Nurettin Can

İçindekiler / Contents Önsöг vii ÉЯa ÁРnОs CsaЭяμ RОsОarМС 1985 Ḳ 1 Aynur Abish BКlКnэŋ НünвОgО ɕОlüаТnО bКвlКnэstэ sКlttКr ‘CЮsЭoms МonМОrnТnР ЭСО bТrЭС oП a МСТlН’ 9 Sema Aslan Demir TürkmОnМОНО eken 21 Ahmet Aydemir TürkçОНО гarП-fiillerin bağlaвıМılık ТşlОЯlОrТ СakkınНa 31 Eвüp BaМanlı Türk DТllОrТnТn şaСıs гamТrlОrТnНО НüгОnsТгlТklОr 43 Ad Backus Living on the edge: Could Dutch Turkish become a peripheral Turkic language too? 63 ←аО BlтsТnР Vol de nuit: Ein sprachlich-kЮlТnarТsМСОs ErlОbnТs übОr den Wolken 79 Bernt Brendemoen Some remarks on the infinitive in -mA in 17th century Ottoman Turkish 103 Christiane Bulut Relics of subordinative structures in Iran-Turkic 115 Joakim Enwall The paradox of ethnic elites: planning and maintenance of minority lanРЮaРОs Тn ЭСО PОoplО’s RОpЮblТМ oП CСТna 127 εОЯlüЭ ErНОm TürkçОНО çОkТm ЯО ЭürОЭТm morПoloУТsТ ЯО söгlüklОrО вansıması 141 Margarete I. Ersen-Rasch NoМС ОТnmal гЮr γ. PОrson PlЮral Тm TürkТsМСОn 151 KarТna FТrkaЯТčТūЭė Endangered Karaim in fight for life 175 Peter B. Golden Qıpčaq 183 εОЯlüЭ GülЭОkТn TürklОr ağaçkakana ne derler? com 1. Introduction Critical sifting of the mass of word comparisons proposed by Altaic scholarship in the past has led to the reconstruction of a simple set of consonant correspondences between the Japonic, Koreanic, Tungusic, Mongolic and Turkic languages in Robbeets (2005: 373-377).1 The correspondences reflect a natural two-fold voicing distinction between three stops (pTEA *p/b, *t/d, *k/g) as well as a single fricative pTEA *s without voicing distinction, in contrast to the reconstruction of an additional fricative *x in Miller (1987) or to the more rigorous system of three-fold con- sonantism (voiceless-voiced-aspirated) reconstructed by Starostin and his colleagues (2003) on the basis of massive amounts of data. As for the velar obstruents, for instance, Starostin et al. (2003) listed the corre- spondences given in Table 1, whereas, restricting myself to a small core of reliable evidence, I arrived at a more limited set given in Table 2. pTEA pJ pK pTg pMo pTk *k- *k- *k- *k- *k- *g- *-k- *-k- Ẕė, Ẕ-h- / *-k *-k- *-g- / *-g *-k- / *-g(V)r *g- *k- *k- *g- *g- *g- *-g- *-k-, (Т↑ẓė Ẕė, Ẕ-h- / *-k *-g- *-h-, *-g(Vh) *-g- - *k- *k- *x- *k- *k- *-k‘- *-k- *-k-, -h- *-k-, -x- *-k-, *-g[Vh] / -g *-k- Table 1: Correspondences for velar obstruents according to Starostin et al. (2003: 25) 1 TСО labОl “TransОЮrasТan” аas МoТnОН bв JoСanson & RobbООЭs (β010μ 1‒βẓ Эo rОПОr Эo a large group of geograpСТМallв aНУaМОnЭ lanРЮaРОs, ЭraНТЭТonallв knoаn as “AlЭaТМ”, ЭСaЭ include up to five different linguistic families: Japonic, Koreanic, Tungusic, Mongolic, and Turkic. 376 Martine Robbeets pTEA pJ pK pTg pMo pTk *k- *k- *k- *k-, *x- *k- *k- *-k- *-k- *-k- *-k- *-k- *-k- *g- *k- *k- *g- *g- *k- *-g- *-k- *-k- *-g- *-g- *-g- Table 2: Correspondences for velar obstruents according to Robbeets (2005: 373) Although Benzing (1955a: 990) reconstructed pTg *x- as a distinct phoneme, in Robbeets (2005: 311), I refrained from reconstructing an original initial fricative pTEA *x- on the basis of the Tungusic correspondence alone because else, Tungusic would be the only branch in which the fricative was preserved. Moreover, the reconstruction of pTg *x by Benzing was restricted to initial position, which can be taken as an indication that it developed secondarily from pTg *k-. Furthermore, I rejected the correspondence between pJ *p- and pTg *x- proposed by Miller (1987) because only a single etymology reflecting it stood the sifting criteria (Robbeets 2005: 292). As sЮМС, mв ПТnНТnРs sООmОН Эo sЮpporЭ RamsЭОНЭ’s (1957μ 7βẓ aРnosЭТМ ЯТОа ЭСaЭ “DТО FraРО, ob ЮnН Тn аТО аОТЭ ОТn anlaЮЭОnНОs h- in den altaischen SpraМСОn ЮrsprünРlТМС ОбТsЭТОrЭ СaЭ oНОr nТМСЭ, mЮß аoСl Пür ТmmОr ЮnbОanЭаorЭОЭ blОТbОn,” ОЯОn ТП ОЯТНОnМО for pTEA *x was ḳ МonЭrarв Эo PoppО’s (1960μ γβ-33) oЯОrsЭaЭОmОnЭμ “Es ТsЭ kОТn GrЮnН ЯorСanНОn, ОТn ЮrsprünРlТМСОs Ẕh- oder *x- zu postulТОrОn. DТОsОr δaЮЭ ТsЭ nтmlТМС sОkЮnНтrОr NaЭЮr.” ḳ not completely lacking. In this article, I will advance evidence for the reconstruction of a medial velar fricative pTEA *-x-, thus refuting my initial objection against the addition of a velar fricative phoneme to the Transeurasian sound inventory. Interestingly, the main indications of the presence of such a proto-phoneme in Transeurasian come from morphological rather than lexical comparison. In Section 2, I will present comparative evidence for the reconstruction of a resultative nominalizer having two allomorphs, i.e. pTEA *-xA ~ -kA, which require the reconstruction of a velar fricative. In Section 3, I will complement morpho-phonological evidence with indications obtained from lexical comparison. SТnМО ЭСaЭ mОmorablО Нaв Тn ЭСО Пall oП β004, аСОn I ПТrsЭ mОЭ ProПОssor CsaЭя and her husband Professor Johanson near Akamon, the red entrance gate of the University of Tokyo, both scholars have repeatedly encouraged me to focus my attention on the comparison of morphology. Back in Europe, I had the pleasure to be among the students and researchers who are regularly invited at their home in Mainz to discuss various theoretical problems. One of these discussions, concerning the possible phonemic status of the velar fricative in early Old Turkic proposed in Johanson (1979) made me change my mind about the reconstruction of pTEA *-x- and is at the basis of the ideas that I would like to present in this article. As a sign of rОspОМЭ Пor ProПОssor CsaЭя’s sМСolarsСТp anН РraЭТЭЮНО Пor СОr mТlН РЮТНanМО oЯОr ЭСО past years, this article intends to show how comparative morphology can contribute to solving problems of phonological reconstruction. A velar fricative in proto-Transeurasian 377 2. Indications in bound morphology: a resultative nominalizer pTEA *-xA- 2.1. pTk *-xA ~ *-kA 2.1.1. Old Turkic The Old Turkic deverbal noun suffix -gA derives nominal and adnominal forms from verb stems, as illustrated in the derivational pairs in (1a) (Erdal 1991: 376-382). Since the suffix mainly derives action nouns and subject nouns from intransitive verbs (e.g. OTk. ɕüɕrОgО ‘ЭСЮnНОrТnР’ν öpɕО ‘lЮnР’ẓ, obУОМЭ noЮns Пrom ЭransТЭТЯО verbs (e.g. OTk. tilge ‘sЭrТp’ẓ anН sЭaЭТЯО propОrЭв noЮns Пrom МСanРО oП sЭaЭО ЯОrbs (e.g. ɕэsgК ‘sСorЭ’ẓ, ТЭ Мan bО МСaraМЭОrТгОН as a rОsЮlЭaЭТЯО nomТnalТгОr. TСОrО arО somО ТnsЭanМОs oП allОРОН “НТssТmТlaЭТon” Тn аСТМС the suffix OTk -gA seems to devoice to -kA following /r, p/, e.g. in ötürɕО, tКrɕК and öpɕОṬ If pre-OTk *yupka ‘ЭСТn, slОnНОr, ЮnsЮbsЭanЭТal’ rОПlОМЭОН Тn OTk. yuyka, Kharakh. yupka, Turkish yufka, Uzbek yupqa, etc. can be derived from an obsolete verb *yup- ‘Эo bОМomО ЭСТn’, ТЭ may also be an example of this phenomenon. (1) Reflexes of the resultative deverbal noun suffix pTk *-xA in Old Turkic (1a) lexical (ad)nominalization OTk. bil- ‘Эo knoа (Эr.ẓ’ bilge ‘аТsОν a аТsО pОrson, МoЮnsОllor’ OTk. ɕölТ- ‘Эo bО sСaНв, sСaНОНν Эo sСaНО, РТЯО sСaНО Эo (Эr./ТnЭr.ẓ’ OTk. ɕölТgО ‘sСaНoа, НООp sСaНО’ OTk. ɕöšТ- ‘Эo obsЭrЮМЭ lТРСЭ (ТnЭr.ẓ’ ɕöšТgО ‘lТРСЭ sСaНoа’ OTk ɕэs- ‘Эo pТnМС, sqЮООгО, rОНЮМО (Эr.ẓ’ ɕэsgК ‘sСorЭ’ OTk. ɕüɕrО- ‘Эo ЭСЮnНОr (ТnЭr.ẓ’ ɕüɕrОgО ‘ЭСЮnНОrТnР’ OTk. öp- ‘Эo kТss, sТp or sЮМk Тn ЭСО aТr or a lТqЮТН’ öpɕО ‘РОnОraЭОН Тn ЭСО lЮnРν lЮnР, anРОr’ OTk ötür- ‘Эo МaЮsО or ПorМО Эo pass ЭСroЮРС’ ötürɕО ‘pЮrРaЭТЯО’ OTk. sal- ‘to moЯО, pЮЭ Тn moЭТon, aРТЭaЭО’ salga ‘rОsЭТЯО’ OTk. tar- ‘to НТspОrsО, sОnН aаaв (Эr.ẓ’ tarka ‘alonО, lonОlв’ OTk. til- ‘Эo МЮЭ ТnЭo sЭrТps (Эr.ẓ’ tilge ‘sЭrТp’ OTk. tut- ‘Эo СolН, МaЭМС (Эr.ẓ’ tutga ‘СanНlО’ (1b) clausal adnominalization nirvan-ka bar-ga-sok-ta nirvana-DAT go-ADN-one.time-LOC ‘АСОn onО РoОs Эo NТrЯana’ (ErНal 1991μ 159ẓ 378 Martine Robbeets (1c) finite čОЯТš ay-u bẹr-ge men method explain-CONV give-FIN 1SG ‘I аТll ОбplaТn ЭСО mОЭСoН Пor вoЮ’ It is important, however, to realize that the rendering of the distinction between /g/ and /k/ in Old Turkic sources is imprecise. In traditional Turcological transcription (a.o. Erdal 2004: 78), the Old Turkic sources are thought to represent a system whereby voiced /g/ and voiceless /k/ are distinguished before back vowels, but not before front vowels. The back-harmonic variant of /g/ is thought to have been a fricative [ ], whereas the front variant is thought to have been a voiced stop [ģ]. Traditionally, the East Old Turkic runiform sign <k1>, the Ancient Uighur <ḥeth> sign and the Karakhanid Arabic <qa:f> are thus interpreted as the back voiceless stop [ ], whereas the East Old Turkic runiform sign <k2>, the Ancient Uighur <keph> sign and the Karakhanid Arabic <ka:f> sign are interpreted as either a front voiceless stop [ ] or a front voiced stop [ģ]. TСО EasЭ OlН TЮrkТМ rЮnТПorm sТРn < 1>, the double dots in Ancient Uighur and the Karakhanid Arabic <ghain> sign are taken as the back vocalic alternant of /g/ which is pronounced as a [ ] and stands in opposition to the back vocalic alternant of /k/. Hence, the notations OTk. ɕэsgК ‘sСorЭ’, OTk. yuyka ‘ЭСТn’, OTk. ɕölТgО ‘sСaНoа’ anН OTk. öpɕО ‘lЮnР’ ЭransМrТbО аСaЭ Тs commonly believed to be rendered in the sources as [ a], [yuy a], [ɕölТ e ~ ɕölТģО] and [öp e ~ öpģО], respectively. The determination of a voiced velar stop in the notation of OTk. ɕölТgО ‘sСaНoа’, bЮЭ a ЯoТМОlОss ЯОlar Тn OTk. öpɕО ‘lЮnР’ Тs dependent upon the reflexes in the contemporary Turkic languages, which display a ЯoТМОН ЯОlar Тn ЭСО ПormОr МasО, bЮЭ a ЯoТМОlОss Тn ЭСО laЭЭОrν О.Р., Пor ‘sСaНoа’μ Tk. gölgО, Gag. ölgО, Az. ɕölgт, Tkm. ɕölgО, Tat. ɕülɛgɛ, Bash. ɕülɛgɛ, Uig. ɕölТgɛ, Tuva бölОgО, Tof. бölОgО, ОЭМ.ν Пor ‘lЮnР’μ Tk. öПɕО ‘anРОr’, öвɕОn ‘lЮnР’, GaР. üПɕe, Az. öбbт, Tkm. öвɕОn, Tat. üpɕт, Kir. öpɕö, Bash. üpɕт, Sal. öССОn, Uz. , Uig. öpɕт, SUig. öɕpО, Shor öɕpО, Tuva öɕpО, Tof. öǯɕpО, etc. Johanson (1979, 2012, pc.), however, proposes a different interpretation of the Old Turkic signs for velars. In his view, the East Old Turkic runiform sign <k1>, the Ancient Uighur <ḥeth> sign and the Karakhanid Arabic <qa:f> cannot only stand for a back voiceless stop [ ], bЮЭ also Пor a baМk ЯoТМОН sЭop Д ], аСТlО ЭСО EasЭ OlН Turkic runiform sign <k2>, the Ancient Uighur <keph> sign and the Karakhanid Arabic <ka:f> sign stand for either a front voiceless stop [ ] or a front voiced stop [ģ]. SТnМО ЭСО EasЭ OlН TЮrkТМ rЮnТПorm sТРn < 1> which stands for a back fricative [ ] Сas a ПronЭ МoЮnЭОrparЭ < 2>, he interprОЭОs 2 as a ПronЭ ПrТМaЭТЯО Д ˈ], raЭСОr ЭСan as front voiced stop [ģ] in traditional transcription. As such East Old Turkic mirrors a sОparaЭО ПrТМaЭТЯО pСonОmО / /, аСТМС Тs noЭ ЭСО mОrО baМk-vocalic allophone of /g/ as commonly assumed. In the followi continues to be marked by the double dots in Ancient A velar fricative in proto-Transeurasian 379 Uighur and by the <ghain> sign in Karakhanid. Hence, the notations OTk. ɕэsgК ‘sСorЭ’, OTk. yuyka ‘ЭСТn’, OTk. ɕölТgО ‘sСaНoа’ anН OTk. öpɕО ‘lЮnР’ ЭransМrТbО what Johanson believes to be rendered in the sources as [ a ~ ɕэsẋa], [yuy a ~ вuв К], [ɕölТ e ~ ɕölТģО ~ ɕölТ ˈe ~ ɕölТбˈО] and [öp e ~ öpģО ~ öp ˈe ~ öpбˈО], respectively. Contemporary languages, especially Oghuz Turkic, allow us to determine the Old Turkic voicing, but not the place of articulation (stop or fricative). As for OTk. ɕэsgК ‘sСorЭ’, ЭСО soЮrМОs mark ТЭ as a ПrТМaЭТЯО anН ЭСО МonЭОmporarв languages indicate that it was voiced, i.e. [ a]: Tk. kɨsa, Az. ɨsa, Tkm. ɨs К, Dolg. kɨhalga / Yak. kɨhalga ‘sЭraТЭs, sorroа’. OTk. yuyka ‘ЭСТn’ Тs markОН Тn ЭСО sources as a stop and the contemporary languages indicate that it was voiceless, i.e. [yuy a]: Tk. yufka, Gag. jufqa, Az. yuxa, Tkm. yu:qa, Tat. yuqa, Kirg. upqa, Kaz. žuqК, Bash. yoqa, Krm. yuwɣa ~ yufqa, Uz. yupka, Uig yupqa, SUig. yuqa, Sal. вoχbК, etc. OTk. ɕölТgО ‘sСaНoа’ Тs markОН Тn ЭСО soЮrМОs as ОТЭСОr sЭop or ПrТМaЭТЯО and the contemporary languages mentioned above indicate that it was voiced, i.e. [ɕölТģО ~ ɕölТ ˈe]. Finally, OTk. öpɕО ‘lЮnР’ Тs markОН Тn ЭСО soЮrМОs as ОТЭСОr sЭop or fricative and the contemporary languages mentioned above indicate that it was voiceless, i.e. [öp e ~ öpбˈО]. TСЮs, ТП / / rОallв СaН a pСonОmТМ sЭaЭЮs Тn Оarlв OlН TЮrkТМ anН bОПorО, ЭСО allomorphs of our resultative nominalizer are -ga (< pre-OTk *- К), -ka (< pre-OTk *-ka), -ge (< pre-OTk *- О or *-ge), -ke (< pre-OTk *-ke or *-xe ). The principle of OkСam’s raгor ЭСОn lОaНs Юs Эo rОМonsЭrЮМЭ a ЯoТМОН ПrТМaЭТЯО sЮППТб prО-OTk *- A and a voiceless stop allomorph *-kA in sonorant and labial obstruent environment. It cannot be excluded that the original velar fricative had no phonological voice НТsЭТnМЭТon anН ЭСaЭ ТЭ аas rОalТгОН as Д ] Тn ЯoМalТМ ОnЯТronmОnЭ, bЮЭ as Дб] Тn consonantal positions. In other words pTk *-xA was pronounced [-xa ~ -xe] following consonants, but [- a з - О] Тn ТnЭОrЯoМalТМ posТЭТon, аСТМС аas Тn ЭСО majority of cases, given that most original verb bases ended in a vowel. Assuming that the resultative nominalizer had a fricative onset can explain the voiceless stop allomorph, while the assumption that it had a voiced stop onset cannot. From a phonological point of view, it is difficult to motivate the devoicing of a voiceless velar stop in a sonorant environment. However, de-fricativization is commonly seen in this environment, for instance in Spanish and Icelandic (Johanson 1979: 30).2 The high sonority of the sonorants r, l, n, m makes the articulation of a following consonant possible without an intermediate vowel. Therefore, vowels are expected to drop more easily following sonorants. Vowel syncope leads to the juxtaposition of two continuants, which is articulatory inconvenient and will be avoided by dropping the incomplete oral closure from the second consonant: the fricative will become a stop. Note that the glide in OTk. yuyka, the labiodental fricative reflexes in Tk. yufka, Gag. jufqa, Krm. yuwɣa ~ yufqa and the voiced bilabial stop in Sal. вoχbК may suggest an earlier bilabial fricative in pre-OTk. 2 SpanТsС /Н/ Тs rОalТгОН as a ПrТМaЭТЯО Дð] Тn mosЭ ОnЯТronmОnЭs (О.Р. nada Дnaða] ‘noЭСТnР’ẓ, except in word-initial position, following nasals and [l] where it is realised as a stop [d]. 380 Martine Robbeets *вußɕК ‘ЭСТn’ anН lТkОаТsО ЭСО МonЭТnЮanЭs Тn Tk. öПɕО ‘anРОr’, öвɕОn ‘lЮnР’, Aг. öбbт ‘lЮnР’, GaР. üПɕО, Tkm. öвɕОn, Sal. öССОn may point to pre-OTk. *ößɕО. The continuant ß would then provide an environment for the de-fricativation of the original suffix pTk *-xA. Johanson (1979: 79) finds that vowel syncope following [r] occurred later in pre-Old Turkic than vowel delition following [l] and [n]. This seems to suggest a higher sonority of [l] and [n], which probably yielded the voiced allophone in pre-OTk *- A similar to vocalic positions, whereas following [r] de- ПrТМaЭТЯaЭТon Эook plaМО sТmТlar Эo oЭСОr МonЭТnЮanЭ posТЭТons. SТnМО JoСanson’s assumption of an original fricative phoneme offers a phonological motivation for the observed allomorphy, I am inclined to reconstruct pTk *-xA ~ *-kA as the resultative nominalizer. Although clausal (ad)nominalization of OTk -gA is not overtly attested, there are some indications of such use reflected in the compounding of the suffix with OTk -sOk, аСТМС Тs a НОnomТnal sЮППТб ЭСaЭ sООms Эo ТnНТМaЭО ‘onО (ОlОmОnЭ/plaМО/ЭТmОẓ oЮЭ oП sОЯОral opЭТons’, О.Р. sэngКr ‘onО oП Эаo/morО sТНОs’ sэngКrsoɕ ‘СТnНqЮarЭОrs oП a СorsО, Т.О. ЭСО plaМО аСОrО a sОМonН rТНОr sТЭs’ (ErНal 1991: 157-158). When added to the resultative deverbal noun suffix OTk -gA it designates one specific time out of several future options and it is often used in a МonЭОбЭ oП ‘ЭСО ЭТmО oП ПЮЭЮrО НОaЭС, ЭСО ЭТmО аСОn onО аТll СaЯО НТОН’, as ТllЮsЭraЭОН in (1b). In these constructions OTk -gA can be interpreted as a perfective future adnominalizer. OTk -gA has further developed into a marker of finiteness, denoting future meaning such as in example (1c) (Erdal 1991: 382, 2004: 242). The development from resultative into a finite future can be explained over a perfective with non-past rОПОrОnМО, as Тn GОrman “εorРОn bТn ТМС sМСon abРОПaСrОn”. ComrТО (1976μ 66ẓ gives examples of this development from ancient Greek and Russian. Interestingly, OTk. -gA alternates with OTk. -gAy in the derivation of seemingly adnominal forms in the OTk -gA-sOk construction (Erdal 1991: 157-158) and of future finite forms (RтsтnОn 1957μ 1β4-125, von Gabain 1959: 115, Erdal 2004: 242). The inability of -gAy to derive deverbal nominal forms seems to suggest that it represents a later formation consisting of OTk. -gA in its adnominal use plus a bound noun I ‘ЭСТs (onОẓ’ аСТМС РraНЮallв rОplaМОН ЭСО orТРТnal sЮППТб Тn -gA, at a time when the lexicalizations of the deverbal nouns in -gA were already accomplished. The hypothesis that -gAy is derived from -gA is proposed by Erdal (1979: 89 and 2004: 243), who identifies the final element -y as the nominative form of the archaic demonstrative pronoun *-I, postposed for subject reference. A velar fricative in proto-Transeurasian 381 2.1.2. Chuvash CСЮЯasС ė bОТnР ЭСО rОРЮlar МorrОsponНОnМО oП OTk. -g- (Benzing 1959: 712; Starostin et al. 2003: 143-144)3, the corresponding Chuvash suffix is -a ~ -e, a suffix which derives nominal and adnominal forms from verb stems, as illustrated in the derivational pairs in (2a) (Levickaja 1974: 152-153; Fedotov 1997: 327-328). Since the suffix mainly derives subject nouns from intransitive verbs (e.g. makra ‘Мrв- babв’ẓ, obУОМЭ noЮns Пrom ЭransТЭТЯО ЯОrbs (О.Р. śэrК ‘lТnО’ẓ anН sЭaЭТЯО propОrЭв noЮns from change of state verbs (e.g. śКЯrК ‘roЮnН’ẓ, ТЭ Мan bО МСaraМЭОrТгОН as a resultative nominalizer. The Chuvash cognates of OTk. öpɕО ‘РОnОraЭОН Тn ЭСО lЮnРν lЮnР, anРОr’ anН OTk. yuyka ‘ЭСТn, slОnНОr, ЮnsЮbsЭanЭТal’ arО CСЮ. öpɕО, üpɕО ‘lЮnР’ and Chu. ‘ЭСТn, narroа’, аСОrОbв OTk. -k- regulary corresponds to Chuvash -k- or -x- and goes back to pTk *-k-. Given that the voiceless velar stop allomorph has lexicalized in Eastern Old Turkic, as well as in Chuvash, which is the only surviving relative of the Western Turkic branch, the derivation and de-fricativization must have taken place already in proto-Turkic, the common ancestor of both branches. The optative mood in Chuvash in e.g., vulam ‘I аoЮlН lТkО Эo rОaН’ (< Ẕvula-a-m read-OPT.NML-POSS.1SG) and vular ‘АО аoЮlН lТkО Эo rОaН’ (< Ẕvula-a-r read- OPT.NML-POSS.PL) has been compared to the finite future OTk -gA (RтsтnОn 1957μ 125). Since the imperative plural Яulăr ‘rОaН!’ ТnМlЮНОs ЭСО sОМonН plЮral possОssТЯО suffix -ăr (e.g. КčК ‘МСТlН’ Кč-ăr ‘вoЮr (PLẓ МСТlН’ẓ, ТЭ probablв НОrТЯОs Пrom an optative nominalization *vula-a-ăr (read-OPT.NML-POSS.2PL) as well. It is not unlikely that the optative, illustrated in (2b) reflects the finite use of the Chuvash resultative nominalizer -A, going over a perfective future reading. (2) Reflexes of the resultative deverbal noun suffix pTk *-xA in Chuvash (2a) lexical nominalization mКɕăr- ‘Эo Мrв’ makra ‘Мrв-babв’ śКЯăr- ‘Эo ЭЮrn, аСТrl, Рo roЮnН’ śКЯrК ‘roЮnН’ śКrКt- ‘Эo ОбposО, barО, sЭrТp (Эr.ẓ’ śКrКtК ‘РlaНО, МlОarТnР, opОn spaМО sЮrroЮnНОН bв аooНs’ śэr- ‘Эo Нraа, аrТЭО (Эr.ẓ’ śэrК ‘lТnО’ vit- ‘Эo МoЯОr (Эr.ẓ‘ vite ‘Мoа-house, cattle-sСОН’ (2b) finite lОš t nčО-ne an-sa kur-ar other world-ACC descend see-OPT.1PL ‘δОЭ’s Рo Нoаn anН sОО ЭСО oЭСОr аorlН’ (KrЮОger 1961: 159) On the basis of the above observations, the deverbal noun suffix pTk *-xA can be reconstructed as a resultative suffix that derived nominal and adnominal forms from 3 The sound correspondence is reflected in, for instance, OTk. buzagu and Chu. păru ‘МalП’ and in OTk. boɣuz and Chu. pэr ‘ЭСroaЭ’. 382 Martine Robbeets verb stems; it derived subject nouns from intransitive verbs, object nouns from transitive verbs and stative property nouns from change of state verbs. It was gradually extended to the clausal level, deriving perfective non-past relative clauses, relics of which are still reflected in Old Turkic. Since Chuvash and Old Turkic share the perfective future connotation in their finite reflexes of this suffix, the insubordination was probably already accomplished in proto-Turkic.4 The suffix pTk *-xA used an allomorph *-kA which arose through de-fricativation following the continuants *r and *ßṬ Both Old Turkic and Chuvash preserve some lexicalizations that are suggestive of this original conditioning factor. Since the large majority of proto-Turkic verb stems was thematic, the initial fricative phoneme of pTk *-xA was most frequently rendered by a voiced allophone *- AṬ The allomorph *- A spread into Chuvash as -A and into Old Turkic as - A, where it became transcribed as -gA due to the merger of OTk / / and /g/. 2.2. pMo *-xA ~ *-kA Derivational pairs in Written Mongolian, such as those in (3a) enable the reconstruction of pMo *-xA as a resultative deverbal noun suffix. Since the suffix derives object nouns from transitive verbs (e.g. idege ‘ПooН’ẓ anН sЭaЭТЯО propОrЭв nouns from punctual change of state verbs (e.g. ОНügО ‘МonЭОmporarв’ẓ, ТЭ Мan bО reconstructed as an original resultative marker. The most widespread reflex of pMo *-xA is WMo. - К ~ -ge, MMo. -’a ~ -’e, аСОrОbв Аεo. Д ] is the allophone of /g/ before back vowels a, o, u in the same way as [q] is the allophone of /k/ in that position. However, there are at least two derivational pairs that reflect a voiceless allomorph WMo. -qa ~ -ke following -r- and -b- (< intervocalic [ß]), i.e. WMo. cubu- ‘moЯО onО aПЭОr anoЭСОr Тn sТnРlО ПТlОν Эo Пall Тn Нrops or sТnРlО РraТns’ cubqa ‘lОaЯОs oП a ЭrОО’ Тn čubqК ɔul urК- ‘lОaЯОs arО ПallТnР oП’ anН Аεo. to urТ- ‘Эo Рo aboЮЭ, МТrМlО, sЮrroЮnН (Эr.ẓ to urqК ‘ЭСО ОnМТrМlОmОnЭ oП ЭСО ЭОnЭ’. From a phonological point of view, it would be difficult to motivate the devoicing of a voiceless velar stop in a voiced environment. Assuming, however, that the resultative nominalizer had a fricative onset, i.e. pMo *-xA, can explain the devoicing: given that in intervocalic position pMo *b became a bilabial voiced fricative *ß (Poppe 1955: 99), vowel syncope would have led to the juxtaposition of two original continuants *rx and *ßбṬ Being articulatory inconvenient, these clusters would have been avoided by dropping the feature [+ continuant] from the second consonant, i.e. pMo *-xA became WMo. -kA. In the majority of cases, however, pMo. *-xA would have been attached to a thematic vowel stem, where it became realized as a voiced fricative - A and, velar fricatives merging with velar stops, it is now transcribed as WMo. -gA, a scenario recalling the one proposed for Turkic above. 4 In this article, I ЮsО ЭСО ЭОrm “ТnsЮborНТnaЭТon” Тn lТnО аТЭС EЯans’ (β008μ γ67ẓ НОПТnТЭТon as “ЭСО МonЯОnЭТonalТгОН maТn МlaЮsО ЮsО oП аСaЭ, on prТma ПaМТО РroЮnНs, appОar Эo bО Пormallв sЮborНТnaЭО МlaЮsОs” (sОО also RobbООЭs β009 anН ПorЭСМomТnРẓ. A velar fricative in proto-Transeurasian 383 The suffix also appears as a perfective marker in complement or relative clauses, often modifying temporal expressions derived from a noun in the dative case, such as inaqsi(-da) (this.side-DATẓ ‘as lonР as noЭ’ Тn (γМẓ anН uНu’uТǰ-e) (still.lack-DAT) or ОНügüТ (sЭТll.laМkẓ ‘bОПorО, noЭ вОЭ’ Тn (55bẓ. 5 Note that the suffix occasionally alternates with a voiceless alternant *-kA, such as MMo. -qa in (3c). This allomorphy probably goes back to the allophony between pMo *-xA and -kA, conditioned by certain continuants. If the perfective converb WMo. - КН ~ -ged, MMo. - Кt ~ -get (Weiers 1966: 214-216; Poppe 1954: 97) can be derived from the resultative nominalizer in the dative case WMo. / MMo. -dA ~ -tA, it also reflects the use of the suffix as a nominalizer embedding complement clauses. In lТЭОraЭЮrО, ЭСО sЮППТб Тs knoаn as ЭСО “nomОn ТmpОrПОМЭТ” (PoppО 1954μ 94, АОТОrs 1966μ 197, SпrköгТ β004μ 44ẓ or as ЭСО “ТmpОrПОМЭТЯО pasЭ parЭТМТplО” (Orlovskaya 1999: 101), implying an action that started in the past but continues into the present. In reality, the suffix has perfective aspectual meaning in its nonfinite use, while it has developed imperfect temporal meaning in its finite use. The use of MMo. -’a ~ -’e in independent sentences is illustrated in (3d). (3) Reflexes of the resultative deverbal noun suffix pMo *-gA in Mongolic (3a) lexical (ad)nominalization in WMo. bos- ‘Эo rТsО, sЭanН Юp (ТnЭr.ẓ’ bosu К ‘НoorsТll, ЭСrОshold (n.), vertical, upright, ОrОМЭ (aНУ.ẓ’ ОНü- ‘Эo bОРТn, sЭarЭ, МommОnМО (Эr.ẓ’ ОНügО ‘noа, aЭ prОsОnЭ, МonЭОmporarв (aНУ. anН aНЯ.ẓ’ ide- ‘Эo ОaЭ, МonsЮmО (Эr.ẓ idege ‘ПooН’ ir- ‘Эo ПТll Юp, Эo СОap Юp, Эo bО ПolНОН (as a СОmẓ (ТnЭr.ẓ’ ТrК К ‘rТpplОs on ЭСО sЮrПaМО oП аaЭОr’ ɔКl К- ‘Эo МonnОМЭ (Эr. / ТnЭr.ẓ’ ɔКl К К ‘МonnОМЭТon (n.ẓν МonnОМЭОН (aНУ.ẓ’ kebte- ‘Эo lТО Нoаn, rОМlТnО (ТnЭr.ẓ’ kebtege ‘lвТnР Нoаn, СorТгonЭal (aНУ. anН aНЯ.ẓ’ mede- ‘Эo knoа, ЮnНОrsЭanН, pОrМОТЯО, ПТnН oЮЭ (Эr. / ТnЭr.ẓ’ medege ‘ТnПormaЭТon, messaРО, ТnЭОllТРОnМО (n.ẓ’ melje- ‘Эo makО a bОЭ (Эr.ẓ’ meljege ‘bОЭ, arРЮmОnЭ (n.ẓ’ qolbu- ‘Эo ЮnТЭО, МonnОМЭ, lТnk (Эr.ẓ’ qolbu К ‘ЭТО, lТnk, НoЮblО, paТr (n. / aНУẓ’ sana- ‘Эo ЭСТnk, rОПlОМЭ (Эr. / ТnЭr.ẓ’ sКnК К ‘ЭСoЮРСЭ, ТНОa, rОПlОМЭТon’ tülО- ‘Эo sОЭ on a ПТrО, bЮrn (Эr.ẓ’ tülОgО ‘ПТrОаooН, kТnНlТnР’ (3b) clausal (ad)nominalization in MMo. basa ber nasun-dur ɕürü-ge ОНügüТ a-mu besides PT long.time-DAT reach-NML still.lacking be-FIN ‘AnН bОsТНОs вoЮ НТНn’Э rОaМС ЭСО ПЮll aРО вОЭ’ (OrloЯskaУa 1999μ 10βẓ 5 The negative particle MMo. uНu’uТ, ОНügüТ, WMo ОНüТ ‘noЭ вОЭ, bОПorО’ Тs Мommonlв derived from ОНügО ‘noа’ anН ügОТ ‘absОnМО, laМk’. 384 Martine Robbeets (3c) clausal (ad)nominalization in MMo. Bi ber Manjusiri-yin sayin-tur bayasqulang oron-i I pt Manjusiri-GEN goodness-DAT peace place-ACC olu-qa inaqsi ger-tecegen qar-qu minu find-NML this.side house-ABL go.out-NML my bütü-ɕü bol-tuqai carry.out-CONV become-IMP.3SG ‘As lonР as I СaЯОn’Э ПoЮnН ЭСО plaМО oП pОaМО ЭСroЮРС ЭСО РooНnОss oП ЭСО εanУЮsТrТ, my leaving the house [as a monk] sСoЮlН bО МarrТОН oЮЭ’ (Аeiers 1966: 200) (3d) finite in MMo. Sigi Qutuqu ese abu-’a Sigi Qutuqu NEG take-FIN ‘SТРТ QЮЭЮqЮ Сas noЭ ЭakОn’ (SH β5βν АОТОrs 1966μ 198ẓ The suffix WMo. - КТ ~ gei, MMo. -’КТ ~ -’ОТ, which alternates with WMo. - К ~ ge, MMo. -’К ~ -’О in clausal (ad)nominalization and finite function (Poppe 1954: 94; Weiers 1966:197- 206) is probably a derivation of the adnominalizer plus a reflex of the original bound noun pTEA *i ‘ПaМЭ, ЭСТnР, ЭСТs (onОẓ’.6 The Khitan suffix -hu ~ -ho ~ -gi, which derives deverbal nouns and adnominal forms from verbs is probably cognate with WMo. - К ~ ge (Kane 2009: 155-156). The examples provided by Kane all involve perfective relative clauses of the kind of example (4a). While the allomorph -gi follows front-vocalic stems, the allomorphs -hu ~ -ho follow back-vocalic stems, whereby -ho follows stems with back rounded vowels. As illustrated in (4b), the suffix is also found as a marker of past tense. (4) Reflexes of the resultative deverbal noun suffix pMo *-xA in Khitan (4a) clausal adnominalization oju-hu DAY close-ADN day ‘ЭСО Нaв ДЭСО МoППТn] аas МlosОН’ (KanО β009μ 155ẓ 6 The bound noun pTEA *i ‘ПaМЭ, ЭСТnР, ЭСТs (onОẓ’ Тs rОПlОМЭОН Тn OlН JapanОsО as a boЮnН noun i ‘ПaМЭ (ЭСaЭẓν ЭСaЭ (аСТМСẓ’ Пor ТnsЭanМО Тn ЭСО ОбprОssТon OJ aruiwa ‘pОrСaps, or’ < pJ *ar-u i pa (be-ADN fact TOP), and it is probably related to the demonstrative pronoun pJ *i ‘ЭСТs’ rОПlОМЭОН Тn OJ ima ‘noа’ (< pJ Ẕi-ma this-interval). Middle Korean has a bound noЮn·i ‘ЭСТnР, onО, pОrson’ (εarЭТn 199βμ 548ẓ, аСТМС Тs probablв rОlaЭОН Эo ЭСО nomТnal and adnominal demonstrative pronoun MK ·Т ‘ЭСТs onО, ЭСТs pОrson, ЭСТs ЭСТnРν ЭСТs’. TСО Tungusic languages share a suffix *-i: that derives ordinal from cardinal numbers, e.g. Evk. эlКn ‘ЭСrОО’ -> эlТ: ‘ЭСО ЭСТrН onО, ЭСТrН’ (< pTР Ẕila-i: three-thing) (Benzing 1955a: 1051), which might be related to the third person pronoun Manchu i. The Turkic demonstrative pronoun *i, which was postposed for subject reference may ultimately go back to this form as well (Erdal 1979: 89 and 2004: 207, 243). A velar fricative in proto-Transeurasian 385 (4b) finite qihu ai bКs […] ɔuung śТúlТng o-ho that year then zhongshuling become-FIN ‘TСaЭ вОar, СО аas ЭСОn appoТnЭОН sОМrОЭarТaЭ НТrОМЭor’ (KanО β009μ 8γ, 1β1ẓ 2.3. pTg *-xA: ~ *-kA: 2.3.1. Manchu The resultative nominalizer in Manchu is -ha ~ -he ~ -ho with a voiceless velar fricative, while about 185 verbs take the allomorph -ka ~ -ke ~ -ko with a voiceless velar stop and some 14 verbs take the allomorph -ngka ~ -ngke ~ -ngko (Gorelova 2002: 240-41, 255-256). Vovin (1997: 271-274) proposes that Manchu intervoca- lic -k- goes back to proto-Manchu *-nk-. In many cases, the -kA- allomorph is indeed attached to verbs, which had an original stem final nasal, e.g. Ma. dosi- ‘Эo ОnЭОr’ dosi-ka, whereby the causative Ma. dosim-bu- still preserves a trace of this nasal. However, it seems that the allomorphy goes back to the proto-Tungusic level and that -kA- appears following verb stems with stem-final *-b- (? < *-ß-) as well, e.g. Ma. je- ‘ОaЭ’ je-ke, whereby the Northern and Southern Tungusic languages (Evk. / Even jep- ~ jeb-, Neg. jep-, Solon jeb-, Na. jeb- ~ jep-, Ud. jepte-, Orok deptu-) reflect a stem-final b in pTg *jeb- ‘Эo ОaЭ’, аСТМС аas alrОaНв losЭ Тn proЭo-Manchuric (Sibe je-, Ma. je-, Jurchen je-fu ‘Эo ОaЭ’ẓ.7 When Manchu -hA meets a stem-final nasal, which developed later in the course of Manchu history, for instance through syncope, the suffix undergoes de-fricativation, while the nasal remains and undergoes velarization, resulting in the allomorph -ngkA. In most instances, the allomorph Ma. -ngkA reflects the contraction of the processive suffix in -nA- ‘Эo НОЯОlop sponЭanОoЮslв’ plЮs ЭСО rОsЮlЭaЭТЯО nomТnalТгОr Тn -hA (Ma. -ngkA < *-nkA < *-nhA < *-nA-hA). The verbs Ma. ba- ‘Эo bО laгв, ЭТrОН, Рnaа a СolО’, jo- ‘Эo brТnР Эo mТnН, rОМall, mОnЭТon’, we- ‘Эo mОlЭ’ all Пorm rОsЮlЭaЭТЯОs Тn -ngkA, i.e. bangka, jongko, wengke, but these verbs have corresponding lexicalized derivations with the processive suffix, i.e. bana- ‘Эo bО(МomОẓ laгв’, jono- ‘Эo rОМall’, wene- ‘Эo mОlЭ’ on which their resultative form seems to be built.8 In sЮМС МasОs, МonЭrarв Эo ↑oЯТn’s proposal, Ma. -k- clearly does not go back to proto-Manchu *-nk-. It therefore appears that following *-n- and *-b- stems, the reflex of pTg *-xA underwent de- 7 Note that proto-Tungusic velar clusters, consisting of a sonorant or labial stop p plus a velar stop k, lost the sonorant or the stop and yielded the reflex -k- in Manchu, e.g., pTg *purke:- ‘Эo bО borОН, anРrв’ Тn EЯk. hurke:-, Even СörɕОn- ‘Эo bО borОН’ вТОlНОН εa. fuke- ‘Эo bО anРrв’ or pTР Ẕjapkun ‘ОТРСЭ’ Тn EЯk. japkun, Na. ɔКɕpõ, Ud. jakpun, etc. yielded Ma. jakun. 8 Note that the processive suffix Ma. -nA- ‘Эo spontaneously develop the action denoted by ЭСО basО ЯОrb’ also sЮrПaМОs Тn ЭСО ТmpОrПОМЭТЯО (aНẓnomТnal Пorms εa. bandara, jondoro and wendere (Gorelova 2002: 255). 386 Martine Robbeets fricativization to pTg *-kA and was reflected in Ma. -kA, whereas following other stems the original voiceless velar fricative was retained and reflected in Ma. -hA. Example (5a) illustrates the use of the resultative nominalizer followed by the dative suffix, a construction which denotes an action after which another action starts. As illustrated in (5b), -hA / -kA is also used as a perfective adnominalizer of relative clauses. In (5c), it appears as a finite past marker. (5) Reflexes of the resultative deverbal noun suffix pTg *-kA: in Manchu (5a) clausal nominalization muse ere аКɕšКm-be geli wa-ha-de we this frog-ACC also kill-NML-DAT muke iningdari lakcaraku: eye-mbi water every.day uninterruptedly flow-FIN ‘AПЭОr аО kТll ЭСОsО ПroРs аaЭОr аТll Пloа ЮnТnЭОrrЮpЭОНlв’ (Gorelova 2002: 257-258) (5b) clausal adnominalization ere abala-me gene-he gucu-sa this hunt-CONV go-ADN companion-PL ji-ci ai seme ala-mbi come-CONV what reason tell-FIN ‘АСaЭ аТll (oЮrẓ МompanТons, аСo СaЯО РonО СЮnЭТnР, ЭОll (Юsẓ аСОn ЭСОв rОЭЮrn?’ (Gorelova 2002: 257) (5c) finite ahu:n ji-he turgun-de, deo gene-he elder.brother come-ADN reason-DAT, younger.brother go-FIN ‘SТnМО ЭСО ОlНОr broЭСОr МamО, ЭСО вoЮnРОr broЭСОr аОnЭ aаaв’ (GorОloЯa β00βμ 488ẓ 2.3.2. Even Although Benzing (1955b: 38) labels Even -ka: ~ -ke: as an instrumental deverbal noun suffix, the derivational pairs in (6a) suggest that the suffix can also derive action nouns (e.g. hi:lka ‘sЮППОrТnР’ẓ, sЮbУОМЭ noЮns (О.Р. etka ‘poаОr, РoЯОrnmОnЭ, regime) and object nouns (e.g. ewi:ke: ‘РamО, Эoв’ẓ as аОll as aНУОМЭТЯОs (О.Р. no:dikaμ ‘bОaЮЭТПЮl, appОalТnР’ẓ. TСО НОrТЯaЭТon of object nouns from transitive verbs suggests that the original meaning of the suffix is resultative. As a finite suffix, Even -kA Тs ЮsОН Эo ОбprОss аarnТnРs Тn ЭСО sОnsО oП ‘АaЭМС oЮЭ or вoЮ аТll … !’ (BОnгТnР 1955aμ 1089ν 1955bμ 99, 101-102); see (6b). These expressions are probably derived from an original perfective meaning, comparable to ЭСО ЮsО oП EnРlТsС ‘вoЮ arО НОaН’ Тn ЭСО mОanТnР oП ‘АaЭМС oЮЭ or вoЮ аТll НТО!”. TСО future suffix -ji- ~ -ci- can optionally precede the ending, without observable change in meaning, e.g. Ma:-k (kill-FIN) and Ma-ji-k (kill-FUT-FINẓ boЭС mОan ‘(АaЭМС oЮЭ orẓ СО аТll kТll вoЮ!’ A velar fricative in proto-Transeurasian 387 (6) Reflexes of the resultative deverbal noun suffix pTg *-kA: in Even (6a) lexical (ad)nominalization el- ‘Эo rТsО, ОlОЯaЭО (ТnЭr.ẓ’ elke: ‘appОaranМО, ОбСТbТЭТon’ Тn elke: o:- ‘Эo appОar, ОбСТbТЭ’ et- ‘Эo МonqЮОr, masЭОr, РОЭ ЭСО ЮppОr СanН (ТnЭr.ẓ’ etka ‘poаОr, Рovernment, regime, laаν lОРal, laаПЮl, laаРТЯТnР’ ewi:- ‘Эo plaв (РamО, ТnsЭrЮmОnЭẓ (Эr.ẓν Эo ОnУoв onОsОlП, to pass ЭСО ЭТmО (ТnЭr.ẓ’ ewi:ke: ‘РamО, Эoв’ hi:l- ‘Эo sЮППОr, bО nООНв, aРonТгО (ТnЭr.ẓ’ hi:lka ‘sЮППОrТnР, aРonв, nООНν mТsОrablО, nООНв, poor’ hi:ral- ‘Эo bОМomО anРrв’ СТ:rКŋɕК ‘anРrв, ПЮrТoЮs’ no:diw- ‘Эo РloaЭ oЯОr ЭСО sТРСЭ oП somОЭhing, observe something with joy, enjoy; to appОal Эo’ no:dikaμ ‘bОaЮЭТПЮl, appОalТnР’ te:w- ‘Эo pЮЭ (ТnЭoẓ, Эo ТnsОrЭν Эo loaН, Эo ПТll (Эr.ẓ’ tepke ‘МoЯОr, baР, МasТnР, СЮsk, МonЭaТnОr’ (6b) finite mu-le: tik-ci-ke-riw water-DAT fall-FUT-FIN-1SG ‘ДАaЭМС oЮЭ or] I аТll Пall ТnЭo ЭСО аaЭОr!’ (BОnгТnР 1955bμ 10βẓ 2.3.4. Udehe In Udehe, the resultative nominalizer pTg *-xA: ~ *-kA: is reflected in the finite perfect paradigm, which refers to perfective events in the past (Nikolaeva 1999: 147, 147ν BОnгТnР 1955aμ 1089μ “PrтЭОrТЭЮm II”ẓ. As opposОН Эo ЭСО pasЭ ЭОnsО, ТЭ ТmplТОs perfective meaning in the sense of relevance for the current situation, adding the sОmanЭТМ ОlОmОnЭ ‘alrОaНв’. TСО sЮППТб -ga-/ -’К- ~ -ge-/ -’О- is added after most vowel stems as in (7b), while -ka- ~ -ke- follows consonant stems, e.g. Ud. diang-ka- i ‘I СaЯО (alrОaНвẓ saТН ТЭ’. HoаОЯОr Тn ЭСО small Мlass oП Тrregular verbs that conjugates like Ud. ga- ‘Эo ЭakО’ anН ПЮrЭСОr ТnМlЮНОs ←Н. o- ‘Эo bОМomО’, bu- ‘Эo НТО’, ne- ‘Эo ТnsОrЭ’ anН nag- ‘Эo mООЭ’, ЭСО pОrПОМЭ appОars as -xa ~ -xe, e.g. Ud. ga-xa-mi ‘I СaЯО (alrОaНвẓ ЭakОn’. TСТs sООms Эo ТnНТМaЭО ЭСaЭ an orТginal velar fricative suffix pTg *-xA: was preserved as such in a limited set of high-frequency ḳ and therefore irregular ḳ verbs, but devoiced to * - A- between vowels (*..u-xa > *..u К > ṬṬugК, *..i-xa > *..Т К > ṬṬТgК, *..a-xa > *a- К > ..’К, *..e-xe > *..e- О > ..’О), while it underwent de-fricativization following sonorants and fricatives (*..n-xa > *..n-ka > ..ngka, *.. -xa > *.. -ka > ..gka, *..ß-xa > *..ß-ka > ..kpa, *..m-xa > *..m-ka > ..ngka). Note that Menges (1968: 138) and Simonov (1988: 55) postulate a voiced fricative / / in opposition to the voiced stop /g/ for some Udehe dialects, but the phonological contrast seems to have merged in the southern varieties of Udehe, described by Nikolaeva (1999: 29). As illustrated in (7a), the Udehe converb suffix for same-subject forms -si is added to the perfect verb stem, e.g. diang-ka-si ‘СaЯТnР saТН’, eme-ge-si ‘СaЯТnР 388 Martine Robbeets МomО’ (NТkolaОЯa 1999μ 147, 147ẓ. DТaМСronТМallв, ЭСТs sООmТnРlв ЮnЮsЮal НОrТЯaЭТon can be explained by tracing -si back to pTg *-pA-i, the accusative suffix in its reflexive form. In this way the converbial clause goes back to an original complement clause. (7) Reflexes of the resultative deverbal noun suffix pTg *-xA: in Udehe (7a) clausal nominalization Uta xegise-li kongko dieli:-ni, xegise-li emegi-ge-si. that above-PROL croak.croak fly.FIN-3SG above-PROL come-NML-CONV ‘TСaЭ ДraЯОn] Мroak, Мroak, ПlТОs Нoаn Пrom aboЯО, СaЯТnР МomО Пrom aboЯО. (Nikolaeva 1999: 507) (7b) finite Bi emegi-ge-i I come-FIN-1SG ‘I СaЯО (alrОaНвẓ МomО baМk’ (NТkolaОЯa 1999μ 148ẓ In sum, the Tungusic languages provide evidence for the reconstruction of two allomorphs for the resultative deverbal noun suffix: pTg *-xA: and pTg *-kA:. The allomorphs show signs of being variants of the same suffix in two different phonological positions in Manchu and Udehe. In the Northern Tungusic languages, where ḳ as illustrated in the sound correspondences below (Benzing 1955a: 976- 977; Starostin et al. 2003: 158, 160) ḳ the reflex of pTg *-k- and *-x- has merged into Evk. -k- and Even -k-, it is not possible to distinguish between both allomorphs. BОnгТnР’s (1955aμ 989-991) reconstruction of pTg *x- was limited to word-initial position, but he already hinted at the possibility of reconstructing pTg *-x- in a few words. The reconstruction of intervocalic pTg *-x- has been elaborated by Dybo (1990) and it was followed by Starostin et al. (2003: 158). pTg Ma. Na. Olč. Orok Oroč. Ud. Sol. Neg. Ev. Even *-k- k~h k k k k k~x kзбзРз‘ k~g~x k~x k k *-x- k~h бзė бзė бзė kзбзė kзбзРз‘ зРзбзė k~x k k The Manchu and Udehe reflexes of pTg *-xA: ~ *-kA: show variation in a similar environment: in Manchu most verb stems are thematic and take -hA, whereas verbs originally ending in *n or *b (? < *ß) take -kA; in Udehe most thematic verbs stems take -gA, whereas verb stems ending in an original continuant *n, *m, * or *ß take -kA, but there is a limited set of high-frequency verbs, which has preserved -xA. This suggests that the de-fricativization of *-xA in continuant environment was already accomplished in proto-Tungusic. Tracing the allomorphy back to the proto- Tungusic level is further supported by the observation that -kA- can even follow verb stems with a stem-final vowel in Manchu and elsewhere in Manchuric, that must be reconstructed with a stem-final continuant in proto-Tungusic. A velar fricative in proto-Transeurasian 389 Lexicalizations in Even and Evenki suggest that pTg *-xA: ~ *-kA: originally functioned as a resultative deverbal noun suffix, deriving subjects of intransitive verbs and objects of transitive verbs. Juxtaposed to other nouns, some deverbal nouns developed into nominal adjectives. The use of the Manchu suffix in complement and relative clauses indicates that the resultative nominalizer was extended to the clause level. This observation is confirmed by the analysis of converbial forms in Udehe reflecting pTg *-xA-pi ~ *-kA-pi as compounds of the resultative nominalizer and the accusative suffix in the reflexive (pTg *-pi < *-pA-i ACC-REFL; Benzing 1955b: 1091). The perfective adnominalizer probably underwent insubordination, resulting in the finite use preserved in Manchu, Even and Udehe. The insubordination process triggered the development of temporal from aspectual distinctions, i.e. a grammaticalization from resultative nominal to perfective adnominal to perfect finite marking. 2.4. pK *-kAi Among the innovations that characterize the proto-Koreo-Japonic branch of the Transeurasian languages are the reduction of voicing distinction and probably also velar de-fricativization, resulting in the merger of pTEA *-k-, *-g- and *-x- in pKJ *-k-. Although the Japanese and Korean data are telling from a functional perspective and their attestation is relevant as an argument against borrowing, they can not add much to the phonological question about the reconstruction of a velar fricative. Korean reflects a deverbal resultative and instrumental noun suffix K -kay / -key, -ay/ -ey, MK -·ɕКв ṭ -·ɕОв, -·GКвṭ -·GОв (Martin 1992: 429, 600). As illustrated in (8), the suffix derives object nouns from transitive verbs (e.g. kkalkay ‘МЮsСТon’ẓ as аОll as subject nouns from intransitive verbs (e.g. kalikay ‘a ЭаoПolН sМrООn’ẓ anН ЯОrbal adjectives (e.g. killay ‘Пor lonР (aНЯ.ẓ’ẓ. NoЭО ЭСaЭ aНЯОrbТal ЮsО НОrТЯОs Пrom instrumental nouns. (8) The use of the deverbal noun suffix K -kay / -key, -ay/ -ey K cci- ‘Эo sЭОam (Эr.ẓ’ ccikey ‘ЭСТn sЭОа’ K elk- ‘Эo аОaЯО, makО (Эr.ẓ’ elkay ‘sЭrЮМЭЮrО’ K kkal- ‘Эo sprОaН oЮЭ (О.Р. a bОН, a maЭẓ, laв oЮЭ, sТЭ on (Эr.ẓ’ kkalkay ‘МЮsСТon’ K kali- ‘Эo СТНО, sСТОlН (Эr.ẓ’ kalikay ‘a ЭаoПolН sМrООn’ K kalu- ‘Эo splТЭ, МЮЭ, НТЯТНО (Эr.ẓ’ kallay ‘НТЯТsТon’ K ki:l- ‘Эo bО lonР (ТnЭr.ẓ’ killay ‘Пor lonР (aНЯ.ẓ’ K kkwumi- ‘Эo НОМoraЭО, ornamОnЭ (Эr.ẓ’ kkwumikay ‘ornamОnЭ’ K nal- ‘Эo Пlв (ТnЭr.ẓ’ nalkay ‘a аТnР’ K nol- ‘Эo plaв, ОnУoв onОsОlП (ТnЭr.ẓ’, nolli- ‘Эo lОЭ/ makО somОonО plaв’ nolikay ‘a plaвЭСТnР, a Эoвν a pОnНОnЭ ЭrТnkОЭ аorn bв laНТОs aЭ ЭСОТr аaТsЭ’ K teph- ‘Эo МoЯОr аТЭС, pЮЭ on (О.Р. bОНМloЭСОsẓ (Эr.ẓ’ tephkay ‘bОННТnР, bОН clothes, qЮТlЭ, (bОНẓ МoЯОrs’ 390 Martine Robbeets Given that the contemporary monophthongs ay and ey go back to Middle Korean diphtongs, in which the y offglide commonly derives from contraction of a syllable with a high front vowel (e.g K ka:y < MK ɕК·СТ ‘НoР’ẓ, ТЭ Тs noЭ ЮnlТkОlв ЭСaЭ MK -·ɕAв ultimately derives from pK *-kA-i, in which *-kA would be a resultative adnominalizer and *-i ЭСО boЮnН noЮn ‘ПaМЭ (ЭСaЭẓν ЭСaЭ (аСТМСẓ’ rОПlОМЭОН Тn εK i. 2.5. pJ *-ka 2.5.1. Ryukyuan The common pattern in Ryukyuan verbal adjective inflection is to contract forms of a structure that consists of the nominalizing *-sa plus the auxiliary a(r)- ‘Эo bО’. However, in certain Southern Ryukyuan languages, such as in the Miyako and Yonaguni dialects and in some Northern Ryukyuan languages, such as in Shodon, adjective inflection is built upon a nominalizing *-ka. In the Miyako dialects and in Yonaguni, this construction involves an auxiliary a(r)- ‘Эo bО’, bЮЭ Тn SСoНon, ТЭ appears without copula. As illustrated in example (9a), Shodon has three competing possibilities for adnominal modification: the bare adjective base in its nominal encoding without verbal ending; the adnominal form in -kha or; most commonly, the contracted construction consisting of nominalizing -sa plus the adnominal form of the copula a-n. In finite position, -kha is used more frequently than the periphrastic construction consisting of -sa plus the present form of the copula a-m. However, further inflections for tense, aspect, mood etc. build on the periphrastic construction, e.g. QКСпп-sa-ta-m (red-NML-COP.PERF-FINẓ ‘ТЭ аas rОН’. (9) The use of the Shodon adnominalizer -kha as compared to other adnominalizing constructions (Martin 1970: 134) (9a) adnominalization QКСпп mun red thing QКСпп-kha mun red-ADN thing QКСпп-sa-n mun red-NML-COP.ADN thing ‘a rОН ЭСТnР’ (9b) finite QКСпп-kha red-FIN QКСпп-sa-m red-NML-COP.FIN ‘ТЭ Тs rОН’ A velar fricative in proto-Transeurasian 391 It is commonly suggested that the Shodon ending -kha is a reduction from the converb -ku and the copula a- ‘Эo bО’ (εarЭТn 1970μ 1γγν ↑oЯТn β009μ 460-461), but this derivation has two problems: first, it cannot account for the aspiration in the Shodon suffix and, second, it fails to explain why -kha, in contrast to -sa, is not followed by inflected forms of the contracted copula. In Shodon, voiceless stops were aspirated before mid and low vowels, i.e. /k/ became /kh/ before /e, o, a/ but unaspirated before high vowels, i.e. /k/ remained unchanged before /i, u/. Consequently, pR *-ka is expected to yield Shodon -kha, while pR *-ku will yield Shodon *-ku. Besides, the forms in -kha do not bear traces of copula inflection such as -m, -n, -tam, etc. in the -sa constructions. Therefore, I derive Shodon -kha from the adnominalizer pR *-ka < pJ *-ka without copula intervention. Even if the Miyako adjective inflections in -k-ar- and the zero affixed Yonaguni cognates are combined with the auxiliary ar- ‘Эo ОбТsЭ’ (BОnЭlОв β008μ βγ8-239, 243- 301), there is no evidence that they should be derived from *-k[u] ar- rather than from *-k[a] ar-, the latter being phonologically somewhat more plausible because of the juxtaposed low vowels. Although in Old Ryukyuan the construction -sa ar- was predominantly used, Vovin (2009: 460-461) provides some relic examples such as yo-k-ar-u (good-NML-EXIST-ADN) and kiya-k-ar-u (bright-NML-EXIST-ADN) testifying to the gradual replacement of an older competing periphrastic construction *-ka ar-. 2.5.2. Mainland Japanese Old Japanese preserves evidence for the lexicalization of a resultative adnominalizer pJ *-ka, which derives adnominal forms from verbal adjectives and verb stems, as illustrated in (10a). Apart from OJ nipaka ‘sЮННОn’, OJ ogosoka ‘solОmn, maУОsЭТМ’, OJ paruka ‘Пar, НТsЭanЭ, rОmoЭО’ anН OJ sayaka ‘МlОar, brТРСЭ’ below, deverbal derivation is also frequently seen in verb stems derived with the fientive verbalizer pJ *-ya-: OJ awoyaka ‘blЮО’ (< OJ awo ‘blЮО’ẓ, OJ ko1mayaka ‘НОnsОlв (РroаТnРẓ’ (< pJ *koma- ‘Эo bО small, ПТnО’ẓν OJ matoyaka ‘roЮnН’ (< OJ mato ‘roЮnН’ẓν OJ mame2yaka ‘sОrТoЮs’ (< OJ mame2 ‘sТnМОrО’ẓν OJ nagoyaka ‘soПЭ’(< pJ Ẕnanko- ‘Эo Мalm Нoаn’ Тn OJ nag- ‘Эo moа (Эr.ẓ, bОМomО Мalm (ТnЭr.ẓ’, OJ nagi2- ‘Эo bОМomО Мalm’, OJ nago1m- ‘Эo Мalm Нoаn’ẓν OJ nikoyaka ‘РОnЭlО’ (< OJ niko1(-ẓ ‘soПЭ’ẓν OJ panayaka ‘РorРОoЮs, РraМОПЮl’ (< OJ pana ‘ПloаОr’ẓν OJ sumiyaka ‘sаТПЭ, МlОar’ (< OJ sum- ‘Эo bОМomО МlОar’ + OJ -i NML); etc. Since pJ *-ya- derives fientive mОanТnР Тn ЭСО sОnsО oП ‘Эo bОМomО lТkО ЭСО propОrЭв oП ЭСО basО’, ТЭ rОqЮТrОs a resultative suffix pJ *-ka Эo НОrТЯО a sЭaЭТЯО aНУОМЭТЯal noЮn ‘bОТnР ЭСО propОrЭв’. TСО resultative meaning of the adnominal suffix is also evidenced by the observation that it derives object nominals from transitive verbs, e.g. OJ paruka ‘Пar, НТsЭanЭ, rОmoЭО’ ОбprОssОs ЭСО mОanТnР ‘аСaЭ Тs rОmoЯОН’ raЭСОr ЭСan ‘rОmoЯТnР’. Since the suffix is lexicalized in only a few adjectival nouns and thus no longer productive in Old Japanese, there is no evidence left of its original syntactic behavior. Nevertheless, the etymological analysis of OJ me2duraka ‘sЭranРО, rarО, prОМТoЮs’ Тn (10bẓ sЮРРОsЭs, ЭСaЭ ЭСО aНnomТnal ЯОrb Пorms orТРТnallв МoЮlН ЭakО 392 Martine Robbeets arguments such as OJ me2 ‘ОвО’ Тn ЭСО НaЭТЯО, mЮМС lТkО a rОlaЭТЯО МlaЮsО. FТnТЭО forms ending in -ka such as yo-ka ‘ТЭ Тs РooН’, na-ka ‘ТЭ Тs noЭ’ arО ПoЮnН Тn parЭs oП Kyushu (Martin 1987: 803); see (10c). (10) Reflexes of the use of pJ *-ka in Mainland Japanese (10a) lexical adnominalization pJ *ata- ~ atu- ‘Эo bО аarm’ Тn plaМО namО AЭamТ (< Ẕata-umi ‘аarm sОa’ẓ, OJ atu- B ‘Эo bО аarm’ OJ atataka (4.11ẓ ‘аarm’ pJ *isasa- ‘Эo bО lТЭЭlО, ПОа’ Тn KТkaТ qisaa- B ‘Эo bО lТЭЭlО, ПОа’ OJ isasaka (4.бẓ ‘a lТЭЭlО, somОаСaЭ, slТРСЭ’ pJ *koma- ‘Эo bО small, ПТnО’ Тn SСЮrТ guma- ‘small’, Yo. kuma- B ‘ПТnО’, OJ ko1mayaka ‘НОnsОlв (РroаТnРẓ’ εJ komaka (γ.7bẓ ‘ПТnО, small, НОЭaТlОН’ pJ *nipa- ‘Эo appОar sЮННОnlв’ Тn OJ nipasi- ‘sЮННОn’, OJ nipi1 ‘nОа’ (OJ -i NML, see CСapЭОr 7ẓ OJ nipaka (γ.7bẓ ‘sЮННОn’ pJ *әnkә-sә- (be.majestic-CAUS-ẓ ‘makО maУОsЭТМ (Эr.ẓ’ Тn OJ ogo2r- A ‘Эo bО extravagant, РОЭ proЮН, bО arroРanЭ’ OJ ogosoka (4.11ẓ ‘solОmn, maУОsЭТМ’ pJ *paru- ‘Эo МlОar, rОmoЯО (Эr.ẓ’ Тn OJ par- B ‘Эo opОn РroЮnН, МlОar lanН (МЮlЭТЯaЭТonẓ’ OJ paruka (γ.7bẓ ‘Пar, НТsЭanЭ, rОmoЭО’ pJ *saya- ‘Эo bОМomО МlОar, pЮrО’ Тn OJ saye- ?B ‘Эo РОЭ МlОar, brТРСЭν РОЭ МolН’, OJ sayame2- ?B ‘Эo МlОan, pЮrТПв (Эr.ẓ’ OJ sayaka (γ.7bẓ ‘МlОar, brТРСЭ’ OJ tura- A ‘Эo bО ЭoЮРС’ OJ me2duraka (4.11ẓ ‘sЭranРО, rarО, prОМТoЮs’ (10b) clausal adnominalization OJ me2duraka pi1to2 pre-OJ *me2-n(i) tura-ka pi1to2 eye-DAT be.tough-ADN person ‘sЭranРО pОrson’ (10c) finite Kyushu dialects yo-ka good-FIN ‘IЭ’s РooНν OK.’ In sum, the resultative adnominalizer pJ *-ka, which derives resultative adnominal forms from verbal adjectives and verb stems can be reconstructed on the basis of Ryukyuan and Old Japanese data. In Ryukyuan, the derivations are restricted to verbal adjective stems. Whereas Shodon testifies to the original use of this suffix without the intervention of a copula, the periphrastic constructions found in Old Ryukyuan, the Miyako dialects and in Yonaguni are probably secondary. In Mainland Japanese, the derivations with *-ka seem to involve original verb stems as well as verbal adjectives. The finite use of the suffix, observed in some Kyushu dialects and in Shodon, may have developed over clausal adnominalization, as suggested by one Old Japanese adjective etymology. A velar fricative in proto-Transeurasian 393 2.6. Historical development The comparative evidence, summarized in Table 3, indicates that the resultative nominalizer pTEA *-xA ~ *-kA originated as a resultative deverbal noun suffix, marking a derivational process at the lexical level, was then extended to function as a perfective (ad)nominalizer in dependent clauses at the syntactic level, and was eventually extended still further to mark finite perfective past/future forms in independent clauses.9 pTEA pJ pK pTg pMo pTk *-xa ~ *-ka *-ka *-ka(-)i *-xa: ~ *-ka: *-xa ~ *-ka *-xa ~ *-ka lexical RES.NML lexical RES.ADN lexical lexical RES.NML lexical RES.NML lexical RES.NML - RES.NML clausal PFV.NML clausal PFV.NML clausal clausal RES.ADN clausal PFV.ADN clausal PFV.ADN PFV.FUT.ADN RES.FIN PST.FIN PST.FIN FUT.FIN Table 3: Comparative evidence for the reconstruction of the resultative nominalizer pTEA *-xA ~ *-kA For a detailed functional analysis of this change within the framework of grammati- calization theory, I refer to Robbeets (forthcoming). In this article, however, I have focussed on the formal aspects of the change in so far as they reflect an original velar fricative pTEA *-x-. The Tungusic, Mongolic and Turkic languages display traces of a peculiar allomorphy whereby velar fricative (pTg *-xA:) or voiced velar (pMo *-gA, pTk *-gA) suffixes sporadically alternate with voiceless allomorphs *-kA(:) in certain continuant environments. From a phonological viewpoint, the most plausible explanation for this alternation is to reconstruct pTg/ pMo/ pTk *-xA(:) as the original morpheme. The evidence suggests that pTg/ pMo/ pTk *-kA(:) arose through de-fricativation following the continuants *r and *ß and in Tungusic also following *n, *m, * Ṭ This behavior is probably the result of a regular allomorphy in the common ancestor, whereby pTEA *-xA underwent de-fricativization following the continuants *n, *m, *r, *x and *ßṬ In non-continuant environment, the main allomorph pTg *-xA: was retained as a fricative in Ma. -hA and in some Udehe high-frequency verbs in -xA or, alternatively in Udehe, it voiced to * - A- and ḳ through merger of / / an /g/ ḳ yielded Ud. -gA- following vowels. The reflex of pTg *-x- being -k- in the Northern Tungusic languages, however, pTg *-xA: following non-continuants there merged with its allomorph *-kA: following continuants, yielding only -kA(:) in Even and Evenki. 9 Previous research including some elements of the etymology proposed here ḳbe it under a different configuration and not including the Japanese memberḳ include Ramstedt (1952: 89-92), Poppe (1955: 273; 1972: 45-46), Menges (1984: 261), Baskakov (1981, 72-73), Choi (2002: 29), Vovin (2001: 189-190; 2008: 84-85) and Starostin et al. (2003: 227). 394 Martine Robbeets Environments other than stem-final *r and *ß in Mongolic and Turkic were in the majority of cases represented by vowel-final verbs. There, pMo. *-xA and pTk *-xA became realized as a voiced fricative - A and, velar fricatives merging with velar stops, the suffix became transcribed as WMo. -gA and OTk. -gA. By analogy, these forms spread to verbs ending in a consonant other than *r and *ß. Assuming that just like the feature of voice distinction, the fricativation of pTEA *-xA became neutralized in proto-Koreo-Japanic, we can thus reconstruct the allomorphy between pTEA *-xA and *-kA following the continuants *n, *m, *r, *x and *ßṬ The shared allomorphy between the resultative nominalizers pTg *-xA: ~ *-kA:, pMo *-xA ~ *-kA and pTk *-xA ~ *-kA is a strong argument in favor of genealogical relatedness because (i) the allomorphs appear in a comparable, but peculiar phonological environment, (ii) the allomorphy is a sporadic, seemingly irregular phenomenon in the individual Turkic, Mongolic and Tungusic branches, but it can be traced back to regular phonological behavior at the Transeurasian level and (iii) sharing unreduced allomorphy is an indication against borrowing (Robbeets 2012: 433). 3. Indications in lexicon: etymologies reflecting pTEA *-x- In this article, I have followed Dybo (1990) and Starostin et al. (2003: 158) in their proposal that the medial velar fricative *-x- can be reconstructed as a separate phoneme in Tungusic, a possibility already hinted at by Benzing (1955a: 989-991). Since the number of Tungusic words reflecting pTg *-x- is relatively low to begin with, very few of them have an acceptable Japanese cognate. Establishing sound correspondences in Robbeets (2005), I have only taken etymologies with a Japanese member into account. This approach has the advantage of reducing the probability of borrowing, but the drawback of overlooking sound correspondences that are not sufficiently reflected in Japanese. The reconstruction of the velar fricative pTEA *-x- probably represents such a case. As shown in Table 4, some words reflecting pTg *-x- correspond to proto- Mongolic and proto-Turkic lexemes reflecting an alternation between voiced *-g- and voiceless *-k- in a particular cognate set of words. This alternation in lexemes cannot be ascribed to a particular phonological environment, but it recalls the allomorphy between the Mongolic and Turkic reflexes of the resultative nominalizer *-xA, in which a voiced velar suffix *-gA alternates with a voiceless allomorphs *-kA in certain continuant environments. The free alternation between *-g- and *-k- reflexes in word stems is similar to the allomorphy of the perfective adnominalizer in complement and relative clause in Middle Mongolian, freely switching between the suffixes -GA and -KA regardless of conditioning by a continuant. The etymologies in Table 4 are obtained by screening verb stems reflecting pTg *-x- and trying to match them with cognates in other Transeurasian languages. A more elaborate search for cognates, including nominal stems as well, might yield more evidence. The etymologies underlying the comparisons in Table 4 are given below. A velar fricative in proto-Transeurasian 395 TEA Japonic Koreanic Tungusic Mongolic Turkic *-x- *-k- *-k- *-x- *-g- ~ *-k- *-g- ~ *-k- Na. daxa- WMo. НК К- OTk. yak- MMo. daqa- OTk. yagu- pTEA *daxa- ‘Пolloа’ ‘Пolloа’ ‘bОМomО nОar’ 1. pTg *daxa- pMo *daga- ~ pTk *yagu- ~ ‘МomО nОar’ *daka- *yak- Evk. tokor- WMo. to urТ- Balk. tögОrОɕ ‘sЮrroЮnН’ ‘roЮnН’ pTEA *t xʊ- WMo. tokir Balk. toxun 2. ‘ЭЮrn aroЮnН’ ‘ЭЮrn rounН’ ‘bОnЭ’ ‘аСООl’ pTg *toxo- pMo *togu- pTk *togu- ~ ~ *toki- *toku- Na. moxo- WMo. moqu- ‘bОМomО ‘bОМomО pTEA *m xʊ- ОбСaЮsЭОН’ ОбСaЮsЭОН’ WMo. mo utur 3. ‘Эo bОМomО rЮТnОН’ ‘blЮnЭ’ pMo *mogu- ~ pTg *moxo- *moku- Table 4: Verbal etymologies reflecting pTEA *-x- 1. pTEA *daxa- ‘Эo МomО nОar’ Neg. daxaw-, Sibe НКСǝ-, Ma. daha-, Jur. tai-xa, Olč. -, Orok -, Na. daxa- Oroč daxu- ‘Эo Пolloа, obОв’, ←Н. dahala- ‘Эo aРrОО’, pTР Ẕdaxa- ‘Эo Пolloа’ WMo. НК К-, MMo. daqa-, Khal. daga-, Bur. daxa-, Kalm. НКбǝ-, Ordos daGa-, Dong. daGa-, Bao. daGa-, Dag. daga-, SYug. ta a-, Mgr. daGa:- ‘Эo Пolloа’, pMo *daka- ~ *daga- ‘Эo Пolloа’ OTk. yak- ~ yagu- ‘1 Эo approaМС, bО(МomОẓ nОar Эo’, вКɕэn ~ yagru з вaРЮk ‘ β nОar’, вКɕэš- ~ вКguš- ‘γ Эo Нraа nОar Эo ОaМС oЭСОr’, KarakС. вКбšэ ‘4 РooН’, εTk. yavu- 1, вКɕэn ~ yavuk 2, вКбšэ 4, Tk. вКɕэn ~ yavuk 2, вКɕэš- 3, вКɕšэ 4, Az. вКЯэ- 1, yaxɨn ~ yowuG 2, вКбšɨ 4, Tkm. yak- ~ yovu- 1, yowuq 2, вКGšэ 4, Tat. вКɕэn ~ yawuk 2, вКбšэ 4, Kirg. ū- 1, Кɕэn ~ u:q 2, Кɕšэ 4, Kaz. žuа- 1, žКɕэn ~ žuаэɕ 2, žКɕsэ 4, Nog. вuаэ- 1, вКɕэn 2, вКбšэ 4, KKalp. žuа- 1, žКɕэn ~ žuаэɕ 2, žКɕsɨ 4, Uz. yakin ~ j vuq 2, ɔКбšТ 4, Uigh. yak- 1, yekin 2, вКɕšТ 4, Chu. 2, pTk *yak- ~ yagu- ‘Эo bОМomО nОar’ 2. pTEA *t xʊ- ‘Эo ЭЮrn aroЮnН’ Ma. tohoro ‘1 аСООl, Сoop’, EЯk. tokor- ‘β Эo Рo roЮnН, ЭЮrn roЮnН’, toɕčТɕК ‘γ МЮrЯОН, bОnЭ’, NОР. toxoy- 2, toɕčoɕК 3, Orok - ‘Эo РrТnН’, Na. toбorТqõ ‘pЮllОв, a аСООl ЮsОН Эo ЭransmТЭ poаОr’, pTg toxo- ‘Эo ЭЮrn roЮnН’ WMo. to urТ- ‘1 Эo Рo aboЮЭ, МТrМlО, sЮrroЮnН, ЭoЮr, roll (Эr.ẓ’, to urТg ‘orbТЭ, МТrМЮmПОrОnМО’, to urТgu ‘roЮnНaboЮЭ (aНЯ. /aН.ẓ’, to unК/ to unu ‘smokО-hole in the top oП a вЮrЭ’, Аεo. tokir, takir ‘bОnЭ, МrТpplОН’, tКɕТ Кr ‘МrookОН, МrТpplОН’, takirla- ‘Эo bО bОnЭ, МrТpplОН’, εεo. to’orТ- ‘Эo ЭЮrn aroЮnН (Тn МombaЭẓ’, to’orТqК- ‘Эo ОnМТrМlО, orbТЭ’, togorigai, togarik ‘β roЮnН’, Khal. tögrög, dugarig 2, toxir ‘γ МЮrЯОН, bОnЭ’, BЮr. 396 Martine Robbeets tüбОrТg, toxir 3, Kalm. tögǝrǝg, Нu ǝr ǝ 2, ‘МrТpplОН, mЮЭТlaЭОН, НТsloМaЭОН’, OrНos tögöröɕ 2, Dag. tukurin 2, pMo *togu- ~ *toki- ‘Эo ЭЮrn roЮnН’ OTk. tegirmi ‘1 roЮnН’, KarakС. tegirme 1, Az. НтɔТrmТ 1, Tk. degirmi ‘МТrМlО’, Tkm. tegelek, toGalaq, Uz. 1, ɣin ‘β аСООl, Сoop’, ←ТРС. Нüglтɕ 1, Tat. tügɛrɛk 1, tu эm 2, Krm. togerek, Kaz. toɣэn 2, Nog. tögОrОɕ 1, to эn 2, Bash. tüŋтrтɕ 1, tu эn 2, Balk. tögОrОɕ 1, toxun 2, KKalp. НöŋgОlОɕ 1, to эn 2, Khak. to эlКб 1, Shor to КlКq 1, Oirat to oloq 1, SUigh. НoGэr 1, Yak. tier- ‘γ Эo ЭЮrn roЮnН’, DolР. tier- 3 Chu. togăn 2, pTk *togu ~ toku- ‘Эo ЭЮrn aroЮnН’ 3. pTEA *m xʊ- ‘Эo bОМomО ОбСaЮsЭОН’ Ma. moho- ‘1 Эo losО poаОrs, bО(МomОẓ ОбСaЮsЭОН’, Olč. moxo- 1, Na. moxo- 1, Even mukay ‘nООН, poЯОrЭв, НТsЭrОss, СarНsСТp’, mukay- ‘Эo lТЯО Тn poЯОrЭв’, EЯk. muɕčОrО:- ‘Эo НТО’, pTР Ẕmoxo- ‘Эo bОМomО ОбСaЮsЭОН’ WMo. moqu- ‘Эo РОЭ ЭТrОН, аОakОn, ОбСaЮsЭ onО’s sЭrОnРЭС, bО blЮnЭ (ТnЭr.ẓ’, Аεo. mo ulМКr, mo utur ‘blЮnЭ, аТЭСoЮЭ poТnЭ, СornlОss’, Аεo. möɕü- ‘1 Эo bОМomО ОбЭТnct, НТО oЮЭ, МollapsО, pОrТsС’, KСal. möбö-1, Bur. müбО- 1, Bao. mɛgǝ- 1, Dag. muku- 1, Kalm. möɕr- ‘bО ЮnablО’, OrНos möбö- ‘Эo РОЭ ТnЭo mТsОrв’, pεo Ẕmogu- ~ *moku- ‘Эo bОМomО ОбСaЮsЭОН’ Note that Korean has K muk-, MK muk- ‘Эo РОЭ olН, oЮЭНaЭОН, oЮЭаorn, sЭalО’, pK Ẕmuk- ‘Эo bОМomО olН, sЭalО’, bЮЭ ЭСО ЯoаОl НoОs noЭ МorrОsponН rОРЮlarlв. 4. Conclusion The morphological and lexical comparisons made in this article indicate that the correspondences for velar obstruents presented in Table 2 should be reviewed to allow for the reconstruction of a velar fricative in proto-Transeurasian. The resulting new set of correspondences for velar obstruents is presented in Table 5. pTEA pJ pK pTg pMo pTk *k- *k- *k- *k- *k- *k- *-k- *-k- *-k- *-k- *-k- *-k- *g- *k- *k- *g- *g- *k- *-g- *-k- *-k- *-g- *-g- *-g- *x- *k- *k- *x- *k- *k- *-x- *-k- *-k- *-x- *-k- ~ *-g- *-k- ~ *-g- Table 5: Review of the correspondences for velar obstruents in Table 2 The main objections against the reconstruction of an original velar fricative were the lack of evidence of a symmetrical medial velar fricative series and the observation that a distinct reflex was restricted to the Tungusic branch only. The comparison of the resultative nominalizer, however, reveals traces of an ancient allomorphy in three branches, i.e. Turkic, Mongolic and Tungusic, the phonology of which can only be explained through the assumption of an original velar fricative suffix. A screening of lexical data seems to support this assumption, pointing to distinct reflexes of pTEA A velar fricative in proto-Transeurasian 397 *-x- as *-x- in Tungusic and as the alternation between *-g- and *-k- within a single etymon in the Turkic and Mongolic languages. As such, the answer to the phonological question about the existence of a velar fricative in Transeurasian, originally posed by Ramstedt and Poppe, is primarily motivated by morphological comparison. Due to restrictions in space, this article has been limited to the comparative examination of a single bound morpheme and the screening of a few lexical etymologies for verb stems. Future comparative research on this topic might be able to reconstruct other bound morphemes with velar fricative initials reflecting a similar allomorphy. It will probably also be possible to elaborate the lexical evidence for pTEA *-x-, adding etymologies for nominal roots as well. An accurate reconstruction of the Transeurasian sound inventory would furthermore benefit from researching the symmetry with the initial velar suffix pTEA *x- and from examining the possibility of reconstructing fricatives other than pTEA *s and *x Тn ЭСО lТРСЭ oП JoСanson’s (1976) findings for Turkic. Abbreviations b) Linguistic forms ABL ablative OPT optative ACC accusative NEG negation ADN adnominalizer NML nominalizer CONV converb PL plural COP copula PFV perfective DAT dative PROL prolative FIN finite PST past FUT future PT particle GEN genitive RES resultative IMP imperative SG singular LOC ocative 398 Martine Robbeets b) Languages Az. Azerbaijanian Na. Nanai Balk. Balkar Neg. Negidal Bao. Bao’an Nog. Noghay Bash. Bashkir OJ Old Japanese Bur. Buriat Olch. Olcha Chu. Chuvash OTk. Old Turkic Dag. Dagur pA proto-Altaic Dolg. Dolgan pJ proto-Japonic Dong. Dongxiang (Santa) pK proto-Koreanic Evk. Evenki pKJ proto-Koreo-Japanic Gag. Gagauz pMo proto-Mongolic J (standard Tokyo) Japanese pR proto-Ryukyuan Jur. Jurchen pTEA proto-Transeurasian K (standard Seoul) Korean pTg proto-Tungusic Kalm. Kalmuk pTk proto-Turkic Karakh. Karakhanide (Old Turkic) Sal. Salar Kaz. Kazakh Sd. Shodon KBalk. Karachay-Balkar Sol. Solon Khal. Khalkha SUig. Sary-Uighur Khak. Khakas SYug. Shira-Yughur Kirg. Kirghiz Tat. (Volga) Tatar KKalp. Kara-Kalpak Tk. Turkish Krm. Karaim Tkm. Turkmenian MK Middle Korean Tof. Tofalar MMo. Middle Mongolian Ud. Udehe Mgr. Monguor Uig. Uighur Mogh. Moghol Uz. Uzbek MTk. Middle Turkic WMo. Written Mongolian References Baskakov, Nikolaj A. 1981. AltКТsɕКɔК sОm’ɔК ɔКгвɕoЯ Т ОО ТгučОnТО. Moskova: Nauka. Bentley, John 2008. A linguistic history of the forgotten islands. A reconstruction of the proto- language of the southern Ryukyus. (Languages of Asia 7.) Folkestone: Global oriental. Benzing, Johannes 1955a. Die tungusischen Sprachen. Versuch einer vergleichenden Gram- matik. Abhandlungen der geistes- und sozialwissenschaftlichen Klasse 11, 949Ḳ1099. Benzing, Johannes 1955b. Lamutische Grammatik, mit Bibliographie, Sprachproben und Glossar. Wiesbaden: Harrassowitz. A velar fricative in proto-Transeurasian 399 BОnгТnР, JoСannОs 1959. Das TsМСЮаasМСТsМСО. Inμ DОnв, JОanν GrönbОМС, KaarОν SМСООl, Helmuth & Togan, Zeki Velidi (eds.) 1959. Philologiae Turcicae Fundamenta. Tomus Primus. Wiesbaden: Steiner. 695Ḳ751. Choi, Han-Woo 2002. A comparative morphology of Altaic languages: Deverbal noun suffixes. International Journal of Central Asian Studies 7, 23Ḳ40. Comrie, Bernard 1976. Aspect: an introduction to the study of verbal aspect and related problems. Cambridge: Cambridge University Press. Dвbo, Anna. 1990. InlaЮЭnвО РЮЭЭЮral’nвО Я TЮnРЮso-εan’čžЮrskom Т pra AlЭaУskom. In: Sravitelno-ТstorТОsɕoО ɔКгвɕoгnКnТО nК soЯrОmОnnom tКgО. Moskova: Nauka. Erdal АТОnОr ZОТtsМСrТПt Пür НТО KunНО НОs Morgenlandes 71, 83Ḳ114. Erdal, Marcel 1991. Old Turkic word formation. A functional approach to the lexicon. (Turcologica 7.) Wiesbaden: Harrassowitz. Erdal, Marcel 2004. A grammar of Old Turkic. Leiden: Brill. Evans, Nicholas 2008. Insubordination and its uses. In: Nikolaeva, Irina (ed.) 2008. Finiteness. Theoretical and empirical foundations. Oxford: Oxford University Press, 366Ḳ431. Fedotov, M. R. 1996. ČuЯКšsɕТɔ ɔКгвɕṬ IstoɕТ otnosšОnТО ɕ AltКɔsɕТm Т FТnno-Ugorskim jazykam ТstorТčОsɕКɔК grКmmКtТɕКṬ ČОboksarвμ IгНaЭОl’sЭЯo ČЮЯašskoРo ←nТЯОrsТЭОЭa. Gabain, Annemarie von 1950. AlttürɕТsМСО GrКmmКtТɕ. Leipzig: Harrassowitz. Gorelova, Liliya 2002. Manchu grammar. Leiden: Brill. Johanson, Lars 1979. AlttürɕТsМС Кls ʿНТssТmТlТОrОnНО SprКМСО’Ṭ (Abhandlungen der Akademie der Wissenschaften und der Literatur, Mainz, Geistes- und sozial-wissenschaftliche Klasse 3.) Wiesbaden: Steiner. Johanson, Lars 2012. From the intimate life of Turkic sonorants. Paper presented at the АorksСop “АОsЭ OlН TЮrkТМμ TЮrkТМ loanаorНs Тn HЮnРarТan” НОНТМaЭОН Эo ProПОssor AnНrпs Rяna-Tas on the occasion of his 80th birthday. The Szeged Committee of the Hungarian Academy of Sciences, March 11, 2012. Kane, Daniel 2009. The Kitan language and script. Leiden: Brill. Krueger, John R. 1961. Chuvash manual. Introduction, grammar, reader and vocabulary. (Uralic and Altaic series 7.) The Hague: Mouton. Levickaja, L. S. 1976. IstorТčОsɕКɔК morПologТɔК čuЯКšsɕogo jazyka. Moskova: Nauka. Martin, Samuel 1970. Shodon: a dialect of the northern Ryukyus. Journal of the American Oriental Society 90.1, 97Ḳ139. Martin, Samuel Elmo 1987. The Japanese language through time. New Haven: Yale University Press. Martin, Samuel Elmo 1992. A reference grammar of Korean. Tokyo: Tuttle. Menges, Karl Heinrich 1968. Die tungusischen Sprachen. Leiden: Brill. Menges, Karl Heinrich 1984. Korean and Altaic. A preliminary sketch. Central Asiatic Journal 28, 234Ḳ295. Miller, Roy Andrew 1987. Proto Altaic *x-. Central Asiatic Journal 31, 19Ḳ63. Nikolaeva, Irina A. 1999. A Grammar of Udehe. Leiden University Ph.D dissertation. Orlovskaya, M. N. 1999. Jazyk mongolskich tekstov XIII-XIV vv. Moskova: Inst. Vostokovedenija RAN. Poppe, Nicholas 1954. Grammar of written Mongolian. Wiesbaden: Harrassowitz. Poppe, Nicholas 1955. Introduction to Mongolian comparative studies. (εцmoТrОs НО la soМТцЭц FТnno-Ougrienne 110.) Helsinki: Suomalais-Ugrilainen Seura. 400 Martine Robbeets Poppe, Nicholas 1960. Vergleichende Grammatik der altaischen Sprachen. Teil 1, Verglei- chende Lautlehre. (Porta Linguarum Orientalium, N. S. 4.) Wiesbaden: Harrassowitz. Poppe, Nicholas 1972. A New Symposium on the Altaic Theory. Central Asiatic Journal 16, 37Ḳ58. Ramstedt, Gustaf J. 1952. EТnПüСrung in die altaische Sprachwissenschaft, II, Formenlehre. (εцmoТrОs НО la SoМТцЭц ПТnno-ougrienne 104, 2.) Helsinki: Suomalai-Ugrilainen Seura. Ramstedt, Gustaf J. 1957. EТnПüСrung Тn НТО КltКТsМСО SprКМСаТssОnsМСКПt, I, Lautlehre. (εцmoТrОs НО la SoМТцЭц finno-ougrienne 104, 1.) Helsinki: Suomalai-Ugrilainen Seura. RтsтnОn, εarЭЭТ 1957. MКtОrТКlТОn гur MorpСologТО НОr türɕТsМСОn SprКМСОnṬ (Studia Orientalia 21.) Helsinki: Suomalais-Ugrilainen Seura. Robbeets, Martine 2005. Is Japanese related to Korean, Tungusic, Mongolic and Turkic? (Turcologica 64.) Wiesbaden: Harrassowitz. Robbeets, Martine 2009. Insubordination in Altaic. Journal of Philology 31. Ural-Altaic Studies 1, 61Ḳ79. Robbeets, Martine 2012. Shared verb morphology in the Transeurasian languages: copy or cognate? In: Johanson, Lars & Robbeets, Martine (eds.) 2012. Copies vs. cognates in bound morphology. (BrТll’s SЭЮНТОs Тn δanРЮaРО, CoРnТЭТon anН CЮlЭЮrО γ.ẓ δОТНОnμ BrТll. 427Ḳ446. Robbeets, Martine (forthcoming). Insubordination and the establishment of genealogical rela- tionship. In: Evans, Nicholas & Watanabe, Honore (eds.) Dynamics of insubordination. Amsterdam: Benjamins. SпrköгТ, AlТМО β004. Classical Mongolian. εünМСОnμ δINCOε. SТmonoЯ, εТМСaТl 1988. FonoloРТčОskaУa sТsЭОma ←НОРОУskoРo Уazyka. In: Simonov, Michail & Anikin, Aleksandr (eds.) 1988. Istoriko-tТpologТčОsɕТО ТsslОНoЯКnТɔК po TungusТ- Tunguso-MКn’čžursɕТm ɔКгвɕКm. Novosibirsk: Nauka. 44Ḳ88. Starostin, Sergej, Dybo, Anna & Mudrak, Oleg 2003. Etymological dictionary of the Altaic languages. Leiden: Brill. Vovin, Alexander 1997b. Voiceless velars in Manchu. JournКl НО lК SoМТцtц FТnno-Ougrienne 87, 263Ḳ280. Vovin, Alexander 2001. Japanese, Korean and Tungusic. Evidence for genetic relationship from verbal morphology. In: Honey, David B. & Wright, David C. (eds.) 2001. Altaic affinities (Proceedings of the 40th meeting of the PIAC, Provo, Utah 1997.) Indiana University: Research Institute for Inner Asian Studies. 183Ḳ202. Vovin, Alexander 2009. A descriptive and comparative grammar of Western Old Japanese. Part 2: adjectives, verbs, adverbs, conjunctions, particles, postpositions. (Languages of Asia 8.) Folkestone: Global Oriental. Weiers, Michael 1966. UntОrsuМСungОn гu ОТnОr СТstorТsМСОn GrКmmКtТɕ НОs prтɕlКssТsМСОn Schriftmongolisch. Bonn: Rheinischen Friedrich-Wilhelms-←nТЯОrsТЭтЭ PС.D НТssОrЭaЭТon. 

Игры против ...

В данной статистике представлены игроки, против которых футболист сыграл наибольшее количество матчей. Перечислены текущие и бывшие соперники согласно количеству матчей друг против друга. Список также можно представить согласно количеству набранных очков за матч, побед, ничьих, поражений и минут, сыгранных в матчах друг против друга.

С помощью фильтров можно отобразить определенных соперников, матчи в определенных турнирах и клубах, амплуа соперников и другие дополнительные параметры.

Если щелкнуть на цифру игр, сыгранных друг против друга, или выбрать соперника через фильтр, то отобразятся в виде списка все игры двух футболистов. В списке представлено количество минут, сыгранных вместе на поле в каждом мачте друг против друга. Игры, в которых оба игрока сидели на скамейке или не были на поле одновременно, отмечены красным и желтым соответственно.

Проф. Др. Нуреттин Демир
Специалист по гинекологии и акушерству,

Он родился в 1953 году в деревне Гёчек в Фетхие.
Он закончил начальную школу в Гёджеке и среднюю школу в Измире Бука.
Он поступил в высшую военно-морскую школу Хейбелиада в 1967 году, но предпочел гражданскую жизнь и в 1970 году основал медицинский факультет Эгейского университета.
В период с 1977 по 1981 год она работала ассистентом и специалистом в Эгейском университете на кафедре акушерства и гинекологии.
В 1983 году она начала работать специалистом по гинекологии и акушерству в родильном доме и гинекологической больнице SSK Ege.
В 1989 г. получил звание доцента кафедры акушерства и гинекологии.
В 1992 году он основал Медицинский центр Иренбе, цель которого - оказывать успешные услуги в Измире и Эгейском регионе.
В 1998 году в Иренбе был открыт первый частный центр ЭКО в Измире.
Покинув Иренбе в период с 2000 по 2004 год, он работал преподавателем на медицинском факультете Харранского университета, заведовал кафедрой акушерства и гинекологии, заведовал кафедрой хирургических медицинских наук и главным врачом.
Он ушел с медицинского факультета Харранского университета в 2004 году в качестве профессора.
С 2004 по 2009 год он работал специалистом по гинекологии и акушерству и председателем правления частного центра гинекологии и акушерства Иренбе, Центр ЭКО.
В 2009 году она работала преподавателем на медицинском факультете Мугла Сытки Кочман, кафедрой акушерства и гинекологии и деканом-основателем медицинского факультета.
Он занимал пост заместителя Мугла 24 и 25 семестр и вышел на пенсию.
Она начала работать специалистом по гинекологии и акушерству в частном центре гинекологии и акушерства Иренбе, Центр ЭКО.
Женат, имеет двоих детей.