and resultant state participles(Kratzer, 2000). Target state participles denote reversible, transitory states while resultant state participles denote states thathold forever after the event that brings them about. Parsons (1990) defines these terms in the following way:

Resultant States

‘‘For every event e that culminates, there is a corresponding state that holds forever after. This is ‘‘the state of e’shaving culminated’’, which I call the ‘‘Resultant state of e’’ or ‘‘e’s R state’’. If Mary eats lunch, then there is a statethat holds forever after: The state of Mary’s having eaten lunch.’’

A. Gürer / Lingua 149 (2014) 166--187176

Target States

4 On

(1)

The reproblemthe evescope

‘‘It is important not to identify the Resultant-state of an event with its ‘‘target’’ state. If I throw a ball onto the roof, thetarget state of this event is the ball’s being on the roof, a state that may or may not last for a long time. What I amcalling the Resultant-state is different; it is the state of my having thrown the ball onto the roof, and it is a state thatcannot cease holding at some later time.’’. . .. . ...

(Parsons, 1990, cited in Kratzer, 2000)

Out of the four main uses of present perfect (Comrie, 1976, Iatridou et al., 2001) as (i) perfect of result, (ii) experientialperfect, (iii) perfect of persistent situation, (vi) perfect of recent past, resultant state participles are identified with theperfect of result. Comrie (1976) suggests that with the perfect of result ‘a recent present state is referred to as being theresult of some past situation’ and the resulting state is not transitory but holds ever after the event that brings them about.

This line of an analysis predicts that only target state participles are compatible with the adverbial ‘still’ which impliesthat the modified state is transitory in nature. Anagnostopoulou (2003) makes a two-way distinction for the phrasaladjectival participles in Greek based on the following examples.

(41)

e o

Duyo‘Y

vieantof

a.

*

*

f the r

kanu canou can

wer notic for t, the grthis stu

Ta

ewieve

nst die thegreet t

tes thathe semammatdy and

pedhia

r notes that

Gäste e guests ahe guests fo

as this pair

antic distincicality of the

we leave i

ine

the s

ine

n

r an

is semtion o

struct for fu

akoma

tructure in

Stunde lahour lohour’.

antically ef phrasal ature in (1)

rther inve

krimena

(42b) is bet

ng begrüssng greet

quivalent, gdjectival parcan be due tstigation.

The

children are still hidden b. To theorima ine (*akoma) apodedigmeno

ter when us

en.

rammaticalitticiples. As

o modificati

The

theorem is (still) proven (Anagnostopoulou, 2003, 52a-53a)

Although both of the forms are phrasal adjectival participles with --menos; in (41a) the target state participle denotes areversible state and it is grammatical with the adverbial ‘still’ but in (41b) the resultant state participle which denotes anirreversible state is not grammatical with the same adverbial.

The other testing ground is ‘for adverbials’ which is expected to be compatible only with the verbs that can form targetstate participles by virtue of being reversible in nature. This is illustrated below for German.

(42)

a. Die Mutter hat die Geisslein für ein paar Stunden

ed with the ex

y of this structuthese adverbiaon of the even

versteckt.

The mother has the little goats for a few hours hidden Implies: the little goats were hidden for a few hours.

b.

Du kannst die Gäste für eine Stunde begrüssen. You can the guests for an hour greet ‘You can greet the guests for an hour’. (Kratzer, 2000, 10a-11a)

The verbs that can surface as target state participles can be modified by ‘for adverbials’ as in (42a) but the verbs that cansurface as resultant state participles are incompatible with ‘for adverbials’ as in (42b). The irreversible resultant statenamely the guests’ holding the property of being greeted for an hour cannot be modified by ‘for adverbial’. However thestructure is fine when the adverbial modifies the event of greeting the guests.4

Now we will see whether a similar distinction is also observed in Turkish with phrasal participles taking --(I)lI or --mIşmorphology. We will start with the first diagnostics.

(43)

a. saç-ı hala yap-ılı. hair-3SG POSS still make-PTPL ‘Her hair is still done’

b.

saç-ı hala yap-ıl-mış. hair-3SG POSS still make-VOICE-PTPL Intended reading: ‘Her hair is still done’

pression ‘lang’ instead of ‘für’ as illustrated below.

re and the ungrammaticality of the structure in (42b) isls can modify both the event and the resultant state oft itself not the resultant state. This issue is beyond the

A. Gürer / Lingua 149 (2014) 166--187 177

While the form with the participle --(I)lI is compatible with ‘still’ (43a), the one with the --mIş participle is not (43b). The formswith --mIş, which are incompatible with the adverbial ‘still’, cannot also be modified by ‘for adverbials’.

(44)

a.

*

Çocug-un

kol-u iki saat boyunca sar-ılı-ydı. child-GEN arm-3SG POSS two hour during bandage-PTPL-PAST ‘The child’s arm was bandaged for two hours’

b.

Çocug-un kol-u iki saat boyunca sar-ıl-mış-tı. child-GEN arm-3SG POSS two hour during bandage-VOICE-PTPL-PAST Intended reading: ‘The child’s arm was bandaged for two hours’

In (44a) the adverbial modifies the state of being bandaged for a period of time. The structure in (44b) is not compatiblewith adverbial modification when the adverb modifies the resultant state of the arm’s being bandaged for 2 h. The samestructure is grammatical only when the adverbial modifies the length of the bandaging process.

Based on these data we conclude that the participles with --mIş are resultant state participles denoting irreversiblestates and the ones with --(I)lI are target state participles denoting transitory states. We suggest that followingrepresentations for the phrasal adjectival participles in Turkish which will be modified for resultant state participles inSection 5.1.

(45)

ASPTARGET ASP RESULTATIVE

vP ASP vP ASP-ılı -mış

√ v √ v sar sarıl

The aspectual stativizers ASPTARGET and ASPRESULTATIVE do not directly attach to the root but to the eventivelayer. This additional layer makes adverb modification possible with phrasal adjectival participles in contrast to thelexical adjectival participle given in (40). However the semantic nature of Asp in both cases differ and hence we getdifferent grammaticality judgments for ‘for adverbials’ and ‘still’. The next section presents further evidence for thisclassification.

4.1. Compatibility with agentive phrases

In Greek, the forms analyzed as resultant state participles are not only compatible with event oriented manner adverbsbut also with agentive phrases, purpose clauses and agent oriented adverbs while target state participles are not(Anagnastopoulou 2003). In German both resultant and target state participles are not compatible with agentive phrases,purpose clauses and agent oriented adverbs (Kratzer, 2000). This diagnostic is important in the sense that as eventoriented manner adverbs indicates an eventive projection, agentive phrases, purpose clauses and agent orientedadverbs all indicate the presence of a voice projection that hosts the implicit external argument.

Now we will see how Turkish adjectival participles behave with respect to this diagnostic.

-mIş participles

(46) a. çocuk-lar tarafından kes-il-miş kagıt-lar

child-PL

by cut-VOICE-PTPL paper-PL ‘The papers cut by the children’

b.

bu musluk başı kullanışlı ol-ma-sı için bük-ül-müş. this faucet knob handy be-NOML-3SG POSS for bend-VOICE-PTPL ‘This faucet knob is bent to be handy’

c.

kolayca al-ın-ma-sı için kasıtlı olarak diz-il-miş kitap-lar easily take-VOICE-NOML-3SG POSS for deliberately order-VOICE-PTPL book-PL ‘The deliberately ordered books to be taken easily’

d.

pergel-le çiz-il-miş şekil-ler compasses-INST draw-VOICE-PTPL shape-PL ‘The shapes drawn with compasses’

A. Gürer / Lingua 149 (2014) 166--187178

Table 1distinctive

Tests for

Adjective

The even

The (ir)reThe voice

a As illuinclude o

--(I)lI participles

*

*

*

*

*

*

*

*

properties of participles in Turkish.

Lexicaladjectivalparticiples-Ik

hood Appear in pre-nominal position +

Complements of ‘act, become, seem, look’ +

tive layer Event-oriented manner dverb modification �

Complements of ‘build,make’ that denote characteristic states

+

versible states Compatibility with ‘still’or ‘remain’ +

projection Agent PP �

Purpose clauses �

Agent oriented adverbials �

Instrumental Phrases �

strated in Section 2, relative clause participles can also appear in pre-nominal positinly the verbal-eventive passives exemplified in (15d).

Targepartic

-(I)lI

+

+

+

�

+

�

�

�

�

on, how

t stateiples

ever

(47)

a. hemşire tarafından sar-ılı kol nurse by bandage-PTPL arm Intended reading: ‘The arm bandaged by the nurse’

b.

bu musluk başı kullanışlı ol-ma-sı için bük-ülü. this faucet knob handy be-NOML-POSS for bend-PTPL Intended reading: ‘This faucet knob is bent to be handy’

c.

kimse-nin içeri gir-me-me-si için kasıtlı olarak

verb

kapa-lı

Resultantstateparticiples-mIş

+

+

+

�

�

+

+

+

+

al eventive parti

kapı

Verbapartici

�a

�+�

�++++

ciples in

somebody-GEN

inside enter-NEG-NOML-3SG POSS for deliberately close-PTPL door Intended reading: ‘The deliberately closed door to make nobody enter’

d.

dolmakalem-le yaz-ılı mektup-lar fountain pen-INST write-PTPL letter-PL Intended reading: ‘the letters written with a fountain pen’

-Ik participles

(48) a. belediye tarafından yık-ık bina

municipality

by demolish-PTPL building Intended reading: ‘The building demolished by the municipality’

b.

bu musluk başı kullanışlı ol-ma-sı için bük-ük. this faucet knob handy be-NOML-3SG POSS for bend-PTPL Intended reading: ‘This faucet knob is bent to be handy’

c.

insan-lar-ı acındırmak için kasıtlı olarak sök-ük elbise-ler person-PL-ACC arouse pity for deliberately unravel-PTPL clothe-PL Intended reading: ‘The deliberately unraveled clothes to arouse pity of people’

d.

makas-la kes-ik kagıt-lar scissors-INST cut-PTPL paper-PL Intended reading: ‘The papers cut with scissors’

The grammaticality contrasts clearly indicate that only the resultant state participles with --mIş are licit with agent PPs,purpose clause, agent oriented adverbials and instrumental phrases which indicates the presence of a voice projection forthese participles the discussion of which we leave for Section 5.1.

The following table summarizes the distinctive properties of lexical adjectival, target state, resultant state participlesand verbal-eventive participles in Turkish (Table 1).

The next section focuses on the restrictions on participle formation in Turkish which further sheds light on the internalstructure of participles.

l-eventiveples

the table

A. Gürer / Lingua 149 (2014) 166--187 179

5. Restrictions on participle formation

Most of the roots that form lexical adjectival participles with the --(I)k suffix do not have a phrasal adjectival counterpartwith --(I)lI.

(49)

a. aç-ık open-PTPL ‘opened’

b.

aç-ılı *open-PTPL

(50)

a. eg-ik slant-PTPL ‘slanted’

b.

eg-ili *slant-PTPL

However there are some forms that can surface with both forms.

(51)

a. bük-ük kol bend-PTPL arm ‘bent arm’

b.

bük-ülü musluk başı bend-PTPL faucet knob ‘bent faucet knob’

(52)

a. çiz-ik yüzey scratch-PTPL surface ‘scratched surface’

b.

alt-ı çiz-ili satır-lar down-3SG POSS draw-PTPL line-PL ‘the underlined lines’

Note that the forms in (51--52) have different interpretations with --(I)k and -(I)lI forms.

All the forms that can form a lexical participle with the --(I)k suffix or target state participle withthe --(I)lI suffix can surface as resultant state participle with the --mIş suffix.

(53)

a. aç-ık kapı open-PTPL door ‘the open door’

b.

aç-ıl-mış kapı open-VOICE-PTPL door ‘the opened door’

(54)

a. sar-ılı kol wrap-PTPL arm ‘the bandaged arm’

b.

sar-ıl-mış kol wrap-VOICE-PTPL arm ‘the bandaged arm’

Now we will investigate the compatibility of --(I)k, -(I)lI and --mIş morphology with the aspectual verb classes as illustrated

below (Vendler, 1957; Dik, 1989; Smith, 1997).

(55)

States [+durative, -dynamic, -telic] Actions [+durative, +dynamic, -telic] Accomplishments [+durative, +dynamic, +telic] Achievements [-durative, +dynamic, +telic]

A. Gürer / Lingua 149 (2014) 166--187180

Accomplishments and achievements are inherently bounded and have an end point and hence they are marked as[+telic]. States do not require energy for continuation and they do not have an end point; hence they are [--dynamic] and[-telic]. Achievements do not encode duration in time but they denote instantaneous situations and hence they are[--durative]. As discussed in detail in Sections 3 and 4, lexical adjectival participles denote characteristic states not derivedfrom prior events while phrasal adjectival participles denote states resulting from prior events and all participles include anaspectual stativizer. Hence we predict both lexical and adjectival participles to be derived from verbs that have an inherentend-point namely which are [+telic].

Nakipoglu-Demiralp (1998) independently notes that a verb can bear the --(I)k participle only if it is telic, aspectually delimited.This is also the case with participles bearing -(I)lI morphology. Hence --(I)k and -(I)lI cannot attach to activity verbs, states whichare not bounded in nature and do not encode an endpoint, and also to achievements which indicate an instantaneous state.Both --(I)k and -(I)lI participles can be derived from accomplishment verbs which denote events with a natural endpoint.5

*

(56)

5 Althodenote c

a.

*

*

*

*

*

*

*

*

ugh lexharacte

koş-uk

ical adjectival

ristic states w

çocuk

run-PTPL child Intended reading: ‘the child who has run’

b.

bil-ik

paith

çocuk

know-PTPL child Intended reading: ‘the child who knows’

c.

var-ık

r

yolcu

reach-PTPL passenger Intended reading: ‘the passenger who has reached’

d.

yık-ık

ticipout m

ev

demolish-PTPL house ‘the demolished house’

les with --(I)k araking referen

(57)

a. koş-ulu çocuk run-PTPL child Intended reading: ‘the child who has run’

b.

bil-ili çocuk know-PTPL child Intended reading: ‘the child who knows’

c.

var-ılı yolcu reach-PTPL passenger Intended reading: ‘the passenger who has reached’

d.

kapa-lı ev close-PTPL house ‘closed house’

In a similar vein, not all types of verbs can form phrasal participles with --mIş suffix. Activity verbs (58a-b), statives (58c)and achievements (58d) are also not compatible with phrasal participle

--mIş but accomplishments are fully acceptable (58e).

(58)

a. ögren-miş adam learn-PTPL man Intended reading: ‘a man who has learned’

(Slobin and Aksu, 1982, 6, with my modifications)

b. koş-muş atlet

run-PTPL

athlete Intended reading: ‘the athlete who has run’

(Nakipoglu-Demiralp, 1998, 43a, with my modifications)

c. bil-miş ögrenci

know-PTPL

student Intended reading: ‘a student who knows’

e derived from accomplishment verbs, which encode a process with a natural end point, theyce to a prior event as illustrated with ample data in Section 3.

*

A. Gürer / Lingua 149 (2014) 166--187 181

6 Unacmodified

cannot b

(1) Teda‘T

(2) *BibuIn

(3) *çosoIn

(4) *GöeyIn

d.

cusativby evee the co

hlikeli

ngerouhe buildna

ilding

tended

rba haup stitended

z-ler-ime-PL-Ptended

var-mış

es with -mIş mornt-oriented mannmplement of ‘re

bir biçim-dsly a style --Ling that has dançök-müş

collapse-PTPL

reading: ‘The bula sogu-muşll cool-PTPLreading: ‘The so

şiş-miş

OSS swell-PTPreading: ‘My eye

yolcu

phology are reser adverbs, (ii)main’.

e/aniden

OC/suddenly

gerously/suddol-du.become-PASTilding became

up is still coolekal-dı

L remain-PASs remained sw

reach-PTPL

passenger Intended reading: ‘a passenger who has arrived’

e.

inşa ed-il-miş

ul ca

çöcoen

co

d’

Tol

ev

tant statennot surf

k-müş

llapse-PTly collaps

llapsed’

len’

build

make-VOICE-PTPL house ‘the house which is built’

The same restrictions are imposed on --mIş participial formation in that --mIş participles can be derived fromaccomplishment verbs. However, in contrast to the structures with the --(I)k and -(I)lI participles exemplified in (56--57), theungrammatical structures with --mIş participles given in (58) become grammatical when they are accompanied by [+telic]expressions.

(59)

a. dil bilim-i ögren-miş adam Linguistics-ACC learn-PTPL man ‘a man who has learned linguistics’ (Slobin and Aksu, 1982, 7, with my modifications)

b.

marathon koş-muş atlet marathon run-PTPL athlete ‘the athlete who has run a marathon’ (Nakipoglu-Demiralp, 1998, 43b, with my modifications)

c.

terminal-e var-mış yolcu station-DAT reach-PTPL passenger ‘the passenger who has reached the station’

The event measuring DPs mark telicity in VP internal structure and have an effect on the inner aspect of the verb aspointed out by Travis (2005). The additional DPs ‘dil bilimi’,‘marathon’, ‘terminal’ force a telic interpretation. The processverbs forming --mIş participles in (59a-b) which are inherently [-telic] are interpreted to have an end-point, a resultant state.Hence the ungrammatical adjectival participles in (58a-b) are fully grammatical in (59a-b). The achievement verb which isalready telic becomes acceptable with --mIş morphology with an overt telic expression. We suggest that as resultant stateparticiples denote states resulting from prior events, the telic expression in (59c) shifts the instantaneous implication of theverb to a process with an end point and the adjectival participle becomes grammatical.

Statives do not encode an endpoint and hence they are unacceptable with --mIş participle. In the grammaticalcounterpart of (58c) given in (60) with the additional DP the stative verb is interpreted as an accomplishment having anendpoint.

(60)

cevab-ı bil-miş ögrenci answer-ACC know-PTPL student ‘The student who has known the answer’

Unaccusative verbs which form participles with --mIş show the same property.6 Telic change of state verbs can formparticiples with --mIş as in (61), while [-telic] change of state verbs can form --mIş participles only when accompanied bytelic expressions as in (62).

participles as the following diagnostics indicate. The adjectival participles (i) can beace as complements of ‘become’ (iii) cannot be modified by the adverbial ‘still’, (iv)

binaPL buildinged’

(61) a. çürü-müş elma

A. Gürer / Lingua 149 (2014) 166--187182

7 In Greand Anag

(1) asko

ek alsonostop

pr-iz-o

kin-iz-o

the suffix --izoulou, 2008)

‘whiten’ a ‘redden’ k

rot-PTPL

apple ‘rotten apple’

b.

eri-miş

w.

spok

mum

melt-PTPL candle ‘melted candle’

hich verbalizes

r-iz-men-os ain-iz-menos k

(62)

a. ? büyü-müş çiçek grow-PTPL flower ‘grown flower’

b.

iki hafta iç-in-de

the roots

spr-os/i/o

okin-os/i/o

büyü-müş

form phrasal pa

‘white’ *tos ‘red’ *tos

çiçek

two week in-3SG POSS-LOC grow-PTPL flower ‘flower that grew up in two weeks’

Based on an experimental study on unaccusative/unergative distinction, Acartürk (2005) independently shows thatchange of location verbs are more compatible with --mIş participle when they are used with dative or locative expressionsthat denote an end point as indicated in (63).

(63)

a. ?? gel-miş misafir come-PTPL guest Intended reading: ‘a guest who has come’

b.

Kadın, tehlikeli bir biçim-de yer-e

rticiple

düş-müş

s with the su

adam-a

ffix --menos on

(Alexiad

yardım

ly, and de

ou and A

et-ti.

woman dangerously a style--LOC ground-DAT fall-PTPL man-DAT help make-PAST ‘The woman helped the man who has dangerously fallen onto the ground.’

(Acartürk, 2005, 1 pg. 105, with my modifications)

The fact that verbs which are [-telic] or [+telic] but [-durative] in nature can form --mIş participle only when they areaccompanied by [+telic] expressions clearly indicates that the participles with --mIş encode a process with its resultantstate.

This is further supported by suffixes that form verbs from adjectives. There are some suffixes such as --lAş and --(A)rthat attach to adjectives to form verbs that encode the process of attaining a particular state (Göksel and Kerslake, 2005).7

All these derived verbs form phrasal participles with --mIş.

(64)

a. sarı yellow ‘yellow’

b.

sar-ar yellow-VERBL ‘become yellow’

c.

sar-ar-mış yaprak-lar yellow-VERBL-PTPL leaf-PL ‘leaves that became yellow’

(65)

a. güzel beautiful ‘beautiful’

b.

güzel-leş beautiful-VERBL ‘become beautiful’

notes eventivity (Alexiadou

nagnostopoulou, 2008, 32)

A. Gürer / Lingua 149 (2014) 166--187 183

8 Slobi--mIş is a

c.

*

n and An exte

güzel-leş-miş

*

ksu (1982) suggest that as thnsion of participial function.

kız-lar

beautiful-VERBL-PTPL girl-PL ‘girls that became beautiful’

In Turkish linguistics literature, --mIş is studied extensively and many functions have been attributed to this suffixsuch as inference, hearsay, perfect, narrative and evidential mood (Johanson, 1971; Banguaoglu, 1974; Underhill,1976; Yavaş, 1980; Slobin and Aksu, 1982; Aksu-Koç, 1988; Taylan, 1996; Kornfilt, 1997; Kelepir, 2000; Johanson,2000; Taylan, 2001; Göksel and Kerslake, 2005; Arslan-Kechriotis, 2006 among many others). Johanson (1971, 2000)suggests that --mIş in Turkish is not a resultative perfect but it encodes inferential and experiential meanings. Slobin andAksu (1982) and Aksu-Koç (1988) on the other hand note that ‘past participle must encode both process and resultantstate.’8

As illustrated above, verbs from different aspectual groups can form participles with --mIş only when they are restrictedby telic expressions that force the verb to have and end point or to be interpreted as a process with an end point. Hence,within this study --mIş is analyzed as denoting a process with its resultant state in line with Slobin and Aksu (1982). Thedata discussed in this section has further shown that the Asp head in adjectival participles differ semantically in adjectivalparticiples as the addition of telic expressions save the structures with --mIş participles, this is not the case with --(I)k and-(I)lI participles as illustrated in (56--57).

In what follows, we turn to the suffixes --(I)l/(I)n that surfaces with resultant state participles.

5.1. Voice projection in resultant state participles

Participles with --mIş are also not compatible with inherently causative verbs as in (66a). This structure becomesgrammatical only when --(I)l/(I)n is attached to the root as in (66b).

(66)

a. kır-mış bardak break-PTPL glass Intended reading: ‘broken glass’

b.

kır-ıl-mış bardak break-VOICE-PTPL glass ‘broken glass’ (Slobin and Aksu, 1982, 8--9, with my modifications)

We suggest that there is an implicit agent implication with the verb kır-. Hence in order to denote the state of beingbroken resulting from a prior event, the initiator of the process, the implicit causer, has to be specified. As illustrated withample data in Section 4.1, the participles with --mIş allow an implicit agent that is evidenced by the compatibility ofAgent PPs, purpose clauses, agent oriented adverbials and instrumental phrases with these participles. Based onthese data we suggest that --(I)l/(I)n is the head of the Voice projection that hosts the implicit causer or the implicitexternal argument.

This is further supported by unaccusatives that bear --mIş morphology. The sole arguments of unaccusatives aresuggested to be the internal argument of the verb not the agent of the verb (Perlmutter, 1978). Note that unaccusativeverbs forming --mIş participles become ungrammatical with the addition of --(I)l/(I)n morphology.

(67)

a. sol-muş çiçek-ler wilt-PTPL flower-PL ‘wilted flowers’

b.

sol-un-muş çiçek-ler *

(68)

a. don-muş göl freeze-PTPL lake ‘frozen lake’

b.

don-ul-muş göl

e participial function of --mIş includes both the process and the resultant state, past tense use of

A. Gürer / Lingua 149 (2014) 166--187184

These structures become grammatical only when causative marker is attached to the root.

(69)

9 Uyguwhich ma

a.

n (2009y not b

sol-dur-ul-muş

) takes the --Il/In suffix not as mae phonologically realized as a

çiçek-ler

wilt-CAUS-VOICE-PTPL flower-PTPL Literally: ‘flowers that are made wilted’

b.

don-dur-ul-muş

rkenull

göl

freeze-CAUS-VOICE-PTPL lake Literally: ‘lake which is made frozen’

As mentioned in Section 2, in addition to the --mIş participle formation, the suffix --(I)l/(I)n appears with verbal-eventivepassives, reflexives, middles, and in some reciprocals in Turkish. Taneri (1993) suggests that the --(I)l/(I)n suffix is theexpression of an implicit argument in passives, reflexives and inchoatives. Zeyrek (2006) also claims that the --(I)l/(I)nsuffix denotes an implicit argument in inchoatives.9

Based on the above mentioned examples, in line with Taneri (1993) and Zeyrek (2006), we take (I)l/(I)n as the markerof voice that projects VoiceP and hosts agent PPs, instrumental phrases and agent oriented adverbs with --mIş participles.The causative marker in (69) introduces an implicit causer which the VoiceP hosts and hence the structures becomegrammatical.

6. Internal structure of participles

The ample data we have discussed have shown that Turkish shows a tripartite classification for adjectival participleswith different morphological markers in each case. Lexical adjectival participles with --(I)k denote characteristic states,without making reference to a prior event. Phrasal adjectival participles denote states resulting from a prior event.

Based on the presence of an event implication with phrasal adjectival participles and the absence of an eventimplication with lexical adjectival participles, lexicon and syntax have been proposed to be the formation domain of lexicaland phrasal adjectival participles respectively (Wasow, 1977; Aya Meltzer-Asscher, 2010). Within this study, in line withKratzer (1994, 2000), Anagnostopoulou (2003), Alexiadou and Anagnostopoulou (2008), participles are not differentiatedbased on the domain of formation but based on different attachment sites of participial morphology in syntax.

Lexical adjectival participles with --(I)k are illicit with event-oriented manner adverb modification, agent PPs, purposeclauses, agent oriented adverbs and instrumental phrases which indicate that the aspectual stativizer do not attach to aneventive layer or VoiceP but it is directly attached to the root. Event oriented and agent oriented adverbial modification isnot licit because there is not a verbal projection to host adverbs or a Voice projection to host an implicit agent as illustratedin (40) repeated below as (70).

(70)

ASPLEXICAL

√ ASPçiz –ik

‘scratch’

The internal structure of phrasal adjectival participles is distinguished from the internal structure of lexical adjectivalparticiples based on the additional eventive layer in phrasal adjectival participles as evidenced by event-oriented adverbmodification. We suggest that ASP head in phrasal adjectival participles does not directly attach to the root but selects forvP as is the case in Greek (Alexiadou and Anagnostopoulou, 2008).

However there are some syntactic and semantic differences between the phrasal adjectival participles. Resultantstate participles with --mIş morphology differ from target state participles with --(I)lI morphology in that only resultant stateparticiples are licit with agent PPs, purpose clauses, agent oriented adverbs and instrumental phrases which indicatesthe presence of a VoiceP. Similar to its Greek counterparts, resultant state participles in Turkish attach above VoiceP,

r of implicit causer but as phonologically realized head of BECOMEP in inchoative formation-head with certain group of verbs.

(Uygun, 2009, 47-48b)

A. Gürer / Lingua 149 (2014) 166--187 185

while target state participles attach below VoiceP as illustrated in (71) for target state participles and in (72) for resultantparticiples.

(71)

√

k‘

ASPTARGET

vP ASP-lı

v

apaclose’

(72)

ASPRESULTATIVE

VoiceP ASP-mış

vP -ıl

√ v aç

‘open’

Additionally, the semantic compositionality of the ASP head in resultant and target state participles differ in that (i)resultant state participles denote irreversible states and hence they are not compatible with ‘for adverbials’ and ‘still’ whichentails the modified state to be transitory, (ii) resultant state participles derived from [-telic] activity verbs become licit whenaccompanied by telic expressions while this is not possible with target state participles.

As for the representation of unaccusative verbs forming adjectival participles with--mIş morphology, we suggest thefollowing structure in syntax.

(73)

ASPRESULTATIVE

vP ASP-müş

√ vçürü‘rot’

The participles with --mIş do not have a voice projection but they attach to an eventive layer. These participles are notcompatible with the expressions ‘still’ and ‘remain’ which differentiates them from target state participles.

However the structure in (73) is the same with target state participles given in (71) in that there is not a VoicePbelow AspP and the AspP selects for a verbal projection. We can capture the difference between (71) and (73) throughdifferent semantic natures of the target and resultant state aspects in each case. The discussion in Section 5 hasalready shown that even in the absence of a VoiceP, resultant state aspect is different from target state aspect in thatwith the addition of [+telic] expressions, verbs from different aspectual groups can form resultant state participles with--mIş morphology while this is not possible with target state participles with --(I)lI morphology. Hence the distinctionbetween target state and resultant state participles cannot be reduced to the presence of the VoiceP with resultantstate participles.

The other alternative is to assume that VoiceP projects in all cases with --mIş participles but it is overtly realized withunaccusatives only when there is a causative marker that evokes an implicit causer interpretation. Although both of thealternatives can capture the Turkish data, we will choose the first option and do not assume a vacuous Voice projectionwith unaccusatives. Even in the absence of VoiceP, different behavior of the aspectual heads with respect to compatibilitywith still, for-adverbial and compatibility with different aspectual verbs indicate that semantic composition of stativizersdiffer for phrasal adjectival participles.

Within this analysis lexical and phrasal adjectival participles are not formed in different domains. Adjectival participlesare all formed in syntax and it is the internal structure of the participles together with the different semantic compositionalityof ASP heads that allows participles to have distinctive properties.

7. Conclusion

In this paper, I have proposed a typology of adjectival participles in Turkish based on a variety of diagnostic tests. Thestudy shows that Turkish marks the three types of adjectival participles with different morphological markers. Instead ofsuggesting lexicon as the formation domain of lexical adjectival participles and syntax as the formation domain of phrasaladjectival participles (Wasow, 1977; Aya Meltzer-Asscher, 2010), syntax is proposed as the single formation domain

A. Gürer / Lingua 149 (2014) 166--187186

(Kratzer, 1994, 2000; Anagnostopoulou, 2003; Alexiadou and Anagnostopoulou, 2008). The differences between theparticiples are captured via different attachment sites of the participle morphology and different semantic properties of theaspectual stativizers.

Based on the syntactic and semantic diagnostics discussed in the paper, we conclude that adjectival participles with--Ik morphology are lexical adjectival participles denoting characteristic states without a prior event implication. They arederived in syntax but do not allow event-oriented adverb modification as the Asp head attaches directly to the root in theabsence of an eventive layer. Phrasal adjectival participles with -(I)lI and --mIş differ from lexical adjectival participles inthat they allow event-oriented adverb modification which is taken as an indication of a verbal projection above the root.However the phrasal adjectival participles do not form a homogenous group from semantic and syntactic point of view.The participles with --mIş denote irreversible states, participles with -(I)lI denote reversible states. This is supported bycompatibility with ‘still’ and ‘for adverbials’. Additionally participles with --mIş allow agent implication as is the case in Greekwhich indicates a Voice projection in the structure.

Acknowledgements

I would like to thank Balkız Ӧztürk for her insightful discussions and invaluable comments on the study. I have alsobenefitted from the invaluable comments of Aslı Göksel and the anonymous reviewers of the journal. The first version ofthis paper was presented at the Workshop on Aspect and Argument Structure of Adjectives and Participles, at theUniversity of Greenwich. I would like to thank the audiences of the workshop for their questions and comments on thestudy. All errors are mine.

References

Acartürk, C., 2005. Gradient Characteristics of the Unaccusative/Unergative Distinction in Turkish: An Experimental Investigation. (UnpublishedMA thesis). The Middle East Technical University.

Aksu-Koç, A., 1988. The Acquisition of Aspect and Modality: The Case of Past Reference in Turkish. Cambridge University Press, Cambridge.Alexiadou, A., Anagnostopoulou, E., 2008. Structuring participles. In: Chang, C.B., Haynie, H.J. (Eds.), Proceedings of the 26th West Coast

Conference on Formal Linguistics. Cascadilla Proceedings Project, Somerville, MA, pp. 33--41.Anagnostopoulou, E., 2003. Participles and voice. In: Alexiadou, A., Rathert, M., von Stechow, A. (Eds.), Perfect Explorations. Mouton de Gruyter,

Berlin, pp. 1--36.Arslan-Kechriotis, C., 2006. Perfect in Turkish. Turk. Lang. 10, 246--270.Aya Meltzer-Asscher, 2010. Present participles: categorical classification and derivation. Lingua 120, 2211--2239.Banguaoglu, T., 1974. Türkçe’nin Grameri. Baha Publications, Istanbul.Bresnan, J., 1982. The passive in lexical theory. In: Bresnan, J. (Ed.), The Mental Representation of Grammatical Relations. MIT Press,

Cambridge, MA.Comrie, B., 1976. Aspect. Cambridge University Press, Cambridge.Dik, S., 1989. The Theory of Functional Grammar, Part 1: The Structure of the Clause. Foris Publications, Dordrecht.Embick, D., 2004. On the structure of resultative participles in English. Linguist. Inq. 35 (3), 355--392.Göksel, A., Kerslake, C., 2005. Turkish: A Comprehensive Grammar. Routledge, London-New York.Iatridou, S., Anagnastopoulou, E., Izvorski, R., 2001. Observations about the Form and Meaning of the Perfect. In: Ken Hale: A Life in Language.

MIT Press, Cambridge, pp. 189--238.Johanson, L., 1971. Aspekt im Türkischen. Almqvist & Wiksell, Uppsala.Johanson, L., 2000. Viewpoint operators in European languages. In: Dahl, Ӧsten, (Eds.), Tense and Aspect in the Languages of Europe. Mouton

de Gruyter, Berlin/New York, pp. 27--187.Kelepir, M., 2000. Perfect Constructions in Turkish. MIT, MS.Kornfilt, J., 1997. Turkish. Routledge, London.Kornfit, J., 2001. Functional projections and their subjects in Turkish clauses. In: Erguvanlı Taylan, E. (Ed.), The Verb in Turkish. John Benjamins,

Amsterdam, pp. 183--213.Kratzer, A., 1994. The Event Argument and the Semantics of Voice. University of Massachusetts at Amherst, MS.Kratzer, A., 2000. Building statives. Paper presented at the Berkeley Linguistic Society Annual Meeting 26.Nakipoglu-Demiralp, M., 1998. Split Intransitivity and the Syntax-Semantics Interface in Turkish. (PhD dissertation). University of Minnesota.Parsons, T., 1990. Events in the Semantics of English. A Study in Subatomic Semantics. MIT Press, Cambridge, MA.Perlmutter, D., 1978. Impersonal passives and the unaccusative hypothesis. In: Proceedings of the Fourtth Annual Meeting of the Berkeley

Linguistics Society. University of California, Berkeley, pp. 157--189.Slobin, D.I., Aksu, A.A., 1982. Tense, aspect, and modality in the use of the Turkish evidential. In: Hopper, P.J. (Ed.), Tense-Aspect: Between

Syntax and Pragmatics. John Benjamins, Amsterdam.Smith, C., 1997. The Parameter of Aspect. Dordrecht, Kluwer.Taneri, M., 1993. Morpheme --(I)l(I)n: The syntax of personal passives, impersonal passives and middles in Turkish. (Unpublished PhD

dissertation). Dept. of Linguistics, University of Kansas.Taylan, E., 1996. On the parameter of aspect in Turkish. In: Konrot, A. (Ed.), Proceedings of the sixth International Conference on Turkish

Linguistics. Eskişehir, Anadolu Üniversitesi, pp. 153--168.

A. Gürer / Lingua 149 (2014) 166--187 187

Taylan, E., 2001. On the relation between temporal/aspectual adverbs and the verbb form in Turkish. In: Erguvanlı Taylan, E. (Ed.), The Verb inTurkish. John Benjamins, Amsterdam, pp. 97--128.

Travis, L.D., 2005. Articulated vPs and the computation of aspectual classes. In: Kempchinsky, P., Slabakova, R. (Eds.), Aspectual Inquiries.Springer, Dordrecht, pp. 69--93.

Underhill, R., 1976. Turkish Grammar. MIT Press, Cambridge.Uygun, D., 2009. A Split Model for Category Specification: Lexical Categories in Turkish. (Dissertation). Bogaziçi University.Vendler, Z., 1957. Verbs and times. Philos. Rev. 66 (2), 143--160.Wasow, T., 1977. Transformations and the Lexicon. In: Culicover, P., Wasow, T., Akmajian, A. (Eds.), Formal Syntax. Academic Press, New York,

pp. 327--360.Williams, E., 1981. Argument structure and morphology. Linguist. Rev. 1 (1), 81--114.Yavaş, F., 1980. On the Meaning of the Tense and Aspect Markers in Turkish. (Dissertation). University of Kansas.Zeyrek, D., 2006. Anticausatives in Turkish. In: Paper Presented at the XIII International Conference on Turkish Linguistics. Uppsala University,

Sweden.