Journal of Language Relationship: Volume 14/3-4 9781463237745

Table of contents :
Table of Contents / Содержание
Contributors
Сведения об авторах
Note for Contributors
Будущим авторам
Issue 3 / Часть 3
Articles / Статьи
S-singulatives in Ket
Historical phonology of Proto-Northern Jê
Discussion Articles / Дискуссионные статьи
Initial *sp- in Hittite and šip(p)and- ‘to libate’
Response to C. Melchert
Reports / Хроника
Towards Proto-Niger-Congo: Comparison and Reconstruction, Paris, LLACAN, September 1–3, 2016
Book Reviews / Рецензии
С. В. Кулланда [Sergei Kullanda]. Скифы: язык и этногенез [The Scyths: language and ethnogenesis]. М.: Университет Дмитрия Пожарского, 2016. 215 с.
Carlotta Viti (ed.). Perspectives on historical syntax. Studies in Language Companion Series 169. Amsterdam: John Benjamins, 2015. 158 p.
Issue 4 / Часть 4
Articles / Статьи
Genealogical classification of New Indo-Aryan languages and lexicostatistics
К вопросу о точности глоттохронологии: датирование процесса лексических замен по данным романских языков
Славянский чертежник: этимология слав. *čьrtъ ‘черт’
The origin of Khanty retroflex nasal

Citation preview

ВЕСТНИК РГГУ Научный журнал

Серия «Филология. Вопросы языкового родства»

Российский государственный гуманитарный университет Институт языкознания Российской Академии наук

Вопросы языкового родства Международный научный журнал № 14, часть 3—4 (2016)

Москва 2016

Russian State University for the Humanities Institute of Linguistics of the Russian Academy of Sciences

Journal of Language Relationship International Scientific Periodical Nº 14, issue 3–4 (2016)

Moscow 2016

Редакционный совет: Вяч. Вс. ИВАНОВ (Москва – Лос-Анджелес) / председатель Х. АЙХНЕР (Вена) В. БЛАЖЕК (Брно) У. БЭКСТЕР (Анн Арбор) В. Ф. ВЫДРИН (Париж) М. ГЕЛЛ-МАНН (Санта-Фе) Ф. КОРТЛАНДТ (Лейден) А. ЛУБОЦКИЙ (Лейден) Дж. МЭЛЛОРИ (Белфаст) А. Ю. МИЛИТАРЕВ (Москва) Л. ХАЙМАН (Беркли)

Редакционная коллегия: В. А. ДЫБО (главный редактор) Г. С. СТАРОСТИН (заместитель главного редактора) Т. А. МИХАЙЛОВА (ответственный секретарь) А. В. ДЫБО С. В. КУЛЛАНДА М. А. МОЛИНА И. С. ЯКУБОВИЧ

Журнал основан К. В. БАБАЕВЫМ

© Российский государственный гуманитарный университет, 2016

Advisory Board: Vyach. Vs. IVANOV (Moscow – Los Angeles, California) / Chairman W. BAXTER (Ann Arbor, Michigan) V. BLAŽEK (Brno) H. EICHNER (Vienna) M. GELL-MANN (Santa Fe, New Mexico) L. HYMAN (Berkeley) F. KORTLANDT (Leiden) A. LUBOTSKY (Leiden) J. P. MALLORY (Belfast) A. YU. MILITAREV (Moscow) V. F. VYDRIN (Paris)

Editorial Staff: V. A. DYBO (Editor-in-Chief) G. S. STAROSTIN (Managing Editor) T. A. MIKHAILOVA (Editorial Secretary) A. V. DYBO S. V. KULLANDA M. A. MOLINA I. S. YAKUBOVICH

Founded by Kirill BABAEV

© Russian State University for the Humanities, 2016

УДК 800(05) ББК 80/84я5

Вопросы языкового родства: Международный научный журнал / Рос. гос. гуманитар. ун-т; Рос. акад. наук. Ин-т языкознания; под ред. В. А. Дыбо. ― М., 2016. ― № 14, ч. 3—4. ― x + 146 с. ― (Вестник РГГУ. Серия «Филология. Вопросы языкового родства»: Научный журнал).

Journal of Language Relationship: International Scientific Periodical / Russian State University for the Humanities; Russian Academy of Sciences. Institute of Linguistics; Ed. by V. A. Dybo. ― Moscow, 2016. ― No. 14, issue 3–4. ― x + 146 p. ― (RSUH Bulletin. Philology Series: Scientific Periodical).

ISSN 2073-6320

http://www.jolr.ru/ [email protected]

Дополнительные знаки: С. Г. Болотов Add-on symbols by S. G. Bolotov

Подписано в печать 01.12.2016. Формат 60×90/8. Бум. офсетная. Печать офсетная. Тираж 1050 экз. Заказ №

Отпечатано в полном соответствии с качеством предоставленного оригинал-макета в «Наша Полиграфия», г. Калуга, ул. Грабцевское шоссе, 126 Лиц. ПЛД № 42-29 от 23.12.99

Table of Contents / Содержание

Table of Contents / Содержание

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Contributors / Сведения об авторах

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Note for Contributors / Будущим авторам .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

vii ix x

Issue 3 / Часть 3 Articles / Статьи Eugene Helimski. S-singulatives in Ket

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

157

[Е. А. Хелимский. Сингулятивы на -s в кетском языке]

Andrey V. Nikulin. Historical phonology of Proto-Northern Jê

. . . . . . . . . . . . . . . . . . . . . . .

165

[А. В. Никулин. Историческая фонетика северной ветви семьи же]

Discussion Articles / Дискуссионные статьи H. Craig Melchert. Initial *sp- in Hittite and šip(p)and- ‘to libate’

. . . . . . . . . . . . . . . . . . . . . .

187

[Крейг Мелчерт. Начальный кластер *sp- в хеттском языке и глагол šip(p)and- ‘жертвовать’]

Ilya S. Yakubovich. Response to C. Melchert

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

196

[И. С. Якубович. Ответ К. Мелчерту]

Reports / Хроника Towards Proto-Niger-Congo: Comparison and Reconstruction, Paris, LLACAN, September 1–3, 2016 (Galina Sim) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . .

207

Book Reviews / Рецензии С. В. КУЛЛАНДА. Скифы: язык и этногенез [Sergei KULLANDA. The Scyths: language and ethnogenesis], 2016 (П. В. Башарин) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carlotta VITI (ed.). Perspectives on historical syntax, 2015 (Maria A. Molina)

.

211

. . . . . . . . . . . .

217

Issue 4 / Часть 4 Articles / Статьи Anton I. Kogan. Genealogical classification of New Indo-Aryan languages and lexicostatistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . .

[А. И. Коган. Генеалогическая классификация индо-арийских языков и лексикостатистика]

227

М. Е. Васильев, М. Н. Саенко. К вопросу о точности глоттохронологии: датирование процесса лексических замен по данным романских языков

. . . . . . .

259

. . . . . . . . . . . . .

279

. . . . . . . . . . . . . . . . . . . . . . . . . . . .

293

[Mikhail E. Vasilyev, Mikhail N. Saenko. How accurate glottochronology can be? Dating the lexical replacement process in the Romance languages]

И. С. Якубович. Славянский чертежник: этимология слав. *čьrtъ ‘черт’ [Ilya S. Yakubovich. The Slavic draughtsman: etymology of Slav. *čьrtъ ‘devil’]

Mikhail A. Zhivlov. The origin of Khanty retroflex nasal

[М. А. Живлов. Происхождение хантыйского ретрофлексного носового]

Contributors Pavel Basharin — candidate of sciences (Philosophy), senior lecturer, Russian State University for the Humanities (Moscow), [email protected] Anton I. Kogan — candidate of sciences (Philology), researcher, Department of Asian and African languages, Institute of Oriental Studies of the Russian Academy of Sciences (Moscow), [email protected] H. Craig Melchert — PhD, professor of Indo-European studies & professor of linguistics, Emeritus, University of California, [email protected] Maria A. Molina — junior researcher, Institute of Linguistics, Russian Academy of Sciences (Moscow), [email protected] Andrey V. Nikulin — Laboratório de línguas e literaturas indígenas, University of Brasilia, [email protected]

Mikhail N. Saenko — candidate of sciences (Philology), junior researcher, Institute of Slavic studies of the Russian Academy of Sciences, Moscow, [email protected] Galina Sim — postgraduate student, Institute of Linguistics, Russian Academy of Sciences (Moscow), [email protected] Mikhail E. Vasilyev — researcher, School for Advanced Studies in the Humanities, Russian Presidential Academy of National Economy and Public Administration (Moscow), [email protected] Ilya S. Yakubovich — doctor of sciences (Philology), PhilippsUniversität Marburg; Russian State University for the Humanities; researcher, [email protected] Mikhail A. Zhivlov — candidate of sciences (Philology), senior lecturer, Institute for Oriental and Classical Studies, Russian State University for the Humanities; researcher, School for Advanced Studies in the Humanities, RANEPA (Moscow), [email protected]

Сведения об авторах Башарин, Павел Викторович — канд. филос. наук, ст. преп. РГГУ (Москва), [email protected] Васильев, Михаил Евгеньевич — науч. сотрудник ШАГИ Российской академии народного хозяйства (Москва), [email protected] Живлов, Михаил Александрович — канд. филол. наук, ст. преп. Центра компаративистики ИВКА РГГУ, науч. сотрудник ШАГИ РАНХиГС (Москва), [email protected] Коган, Антон Ильич — канд. филол. наук, науч. сотрудник Отдела языков Азии и Африки Института востоковедения РАН (Москва), [email protected] Мелчерт, Крейг — профессор индоевропеистики и лингвистики, Калифорнийский университет, [email protected]

Молина, Мария Александровна — младший научный сотрудник Института языкознания РАН (Москва), [email protected] Никулин, Андрей Владимирович — сотрудник лаборатории по изучению индейских языков и литератур, Бразильский университет, [email protected] Саенко, Михаил Николаевич — канд. филол. наук, мл. науч. сотрудник Института славяноведения РАН (Москва), [email protected] Сим, Галина — аспирант Института языкознания РАН (Москва), [email protected] Якубович, Илья Сергеевич — доктор филол. наук, Марбургский университет имени Филиппа; Российский государственный гуманитарный университет (Москва), [email protected]

Note for Contributors Journal of Language Relationship welcomes submissions from everyone specializing in comparative-historical linguistics and related disciplines, in the form of original articles as well as reviews of recent publications. All such submissions should be sent to the managing editor: G. Starostin Institute for Oriental and Classical Studies Russian State University for the Humanities 125267 Moscow, Russia Miusskaya Square, 6 E-mail: [email protected] Articles are published preferably in English or Russian, although publication of texts in other major European languages (French, German, etc.) is possible. Each article should be accompanied with an abstract (not exceeding 300 words) and keywords. For more detailed guidelines on article submission and editorial policies, please see our website at: http://www.jolr.ru or address the editorial staff directly at [email protected].

Будущим авторам Журнал Вопросы языкового родства принимает заявки на публикацию оригинальных научных статей, а также рецензий от всех, кто специализируется в области сравнительно-исторического языкознания и смежных дисциплин. Рукописи можно высылать непосредственно заместителю главного редактора по адресу: 125267 Москва Миусская площадь, д. 6 Российский государственный гуманитарный университет Институт восточных культур и античности Г. Старостину E-mail: [email protected] Предпочтительные языки публикации — английский или русский, хотя возможна также публикация статей на других европейских языках (французский, немецкий и т. п.). К каждой статье обязательно прикладывается резюме (не более 300 слов) и список ключевых слов. Подробнее о требованиях к оформлению рукописи, редакционной политике журнала и т. п. вы можете узнать на нашем сайте по адресу: http://www.jolr.ru или же непосредственно, обратившись к редакции по электронной почте ([email protected]).

Eugene Helimski † Institute of Finno-Ugrian and Uralic Studies, Hamburg University

S-singulatives in Ket * The paper focuses on an interesting aspect of synchronic and historical morphology of the Ket language and its implications for the reconstruction of Proto-Yeniseian. Based on relic evidence, it is suggested that the component ­sʼ in some Ket nominal stems should be analyzed as a desemanticized singulative marker, possibly still productive at an earlier time stage; internal and external evidence for this hypothesis is presented and discussed. Keywords: Yeniseian languages, Ket language, fossilized morphology, singulatives.

To the memory of Sergei Starostin

1. Introduction An etymological comment to Yen. *ksi (~ x­) ‘tree’ (Ket ōksʼ, pl. aq; Kott atče, atči, pl. ak, ax, āx) in Sergei Starostin’s Comparative Vocabulary of the Yeniseic Languages reads as follows: The form of the plural in this case goes back undoubtedly to Proto-Yen. *xaq ‘trees, forest’ (q.v.). If so, it can be assumed that Proto-Yen. *xksi developed from the original compound *xaq-sV or *xaq-xusa, lit. ‘tree single’ (a similar compound being present e.g. in *de-s ‘eye’, etc.). (Starostin 1995: 198)

The idea of decomposing some Yeniseic stems with singling out the morpheme *­s(V) with singulative meaning can be found, explicitly or implicitly, also in several other entries of this vocabulary, see s.v. *de-s ‘eye’, *pa (> Ket hāsʼ) ‘time (= occurrence)’, *χu-sa ‘one’ (Starostin 1995: 220, 244, 306). The analysis suggested by Starostin differs from the treatment of the pair oˑksʼ — aq in many earlier (and later) publications. They are often mentioned as merely suppletive — presumably unconnected — stems, on a line with ket ‘person, man’ — deŋ ‘men, people’ (Kreinovich 1968: 82; Vall, Kanakin 1985: 13). T. I. Porotova also views them as suppletive, adding a comment according to which the last consonant in o-q-s (= oˑksʼ) must be a verbal marker of state corresponding to German ist (sic! — “глагольный показатель состояния, соответствующий немецкому ‘ist’”) which is absent in plural as long as it denotes a singular state (Porotova 1990: 48).1 In the publications by H. Werner one can find both a mention of supple-

* This paper must have been written in 2005 or 2006 and was intended to be published in a volume dedicated to the memory of Sergei Starostin that was being planned in the USA, but has never been completed. The paper was accessible on the internet as a pdf-file with some technical shortcomings — namely, most special symbols were missing. Here they were restored and checked against the published sources; also, a handful of missing references have been added. This publication is a part of the project on publishing the etymological legacy of Eugen Helimski (RFH project No 14-04-00496a). — Valentin Gusev. 1 See fn. 4 on backgrounds of this peculiar comment. Journal of Language Relationship • Вопросы языкового родства • 14/3 (2016) • Pp. 157—163 • © The author, 2016

Eugene Helimski

tivism (Werner 1997a: 68) and classifying oˑksʼ — aq under a big group of words in Ket in which the opposition singular : plural is manifested through a consonant alternaion and/or an epenthesis (Werner 1995: 89–90). His comprehensive Yeniseic dictionary contains no statements concerning the kind of relationship between 1oˑksʼ and 2aq (and even no reference from the latter entry to the first one), see VWJS 1: 86, 2: 50. It has been known since Castrén’s times that the category of number in Yeniseic (both in Ket with Yug and in Kott) abounds in irregularities; using a plural suffix (­ŋ or ­n, with phonetically and lexically determined distribution) is a typical, but by no means the only way of differentiating between singular and plural forms. I would dare to assert that the numerous treatments and materials published in the last decades, including a special monograph by Porotova (Porotova 1990), added a lot to listing such irregularities but, as long as explanations and attempts at formulating at least some rules are concerned, did not contribute much to the classical presentation by Castrén (1858: 16–25) and to solving the problems discussed by Kreinovich (1968: 79–83), Toporov & Civjan (1968: 235–241). With its intriguing yarn of forms, the Yeniseic category of number challenges linguists with one of numerous riddles posed by these typologically unique languages. Addressing only one aspect of this riddle, I am going to show in this paper: — that one of the factors responsible for the complicated sets of number forms in Ket (and in Yeniseic in general) consists in superimposing and intermingling of two oppositions, singular vs. plural and general vs. singulative, the first one being inflectional and the second one — primarily at least — derivational; — that, in accordance with the assumption made by Starostin, *­s(V) (Ket mostly ­sʼ) can be viewed as a diachronically, and partly also synchronically, productive suffix of singulative forms. N o t e s: (1) The structure of the Yeniseic languages makes the differentiation between morpheme borders and word borders, resp. between synthetic and analytic forms, between derivation and word compounding, between suffixes and final elements in compounds embarassing, and probably — diverting from the practical issue of orthography — not obligatory. (2) Ket and other Yeniseic forms are quoted in this paper mainly (unless otherwise indicated) after Werner’s VWJS, partly also from Porotova’s SKS. The phonetic notations are therefore only partly unified (not more than in these sources). It is regretfully impossible to systematically differentiate between very phonetically exact transcriptions, characteristic of Werner’s own records (these transcriptions usually contain the marking 1–4 for prosodic types), and less reliable records which he quotes in VWJS along with his own, as well as between records in which differing graphic/transcriptional systems are used. Under these circumstances it is superfluous to comment on many minor details of phonetics which can result from dialectal or individual variation as well as from the peculiarities (and quality) of transcription.

2. deˑs’-singulatives in Ket The notion of singulatives is by no means new in Yeniseic linguistics. This term has been applied to a large group of compounds in Ket which include a noun (usually denoting substances, masses, foodstuffs, natural phenomena) as their first component and the word (suffix) deˑsʼ (‘eye’) as the second one, cf. eːľ ‘berries’ — eːľdesʼ ‘(a single) berry’ (VWJS 1: 258 — 3ːlʼ, 3ːlʼdesʼ), qoː ‘hail’ — qoːdesʼ ‘hailstone’, etc., see Porotova 1990: 65–66. In her analysis Porotova stresses that (a) there are nouns which form both plural forms and singulatives, cf. hnʼaŋ 158

S-singulatives in Ket

‘sand’ — hnʼaŋan ‘sands’ — hundesʼ (VWJS 1: 338 — h́nʼaŋdisʼ) ‘sandstone, a grain of sand’; (b) singulatives can have plural forms of their own, cf. qoːdesʼaŋ ‘hailstones (≈ hail)’ — it is reasonable to keep both these properties of deˑsʼ-singulatives in mind when dealing with presumably older sʼ-singulatives. It can be added that in several cases Ket sources quote a deˑsʼ-singulative as “singular” and the form without this element as “plural”, cf. Toporov, Civjan 1968: 237 or the entry Pak. qońdes, qoːńdeːsʼ, pl. qon ‘бисерина, зрачок // bead, pupil (of the eye)’ in SKS. In the following the productive and transparent category of deˑsʼ-singulatives is left aside. It can be thought, however, that it arose as a functional replacement of a similar category which, in the course of time, lost its productivity and transparency.

3. Data on s-singulatives 3.1. The Ket pair oˑksʼ — aq, or one of its members, has the following attested correspondences in other Yeniseic idioms: Yug (Sym Ket) 1oks and 2aχ ‘trees, forest, wood’, Kott atći ‘Baum’, ачи ‘дерево’ (cf. also ачихал ‘вершина’, ачичан ‘корень’) and āx (ag, ak, ax) ‘Bäume, Wald’ with plural āgan (! — see below), ак ‘дрова, лес’, Arin отши ‘дерево, лес’, отшил ‘дерево’ (cf. also ошапок ‘вершина’) and oo ‘дрова, лес’, Pump. hóchon ‘sylva, arbor’ (cf. also chógon in chógon-dýpun ‘folium’), see Castrén 1858; Helimski 1986; Starostin 1995: 198, 295; VWJS 1: 86, 2: 50 (for the sake of precision, data from older sources are quoted here with their original spellings and translations). I believe that the reconstructions suggested by S. Starostin and Werner need both a phonetic and a semantic refinement. First, the proto-form for aq (Starostin: *xaq, Werner: *aq) should not contain an internal glottal stop, the latter being a phonetic (or prosodic) segment automatically appearing in Ket and Yug monosyllables with primary consonantal Auslaut (see Helimski 2000).2 Second, the basic meaning of this word should be preferably reconstructed not as plural ‘trees’ (Starostin: ‘деревья’, Werner: ‘Bäume’ > ‘Wald’, ‘Holz’), but rather as general (substance name) ‘wood, chopwood, firewood’.3 This is confirmed also by numerous verbal derivatives such as Yug áχat ‘Holz besorgen, Vorräte an Holz anlegen’, Ket aq…vet ‘Holz haben’, áRasej ‘Holzvorräte anlegen’, Kott agatâqŋ (Nom. act. agat) ‘hauen’. 3.2. The element ­sʼ is present in the singular form and absent from the plural form not only in the pair ōksʼ — aq, but also in at least three or four further stems, all belonging to archaic strata of the Ket (Yeniseic) vocabulary. These are as follows: • Ket 2sʼεsʼ, pl. 1sʼeˑj ‘лиственница // larch’, Yug 2sεs, pl. 1sej (the SKS quotes also Yug plural forms with the plural marker ­ŋ added to either sεs or sej: Sym šʼeeiŋ, Vor. sesʼŋ). Further Yeniseic counterparts can contain different suffixal elements: Kott šêt, pl. šat, Arin čit, Pump. tag. Cf. also a derivative or compound in which a CV-variant of this stem seems to occur: Ket 2sʼεj, pl. sʼέŋnʼiŋ ‘Sitz aus Edeltannehzweigen // seat made of larch twigs’, Yug 2sεj, pl. sέŋnʼiŋ (the suffixal or second part is ­j, pl. (­ŋ­)nʼiŋ < (­ŋ­)jiŋ, as in 2uj, pl. úŋnʼeŋ ‘cradle’, 2qaj, pl. qáŋnʼiŋ/ qáŋnʼeŋ ‘steep bank, hill’, 2kj, pl. kóŋnʼiŋ/ kóŋnʼeŋ ‘bell’). 2

Also in recent Russian loanwords: Ket sʼaj ‘tea’, lʼεs ‘forest’, hp ‘priest’, met ‘copper’, met ‘honey’ (< Russ. чай, лес, поп, медь, мёд). 3 Note the misunderstandings which arise from the somewhat inadequate choice of translation equivalents. For example, it is customary to translate Ru. дерево simply as ‘tree’, though in numerous contexts — statistically, perhaps, even more frequent — it means ‘wood’ and denotes material rather than a natural object. 159

Eugene Helimski

• Ket 2qusʼ, pl. 2quŋ ‘дом, чум // house, nomad tent’, Yug 2χus, pl. 2χuŋ, Kott hûš, pl. huŋ, Arin ­kʼus,-kus, Pump. ­kut (hukùt), see WVJS 2: 140 with the comment: “Nach der Pluralbilding läßt sich ein altes Kompositum vermuten” (NB: Pump. t is the regular continuation of Yen. *s > Ket sʼ). • Ket 2tsʼ, pl. 2tŋ ‘камень // stone’, Yug 2čsʼ, pl. 2čŋ, Kott šîš, pl. šeŋ, Arin qes, Pump. kit (Werner in VWJS 2: 85 tends to view the forms in Arin and Pump., with pl. not attested, as unrelated to 2tsʼ). • Ket 2qεsʼ/2qäsʼ, pl. qέrʼeŋ ‘песчаная отмель // sandbank’, Yug 2χεsʼ, pl. χέdʼiŋ (this example belongs together with the rest of this group if 2qεsʼ/2qäsʼ < 2qεdsʼ/2qädsʼ, which probably cannot be proven). Besides, the same relationship between number forms is attested in several dozen compound words with one of the above stems as the second component: (SKS) Kel. ilʼoksʼ, pl. ilʼaq ‘щепка // wood splinter’, Kel. boqtis, pl. boqtŋ ‘кремень // flint’, Kur. baŋgus, pl. baŋguŋ ‘землянка // dugout’, etc. etc. 3.3. It can be argued that in the above cases the “pure” stem (without *­s or the plural marker *­ŋ) denotes substances: wood (aq), larch wood (sʼeˑ(j)), stone (t­), possibly also river sand (q(d)­), while the “singular” form denotes a unit of this substance (tree, larch tree, rock, sandbank), and the “plural” form with a formal plural marker ­ŋ several or many such units (stones = rocks, sandbanks). A similar relationship can be assumed also for qu­, possibly ‘home, dwelling place’, its single unit being a house, a tent. This semantic analysis is further supported by the fact that the unmarked “plural” forms can build plural forms of their own: Kott âgan (‘forests’), Yug (Sym) šʼeeiŋ (it can be supposed — but not checked any more, since the dialect is by now extinct — that this plural form actually meant ‘larch forests’, as distinct from 1sej ‘larches, larch forest’). By the way, the above observations discard the popular but superficial and groundless comparisons of Yeniseic words for ‘house’ and for ‘stone’ with German *xūs- (> house) and with Turkic *tāš, correspondingly. 3.4. The following Ket and Yug examples serve as further attestations of the element ­s (­sʼ) and its function: • Ket 1oˑk, pl. ́ksʼin (Yug 1ok, pl. ́ksin) ‘sterlet’. The unusual plural marking (­sʼin instead of ­n) finds a reasonable explanation if we assume that 1ōk is a general noun denoting sterlet as a fish species, its singulative (which is even attested in SKS — not in VWJS — as Kel. oks) denotes a sterlet as a single specimen belonging to this species, and ́ksin is the plural form to this singulative. • Ket 1qˑk, pl. q́ksʼeŋ (and 1qˑńeŋ) ‘Fußweg // footpath, track’ (cf. also the compound búlqk, pl. búlqksʼeŋ ‘Fußspur // footprint(s)’, Yug búlχk, pl. búlχnŋ). Here again it is possible, hypothetically expanding the dictionary data, to assume that the form 1qˑk denotes a footpath/track as a sequence of footprints, its non-attested singulative 1qˑksʼ — a single footprint, and among the two plural forms q́ksʼeŋ refers to a plurality of footprints (left e.g. by the same animal) and to a plurality of footpaths/tracks (left by several animals). • Ket 1quˑk, pl. qúksʼenʼ (SKS: quqsʼŋ, the author’s field materials from Kellog [1993]: quksʼeŋ), Yug 1χuk, pl. χuksin / χúnŋ // 3χuːn ‘hole’. The case appears to be very similar to the previous one, with some kind of — at least original — differentiation between and a hole/perforation in general and a single aperture. 160

S-singulatives in Ket

• Ket Kel. tʼitsʼ ‘one generation’, a derivative from Pak. tit ‘root’ (SKS: 241). • Ket Sul. tŋsʼ ‘money, rouble’, a derivative from 2tŋ id.; the data as presented in dictionaries permit to assume that the meaning ‘money’ was primarily associated with the form 2tŋ, and the meaning ‘rouble’ (= ‘a unit of money’) with its sʼ-derivative (and it cannot even be excluded that this distribution is preserved, at least in dialects). On the other hand, 2tŋ must be, according to VWJS, etymologically identical with 2tŋ ‘stones’ (see above): this suggests a scenario of formal and semantic differentiation between the archaic sʼ-singulative 2tsʼ and the innovative tŋsʼ (in which the function of the plural marker ­ŋ is not “recognized”). • Ket ujsʼ, újisʼ ‘birch-bark laid under a baby’, a derivative from 2uj ‘cradle’ (with presumable basic semantics ‘a piece of cradle’). • Ket 1uˑlʼsʼ/ulʼsʼ, pl. úlʼsʼen ‘a big water basin (sea, long and wide river, the Yenisei)’, a derivative from 1uˑlʼ ‘water’ (with presumable basic semantic ‘a unit of water, water as a single whole’).

4. Discussion As is clear from above, it can be assumed that a number of nouns in Ket — first of all, words of general meaning (denoting substances, masses, groups) — must have been able to participate in a binary opposition, functioning both as general nouns (with the ability to form singulatives) and as singular nouns (with the ability to form plurals). Since several such nouns demonstrate the same property in Kott (or have exact correspondences of their s-singulatives in other, poorly attested, Yeniseic languages), this duality must have existed in Proto-Yeniseic. Further development led in some cases to the reinterpretation of former singulatives as singular forms opposed to plural forms (especially if the shorter stem with general meaning was not preserved); in many other cases s-singulatives were perhaps lost or ousted by deˑsʼ-singulatives, so that the abovementioned examples are only scanty relics from the past. Still, it is hardly realistic to believe that the opposition “general : singulative” was ever as developed as to be comparable with the opposition “singular : plural” and to be an inflectional category rather than a productive derivational model. In any case, this development contributed to the formation of the present situation in Ket: “In gewissen Fällen ist die Pluralbildung immer noch ein Grenzfall zwischen Morphologie und Wortbildung” (Werner 1997b: 102). The (historical) morphological analysis of number forms suggested in this paper stands relatively close to the one by Kreinovich (1968: 81–82), who saw in the pairs 2qusʼ — 2quŋ, 2tsʼ — 2tŋ, 2sʼεsʼ — 1sʼeˑj the opposition of a singular suffix ­sʼ vs. plural suffixes ­ŋ and ­j. Werner criticized Kreinovich’s approach as unacceptable; this was made on several occasions — and with varying argumentation. One of his alternative versions says that ­sʼ has nothing to do with the category of number: it belongs to the original root, and the consonant alternation finds its explanation in historical derivation and historical phonetics (Werner 1995: 87). However, there are examples indicating that the stem-final consonant ­sʼ can be well preserved before plural markers (1keˑsʼ, pl. kasʼn ‘burbot’; 2kisʼ, pl. kísʼeŋ ‘leg’), which possibly led him to another explanation which, in my opinion, does not differ much from Kreinovich’s approach: In der Tat har man es in solchen Fällen in der Regel mit historischen Komposita zu tun, bei denen sich in Plural nur die Pluralform des ersten Kompositionsgliedes bewahrt hat; vom zweiten Kompositionsglied ist in der Singularform nur ein konsonantischer Rest übrig geblieben, der den Eindruck eines Reliktelements 161

Eugene Helimski

macht, welches E. A. Krejnovič als Marker des Singulars deutete. […] Wörter mit dem Reliktelement ­s (jug. ­s/-š) in der Singularform. (Werner 1998: 51, 52; similarly: Werner 1997b: 99)

Indeed, even the fact that the s of the singulatives is in some examples attested as part of the stem in all Yeniseic languages (e.g. in Ket 2qusʼ, Yug 2χus, Kott hûš, Arin ­kʼus, Pump. ­kut ‘house’) does not yet prove that it was always a suffix and not an independent word. It was already mentioned above that the structure (and history) of Ket makes the differentiation between derivation and word compounding — and even more, between “synthetic” and “analytic” forms in inflection — often problematic, and, in any case, hardly productive. The following pair of examples illustrates this thesis, but can possibly also shed some light on the connections — if not on the origin — of s-singulatives: (a) Ket Kur. 3qεːγet, pl. qéŋdεŋ ‘chief, boss’, a compound consisting of 4qä ‘big (in attributive function)’ (pl. 4qäŋ) and 2kεt ‘human being, man’ (suppletive pl. 2dεŋ ‘people’) (b) Ket 4qäsʼ ‘big (in non-attributive function)’ and 4qäsʼ, pl. qäŋsʼin ‘chief, boss’, where ­sʼ (pl. ­sʼin), usually treated by Werner (1998: 39) and other representatives of the Tomsk school as the so-called “predicative suffix”, is added to 4qä instead of 2kεt. Note that in (b), as well as in (a), the plural formes are double marked — the first adjectival component is in both cases supplied with the plural suffix ­ŋ. This means that, historically at least, 4qäsʼ must also be viewed as a compound word. The label “predicative suffix”, introduced by A. P. Dulzon (1968) for the element ­sʼ in non-attributive forms of adjectives, numerals, participles etc. which play an extremely important role in Ket grammar (as well as its counterpart ­še/-ši, pl. ­šin in the grammar of Kott), is misleading. The predicative function is only one (maybe the most important or the most frequent) function of the forms with this suffix, which occur, however, whenever an adjective etc. is used independently, without belonging to an attributive syntagm4. This accounts also for the fact that substantivized adjectives (like 4qäsʼ ‘chief, boss’) and participles are systematically marked with this suffix. Can it be that the suffix of non-attributive forms ­sʼ and the singulative suffix ­sʼ are of the same origin? I am not going to immerse here into details of this issue, but already the postpositive use of one in English with nonattributive/substantivized adjectives and participles (a / the big one, a / the standing one) makes such a historical connection extremely probable. Abbreviations Local varieties of Ket: Kel. — Kellog, Kur. — Kurejka, Pak. — Pakuliha, Sul. — Sulomaj.

References Castrén, M. A. 1858. Versuch einer jenissei-ostjakischen und kottischen Sprachlehre. Hrsg. von A. Schiefner. St. Petersburg. Dulzon, A.P. 1968. Ketskiy yazyk. Tomsk: Izd-vo Tomskogo un-ta. Helimski, E.A. 1986. Arkhivnye materialy XVIII veka po eniseyskim yazykam. In: Paleoaziatskiy sbornik. Leningrad: Nauka: 179–212. Helimski, E.A. 2000. Glottalizaciya v yazykakh sredney Sibiri. In: Evraziyskoe prostranstvo: Zvuk i slovo. Mezhdunarodnaya konferenciya 3–6 sentyabrya 2000: Tezisy i materialy. Moscow: 17–18. 4 Too literal interpretation of the label “predicative suffix”, plus the idea about the verbal nature of predicativity, were, as it seems, responsible for the strange analysis of the word oˑksʼ suggested by Porotova (see Introduction).

162

S-singulatives in Ket

Kreinovich, E.A. 1968. Sposoby deystviya v glagole ketskogo yazyka. In: Vyach. Vs. Ivanov, V. N. Toporov, B. A. Uspenskiy (eds.). Ketskiy sbornik: Lingvistika. Moscow: Gl. red. vost. lit.: 75–138. Porotova, T.I. 1990. Kategoriya mnozhestvennosti v eniseyskikh yazykakh. Tomsk: Izd-vo Tomskogo un-ta. SKS = Porotova, T.I. 2002. Slovar’ govornykh form ketskikh suschestvitel’nykh (s formami mnozhestvennogo chisla). Tomsk: TGPU. Starostin, S.A. 1995. Sravnitel’nyy slovar’ eniseyskikh yazykov. In: S. A. Starostin (ed.). Ketskiy sbornik: Lingvistika. Moscow: Yazyki russkoy kul’tury: 176–315. Toporov, V.N., T.V. Civjan. 1968. Ob izuchenii imeni v ketskom (Nekotorye rezul’taty i perspektivy). In: Vyach. Vs. Ivanov, V. N. Toporov, B. A. Uspenskiy (eds.). Ketskiy sbornik: Lingvistika. Moscow: Gl. red. vost. lit.: 229–246. Vall, M.N., I.F. Kanakin. 1985. Kategorii imeni v ketskom yazyke. Novosibirsk: Nauka. VWJS = Werner, Heinrich. 2002. Vergleichendes Wörterbuch der Jenissej-Sprachen. Bd. 1–3. Wiesbaden: Harrassowitz. Werner, Heinrich. 1995. Zur Typologie der Jenissej-Sprachen. (Veröffentlichungen der Societas Uralo-Altaica, Bd. 45). Wiesbaden: Harrassowitz. Werner, Heinrich. 1997a. Das Jugische (Sym-Ketische). (Veröffentlichungen der Societas Uralo-Altaica, Bd. 50). Wiesbaden: Harrassowitz. Werner, Heinrich. 1997b. Die ketische Sprache. (Tunguso-Sibirica, Bd. 3). Wiesbaden: Harrassowitz. Werner, Heinrich. 1998. Probleme der Wortbildung in den Jenissej-Sprachen. (Lincom Studies in Asian Linguistics 25). München / Newcastle: Lincom Europa.

Е. А. Хелимский. Сингулятивы на -s в кетском языке. Статья посвящена любопытному аспекту синхронной и исторической морфологии кетского языка, имеющему важное значение для праенисейской реконструкции. Основываясь на парадигматических характеристиках ряда архаичных форм, автор предполагает, что элемент ­s’, зафиксированный в некотором количестве кетских именных основ, следует анализировать как десемантизированный показатель сингулятива, по-видимому, обладавший продуктивностью на более ранних этапах развития кетского языка. В статье приводятся и подробно обсуждаются внутренние и внешние данные, так или иначе подтверждающие эту гипотезу. Ключевые слова: Енисейские языки, кетский язык, застывшая морфология, сингулятив.

163

Andrey Nikulin University of Brasilia; [email protected]

Historical phonology of Proto-Northern Jê * This is the first paper in a planned series on the historical phonology of Macro-Jê languages. The Jê languages constitute the largest and the most diverse family within the Macro-Jê stock; for this reason, all comparative Macro-Jê studies depend heavily on Jê data. However, the only attempt at a systematic reconstruction of Proto-Jê phonology and lexicon (Davis 1966) has been severely criticized in subsequent works (Ribeiro and Voort 2010, Nikulin 2015b). In this paper, I propose a reconstruction of the proto-language of Northern Jê, the largest branch of the family. Keywords: Jê languages, Macro-Jê languages, language reconstruction, comparative method.

1. Jê family The Jê family1 comprises ten extant languages, all of which are spoken in Brazil, and approximately four extinct, poorly attested languages (one of which was spoken in the Misiones province of Argentina and in the extreme east of Paraguay). Preliminary lexicostatistical calculations and the distribution of sound changes, lexical and morphological innovations point to the following phylogenetic structure of the family: Cerrado2 Northern Jê Panará3 (PAN) Core Northern Jê AMT: Apinayé (Apinajé, API), Kayapó (Mẽbêngôkre, KAY), Timbira (TIM) Tapayúna (TAP), Suyá (Kĩsêdjê, SUY) Central Jê: Xavánte (XAV), Xerénte (XER), Acroá (†), Xakriabá (†) Southern Jê Ingain (†) Kaingáng (KGG), Xokléng (XOK) (?) Jeikó (†) * I am grateful to CAPES (Coordenação de Aperfeiçoamento de Pessoal de Nível Superior) for providing a scholarship to carry out the present study. 1 Traditionally the term ‘family’ is used in South American linguistics to refer to low-level phyla (roughly equivalent to the term ‘group’ in European linguistics), while deeper phyla are commonly referred to as ‘stocks’ (roughly equivalent to ‘families’ in European linguistics). 2 This phylum has been previously called Amazonian Jê (Ribeiro and Voort 2010: 549) and Northern Jê (Ramirez, Vegini and França 2015: 261); the latter source inappropriately treats what we call Northern Jê as if it were a dialect continuum of a sole language (“Proper Jê”). The choice of the term Amazonian Jê is infelicitous, since the geographical distribution of these languages corresponds much better to the region of Cerrado than to the Amazon. 3 Called Southern Kayapó in older sources. Journal of Language Relationship • Вопросы языкового родства • 14/3 (2016) • Pp. 165—185 • © The author, 2016

Andrey Nikulin

Of these, Timbira is actually a dialect continuum with at least six divergent dialects: Pykobjê, Ramkokamekrá, Krahô, Apãniêkrá, Pará Gavião (Parkatêjê), Krikati. Kaingáng is subdivided into five dialects: Paraná, Central, South-Western, South-Eastern and São Paulo (the latter is considered an independent language in some sources). Minor dialectal differences have also been described for Kayapó as spoken by the Kayapó and Xikrín ethnic groups. A comprehensive overview of the state of affairs in comparative and synchronic studies in Jê is offered by Rodrigues (2012). All data are cited using UTS (Unified Transcription System), based on the IPA with minor differences and currently used as the default standard for the Global Lexicostatistical Database (http://starling.rinet.ru/new100)4. Broad phonetic transcription is preferred over phonemic representation or practical orthography with the exception of Timbira, for which a normalized supradialectal phonemic representation (Nikulin 2016b) is used. The data used in this paper are extracted from the following sources: Panará: Apinayé: Kayapó: Pykobjê: Ramkokamekrá: Krahô: Apãniêkrá: Parkatêjê: Tapayúna: Suyá:

Dourado 2001, Bardagil-Mas et al. 2016, Lapierre et al. 2016a Oliveira 2005, Ham et al. 1979 Costa 2015, Jefferson 1989, Stout and Thomson 1974, Salanova 2001, Salanova p.c. Sá 1999, Amado 2004 Popjes and Popjes 1971 Miranda 2014 Alves 2004 Araújo 2016, Ferreira 2003 Camargo 2010, Rodrigues and Ferreira-Silva 2011 Santos 1997, Nonato 2014, Guedes 1993

Old (late XVIII–early XX century) sources cover some Southern Kayapó, Kayapó, Timbira and Xavánte dialects which are now extinct. The most remarkable of them are: a) the dialect of Southern Kayapó once spoken in Paranaíba and Triângulo Mineiro, unique in that it retained *ɾ (*ɾ > y before back vowels in the dialect of Vila Boa, which apparently evolved into Panará) (Vasconcelos 2014); b) the variety of Xavánte recorded by Ehrenreich (1895), peculiar in that it had undergone the sound changes *c > θ, *-kw- > -ŋw- and *ɾ > y, w, , ɾ (Nikulin 2015a: 27–29); c) Timbira varieties called “Menren” and “Krao” and the Kayapó variety called “Gorotiré” by Loukotka (1963), where r is found in place of earlier * (in modern Timbira h is found, whereas in Kayapó it yielded ʔ or disappeared) (Nikulin 2015a: 25–27). Akroá-Mirim, Xakriabá, Ingain and Jeikó data are limited to low-quality wordlists. They might eventually turn out to be important for further comparative Jê studies (at least Xarkiabá and Ingain show some interesting phonological retentions); however, their data are not taken into account in the present series.

4 Since back and central unrounded vowels do not contrast in any Jê language, back unrounded vowels , ɤ, , are written here as , ,  in order to facilitate the reading.

166

Historical phonology of Proto-Northern Jê

2. Overview The first and only work dedicated to the reconstruction of Proto-Jê phonology is (Davis 1966). Davis considers data from five languages (Apinayé, Timbira, Suyá, Xavánte and Kaingáng) and proposes a reconstruction of the Proto-Jê phonological system. Even though he recognizes that Kaingáng and Xokléng are the most divergent members of the family, he does not attempt to postulate any phonological differences between Proto-Jê, Proto-Cerrado and Proto-Northern Jê. He reconstructs a system of 11 consonant phonemes, 9 oral and 6 nasal vowel phonemes. He also reconstructs 112 lexical items, whose distribution varies from Northern Jê to Jê (in my terminology). Davis’ reconstruction relies on false cognates, especially when it comes to Kaingáng (cf. 35, 55, 59, 86, 100) and fails to account for many sound correspondences, treating many developments as unexplained splits. Other shortcomings in Davis’ work include listing multiple unrelated roots under one etymology (cf. 49) and absence of systematic treatment of Jê morphophonology (e.g. relational prefixes, long verb forms, utterance-internal allomorphs in Xavánte). The correspondences postulated by Davis are presented below as Tab. 1–2 (the notation is modified for Apinayé, Timbira, Xavánte and Kaingáng to match UTS). Table 1. Proto-Jê consonants according to Davis (1966). API

PJ

TIM

SUY

XAV

KGG

*p

p

p

w ~ hw ~ p, h before ɾ

p~b/m~w

p

*t

t

t

t, t, ɾ, n

t ~ d / n,  before w

t, d / n, ɾ

*c

č,  before w

c-, -y

t, y, n

c ~ ʒ ~ y, ʔ before w

y,  in coda k, g,  wordfinally

*k

k

k ~ k

k ~ k

ʔ, h (_ ), sometimes u, w (#_a),  (C_C)

*m

m / b

m/p

m

p~b/m

b / m, p, -g / -ŋ, ­d

*n

n / d

n/t

n

t~d/n

d / n, t

*ɲ

ɲ / 

c, h, -n

n, ɲ

c, ʒ / ɲ, -y

y, n, -ŋ

*ŋ

ŋ / g

ŋ/k

ŋ

ʔ

ŋ / g, k

*w

w

w

w

w, 

, -ŋ

*ɾ

ɾ

ɾ, n

ɾ

ɾ,  (C_ )

ɾ, -n

s, y

c, ʒ / ɲ, h,  wordfinally

ɸ, y, h,  (C_), n (_C)

*z

ʔ, y, ɲ

h, y

Table 2. Proto-Jê vowels according to Davis (1966). API

PJ

TIM

SUY

XAV

KGG

PJ

API

TIM

*a

a

a

a

a

a, ̃

*ã

̃

̃

*

, e, a

, o

, a,

, , a

a, ã

*



̃

SUY

ẽ

XAV

KGG

ã

̃

ã

ĩ

*









, ĩ, i, e

*õ

̃

̃

õ

õ

ũ, ã

*









̃

*ũ

ũ

ũ

ũ

ũ

ũ

*o

o

o

o

u

?

*ẽ

̃

̃

ẽ, e, 

ẽ

̃

*u

u

u

u

u

u

*ĩ

ĩ

ĩ

ĩ

ĩ

ĩ

*







e



*e

e, 

e

e, 

e, , i

e

*i

i

i

i

i

i

167

Andrey Nikulin

The reconstruction by Davis has been heavily criticized, notably by Ribeiro and Voort (2010) and Nikulin (2016a). However, an alternative detailed description of Proto-Jê phonology has never been proposed to date. Many stems in Cerrado languages have two allomorphs: one is used when the word immediately follows its syntactic dependant, another is found in non-contiguous position. The difference between these allomorphs usually affects the initial consonant or the initial syllable. In synchronic descriptions it is practically useful to treat these alternating segments as independent morphemes (‘relational prefixes’, as described by Rodrigues (1952, 1953, 2010 [1981]). In comparative work, however, it is more appropriate to consider entire stems for the following reasons: (a) bare (prefix-less) roots do not occur; (b) the shape of the prefixes is very diverse in individual languages and this diversity can be traced back to PNJ and further; (c) in some instances the prefixes are fossilized and no longer segmentable. Henceforth the stems containing relational prefixes will be notated as follows: “non-contiguous allomorph / = contiguous allomorph”. All verbs in Jê languages can be nominalized (so-called ‘long form’). Since the allomorphy of the nominalization suffix is lexically determined, I systematically provide both the finite (‘short’) and the nominalized forms of the verbs when this information is available. This is notated as follows: “short form(-nominalization suffix)”. Whenever the addition of the suffix causes alternations to the stem, both forms are written separately: “short form / long form”. Finally, in most Jê languages words may surface differently in utterance-final position. In Northern Jê languages the differences are restricted to the presence of echo vowels and are not written out. In Central Jê the differences are sometimes very noticeable (cf. XAV tu // ñm̃ ‘belly’) and not entirely predictable; both allomorphs will be systematically written out separated by a double slash. In Southern Jê languages the vowels of certain roots are affected. I have shown that this phenomenon was present in PSJ and involved lowering of oral close-mid and open-mid vowels in final open syllables with an optional continuant coda (Nikulin 2015b). In the daughter languages (Kaingáng and Xokléng) this process was obscured by a number of sound changes. PSJ syllables containing low, high or nasal vowels, as well as syllables with a nasal coda, were not affected. For roots that match said conditions, I systematically mark whether they were subject (#) or prone (ʔ) to this phenomenon.

3. Proto-Northern Jê 3.1. Syllable structure and echo vowels. The maximal syllable structure of most Northern Jê languages is CRVC, where R is a liquid or a glide. An interesting phenomenon found to a varying extent in all Core Northern Jê languages is the existence of so-called echo vowels. Echo vowels (EV) occur after the coda consonants of final (stressed) closed syllables, mostly in utterance-final position. Their quality depends on the vowel in the syllable nucleus (V1) and on the syllable coda:

168

Apinayé:

EV = V1 (i after palatal -č; i in finite verb forms only after -aɾ; suppressed in non-finite verb forms)

Oliveira 2005: 78–79: 191

Kayapó:

EV = V1 (i if V1 = e; o ~ u if V1 = o; i after d , ;  if V1 = a; i after -č if V1 is not rounded) EV = V1 (i if V1 = a, , in non-finite verb forms, a in nouns), only if the coda is ɾ

Stout and Thomson 1974

Salanova 2001

Historical phonology of Proto-Northern Jê

Ramkokamekrá: EV = V1 (i if V1 = a)

Popjes and Popjes 1971

Krahô:

EV = V1, only if the coda is ɾ

Miranda 2014

Tapayúna:

EV = V1

Camargo 2010: 100–101

Suyá:

EV = V1 (i/ if V1 = a or after m, n, y if V1 is oral; Nonato 2014: 129 ĩ in some words following ̃n; i occurs after coronals and  elsewhere)

Echo vowels are sometimes manifested as a final i in Panará, but Core Northern Jê languages appear to be much more conservative in this respect. Apparently word-final echo vowels were present in all PNJ stems ending in a consonant, except for non-finite verb forms (hence different outcomes in Apinayé and Kayapó and a different correspondence in Central Jê, see below). Thus the presence of echo-vowels was marginally phonemic or quasi-phonemic in PNJ. It should be noted that they may have been suppressed in utterance-internal position for prosodic reasons. In most cases, its quality must have been identical to the quality of the syllable nucleus vowel. The dissimilation with a was apparently operative already in PNJ and persisted in Apinayé, Kayapó, Ramkokamekrá and Suyá; i must have surfaced after palatals and voiced post-nasalized codas. Several rhymes may be optionally analyzed as a sequence of a vowel and a glide (followed by an echo vowel) or a sequence of two vowels. These will be treated in the Vowels section. Syllable-initial clusters involving a liquid (CR) always have a labial or a velar onset in all Northern Jê languages (except for Tapayúna and Suyá, where hɾ, hɺ < *pɾ). It is practically useful to treat them as independent onsets for our purposes. Syllable-initial clusters involving a glide (Cw, Cy; in some languages y yielded a fricative) have a much more restricted distribution: Cw sequences occur mostly before a or  (Pykobjê , Suyá , Panará , ), whereas Cy sequences are relatively frequent only before e (Pykobjê i). For this reason, the glides are better analyzed as parts of raising diphthongs (like Chinese medials). Note that the glides still do interact with the syllable onsets in some cases (while plain vowels do not). In Core Northern Jê languages final syllables are stressed, except certain suffixes (which might be better nalyzed as clitics for this reason). This stress pattern can be securely traced back to PNJ. 3.2. Onset. Many voiced consonant phonemes had two allophonic realizations: one surfaced in oral syllables, another in nasal syllables (the syllable nasality was, and still is, governed by the nucleus vowel). This system is maintained in Apinayé and Kayapó, Tapayúna and Suyá with minimal changes. The following pairs of PNJ consonants occurred in complementary distribution: *m ~ * b, *n ~ * d, *ŋ ~ * g. In addition, *ɲ did not contrast with any other voiced palatal (*y, * and * 5). Since the allophony in question undeniably existed in PNJ (it is paralleled by very similar phenomena in other Jê languages as well as in related Maxakalían, Krenák and Jabutí language families), I chose to represent these allophones in my reconstructions. See Tab. 3 for the summary. Major differences between Davis’ reconstruction of PJ onsets and my reconstruction of PNJ onsets include the reconstruction of a voiced stop series and of a richer set of palatal consonants (four phonemes, five allophones).

5 Except for one very specific environment (namely, before a secondarily nasalized vowel), in which a minimal pair involving * and *ɲ is attested, see 3.3.

169

Andrey Nikulin

Table 3. Onset consonants in Northern Jê languages. PNR

PNJ

API

KAY

TIM

TAP

SUY

*p

p

p

p

p

h, h †

hw, h †

*pɾ

py, pɾ ‡

pɾ

pɾ

pɾ

hɾ

hɺ

*t

t

t (*ty > č)

t (*ty > č)

t (*ty > c)

ʈ (*ti > či, *ty > č)

t (*ti > či, *ty > s)

*

s

ʔ, 

ʔ, 

h (*w > w)

t

s

*k

k (*ka > n ̃, =ɾ ̃ ~ a, *ku > i §)

k

k

k, k §

k (*ky > č, *uka > *ua)

k()

*kɾ

ky, kɾ ‡

kɾ

kɾ

kɾ, kɾ §

kχ

k()ɹ, k ‡

*b

p

p

b

p

w (oral), m (nasal)

p, w §

*

s (* i > ti)

č

ǯ

c

t

t

g

k

m

m

m

m

*g *m

m

*mɾ

k m

mɾ

mɾ

mɾ

ɾ

*n

n-, =ɾ-

n

n

n

n

n

*ɲ

y

ɲ

ɲ

y

ɲ

ɲ

*ŋ

k

ŋ

ŋ

ŋ ~ g

ŋ

ŋ

*ŋɾ

y

ŋɾ

ŋɾ

ɾ

ŋɽ

gɹ

*b

p

b

m

[m]p

b ~ m (*by > ǯ ~ y)

b (*by > my ~ mǯ)

*bɾ

py, pɾ ‡

bɾ

mɾ

[m]pɾ

nɾ

bɺ

*d

t

d

n

[n]t

d ~ n

d

*

s

ǯ

ɲ

[n]c

t (~ d)

t (~ d)

*g

k

g

ŋ

[ŋ]k

g

g

*gɾ

ky, kɾ ‡

gɾ

ŋɾ

[ŋ]kɾ

g

gɹ

*y

y

ž

y

y

ǯ ~ y

y ~ y ~ ǯ

ɾ

ɾ

ɾ

ɾ

ɾ

v

w

w

w

w

*ɾ *w

y,

ɾ†

N o t e s: † Before rounded vowels. ‡ Before front vowels. § In unstressed syllables.

Major differences between Davis’ reconstruction of PJ onsets and my reconstruction of PNJ onsets include the reconstruction of a voiced stop series and of a richer set of palatal consonants (four phonemes, five allophones).

3.2.1. Panará. Non-trivial developments in Panará include: •

170

*ɾ > y before back vowels (did not affect the southernmost dialects of Southern Kayapó): PNJ *ka= gɾ ‘warm’ > PNR =ɾ̃=ky ; PNJ *ɾ̃ ‘flower’ > PNR iỹ; PNJ *kɾ̃ ‘head’ > PNR ikỹ; PNJ *cĩp=kɾa / *ɲĩp=kɾa ‘hand’ > PNR si=kya / yĩ=kya; PNJ *kɾ ‘cold’ > PNR ky; PNJ *caɾa / *yaɾa ‘wing, feather’ > PNR saya ‘flight feather’; PNJ *ka bɾo ‘blood’ > PNR =ɾ̃pyu;

Historical phonology of Proto-Northern Jê

PNJ *kukɾt ‘tapir’ > PNR kyti; PNJ *ɾ ‘anaconda’ > PNR y -ti; PNJ *pɾ (-ɾ) ‘to cover’ > PNR pyo-ɾĭ; PNJ * bɾo-ti ‘Genipa americana’ > PNR pyu-ti, etc. This change did not take place before front vowels: PNJ *kɾ̃(-ɾ) ‘to eat’ > PNR kɾ̃; PNJ *=kɾ ‘house’ > PNR ku=kɾ; PNJ *kɾĩ ‘short (of height), child’ > PNR ku=kɾĩ, etc. •

There are reasons to suspect that PNJ (and Proto-Cerrado) *k in unstressed syllables was phonetically voiced, at least before *a (this is still the situation in Apinayé and Tapayúna; the reflexes are distinct in Central Jê). Panará seems to corroborate this hypothesis: o

*ka [ga] > ñ in unstressed syllables before prenasalized consonants with subsequent flapping of n in intervocalic position: PNJ *ka= gɾ ‘warm’ > PNR ñ= ky / =ɾ̃= ky ; PNJ *ka bɾo ‘blood’ > PNR ñ= pyu / =ɾ̃ pyu; PNJ *kaŋ̃ ‘blood’ > *ka g̃ > PNR ñk̃; PNJ *u=ka ga ‘lazy’ > PNR s=wa ka, etc.;

o

*ka [ga] > a in unstressed syllables before voiceless consonants: PNJ *ka t ‘cotton’ > PNR astĭ ‘cord’; PNJ *kauwă ~ *kawa ‘mortar’ > PNR asuă ‘pestle’; PNJ *kapɾĩ ‘sad’ > PNR apɾĩ-p; PNJ *kapɾ̃t ˜˘ ‘turtle’ > PNR apỹn, etc.;

o

*ku > i in unstressed syllables before voiceless consonants: PNJ *ku ‘fire’ > PNR is; PNJ *kukɾt ‘tapir’ > PNR ikytĭ; PNJ *kub̃ ‘barbarian’ > PNR ip̃; PNJ *kũmtm˘ ‘capybara’ > PNR intŋ, etc.

•

Voiced stops (both plain and prenasalized) underwent devoicing. Intervocalic prenasalized stops seem to have nasalized preceding vowels. In case of monosyllabic roots ĩ was added word-initially (probably for prosodic reasons, as proposed by Lapierre et al. 2016b): PNJ * ba ‘liver’ > PNR ĩ pa; PNJ * bt ‘sun’ > PNR ĩ ptĭ; PNJ * d ‘eye’ > PNR ĩ t , etc.

•

Since CCC onsets are not allowed in Panará, such PNJ clusters were simplified: PNJ * gɾwă ~ * gɾuwa ‘moriche palm’ > PNR ĩ kwa ~ kwa-.

•

A sole example of PNJ *ŋɾ is available, in which ŋ disappears: PNJ *ŋɾ̃C˜ ˘ ‘toucan’ > PNR ỹ-kwekwe, ỹ-s.

It is unclear whether the phonemes g and w existed in Proto-Northern Jê or whether they emerged in Proto-Core Jê after the split of Panará. 171

Andrey Nikulin

3.2.2. Apinayé, Kayapó and Timbira. These languages are relatively conservative phonologically. •

PNJ * yielded ʔ or disappeared in Apinayé and Kayapó (the distribution is not clear); the Timbira reflex is h ( before w): PNJ * ‘seed’ > API  ~ ʔ, KAY ʔ, TIM h; PNJ *o ‘leaf, bodily hair’ > API o, KAY ʔo, TIM ho; PNJ *ku ‘fire’ > API kuv, KAY kuw, TIM kuh; PNJ *wa / *= wa ‘tooth’ > API wa / =čwa, KAY wa / =ǯwa, TIM wa / =cwa; PNJ *kauwă ~ *kawa ‘mortar’ > API kauv ~ kaʔu ~ kauɾŭ, KAY kawa, TIM kahuwă, etc.

•

Another development that affected all these languages is the affricatization of PNJ *ty (API, KAY č, TIM c), though only one example is currently known: PNJ *tyetĕ ‘to burn’ > API četĕ, KAY čet / čeɾĕ, TIM cet.

•

The voiced stop series remains unchanged in Kayapó; in Apinayé and Timbira all of them were devoiced (which is probably why Davis does not reconstruct it for PJ): PNJ *bitĭ ‘only’ > API pič, KAY bit, TIM pit; PNJ *b ‘forest’ > API p, KAY b ; PNJ *boĭ ‘to arrive’ > API poy, KAY boyč, TIM poy; PNJ *ka t ‘cotton’ > API kačt, KAY kaǯ t, TIM kac t; PNJ *wa / *= wa ‘tooth’ > API wa / =čwa, KAY wa / =ǯwa, TIM wa / =cwa; PNJ *ga ‘thou’ > API ka, KAY ga, TIM ka; PNJ *ga / * -ɾ / * -ɾ ‘to fry’ > API =ka / = ɾ ~ =ɾ, KAY =ga / ǯ -ɾ, TIM ka / h -ɾ / c -ɾ.

•

In Kayapó voiced prenasalized consonants became fully nasal. This has no consequences for the phonologic representation, since nasal and prenasalized consonants were allophones already in PNJ (as well as in PJ and probably in PMJ). However, in some exceptional cases the nasality propagated to the following vowel: PNJ * bɾa(-ɾ) ‘to walk’ > KAY mɾã(-yɲ); PNJ *ka bɾo ‘blood’ > KAY kamɾõ ‘blood’, kamɾo ‘spleen’; PNJ * a(-ɾ) ‘to bite’ > KAY ɲã(-yɲ). One case of nasality assimilation is attested: PNJ *yu ĭ ‘hummingbird’ > KAY ɲuy (instead of expected *yuy ).

•

After prefixes ending in -m (< *m, *p) in Kayapó *( ) > y: PNJ *am= o ‘rat’ > KAY am=yo; PNJ *am=  ‘bumblebee’ > KAY am=y; PNJ *=m= a(-ɾ) ‘to chew, to gnaw’ > KAY =m=yã / =m=yã-ɲ, etc. PNJ * sometimes yield my through analogy: PNJ * op opŏ ‘itchiness’ > KAY myomyop (analogy with the next syllable); Proto-Core Jê *p= uwă / *p= w-ɾ ‘to put vertically.PL’ > KAY p=myuw / p=my-ɾ  (analogy with ʔu=m=yuw / ʔu=m=y-ɾ  < *u=m= uwă / *u=m= w-ɾ).

•

All instances of *ɾw were subject to metathesis in Apinayé and Timbira; interconsonantal w was removed in Timbira. In some cases the metathesis was blocked in Timbira via vowel epenthesis:

172

Historical phonology of Proto-Northern Jê

Table 4. Velar k and k in Timbira lects. Cases with variation or unexpected reflexes are shadowed. PNJ

Common TIM

Krahô

Ramkokamekrá

Pykobjê

*go ‘water’

/ko/

ko

ko

ku

*gɾa ‘paca’

/kɾa/

kɾa

kɺa

kɾaː

*gɾwa ~ *gɾuwă ‘moriche log’

/kɾwă/

kɾw ~ kɾw

kɺowă

kɾow

*g ‘yard’

/k/

k

k

k ː (irreg.)

*gɾ ‘dry’

/kɾ/

kɾ ~ kɾ

kɺ (irreg.)

kɾ

*kagɾ ‘warm’

/kakɾ/

kakɾ ~ kakɾ

—

kakɾo

*gɾ ‘sing’

/kɾ /

kɾ ~ kɾ

kɺ

kɾe

*gɾo ‘pig’

/kɾo/

kɾo

kɺo (irreg.)

kɾuː ~ kɾuː

*gokõn (PAMT) ‘squash’

/koʔk̃n/

kuʔkõn ~ kuʔkõn

—

kuʔkõn

*ga ‘thou’

/ka/

ka

ka

ka

*k ‘skin’

/k/

k

k

k

*kɾa ‘offspring’

/kɾa/

kɾa ~ kɾa

kɺa

kɾa

*kɾ ‘hole’

/kɾ /

kɾ ~ kɾ

kɺ

kɾe

*k̃n˜˘ ‘stone’

/k̃n/

kẽn

k̃n

ken

PNJ *ɾuwă / *ɾw-k ‘to descend’ > API vɾ / vɾ, TIM wɾ / wɾ-k; PNJ * gɾwă ~ * gɾuwa ‘moriche palm’ > API gvɾa, TIM kɾ wă ‘moriche log’; PNJ *kɾwt  ‘beak’ > API kvɾt , TIM kɾt; PNJ *ɾw-i ‘rib’ > API vɾ-ʔi, TIM wɾʔ-hi. •

PNJ *ŋɾ is preserved in Apinayé and Timbira; for Timbira, only two examples are available, in which ŋ disappears (note that no cognates outside Core Jê have been identified for any other words containing *ŋɾ in Proto-Core Jê): PNJ *ŋɾ̃C˜ ˘ ‘toucan’ > API ŋɾ̃yɲ, KAY ŋɾ̃t, TIM ɾ̃; PAMT *iŋɾ̃t˜ ˘ ‘sprout’ > API iŋɾ̃t˜ ˘, TIM hiɾ̃t.

•

Voiced prenasalized stops were devoiced in Timbira; the prenasalization was lost except at morpheme boundaries. Lapierre et al. (2016b) took this as evidence to group Timbira and Panará against other Northern Jê languages; however, the innovations shared by Core Northern Jê and not shared by Panará clearly outnumber the number of features common to Timbira and Panará.

•

In most Timbira varieties there are two contrasting voiceless velars: k and k (Sá 1999: 52– 53, Popjes and Popjes 1971: 9, Miranda 2014: 30). This opposition is not rendered consistently in the transcriptions, which points to a considerable degree of variation already in Proto-Timbira. Apparently this opposition survives mainly in Pykobjê and Ramkokamekrá, whereas it is obsolescent in Krahô and non-existent in Apãniêkrá and Parkatêjê. Timbira k goes back to PNJ *k in stressed syllables, while Timbira k goes back to PNJ * g, *g and *k in unstressed syllables. A non-exhaustive list of Timbira etymologies illustrating this situation is provided in Tab. 4.

3.2.3. Tapayúna and Suyá. These two share some important innovations that suggest that these languages are very closely related (Rodrigues and Ferreira-Silva 2011): 173

Andrey Nikulin

•

debuccalization of *p (TAP h!, SUY hw) and further delabialization in complex onsets: PNJ *pa ‘arm’ > TAP h!a, SUY hwa 6; PNJ *puɾŭ ‘field’ > TAP, SUY huɺŭ; PNJ *pɾ̃ ‘wife’ > TAP hɾ̃, SUY hɺ̃; PNJ *pɾ ‘road’ > TAP hɾ, SUY hɺ, etc.

•

affricatization and optional prenasalization of PNJ *y (non-phonemic): Proto-Core Jê *yt  ~ *y t ‘sweet potato’ > TAP yɾ ~ ǯɾ ~ ǯɾ, SUY yɾ ~ yɾ ~ ǯɾ, etc.

•

alveolarization of PNJ * and * (TAP t and t ~ d, SUY t and t ~ d): Proto-Core Jê *u e / *= u e ‘bow’ > TAP tute, SUY sute / =tute; PNJ *a= / * -ɾ / * -ɾ ‘to enter’ > SUY a=t / t -ɺ / s -ɺ ; PNJ *= a / * ã-m / *ã-m ‘to stand’ > SUY =ta / tã-m / sã-m; PNJ *ka wa ‘salt’ > TAP kat!a, SUY katwa; PNJ *= wa / *w-ɾ / * w-ɾ ‘to bathe’ > SUY tw ~ tw; PNJ *ka e ‘star’ > TAP ka te-či ~ ka de-či, SUY kãte-či; PNJ * i ‘mother’ > TAP ti-ɾe; PNJ * a / * a-ɾ ‘to bite’ > TAP kũ=ta, SUY ta; PNJ * o / * o-ɾ ‘to hang’ > SUY to / to-ɺŏ; PNJ * epĕ ‘bat’ > TAP tewĕ, SUY dewĕ; PNJ * om opŏ ‘itchiness’ > TAP do dowŏ, etc.

•

affricatization of PNJ *t before *t (TAP či, SUY či): PNJ *akati ‘day’ > TAP agači, SUY akači; PNJ *=ti ‘augmentative’ > TAP =či, SUY =či, etc.

Individual straightforward developments in Tapayúna and Suyá include: • PNJ *t > TAP ʈ, SUY t: PNJ *tp ‘fish’ > TAP ʈw, SUY tw; PNJ *kat / *kat -ɾ ‘to leave / to be born’ > TAP kaʈ , SUY kat / kat -ɺ; PNJ *tikĭ ‘belly’ > SUY tikĭ, etc. In one case, one can suspect Kayapó or Suyá influence in Tapayúna: PNJ *tk ‘black’ > TAP tg, SUY tk. •

PNJ * > TAP t, SUY s: PNJ * ‘seed’ > TAP t, SUY s; PNJ *wak̃ ‘coati’ > TAP toak̃, SUY swak̃; PNJ *ku ‘fire’ > TAP kut, SUY kwis; PNJ * k ‘hawk, bird’ > TAP t g, SUY s k, etc.

•

PNJ *b > TAP w/m (per nasality), SUY p, w (in unstressed syllables?): PNJ *b̃ ‘grass’ > TAP mõ, SUY p̃; Proto-Core Jê *b ‘forest’ > TAP w , SUY p ‘grass, bush’; PNJ *b - ~ *b̃- ‘corn’ > TAP w -t ~ mõ-t, SUY w -s; PNJ *boĭ ‘to arrive’ > SUY pyĭ / poɾŏ; 6

174

Note that Guedes (1993) systematically writes ! and !w where other authors write hɾ and hw.

Historical phonology of Proto-Northern Jê

PNJ *=bĩ / *bĩ-ɾ ‘to kill’ > SUY pĩ / pĩ-ɺı˜˘; PNJ *ba ‘1SG.NOM, 1INCL.ABS’ > TAP wa, SUY pa ~ wa; PNJ *b ɾ- ‘pepper (tree-seed)’ > TAP w y-t; PNJ *bi / bi-ɾ ‘to ascend’ > SUY pi; PNJ *b / b-ɾ ‘to take’ > TAP w, SUY p; PNJ *bãma˜˘ ‘father’ > TAP m̃m ˜˘, SUY p ̃m ̃, etc. The suggested distribution is violated in PNJ *bitĭ ‘only’ > SUY wiɾĭ ‘always’, if the comparison is correct. In isolated cases TAP, SUY w is found as an irregular reflex of other PNJ stops: PNJ *(a=)ka b t ‘night’ > TAP a=gawɾ, but SUY (a=)ka bɺ; PNJ * b ĭ ‘honey’ > TAP wy, but TAP by-t ‘bee’, SUY bnĭ; PNJ *p i ‘one’ > TAP, SUY wti; PNJ *p- ‘verbal prefix with unclear meaning’ > SUY w-. •

PNJ *mɾ > TAP ɾ; PNJ * bɾ > TAP nɾ, SUY bɺ; PNJ *kɾ > TAP kχ, SUY k()ɹ; PNJ *ŋɾ > TAP ŋɽ, SUY gɹ; PNJ * gɾ > TAP g&, SUY gɹ: ˜˘ / ɾũm-; PNJ *mɾũmu ˜˘ ‘ant’ > TAP ɾũwu Proto-Core Jê * bɾ ‘animal, game’ > TAP nɾ, SUY bɺ; PNJ * bɾo-ti ‘Genipa americana’ > TAP nɾo-či; PNJ *ka bɾi ‘heron’ > TAP kanɾi; PNJ *kɾa ‘offspring’ > TAP kχa, SUY kɹa; PNJ *kukɾt ‘tapir’ > TAP kukχɾ, SUY kuk()ɹɾ; PNJ *ŋɾ̃ŋɾ̃ ~ *ŋɾ̃ ‘green’ > TAP ŋɽ ̃ŋɽ ̃ ~ ŋɽ ̃ ‘blue, green, yellow’, SUY gɹa gɹa-nĭ ‘yellow’; PNJ * gɾ ‘egg’ > TAP g&, SUY gɹ; PNJ * gɾotŏ ‘Pleiades’ > SUY gɹoɾŏ; PNJ * gɾ ‘to warm up’ > TAP ka= g& ‘warm’, SUY gɹ , etc.

•

PNJ b > TAP b ~ m, PNJ d > TAP d ~ n: PNJ * ba ‘liver’ > TAP ba ~ ma; PNJ * bt ‘sun’ > TAP bɾ ~ mɾ; PNJ * de ‘giant otter’ > TAP de ~ ne; PNJ * da ‘rain’ > TAP da ~ na; PNJ * d ‘eye’ > TAP d ~ n , etc.

•

PNJ Cw > TAP C!: PNJ *ka wa ‘salt’ > TAP kat!a; PNJ *kwɾ  ‘manioc’ > TAP k!ɾ; PNJ *wa ‘sour’ > TAP t!a-či, etc.

•

PNJ *ky > TAP č, PNJ *ty > TAP č, SUY s, PNJ * by > TAP y ~ ǯ ~ ǯ, SUY mǯ: PNJ *ky ‘thigh’ > TAP č; PNJ *tyetĕ ‘to burn’ > TAP čeɾĕ, SUY seɾĕ; PNJ * byed ĭ ‘husband’ > TAP yeɾĕ ~ ǯeɾĕ ~ ǯeɾĕ, SUY mǯenĭ, etc.

•

In two words PNJ *k disappears in Tapayúna; in both cases, the root is preceded by the same prefix (TAP tu- < PNJ *u): PNJ *u=ka d ‘medicine’ > TAP tu=an, SUY su=ka d; PNJ *u=ka ga ‘lazy’ > TAP tu= ̃nga. 175

Andrey Nikulin

•

According to Nonato (2014), t and k contrast with t and k in Suyá. This contrast is not recognized by Santos (1997) and Guedes (1993). Even throughout Nonato’s recordings the contrast is inconsistent (e.g. i=t̃-m ˜˘ ~ i=t̃-m ˜˘ ‘my going’). As demonstrated above, SUY t more often goes back to PNJ *t, whereas SUY t usually goes back to PNJ * . I was not able to find any similar correlations for SUY k and k: PNJ *kuke ĭ ‘agouti’ > SUY kukenĭ; PNJ *wak̃ ‘coati’ > SUY swak̃, etc.

Note that TAP k is realized as [g] in unstressed syllables (this is reflected in my transcription) and is aspirated before back vowels (this is not reflected in my transcription). This is likely to be a retention from PNJ. However, this does not seem to be related to the aspiration contrast in Suyá. Further studies are needed to determine the status of the contrast in question in Suyá as well as its origins. •

PNJ *g > SUY k (might have also happened in Tapayúna but the words in question are not attested in available sources on that language): PNJ *ga ‘2SG.NOM’ > SUY ka; PNJ *ga / * -ɾ / * -ɾ ‘to fry’ > SUY ka; PNJ *gu ‘1INCL.NOM’ > SUY ku, etc.

•

In several isolated words, PNJ *kɾ > TAP, SUY k (Guedes: č) before front vowels: PNJ *kɾĩ ‘village’ > SUY kĩ (Guedes: čĩ); PNJ *kɾitĭ ‘pet’ > TAP, SUY kiɾĭ; PNJ *kɾ̃ ‘parakeet’ > TAP kχẽ, SUY k̃ (Guedes: čẽ); PNJ *kɾitĭ ‘grasshopper, cricket’ > TAP kχit-čĭ ~ kit-čĭ.

Given that this irregular process affected different words in Tapayúna and Suyá, it must have taken place after their split. Note that in other words satisfying these conditions PNJ *kɾ developed normally: PNJ *kɾ ‘hole’ > TAP kχ, SUY kɹ; PNJ *kɾĩ (/ *kɾĩ-ɾ ?) ‘to sit.PL’ > SUY kɹĩ, etc. •

Apparently ɾw-like clusters are not tolerated in Tapayúna: PNJ * gɾwa ~ * gɾuwa ‘moriche palm’ > TAP g&uwa; PNJ *kɾwy  ‘Amazon parrot’ > TAP kχ tkχ ; PNJ *akɾwt  ‘cashew’ > TAP akχ y-tĭ.

3.3. Nucleus. Northern Jê languages typically have large vowel inventories and little to no vowel allophony. I assume that PNJ vowels have been most faithfully preserved in Kayapó and Common Timbira. The correspondences are summarized in Tab. 5. Of these, *ũ and *ã were not phonemic, and * and * were very rare. *ye and *iyă, as well as *wa and *uwă, were frequently in variation, whose nature is yet to be discovered. •

*ũ (~ *) and *ã were allophones of PNJ *u, * and *a before nasal codas:7 PNJ * ũmu ˜˘ ‘father (vocative)’ > PNR sũ, KAY ǯũn, TIM cũm ~ cũ, TAP tu-ɾe; ˘ ˜˘, KAY tũm, TIM tũm, TAP ʈũmu ˜˘, SUY tũmu ˜˘; PNJ *tũmu ˜ ‘old’ > PNR =tũ, API tũmu 7

176

The marginal status of these phonemes in Kayapó has already been noted by Salanova (2001: 24).

Historical phonology of Proto-Northern Jê

Table 5. Vowels in Northern Jê languages. PNR

PNJ

API

KAY

TIM

TAP

SUY

*a

a

a

a

a

a

a

*













*













*



,







 "†

(_y)

o, w" (_y)

*o

o

o

o

o

o,

*e

e

e

e

e

e

e

*"



"

"

"



"

*u

u

u

u

u

u

u

*i

i

i

i

i

i

i

*













*wa

wa

wa

wa

wa

a†

wa

u ~ uɾŭ

uw

uwă

uwă

w"

w"

w"

"I

w"

e‡

e‡

*uwă *w" *ye

w, w, u i, y (?)

*iyă

že,

e‡

ye,

e‡

ye,

e‡

i ~ ža ~ iɾĭ

iy

iyă

*ã

̃ ~ aːŋ

̃ ~

ã

a

a ~ ̃

iyă

̃

*̃

̃

̃

õ

̃

õ

̃

*̃

̃

̃

ẽ

̃

ẽ

̃

* ̃

̃

̃

̃

̃

̃

̃

*ĩ

ĩ

ĩ

ĩ

ĩ

ĩ

ĩ

*ũ

ũ

ũ

ũ

ũ

ũ

ũ

*

ĩ ~ iːŋ











N o t e s: † The onset becomes labialized. ‡ The onset becomes palatalized (see 3.2.).

PNJ *kũmtũm u ˜˘ ~ *kũmtm˘ ‘capybara’ > PNR intŋ, KAY kunũm, TIM kũmtũm, TAP koʈũn u ˜˘ ~ ˘ ˘ koʈũwu ˜ , SUY kutũmu ˜; ˜˘, KAY kũm, TIM kũm, SUY kus=kũmu ˜˘; PNJ *kũmu ˜˘ ‘smoke’ > API kũmu ˘ ˘ ˘ ˜ , KAY mɾũm, TIM pɾũm, TAP ɾũwu ˜; PNJ *mɾũmu ˜ ‘ant’ > API mɾũmu PNJ *ɲũm u ˜˘ ~ *ɲm˘ ‘who’ > API ɲ̃m˜˘ (older speakers), ɲamă (younger speakers) ‘another’, KAY ɲũm (Xikrín), ɲb (Kayapó), TIM yũm, TAP ɲmă, SUY ɲũmu ˜˘; PNJ *bãma˜˘ ‘other person’s father’ > API p̃m˜˘, KAY bãm, TIM a=pam, TAP m ̃m(˜˘, SUY p ̃m˜˘; PNJ *= a / * ã-m / *ã-m ‘to stand’ > PNR s̃ ~ saːŋ, API ča / č̃-m ~ ča-ɾ, KAY ǯa / ǯã-m / ã-m, TIM ca / ca-m / ha-m, SUY =ta / tã-ma˜˘ / sã-ma˜˘; PNJ *ãma˜˘ / *ɲãma˜˘ ‘chin’ > API ɲm, KAY ama, TIM hama; PNJ *ãma˜˘-o / *ɲãma˜˘-o ‘beard’ > API ɲ̃m˜˘, KAY ama-ʔo, TIM hama-ho, TAP tam-to. •

Examples of PNJ * (outside the diphthong *w): PNJ *tĭ ‘hard’ > PNR ttĭ, API tyč / tyt, KAY tyč, TIM ty, SUY tuɾŭ (t ɾ  ?); PNJ * / *  ‘bitter’ > API  /  y / č, KAY , TIM h / c, TAP t; PNJ * buwă / * b-ɾ ‘to cry’ > API bu-ɾ ~ bu / b-ɾ, KAY mu / m-ɾ , SUY b-ɺ ; PNJ *ku  ‘bad smell’ > KAY kuǯ, TIM kuč, TAP kut; PNJ *kuɾ ‘smooth’ > API, TIM kuɾ. 177

Andrey Nikulin

The same correspondence is attested in a number of roots whose distribution is limited to Apinayé, Kayapó and Timbira: PAMT * b / * b- ~ * b-ɾ ‘to carry’ > API b / b-y ~ b-ɾ, KAY =m / m-y ‘to grab’, TIM p / p-d (may be related to PNR ĩ p-ɾĭ ‘id.’); PAMT *apɾ / *yapɾ ‘to insult, to dishonor’ > API apɾ / yapɾ, KAY apɾ / yapɾ, TIM apɾ / yapɾ ‘to name’; PAMT *pɾ ‘corn husk’ > API pɾ ‘feather’, KAY pɾ, TIM pɾ ‘corn husk / feather’; PAMT *ubb  ‘deep’ > API upm , KAY ubb ; PAMT *k ‘bad smell, fish smell’ > API, TIM k, KAY k, etc. In one case the daughter languages disagree on the exact quality of Proto-Core Jê vowel: KAY y t, TIM y t, SUY yɾ ~ yɾ ~ ǯɾ ‘sweet potato’ point to Proto-Core Jê *y t, whereas API žt  and TAP yɾ ~ ǯɾ ~ ǯɾ ‘id.’ reflect PNJ *yt . •

•

The sole reliable example of PNJ * is: PNJ * / *-ɾ / *ɲ / *ɲ-ɾ ‘to sit.SG’ > PNR siːŋ ~ sĩ / ɲĩ, y / y-ɾ, SUY =ɲ / s / ɲ-ɺ˘.

API ɲ

/ ɲ-ɾ, KAY ɲ / ɲ-ɾ˘, TIM h / h-ɾ /

The alternation between *ye and *iyă can be exemplified by the following etymologies (note that the sequence *ɾy is regularly simplified to *y): PNJ *kɾiyă / *kye-ɾ ‘to raise’ > PNR ky -ɾi (?), KAY kɾiy / kye-ɾĕ. PNJ *kukiyă / *kukye-ɾ ‘to ask’ > PNR ĩ ky -ɾi (?), API kukža / kukže-ɾ, TIM kukiyă ‘to search’, SUY kukiyă; PNJ *kokiyă / *kokye-ɾ ‘to split’ > PNR kye-y ‘to cut’ (?), API kokže ‘to pick, to lift’ (?), KAY kokye ~ kokiy / kokye-ɾĕ (Xikrín: - -), TIM kokye / kokye-d ; Proto-Core Jê *a=kiyă / *a=kye-ɾ ‘to yell, to argue’ > API a=kiɾĭ / ža=kže-ɾ, KAY a=kiy ~ a=kya / ǯ =kye-ɾĕ, TIM a=kye / a=kye-ɾ, a=kiyă ~ kiyă ‘angry’, SUY a=kiyă; Proto-Core Jê * giyă / * gye- ‘to enter.PL, to put into a deep container.PL’ > API gye / gye-y , a= gye / ya= gye, KAY =ŋiy / =ŋye-y , a=ŋye-y, TIM a=kye-y, SUY a=ŋye / ŋye-ɺĕ; Proto-Core Jê *=ɾiyă ~ *=yetĕ / *yet ‘to hang.PL’ > API a=yetĕ / yet, KAY a=ɾiy, SUY =yeɾĕ / a=yet, saɾiyă / yaɾiyă.

In some other cases no such alternation is attested: PNJ *kye / *kye-d ‘to drag’ > PNR kɾ -ɾi (?), API kže / kže-d , KAY kye / kye-d , TIM kye / kye-d ; Proto-Core Jê *kakye / *kakye-d ‘to scratch’ > API kakže, TIM kakye / kakye-d , SUY kak()e-nĭ; PAMT *akye / *yakye / *akye-d ‘to look for water’ > API akže / žakže / akže-d ‘to open a hole’, TIM hakye / yakye / yakye-d ‘to fetch water’; Proto-Core Jê *kiyă ‘fire pit’ > API kiɾĭ ~ ki, TIM kiyă; PNJ *kye ‘thigh’ > API kže, KAY kye, TIM kye, TAP če; PNJ * byed ĭ ‘husband’ > API bžey , KAY myed , TIM pyed , TAP ǯeɾĕ, SUY mǯenĭ; PNJ *tyetĕ ‘to burn’ > PNR titi, API četĕ, KAY čet / čeɾĕ, TIM cet, TAP čeɾĕ, SUY seɾĕ. The distribution, if it ever existed, must have been obscured by numerous paradigmatic analogies (which seem to have operated to a lesser extent in Kayapó). *iyă is restricted to open syllables, *ye is found both in open and closed syllables. It is possible that originally *ye was found exclusively in closed syllables.

178

Historical phonology of Proto-Northern Jê

•

The alternation between *wa, *w and *uwă can be illustrated with the following examples: Proto-Core Jê *kɾuwă ~ *kɾwa ‘arrow’ > API kɾu, KAY kɾuw, TIM kɾuwă, SUY kɹwa; PNJ * gɾwa ~ * gɾuwă ‘moriche palm’ > PNR ĩ kwa ~ kwa-, API gvɾa, KAY ŋɾwa, TIM kɾ wă ‘moriche log’, TAP g&uwa, SUY ŋɹwa; PNJ *kauwă ~ *kawa ‘mortar’ > PNR asuă ‘pestle’, API kauv ~ kaʔu ~ kauɾŭ, KAY kawa, TIM kahuwă; Proto-Core Jê *ɾuwă / *ɾw-k ‘to descend’ > API vɾ / vɾ, KAY ɾuw ~ ɾw / ɾw-k, TIM wɾ / wɾk, SUY ɺw / ɺw-k ; PNJ * wa / *w-ɾ / * w-ɾ ‘to bathe’ > PNR sw -ɾĭ, API čwa / w-ɾ, KAY ǯuw / w-ɾ / ǯw-ɾ, TIM cwa / w-ɾ / cw-ɾ, SUY tw ~ tw, etc.

The medial -w- was (and still is) prohibited in syllables with labial onset. The following examples should be understood as result of elision of *w in the aforementioned environment: PNJ * buwă / * b-ɾ ‘to cry’ > API bu-ɾ ~ bu / b-ɾ , KAY mu / m-ɾ , SUY b-ɾ ; Proto-Core Jê *p= uwă / *p= w-ɾ ‘to put vertically.PL’ > API = ǯw / = ǯw-y , KAY p= m=yuw / p=m=y-ɾ , TIM p=cw / p=cw-ɾ / =m=c / =m=c-ɾ, SUY w=ntw / w=ntw-ɺ . Once again, the original distribution of these nuclei is obscure. *uwă and *wa are restricted to open syllables, whereas *w is found both in open and closed syllable. I assume that originally *w was restricted to closed syllables; in open syllable, *uwă and *wa would have occurred in free variation. This is corroborated by other cases of alternation in individual languages, such as TIM kwa / kw -ɾ ‘to take.PL’. •

Since Proto-Northern Jê vowel inventory was very rich (no less than 15 monophthongs and 2 diphthongs were phonemic), there was little space for allophony. That is why in most cases the reflexes of PNJ vowels in modern languages are quite straightforward (major shifts have occurred in some Timbira varieties after the split of Proto-Timbira, see (Nikulin 2016b)). However, several poorly understood splits have taken place in individual languages, notably PNJ * > API ,  (Nikulin 2015a: 13): PNJ *a b d ĭ ‘piranha’ > API a b n; PNJ *= ‘basket’ > API ka=v; PNJ *k ‘skin; breast’ > API k; PNJ *k ɾ ‘to whistle’ > API kɾ / k ɾ; PNJ *p t ‘southern tamandua’ > API pt, p t-ɾ, p t-ti, etc.

Their phonemic status is demonstrated by Oliveira (2005: 66–67). In most cases,  is found in phonetically open syllables, while is usually found in phonetically closed syllables (including long verb forms, in which echo vowels are typically absent). The issue is further complicated by the fact that Apinayé may be realized as any of these in free variation: [, , ]. •

Irregular nasalization in Kayapó has been treated in 3.2.2.

•

The reflexes of PNJ *w in Panará are uncertain. w is found in verbs (e.g. PNJ *w-ɾ / * w-ɾ ‘to bathe.NMLZ’ > PNR sw -ɾĭ) but is not attested in nouns: PNJ *kwɾ  ‘manioc’ > PNR kw; PNJ * dwd ĭ ‘snail’ > PNR paɾi= tu; PNJ *twb ĭ ‘fat’ > PNR tũmã, etc.

179

Andrey Nikulin

•

I have already discussed possible irregular vowel splits (especially * > ~ ; *i > i ~ ) in Suyá (Nikulin 2015a: 12–14). However, the analysis in question was based on Guedes’s data. Once Santos’s and Nonato’s recordings are taken into account, the problems discussed in the cited work are no longer valid: these authors consistently record where Guedes writes ~  and i where Guedes writes i ~ .

•

In the proto-language of Tapayúna and Suyá, PNJ *o > *(w) before y: PNJ *boĭ ‘to arrive’ > SUY pyĭ / poɾŏ; PNJ *kukoyĭ ‘monkey’ > TAP kuk!y, SUY kukwyĭ.

•

In extremely rare cases the medial -w- is found before front vowels. These words have no known cognates outside Core Jê (like the words having w in the onset position): Proto-Core Jê *kwe ĭ ‘bird, feather’ > API kvey ‘bird’, KAY kwey ‘bird’, TIM kuwed ‘bird’, TAP t =gwey ‘feather’, etc.

3.4. Coda. Except for syllables whose rhymes go back to PNJ *iyă or *uwă in PNJ, the codas of modern Northern Jê languages reflect PNJ codas. The reflexes sometimes differ phonetically depending on whether the coda was followed by an echo vowel (in utterance-internal position) or not (in utterance-final position, long verb forms in any position). These differences are noted here for Tapayúna and Suyá, where they are absolutely regular and systematic. For other languages they are written out as long as they are phonemic. See Tab. 6–7. Basic correspondences can be illustrated with the following examples: PNJ *tp ‘fish’ > PNR ʈpĭ, API tp, KAY, TIM tp, TAP ʈw, SUY tw; PNJ * bt ‘sun’ > PNR ĩ ptĭ, API bt, KAY mt, TIM pt, TAP bɾ ~ mɾ, SUY bɾ; PNJ *tĭ ‘hard’ > PNR ttĭ, API tyč / tyt, KAY tyč, TIM ty, SUY tuɾŭ (t ɾ  ?); PNJ * bĭ ‘good’ > PNR ĩ p, API bč, KAY mč, TIM py, TAP by- ~ my-, SUY bɾĭ; PNJ * k ‘hawk, bird’ > PNR s, API k-ti, KAY k, TIM h k, TAP t g, SUY s k; PNJ *tob ĭ ‘flour, powder’ > API čob // čomŏ, KAY ob / ǯob , TIM hob / čob ; PNJ *t b ĭ ‘raw’ > API t b // t m, TIM t b , SUY t m; PNJ * byed ĭ ‘husband’ > API bžey , KAY myed , TIM pyed , TAP ǯeɾĕ, SUY mǯenĭ; Proto-Core Jê *tod ĭ ‘armadillo’ > API tod // tonŏ, KAY, TIM tod , TAP ʈoɾŏ, SUY mǯenĭ; PNJ * b ĭ ‘honey’ > PNR nã=pyŋ, API by , KAY my , TIM p , TAP wy, by-t ‘bee’, SUY bnĭ; PNJ *kukoyĭ ‘monkey’ > PNR ĩkoː, API kukoy, KAY kukoɲ, TIM kukoy, TAP kuk!y, SUY kukwyĭ; PNJ *puɾŭ ‘field’ > PNR puː, API puɾ, KAY puɾŭ, TIM puɾ, TAP huɾŭ, SUY huɺŭ; PNJ * dw ‘field’ > PNR ĩ tuĭ, API dv, KAY n, TIM [n]tuwă, TAP, SUY dw. Cf. also PNJ, Proto-Core Jê or PAMT *kopŏ ‘fly (insect)’, *ĩp=k p / *ɲĩp=k p ‘claw, nail’, * epĕ ‘bat’, *ɾ p ‘jaguar’, *tyetĕ ‘to burn’, *k t ‘cicada’, *kukɾt ‘tapir’, *kubt ‘howler monkey’, * butŭ ‘neck’, *ketĕ ‘not’, *ka t ‘cotton’, *wt ‘lizard’, *p t ‘southern tamandua’, *yt  ‘sweet potato’, *tutŭ ‘pigeon’, *ka b t ‘night’, *̃=kotŏ / *ɲ̃=kotŏ ‘chest’, * gɾotŏ ‘Pleiades’, * boĭ ‘to arrive’, * boĭ ‘courbaril’, *eĭ / * eĭ ‘to deceive’, *peĭ ‘to make’, *kakĭ ‘cough’, *tk ‘black’, *ku ekĕ ‘vein’, *tikĭ ‘stomach’, *ka bɾekĕ ‘red’, *pokŏ ‘to ignite’, *kokŏ ‘wind’, *atk ‘forest surrounding the village’, *pe-k ‘to fart’, *t-k ‘to die’, *a ba-k / *ya= ba-k ‘to listen’, *ɾw-k ‘to descend’, * bakĭ ‘scorpion’, *twb ĭ ‘fat’, * b d ĭ ‘macaw’, *a b d ĭ ‘piranha’, * ĭ / * ĭ ‘sweet’, *yu ĭ ‘hummingbird’, *kwe ĭ ‘bird, feather’, *kuke ĭ ‘agouti’, *ɾ ĭ ‘grugru palm’, *bayĭ ‘snake sp.’, * yĭ ‘woodpecker’, *ɾoɾŏ ‘termite’, *b ɾ ‘tree’, *kwɾ ‘manioc’, *paɾĭ ‘foot’, *teɾĕ ‘Euterpe sp.’, *at ɾ ‘tinamou’, *ka beɾĕ ‘Turu palm’, etc. 180

Historical phonology of Proto-Northern Jê

Table 6. Coda consonants in Northern Jê languages after non-nasal vowels. PNR

PNJ *p

pĭ

API

p

KAY

p yč ‡ §

t,

yč ‡

TIM

TAP

TAP

SUY

SUY

(internal†)

(final†)

(internal†)

(final†)

p

p

wV

p

wV

t

t

ɾV

t

ɾV, ɾĭ ‡

*t

tĭ

t,

*

tĭ

yč, t #

č

y

y

y

y, t #

y, ɾV #

*k



k

k

k

k

gV

k

kV, k+ ‡

*b

b

b

b

m

m+

*d

d, y 

d

d

ɾV

n

nĩ

d

y

y

n

nĩ

y

y

y

* 

ŋ

y 

ɲ

*y

ː

y

ɲ

*ɾ

ː, ɾĭ # ¶

ɾ

ɾV,

*w

ĭ

w



ɾi $

ɾ

y

y

ɾV,

wă

y‡

wV

yĩ

ɺV,

y‡

p

ɺV, yĩ ‡ wV

N o t e s: † Internal = in the middle of an intonational phrase, final = immediately preceding a pause. ‡ After a. § After i. # In long verb forms. ¶ After . $ After a, in long verb forms also after  or . Table 7. Coda consonants in Northern Jê languages after nasal vowels. PNR

PNJ

API

KAY

TIM

TAP

TAP

SUY

SUY

(internal†)

(final†)

(internal†)

(final†)

*t

t

t

t

n

nV

*

č

č

y

n

nV

mV, wV

m

mV

nV, ɾV

n

nV

n

nĭ

ɺV

ɺV

*k

k

k

k

k

m

m

m

m

*n

n

n

n

*ɲ

ɲ, n

ɲ

n

*y

y



*m

*ɾ



ɾ

ɾ,

 n‡

ɾ

y

y

N o t e s: † Internal = in the middle of an intonational phrase, final = immediately preceding a pause. ‡ After ẽ, ĩ.

Except in long verb forms, where much variation with *ɲ and *ɾ is attested, the examples are not very numerous. No secure etymologies with a nasal nucleus followed by *p are known, though this syllable pattern might have existed, cf. KAY õp / ɲõp ‘elbow’ of unknown origin. The most reliable etymologies are: Proto-Core Jê *pɾ̃t ˜˘ ‘to run’ > API pɾ̃ t ˜˘, KAY pɾ̃ t, SUY hɺ̃n˜ ˘; Proto-Core Jê *t̃ĭ ‘sister’ > API t̃ č, KAY tõ č ‘brother’, TIM t̃y, SUY t̃n˜ ˘; PNJ *kat̃k˜ ˘ ‘firearm’ > PNR at̃, API kat̃ k˜ ˘, KAY katõ k, TIM kat̃k; Proto-Core Jê *k̃k˜ ˘ ‘lizard’ > API k̃ k ˜˘, KAY kõ k, TIM k̃k, TAP kõk-či; PNJ *k̃n˜˘ ‘stone’ > PNR kĩ˜˘y (?), API k̃n˜˘, KAY k̃n, TIM kẽn, TAP kẽne˜˘, TAP k̃n˜˘; PNJ *ĩn˜ı˘ / *ɲĩn˜ı˘ ‘faeces’ > PNR sĩ / yĩ, API ʔĩn˜ı˘ / ɲĩn˜ı˘, KAY ĩn / ɲĩn, TIM hĩn / yĩn, TAP tĩɾ˜ı˘; Proto-Core Jê *k̃n˜ ˘ ‘articulation, knee’ > API k̃n˜ ˘, KAY kõn, TIM k̃n, TAP kõɾõ, SUY k̃n˜ ˘; PNJ *kapɾ̃n ˜˘ ‘turtle’ > PNR apỹn, API kapɾ̃n ˜˘, KAY kapɾ̃n, TIM kapɾ̃n, TAP kahɾ ̃m-či, SUY kahɺ̃-či; PNJ *kut̃yĭ ‘worm, blind snake’ > API kut̃y, KAY kutõ, TIM kut̃, TAP kuʈõy; PNJ *ɾ̃ɾ˜ ˘ ‘Attalea speciosa coconut’ > API ɾ̃ɾ˜ ˘, KAY ɾ̃n, TIM ɾ̃; Proto-Core Jê *tĩɾı˜˘ ‘alive’ > API tĩɾ˜ı˘, KAY tĩn, TIM tĩɾ, SUY tĩɺ˜ı˘. 181

Andrey Nikulin

3.4.1. Notes on echo vowels. 1. The syllables containing the nucleus *a must have contained a high unrounded echo vowel. This is still the case in some Kayapó and Timbira varieties as well as in and Suyá (Stout and Thomson 1974, Popjes and Popjes 1971, Nonato 2014: 129). This vowel must have triggered palatalization of *t (in Apinayé and Kayapó) and of *ɾ (in Tapayúna and Suyá): PNJ *kɾatĭ ‘base, stem, lower part of the body’ > API kɾată ‘waist, leg, beginning, medial part of a long object’ ~ kɾayč ‘wall, stem, stalk’, KAY kɾayč ‘trunk, stump, pelvis’ (cf. SUY kɹaɾĭ); PNJ *paɾĭ ‘foot’ > TAP h!ay, SUY hwayĭ (cf. KAY paɾĭ); PNJ * ba / * ba-ɾ ‘to know’ > * ba / * ba-ɾĭ > SUY ba / ba-yĭ (cf. KAY ma-ɾĭ); PNJ *kapa / *kapa-ɾ ‘to pull out’ > *kapa / *kapa-ɾĭ > SUY kapa-yĭ. Note that the same echo vowel must have existed in syllables with the vowel *i, but in this case it triggered palatalization only in Apinayé: PNJ *bitĭ ‘only’ > API pič, but KAY bit (cf. TIM pit, maybe SUY wiɾĭ ‘always’); PNJ *kɾitĭ ‘pet’ > API kɾitĭ ~ kɾič, but KAY kɾit (cf. TAP, SUY kiɾĭ); PNJ *= i / *i-ɾi / * i-ɾi ‘to put’ > SUY =ti / si-ɺi / ti-ɺi (cf. KAY =ǯi / ǯi-ɾĭ), etc. This does not necessarily suggest that the echo vowels of these two groups of words were phonetically distinct: it is common for palatalization to be blocked when the consonant is both preceded and followed by palatalizing vowels (this is precisely what happens in languages like Paresí (Brandão 2014: 46)). 2. There are numerous reasons to believe that PNJ long verb forms did not contain echo vowels, as it happens today in Apinayé (Oliveira 2005: 191). They are listed below. •

Although echo vowels are present in Kayapó long verb forms, they are chosen in a special way for syllables whose underlying rhyme is ɾ or ɾ. While in nouns with these rhyme the echo vowel is [i] (b ɾĭ ‘tree, horn’), in long verb forms it copies the nucleus (ak -ɾ / yak -ɾ ‘to cut’). This suggests that these words did not rhyme at an earlier stage.

•

The correspondences in Central Jê languages are different for nouns and long verb forms ending in PNJ *ɾ. Compare the following pairs: PNJ *pa / *pa-ɾ ‘to finish, to kill’, Xavánte pa / pa-ɾi ‘to finish, to erase’; PNJ *paɾĭ ‘foot’, Xavánte paɾa ‘id.’.

What matters here is not the quality of PNJ echo vowel but its presence or absence. The ProtoCerrado forms of these words would have been *pa / *pa-ɾ ‘to finish’ and *paɾă ‘foot’ (the dissimilation seems to have occurred in the independent history of PNJ). •

Some Suyá alternations are explainable if we assume that the echo vowels were suppressed in PNJ long verb forms: SUY pyĭ / pot ‘to arrive’ < *boĭ / *bot < *boĭ / *bo; SUY =yeɾĕ / a=yet ‘to hang.PL’ < *=yetĕ / *yet, etc.

The depalatalization of PNJ * through suppression of an echo vowel is attested in tyt ‘hard’. 182

API

tyč /

Historical phonology of Proto-Northern Jê

It is uncertain whether this phenomenon affected PNJ long verb form suffixes other than *ɾ. As a preliminary solution, I reconstruct forms like PNJ *t̃-m ‘to go.SG’, *k̃-m ‘to drink’, *pe­k ‘to fart’, *t-k ‘to die’, *a ba-k / *ya=ba-k ‘to listen’, *ɾw-k ‘to descend’ (with the unproductive suffixes *-m and *-k also found in a handful of other verbs). However, it has not been proven conclusively that these particular suffixes occurred without an echo vowel. The same applies to the productive suffix *-ɲ.

4. Conclusion For the first time, a phonological reconstruction of Proto-Northern Jê has been proposed. Some issues still remain to be clarified, including: — the emergence of long vowels in Timbira; — the status and sources of syllable-final glottal stops in Timbira and preaspiration in Apinayé (Oliveira 2005: 78); — the status and sources of the k / k opposition in Suyá; — the status of stem-initial alternations of palatal consonants and *g (*ŋ in nasal syllables), first observed by A. P. Salanova (p.c.); — the status and sources of word-initial unstressed syllables without an onset. Now that a reconstruction of PNJ is available, we are in position to proceed to the reconstruction of Proto-Cerrado and, subsequently, Proto-Jê and Proto-Macro-Jê. The importance of such intermediate-level reconstructions as demonstrated, e.g., by S. Starostin (1999), cannot be underestimated; ignoring this stage has led to absence of reliable reconstructions of Proto-Jê, which in turn makes further comparative studies in Macro-Jê impossible. I am planning to propose a reconstruction of Proto-Jê in a forthcoming article.

References Alves, Flávia de Castro. 2004. O Timbira falado pelos Canela Apãniekrá: uma contribuição aos estudos da morfossintaxe de uma língua Jê. PhD. Campinas: Universidade Estadual de Campinas. Amado, Rosane de Sá. 2004. Aspectos morfofonológicos do Gavião-Pykobjê. PhD. São Paulo: Universidade de São Paulo. Araújo, Leopoldina. 2016. Dicionário Parkatêjê-Português. Belém. Bardagil-Mas, Bernat, Myriam Lapierre, Perankô Panará, Andrés Salanova. 2016. A Digital Dictionary of Panará. Symposium on American Indian Languages. Rochester: Rochester Institute of Technology. Brandão, Ana Paula Barros. 2014. A reference grammar of Paresi-Haliti (Arawak). PhD. Austin: The University of Texas at Austin. Camargo, Nayara da Silva. 2010. Língua Tapayúna: aspectos sociolingüísticos e uma análise fonológica preliminar. MA thesis. Campinas: Universidade Estadual de Campinas. Costa, Lucivaldo Silva da. 2015. Uma descrição gramatical da língua Xikrín do Cateté (família Jê, tronco Macro-Jê). PhD. Brasília: Universidade de Brasília. Davis, Irvine. 1966. Comparative Jê phonology. Estudos Lingüísticos: Revista Brasileira de Lingüística Teórica e Aplicada 1(2): 20–24. Dourado, Luciana Gonçalves. 2001. Aspectos morfossintáticos da língua Panará (Jê). PhD. Campinas: Universidade Estadual de Campinas. Ehrenreich, Paul. 1895. Materialien zur Sprachenkunde Brasiliens: III. Die Sprache der Akuä oder Chavantes und Cherentes (Goyaz). Zeitschrift für Ethnologie 27: 149–162. Ferreira, Marília de Nazaré de Oliveira. 2003. Estudo morfossintático da língua Parkatêjê. PhD. Campinas: Universidade Estadual de Campinas. 183

Andrey Nikulin

Guedes, Marymárcia. 1993. Swǰa Mẽkapẽra. Suyá: a lingua da gente: um estudo fonológico e gramatical. PhD. Campinas: Universidade Estadual de Campinas. Ham, Patricia, Helen Waller, Linda Koopman. 1979. Aspectos da língua Apinayé. Cuiabá: Sociedade Internacional de Lingüística. Jefferson, Kathleen. 1989. Gramática Pedagógica Kayapó. Anápolis: Associação Internacional de Linguística, SIL – Brasil. Lapierre, Myriam, Bernat Bardagil-Mas, Andrés Salanova. 2016a. The nasal consonants of Panará. WSCLA. Montréal: Université du Québec à Montréal. Lapierre, Myriam, Andrés Salanova, Bernat Bardagil-Mas. 2016b. A reconstruction of Proto-Northern Jê phonemics. Amazónicas VI. Coloquio Internacional AMAZÓNICAS (“La estructura de las lenguas amazónicas”). Leticia/ Tabatinga: Universidad Nacional de Colombia; Universidade do Estado do Amazonas; Instituto Caro y Cuervo: 110–111. Loukotka, Čestmír. 1963. Documents et vocabulaires inédits de langues et de dialectes sud-américains. Journal de la Société des Américanistes 52: 7–60. Miranda, Maxwell Gomes. 2014. Morfologia e morfossintaxe da língua Krahô (família Jê, tronco Macro-Jê). PhD. Brasília: Universidade de Brasília. Nikulin, Andrey. 2015a. Verifikaciya gipotezy o zhe-tupi-karibskom rodstve [On the genetic unity of Jê-Tupí-Karib]. MA thesis. Moscow: Moskovskiy gosudarstvennyy universitet imeni M. V. Lomonosova. Nikulin, Andrey. 2015b. Apofonia e sistema vocálico do Proto-Jê Meridional: contribuição para estudos comparativos das línguas Jê. Revista Brasileira de Linguística Antropológica 7(2): 275–306. Nikulin, Andrey. 2016a. Proto-Jê revisited: phonology, morphophonology and language contact. Amazónicas VI. Coloquio Internacional AMAZÓNICAS (“La estructura de las lenguas amazónicas”). Leticia/Tabatinga: Universidad Nacional de Colombia; Universidade do Estado do Amazonas; Instituto Caro y Cuervo: 108–109. Nikulin, Andrey. 2016b. Fonologia segmental do conjunto dialetal Timbíra (Jê Setentrional) m.s. Brasília. Nonato, Rafael. 2014. Clause Chaining, Switch Reference and Coordination. PhD. Cambridge: Massachusetts Institute of Technology. Oliveira, Christiane Cunha de. 2005. The language of the Apinajé people of Central Brazil. PhD. University of Oregon. Popjes, Jack, Josephine Popjes. 1971. Phonemic statement of Canela. Preliminary version. Anápolis: Associação Internacional de Linguística (SIL Brasil). Ramirez, Henri, Valdir Vegini, Maria Cristina Victorino de França. 2015. Koropó, puri, kamakã e outras línguas do Leste brasileiro: revisão e proposta de nova classificação. Línguas Indígenas Amercianas 15(2): 223–277. Ribeiro, Eduardo Rivail, Hein van der Voort. 2010. Nimuendajú was right: The inclusion of the Jabutí language family in the Macro-Jê stock. International Journal of American Linguistics 76: 517–570. Rodrigues, Aryon Dall’Igna. 1981 [2010]. Estrutura do Tupinambá. In: Cabral et al. (eds.). Línguas e Culturas Tupí II. Brasília: LALI/UnB, Campinas: Curt Nimuendajú: 12–42. Rodrigues, Aryon Dall’Igna. 2012. Para o estudo histórico-comparativo das línguas Jê. Revista Brasileira de Linguística Antropológica 4(2): 279–288. Rodrigues, Aryon DallʼIgna. 1952. Análise morfológica de um texto Tupi. Logos 15: 56–77. Rodrigues, Aryon DallʼIgna. 1953. Morfologia do verbo Tupi. Letras 1: 121–152. Rodrigues, Cíntia Karla Coelho, Marília de Nazaré Ferreira-Silva. 2011. Comparando as consoantes das línguas Tapajúna e Suyá. Alfa: 601–611. Sá, Rosane Muñoz de. 1999. Análise fonológica preliminar do Pykobjê. MA thesis. São Paulo: Universidade de São Paulo. Salanova, Andrés Pablo. 2001. A nasalidade em Mebengokre e Apinayé: o limite do vozeamento soante. MA thesis. Campinas: Universidade Estadual de Campinas. Santos, Ludoviko Carnasciali dos. 1997. Descrição de aspectos morfossintáticos da língua Suyá (Kĩsêdjê), família Jê. PhD. Florianópolis: Universidade Federal de Santa Catarina. Starostin, Sergey Anatol’evich. 1999. O dokazatel’stve yazykovogo rodstva. In: Ya.G. Testelec, E.V. Rakhilina (eds.). Tipologiya i teoriya yazyka: ot opisaniya k ob”yasneniyu (k 60-letiyu A.E. Kibrika): 57–69. Moskva: Yazyki russkoy kul’tury. Stout, Mickey, Ruth Thomson. 1974. Fonêmica Txukuhamẽi (Kayapó). Série lingüística 3. Summer Institute of Linguistics: 153–176. Vasconcelos, Eduardo Alves. 2014. Panará, Cayapó do Sul e a família Jê: primeiro caminho de análise. Lengua y Literatura Latinoamericana 16: 113–130. 184

Historical phonology of Proto-Northern Jê

А. В. Никулин. Историческая фонетика северной ветви семьи же. Статья является первой в планируемой автором серии публикаций по исторической фонологии языков южноамериканской макросемьи макро-же. Поскольку в рамках этой макросемьи самой большой и разнообразной семьей являются собственно языки же, сравнительные исследования по макро-же в первую очередь зависят от степени исторической обработанности данных по семье же; при этом единственная известная на сегодня попытка системной реконструкции фонологической системы и лексического инвентаря пра-же (Davis 1966) подверглась обстоятельной критике в целом ряде работ (Ribeiro and Voort 2010, Nikulin 2015b). В настоящей статье предлагается промежуточная реконструкция для прасеверного же, представляющего крупнейшую из ветвей семьи же. Ключевые слова: языки же, языки макро-же, языковая реконструкция, сравнительноисторический метод.

185

Discussion Articles / Дискуссионные статьи H. Craig Melchert Carrboro, North Carolina; [email protected]

Initial *sp- in Hittite and šip(p)and- ‘to libate’ * The Proto-Indo-European source of Hittite šip(p)and- ‘to libate’ has been the subject of much discussion, due to its implications for the treatment of initial clusters of sibilant plus stop in Hittite and potential implications for the much larger question of the status of the verbal category of the “perfect” in Anatolian: was the perfect, which in the oldest non-Anatolian IE languages expresses an attained state, inherited also in Anatolian and lost there, or is it an “Indo-Hittite” feature, i.e., a common innovation of “Core Indo-European”? Derivation of šip(p)and- from a PIE reduplicated perfect *s(p)e-spónd- has justifiably been rejected on formal and functional grounds, but improvements in our understanding of the outcome of PIE *spin Hittite, as well as recent innovative proposals regarding the phonology of reduplication and its status in PIE verbal morphology call for a reconsideration of the issue.

Keywords: hi-conjugation, Indo-Hittite, Proto-Indo-European perfect, reduplication.

At the colloquium honoring Holger Pedersen in Copenhagen in 1992, Bernhard Forssman proposed that the Hittite stem šipand- ‘libate; consecrate; offer’ reflects a PIE reduplicated perfect stem *spe-spónd­, while its rarer OH variant išpand- continues a root present (published as Forssman 1994). This account was not favorably received by the Anatolian specialists present upon its initial presentation, and it has subsequently with rare exceptions met mostly with rejection: e.g., Kassian and Yakubovich 2002: 34–5; Jasanoff 2003: 78, note 39; Tischler 2006: 1058 (with further literature); Kloekhorst 2008: 405; and Yakubovich 2009. Positive endorsements known to me are by Schulze-Thulin (2001: 384), LIV2: 577, and Hoffner and Melchert (2008: 27), the last of which elicited a renewed rejection by Yakubovich (2010a: 151). All of those who have rejected Forssman’s derivation of šipand- have explicitly or implicitly assumed that šipand- and išpand- represent alternate spellings of a preserved initial cluster /sp-/. This was also the interpretation I adopted in Melchert 1994: 31–2, although with considerable misgivings. We have learned a great deal more about the fate of initial *sp- in Hittite in the last twenty years, and I have for some time believed that the gist of Forssman’s account of šipand- must be correct (hence the cautious reference in Hoffner and Melchert 2008: 27), but still outstanding formal and functional problems that I could not solve prevented me from asserting this in print. The time has now come for a complete review of the matter. As has never been disputed, the development in Hittite of initial sequences of *st- and *skis consistently išt- and išk- respectively: ištantā(i)- ‘linger, be late’ < *steh2­ ‘stand’, ištu(wa)- ‘be* I am much indebted to Jay Jasanoff for making available to me the text of his forthcoming paper on the PIE perfect in advance of its formal publication and to Ryan Sandell and Sam Zukoff for extensive advice regarding the history of reduplication patterns. The standard disclaimer applies here with particular force, and I am solely responsible for any errors in the application of these authors’ views to the case at hand. Journal of Language Relationship • Вопросы языкового родства • 14/3 (2016) • Pp. 187—205 • © The authors, 2016

H. Craig Melchert

come known’ < *steu­, iškalla- ‘slit, tear’ < *skelH­, iškar- ‘prick, stick’ < *sker­. This is also the most common result for *sp-: išpai- ‘be satiated’ < *speh1(i)­, išpant- ‘night’ < *(k)sp-ént­, išpar‘spread out, strew’ < *sper­, išparre- ‘kick, trample’ < *sperH- (on separation of the last two see Kloekhorst 2008: 406–9), išpart- ‘escape’ < *sperdh­. However, we now have solid evidence for two additional though rare outcomes of *sp­. The first is preservation as /sp-/, where the presence of a synchronic cluster is crucially indicated by alternate spellings with ša­, še­, ši-: ša/e/ipe/ikkušta- /spe/ikusta-/ ‘pin, needle’ (see now CHD Š: 397 for attestations). As seen by Poetto (1986: 52–3), Neumann (1987: 282), and Kimball (1999: 108–9), this word clearly reflects a virtual *sp(e)ik-us-to- to the enlarged root *speig/k‘sharp, pointed’ seen in English ‘spike’, Latin spīca ‘ear (of grain)’, etc. The second rare result is anaptyxis of a vowel u: šuppištuwara- ‘adorned with appliqués, decorations’, šuppištuwari- ‘appliqué, decoration’. The meaning is now assured by the occurrence of the i-stem noun in the Hurro-Hittite Bilingual, KBo 32.14 ii 43 (see Neu 1996: 81 and 146). However, the popular etymology (already Neu 1970: 68) as a compound ‘brightly shining’, allegedly consisting of šuppi- ‘ritually pure’ and ištu(wa)- ‘become known’ makes no sense whatsoever either semantically or formally. Hittite šuppi- means ‘ritually pure’, and there is no basis of any kind for a sense ‘shining’. Nor is the role of the purported second member ‘become known’ in a compound allegedly meaning ‘brightly shining’ explained (see the justified doubts of Kloekhorst 2008: 791).1 This derivation also cannot account for the alternate form išpišduwarāš in KUB 42.64 Vo 2, which cannot be dismissed as a scribal error, since iš-piš-du-wa-ra- does not remotely resemble šu-up-pí-iš-tu-wa-ra- visually or aurally. The decorations attached to a copper cup (thus in the bilingual) and the gold and silver adornments added to clothing may well have been shiny (for the latter see refs. in Tischler 2006: 1198), but they were also more fundamentally stuck or stitched onto their respective objects.2 We are thus surely dealing with a derivative of a different form of the PIE root *spei‘pointed, sharp’ seen already above in /spe/ikusta-/: the sense of /supistwara:-/ was ‘appliquéd’, decorated with something ‘stuck on’ (for the semantics compare the history of English ‘stick’ and ‘stitch’ and German stechen). Note, however, that at least one Hittite speaker knew this word in a form with the regular treatment of *sp- as išp­. I had already recognized the existence of these two examples in Melchert 1994: 32, but found them as exceptional and inexplicable as šipand­. It is now clear, at least to me, that these forms do fit into a well-known Hittite pattern: they show the two regular results of prehistoric *sm-: (1) preservation; (2) u-anaptyxis. The first treatment is shown by Hittite ša/e-me-en-zi, šaam-na-an-zi ‘withdraw; relinquish’ where (pace Kassian and Yakubovich 2002: 12) the alternate spelling of the singular stem clearly shows synchronic /smen-/ (thus with Oettinger 1979: 104, Kimball 1999: 117, and CHD Š: 120), in an ablauting root present *smén­, *sn-énti, even if the root etymology remains uncertain (thus also Kloekhorst 2008: 714–15). There are now three examples for the treatment with anaptyctic ­u­, which is quite real (contra Kloekhorst 2008: 782–5): (1) šummittant- ‘axe’ < virtual *smit-ént- ‘(the) cutting (one)’ (already Knobloch 1956: 67, Kimball 1999: 199 et al.); 1

As per Kloekhorst (2008: 790), despite its clear behavior as an inherited word — an ablauting adjective — Hittite šuppi- ‘ritually pure’ has no clear cognates or etymology. Unfortunately, the attractive comparison with Umbrian sopa/supa and interpretation as ‘taboo’ (Watkins 1975) is very doubtful: see the extended critique by Weiss (2010: 358–83). 2 I know of no basis for the meaning ‘animal representation or icon (usually of metal)’ adopted by Yakubovich (2009: 548, note 5). In any case, the word definitely does not contain šuppi- ‘ritually pure’. 188

Initial *sp- in Hittite and šip(p)and- ‘to libate’

(2) šum(m)um(m)a- ‘unite, make one’ < *sm- ‘one’ + ­uman- ‘belonging to’ + factitive ­a(Rieken 2000: 174, modified by Hoffner and Melchert 2008: 60);3 (3) first plural enclitic possessive ­šumma/i- < *s-mé- < aphaeresized *s-mé- (Rieken 2002: 414–15).4 As emphasized by Rieken, the change of initial *sm- > šumm- with anaptyxis and gemination is a genuine Hittite sound law. She herself (2002: 408) left open the question of its precise conditioning versus that of the preservation as /sm-/. However, the contrast between ša/emen< *smén- and enclitic possessive šumma/i- < *sme- suggests that the different outcomes are conditioned by the accent: namely, that initial *sm- was preserved immediately before the accent but developed to *summ- when the following syllable was unaccented. We cannot be as certain about the accent in šummittant- and šum(m)um(m)a­, but their morphological structure is more than compatible with supposing that the accent stood farther to the right than the original initial syllable. Rieken (2002: 408) reasonably derives Hittite išmeri- ‘bridle, rein’ < *s(h2)mér­, but if the root etymology (to *seh2­ ‘bind’) is correct, as it surely is, this example does not prove a development of *sm- > išm­, since it is more likely that it was the *sh2­ that led to iš- (as in išanittar‘relative by marriage, as per Rieken 1999: 283–4). The resulting unsyllabifiable *išme- was then reduced to išme­. Pace Kloekhorst (2008: 394) nothing requires that the verb išamai- ‘sing’ reflect a zero-grade *sh2m­; it may easily continue full-grade *sh2em­, as he himself assumes for the noun išamāi- ‘song’. We may now return to the matter of the Hittite treatment of initial *sp­. The observed vacillation is now explainable. Pre-Hittite language learners were faced with two models for how to treat *sp-: since it consisted of sibilant plus voiceless stop, they could follow the model of *st- and *sk- and add a prothetic i­; however, since *sp- also consisted of a sibilant plus labial stop, speakers could also follow the model of the other sequence of sibilant plus labial stop, namely *sm­, and according to the position of the accent, either preserve the sequence or insert an anaptyctic ­u­. Although m generally behaves as a sonorant in older Indo-European languages (that is, as a continuant), one must not forget that in articulatory terms it is also a stop. It is thus not unreasonable that Hittite speakers did not show absolute consistency in their treatment of initial *sp­, where *p belonged both to the class of labial stops and to the class of voiceless stops. The dominant practice for most lexemes was to follow the model of the other voiceless stops and add a prothetic vowel i­. Contra Melchert 1994: 32, Kimball 1999: 110–11, Kassian and Yakubovich 2002: 33–5, and Yakubovich 2009: 545–7, there is not the slightest justification to doubt the linguistic reality of the prothetic vowel in išT­, as assumed by Kronasser (1966: 48–9), Eichner (1975: 98), Oettinger (1979: 416–17), Kloekhorst (2008: 61), and others. First of all, the alternations in personal names from the Old Assyrian texts of the Colony period cited by Yakubovich (2009: 546) not only all involve *sp­, as he admits, but show exactly the same variation as we have seen in šuppištuwarā- ~ išpištuwara-: Šu-pu-da-a-šu vs. Iš-pu-da-a-šu, Šupu-na-a-šu vs. Iš-pu-na-a-šu, Šu-pu-nu-ma-an vs. Iš-pu-nu-ma-an. I emphasize that we find no spellings in these names of the type †Ša-pu- or †Ši-pu­, which is what we would expect were 3

Since the word is hapax, the objection of Kloekhorst (2008: 784) that the word does not show geminate spelling for either of the two ­mm- is not compelling. 4 I am not persuaded by Rieken’s two proposed examples of the change *­sm- > ­summ- in morpheme-internal position. Hittite šumanzan- (sic!) means ‘(bul)rush’ and has basic single ­m- (see Melchert 2004: 129–31); CLuvian te/iššumma/i- ‘(unfired) clay cup’ contains the Luvian suffix ­umma/i- also seen in annarumma/i- ‘powerful’. 189

H. Craig Melchert

we facing alternate spellings for /spu-/.5 These names actually further confirm that the variation in the Hittite appellative is genuine: /sup-/ vs. /isp-/. Note that the scriptio plena of the stem vowel in šuppištuwarā- ‘appliqué’ suggests that the accent was not on the vowel following the initial *sp­, and therefore the treatment šupp- beside išp- fits the pattern for šumm- < *sm­. Unfortunately, there is no independent evidence for the position of the accent in the personal names or in ša/e/ippe/ikkušta- ‘pin’, but nothing stands in the way of supposing that the names reflect original accent beyond the first syllable, while the appellative was /spékusta-/ like /smén-/.6 Kimball (1999: 110) cites as “very convincing” my own argument (Melchert 1984: 110) that the Hittite adjective išaškant- ‘blood-shot, blood-stained’ must reflect a compound *išanškant- with the participle of iške/a- ‘anoint, smear’, thus showing that the i- of iške/a- must be purely graphic. The argument is not at all compelling, however, since nothing precludes that the compound was formed in pre-Hittite before the addition of the prothetic i­. In any case, the overlooked new example i-is-ke-ez-[zi] in the fragment KBo 34.243:3 (Ritual of Zarpiya) now excludes both my etymology and that of Rieken (1999: 402), approved by Kloekhorst (2008: 402), which start from *(p)s-s e/o- and *sg(h)-yé/ó- respectively.7 The plene spelling (which would be entirely unparalleled for the prothetic vowel) appears to require a return to the etymology of Oettinger (1979: 327), despite the semantic difficulties associated with the root *(h1)eish1­. The first two arguments adduced by Kassian and Yakubovich (2002: 33) against the reality of the prothetic i- in išT- are also without foundation. Their statement that the prothetic vowel is always spelled i- is correct, but their claim that iš-/eš- alternations are frequent in cases with etymological *i- is patently false: Hitt. iškiš- ‘back’, cognate with Grk. ἴχι(ον) ‘loins’ (a quite certain equation, pace Kloekhorst 2008: 402) is spelled exclusively with iš­, while išā- ‘owner; master, lord’ < *h1es-h2­ó- with regular raising of pretonic short *e to *i (see now on this word and its etymology Nussbaum 2014: 244–5) is also spelled exclusively with iš­, with the single exception of the totally aberrant form eš-é in the NS copy KBo 3.34 i 25, a copyist’s error that has no probative value.8 Their second point, that the prothetic vowel is never spelled with plene as i-iš­, makes no sense, since we would expect the prothetic vowel to be unaccented and thus never lengthened.9 The further argument adduced by Yakubovich (2009: 546, note 3) is also less than compelling. He claims that the HLuvian form sà-ma-ra/i-ka-wa/i-ni (URBS) for the city appearing in Hittite cuneiform as URUIš-mi/e-ri-ka- shows that the Luvians learned this city name through the Hittites with /sm-/, since Luvian had eliminated all cases of initial *sC- in their own language. There are two problems here: first, to my knowledge we know only that Luvian eliminated initial *s+stop by deletion of the sibilant (e.g., HLuvian (*261)tapai vs. Hittite 5 One could, of course, argue that the empty vowel used in the spelling for /sp-/ merely copied the following real /u/ vowel, but the evidence from Hittite appellatives for the reality of u-anaptyxis argues decisively against this. 6 The spelling of the “ethnic” suffix ­uma(n)- with plene, as in LÚi-iš-tu-u-ma-aš (KBo 23.99 i 19), provides some indirect support for an accent */Spunóman-/ at least in the one personal name. 7 Contra Kloekhorst (2008: 402), the inflection iškezzi, iškanzi must be older than that of iškiyazzi, since the inflectional type in ­e-/-a- in base verbs is recessive in Hittite, while that in ­ye-/-ya- is notoriously productive. Thus Rieken’s etymology is excluded also on this basis. 8 Contra Kloekhorst (2008: 390) the form e-eš-a-aš-ši-iš is very unlikely to belong to this word (see Otten 1961: 130–1) and is irrelevant. There is thus no basis for appealing to the sporadic New Hittite change of iš- to eš- (see further below.) 9 The claim of Kloekhorst (2008: 61) that the prothetic vowel cannot be identified with the Hittite phoneme /i/ because it fails to undergo the New Hittite lowering to ­e- is also false, since Yakubovich (2010b: 309–15) has made compelling arguments that the very sporadic change of e > i in New Hittite is not a regular sound change.

190

Initial *sp- in Hittite and šip(p)and- ‘to libate’

ištāpi ‘blocks up’). I am not aware of any evidence that tells us the fate of initial *s+sonorant. Second, even if Luvian had no native words with initial *sR­, the argument is not probative. There is no way to exclude that the Hittites adapted the name *Sme/iriga- in their fashion with prothetic i­, while the Luvians dealt with the initial *sm- by anaptyxis of an ­a­. The Luvian form may easily be read as /Samariga-/. We are thus left with šipand- alternating with išpand- as the only basis for doubting the reality of the prothetic i- in išT­. But we have now seen that this orthographic alternation cannot possibly be interpreted to stand for /spand-/, despite the assertions of Kassian and Yakubovich (2002: 33–5) and Yakubovich (2009: 547–8). We now know how a preserved initial /sp-/ was written where it existed, and as we would predict, it is expressed by alternation between šapV­, še-pV- and ši-pV- in ša/e/ipe/ikkušta- ‘pin, needle’. Given that šip(p)and- is spelled several hundred times with absolute consistency as ši-(ip)-pa-an-t/d°, it is not credible that this spelling stands for /sp-/. The first syllable of the word must be read as /si-/. Possible additional evidence for the reality of a stem /sipánd-/ comes from HLuvian and Lycian. Yakubovich (2009: 555) cites the suggestion of Hajnal (1995: 133–4) that HLuvian (CAELUM.*286.x)sá-pa-tara/i-i-sa (KARKAMİŠ A 2+3, §17a) might mean ‘libation priest’ and reflect an earlier */sVpentero/ī-/ also continued by Lycian hppñterus, which is a professional title or institution.10 It is now clear that Lycian hpp- must be derived from a prehistoric *sVp(contra Melchert 1994: 304–5), and the HLuvian may be read /sapandaris/. For Yakubovich (2009: 556) these forms attest a hybrid Luvo-Hittite creation *s pantalli- ‘pertaining to a libation’ that underwent rhoticism in Luvian and was then borrowed into Lycian. The last step is pure speculation, and the very different morphology of hppñterus- argues rather for a native Lycian word that is at best a root cognate with the Luvian. That the verbal stem is not attested in Luvian or Lycian (thus far!) is not a compelling argument against a Proto-Anatolian stem *sepónd- that led by regular phonological developments to šipand­, */sapand-/, and *hppñt­. I must emphasize, however, that I place no weight on this argument, since the meaning of the Luvian is not fully assured, and that of the Lycian is based entirely on the putative etymology. Kassian and Yakubovich (2002: 33) and Yakubovich (2009: 547) argue that one cannot interpret the first vowel of the Old Hittite/Old Script spelling ši-pa-an-t/d-° as real, because this could only imply a reading /siband-/, and voicing of the stop in this environment cannot be motivated by any known Hittite sound change. This argument reflects a fundamental methodological fallacy and a profound misunderstanding of how orthographies devised by and for native speakers work. Such orthographies cannot be compared to the International Phonetic Alphabet. Native speakers know how the words of their language are pronounced and also the grammar that predicts where they will occur, and writing systems (especially those used by a small elite) need only give just enough clues for another native speaker reader to successfully identify the word intended. Examples like the Anatolian hieroglyphs for Luvian and Linear B for Mycenaean Greek show just how much information can be omitted! Many factors determine spelling practices in a given tradition: aesthetics (important in the Anatolian hieroglyphs used for public inscriptions), convention, convenience, and above all simply imitation of one’s teachers. The Hittites knew that /sipand-/ contained a voiceless labial stop; there was no compulsion to indicate this in a word that occurred hundreds of times in Old Hittite ritual texts. Since the first vowel of ši-pa-an-t/d-° has to have been linguistically real, Yakubovich’s attempt (2009: 550–55) to motivate a Luvian-influenced anaptyxis into the non-existent /spand-/ is beside the point, but he does raise the legitimate question of why, beginning in Middle Hittite, the spell10 For a similar independent interpretation of the HLuvian word and comparison with the Hittite hapax šapa-an-ta-al-la (KBo 31.8+ i 7) see Giusfredi 2010: 123–4.

191

H. Craig Melchert

ing ši-ip-pa-an-t/d°- was introduced and in fact became the dominant orthography. Here the increasing role of Luvian native speakers among the Hittite scribes may well be the responsible factor. The Luvian-speaking scribes surely learned fairly quickly the general Hittite scribal practice of distinguishing intervocalic voiceless from voiced stops by ­VC-CV- versus ­V-CV spellings. It would be entirely natural if they chose to apply this to what seemed the unmotivated exception of ši-pa-an-t/d-°. I stress, however, that this scenario is by no means necessary. Since, I must insist, the word was pronounced /sipánd-/ from the beginning of attested Hittite, a senior scribe could have decided at any time that the exception should be eliminated and a new standard spelling be adopted. A number of changes were made in Hittite spelling practices from Old to New Hittite, and this is merely one of them. I may cite as a parallel for the non-writing of a geminate stop in Old Hittite versus its expression in later manuscripts the example of /tarsikke-/, the older iterative of tar- ‘say’. In Old Script we find only tar-ši-kán-zi and tar-ši-ke-ez-zi in KBo 22.2 Ro 8 and Vo 4, but in Middle Script tar-ši-ik-ke-mi (HKM 46:27) and tar-ši-ik-ke-ši (KUB 14.1 Ro 34), and in New Script copies of Old Hittite texts tar-ši-ik-kán-zi (KBo 3.1 ii 33 and 3.16 iii 14). Whatever the motivation may have been for the introduction of the spelling ši-ip-pa-ant/d°­, the absolutely fixed spelling with initial ši- excludes the reading /spand-/ for Old Hittite, and since there is indeed no way to motivate a voicing of the labial stop, ši-(ip)-pa-an-t/d°- must be interpreted as /sipánd-/, while the rarer variant iš-pa-an-t/d-° stands for regular /ispánd-/. The problem then becomes: how do we account for the existence of these two stems and explain their attested shape and use? The source of the stem išpant- is straightforward: it may continue a PIE root present of the h2e-conjugation *spónd-ei, *spénd-ti ‘libate’, yielding regularly attested išpnti, išpantanzi (Jasanoff 2003: 86) — but see below for an alternative account. An ablauting root present *spénd­, *spd- (Forssman 1994: 102) would also lead to išpant- phonologically, but such a reconstruction is morphologically incompatible with a Hittite i-verb root present. That the i-inflection of išpand- is secondary after šipand- (LIV2: 577) is unlikely. Other Hittite root mi-presents standing beside reduplicated i-presents show no such influence: wēkzi beside wewakki ‘demands’. Forssman (1994: 103) proposed to derive šipand- from a reduplicated stem *spe-spond­, *spe-spd­, assuming a full reduplication of the initial *sp- of the root and differing simplifications leading to Hittite šipand- and Old Latin spepondī. The need to assume a complicated double dissimilation for Hittite whereby the first *p but the second *s was lost has undoubtedly been one of the reasons for the widespread rejection of Forssman’s account. However, there is now a growing consensus that the history of reduplication in IndoEuropean should be understood very differently, namely as an inherited synchronic process whose operation is subject to renewal (whatever theoretical approach one takes to its description): see the extensive argumentation of Keydana 2006, followed by Byrd 2015: 118–21 and others. Furthermore, one should in reconstructing the PIE state of affairs follow the standard procedure of giving most weight to isolated archaisms that cannot easily be motivated as innovations. On this basis, following already Brugmann 1897: 40–41(!), Keydana (2006: 107), Byrd (2015: 120) and others argue on the basis of non-productive forms like Latin present sistō ‘(cause to) stand; stop’, Grk. ἵστημι ‘stand’ plus Avestan hi-štaiti ‘stands’ and OIrish se-scaind ‘jumped’ that the PIE reduplication pattern with roots in initial *sT- was *sV-sT­.11 11 Hittite šiš(a)- ‘order, decide’ may also be a relic reflecting *sí-sh ­ to the root *seh ­ ‘bind’ (thus Kloekhorst 2 2 2008: 758–9; cf. tentatively already Melchert 1984: 153, note 125). For the original stem as šiš(a)- see the MH/MS attestations cited by Kloekhorst and the CHD Š: 450–51.

192

Initial *sp- in Hittite and šip(p)and- ‘to libate’

This means that we may suppose that the PIE reduplicated stem behind Hittite šipandwas *se-spónd­, *se-spd- (also considered as an alternative by Schulze-Thulin 2001: 384). These preforms will in terms of vocalism lead regularly to attested šipnti, šipant/danzi, with regular raising of pretonic short *e to i (see Melchert 1994: 101) and lengthening of the accented short *ó to Hittite ā in the strong stem (spelled plene a few times, as in KBo 17.11 iv 4&14, OH/OS). What remains to be accounted for is the deletion of the second *s of the preform *sespVnd­. Once we regard changes in productive reduplication patterns as reflecting renewal of a synchronic process, there are (at least) two ways to account for the loss of *s in this context. The first may be formulated in terms of pre-Hittite constraints on the syllabification of consonants. Synchronically, an [s] in contact with another consonant at a syllable boundary appears to be treated as ambisyllabic in attested Hittite: note spellings such as ti-iš-ša-kán-zi ‘they (usually) step’ (IBoT 1.36 iv 30) beside usual ti-iš-kán-zi for [tis.skan.tsi] or wa-aš-ša-pa-an ‘garment’ beside wa-aš-pa-an for [was.span] (see Bernabé Pajares 1973: 446–7 and passim; Melchert 1994: 150–52). However, we have compelling reasons to think that at an earlier prestage of Hittite there was a constraint against [s]+stop as a syllable onset. For word-initial position, of course, the evidence is the development of the prothetic i- before *sT­. As argued above, this was undeniably the regular treatment of such initial clusters. The (thus far) unique exception of /spekusta-/ ‘pin’ was “licensed” only by the pressure of preserved /sm-/ with [s] plus labial nasal stop. Addition of the prothetic vowel naturally enabled a prehistoric syllabification *[is.TV­]. Evidence for the same prehistoric constraint on [sT] in medial onsets is furnished by the pattern of anaptyxis in marked imperfectives with the suffix *­s e/o­, where a vowel was inserted between a preceding consonant and the *s or in the case of coronals between the *s and the *k: appiške- ‘take’, akkiške- ‘die’, but taršikke- ‘say’ (see Melchert 2012: 179–80). Once again, the anaptyxis solved the prehistoric synchronic syllabification problem, permitting *[ap.pis.kV­], *[ak.kis.kV­] and *[tar.si.kV­].12 I emphasize that the forms with anaptyxis became underlying representations by the time of attested Hittite, leading by then surely to phonetic realizations [ap.pis.skV­] etc. We may therefore assume that likewise there was a stage at which pre-Hittite (arguably Common Anatolian) *sespnd(V)- could no longer be syllabified as *[se.spn.d(V)­], just as the word-initial *[spó/én.d(V)­] of the nominal stem (DUG)išpanduzzi- ‘libation’ and its derivatives could not be syllabified (likewise in the h2e-present if it existed at this point). In this case, solving the problem in the former by anaptyxis, producing *[se.sV.pn.d(V)­] beside the new [is.pó/én.d(V)­] with prothesis, would have seriously disrupted the formal relationship of words that were in semantic terms transparently related. A simpler alternative solution was to resyllabify *[se.spn.d(V)­] as *[ses.pn.d(V)­]. However, there is now reason to believe that the syllabification *[ses.pn.d(V)­] might itself have been problematic. Zukoff (2014: 272–5) has argued for a context-sensitive version of the well-known Obligatory Contour Principle that prohibits identical adjacent segments. Zukoff proposes that there was also operative in early Indo-European an OCP-SYLLABLE (OCPσ) constraint: “Assign one violation mark * for every syllable that contains identical segments.” 13 If we assume that this constraint also applied at some stage of pre-Hittite (or Common Anatolian), then it would have prohibited the syllabification *[ses.pn.d(V)­], which 12

For the assumption that intervocalic voiceless stops spelled double were geminates that closed the preceding syllable see Melchert 1994: 18 with references and also Kloekhorst 2014: 545–6 (with a different phonological analysis). 13 For an extensive discussion of OCP effects in PIE and its descendants (including but not confined to OCP-σ) see Sandell 2016, who also duly notes (2016: 146) the notorious exceptionality of PIE *ses- ‘sleep’ and its reflexes. 193

H. Craig Melchert

would have been solved by deletion of the s in the syllable coda.14 If loss of the coda consonant led as expected to compensatory lengthening, producing a virtual *[se:.pn.d(V)­], the pretonic long vowel could have been shortened in time to undergo the specific pre-Hittite change of pretonic short *e to i. Compare Hittite ippara- ‘serf’ (or sim.) < *h2ēpor-ó- (Eichner 1973: 72).15 Hittite šipand- may thus be derived by regular phonological developments from a reduplicated stem *se-spónd­, *se-spd­, and I stress again that its absolutely fixed i-vocalism cannot be plausibly explained by any other means. There remains, however, the question of whether such a reduplicated stem is a viable source for the Hittite verb in its attested use. One of the few supporters of Forssman’s original proposal, expresses doubts: “Ist ein altes Zustandsperfekt semantisch sinnvoll?” (Kümmel in LIV2: 577, note 5). Yakubovich (2009: 547) also reasonably protests that there is no discernible functional difference between attested šipand- and išpand- (cf. also Kloekhorst 2008: 406). I myself previously looked in vain for any such contrast in usage. I now believe that such a venture failed because we based our search on false premises. A perfect with the standardly assumed value of an “attained state” hardly fits the usage of the Hittite verb, which is clearly eventive: ‘libate’, secondarily ‘consecrate’ (by pouring a libation over), then by metonymy ‘offer X (to a deity)’ and by syntactic change ‘worship (a deity) with X’: see CHD Š: 384–95. I had supposed that the reduplicated stem belonged to what I regarded as the small class of iterative-durative perfects, such as *we-w(o)rt- ‘roll, revolve’ (on such a meaning for at least some instances of Vedic vavart- see Kümmel 2000: 462ff.). But I could find no clear traces of an iterative-durative or even processual value for šipand­. Jasanoff (forthcoming) has now argued that the “attained state” value of the perfect in Core Indo-European is an innovation and that the classical “perfect” originates in a reduplicated h2e-aorist of the shape *Ce-CóC­, *Ce-CC­, whereas the few “perfects” that show iterative semantics reflect rather reduplicated h2e-presents of the form *Cé-CoC­, *Cé-CC­.16 Hittite wewakk- ‘request’ (repeatedly) and mēma/i- ‘speak’ are direct reflexes of the latter category. By this scenario, *se-spónd­, *se-spd- would have been a reduplicated h2e-aorist and should have referred to the act of libating not as an activity (which would have been expressed by the h2epresent), but as a single telic act.17 If one examines all thirty-plus instances of šipand- in Old Hittite/Old Script, one finds that it is consistently used in such a fashion. It is used to refer to the act of libating once at a particular “station”, such as in front of the window (KBo 17.11+ iv 23) or to the hearth (KBo 17.19 14

One may compare typologically for a similar “repair” the Sanskrit weak perfect stem sed- ‘sit’ < *sé-sd- and more broadly other Sanskrit weak perfect stems of the shape CeC- as well as long-vowel preterite formations in Germanic and Celtic: see Schumacher 2005: 601–5, Zukoff 2014: 274, and Sandell: 2016: esp. 142–3 and 156–7. 15 Zukoff (2015) has now refined his account of Indo-European reduplication patterns in terms of what he labels the POORLY-CUED REPETITION PRINCIPLE: “A CVC sequence containing identical consonants (CαVCα) is dispreferred, due to repetition blindness; it is especially dispreferred if one or both of the consonants lack phonetic cues which are important for the perception of its presence (in contrast to zero) in the speech signal.” For reasons he sets forth, this principle applies especially to the second fricative [s] in a sequence #sVsT­. Since this newer formulation will also handle the case of šipand- < *se-spónd­, I forego extensive discussion here and refer interested readers to Zukoff’s own presentation, available online. 16 While verbs of the latter class have mostly been assimilated to the true “attained state” perfects in the attested languages, Jasanoff stresses that in the oldest Greek their separate origin is still betrayed by a different pluperfect inflection. 17 I do follow LIV : 577 and others, against Jasanoff forthcoming, in supposing that the concrete meaning ‘li2 bate’ of Greek and Hittite is original, from which already in PIE developed the secondary sense ‘pledge, dedicate’ (in the middle ‘pledge, dedicate oneself’). 194

Initial *sp- in Hittite and šip(p)and- ‘to libate’

ii 11). It alone (never išpand­) is used with specification of how many discrete times one performs the act of libating: ‘once’ (KBo 17.11+ iv 33, KUB 43.30 ii 11&15 and often), ‘twice’ (KBo 20.10 i 9), ‘three times’ (KUB 43.30 ii 14), ‘seven times’ (KBo 25.127 ii 25). It alone is attested in the telic sense ‘consecrate’ a sacrificial animal or other object (KBo 17.36+ iii 9 and 17.33+ i 14). Finally, it may be used of worshipping a deity (in the accusative) by libating into a bowl (KBo 25.61 Vo 9). Trying to determine whether the stem išpand- has a synchronically distinct sense and whether its absence in the contexts just cited for šipand- is systematic or merely due to chance is made extremely difficult by the very small number of examples, especially of examples with full context. Aside from the “Ritual for the Royal Couple”, which uses only išpand- in its attested portions (see Otten and Souček 1969: 97), there are a mere handful of other attestations, either in Old Script or later copies. However, the examples in KBo 20 ii 5&6 (OH/OS), where išpanti ‘performs a libation’ occurs in the immediately context of inga ‘bows’ is strongly reminiscent of that of KBo 25.104 ii 12–13 (OH/OS?), where we read LUGAL-uš dKuwaššaš UŠKE[N…] šipanti. Similarly, the phraseology [… ]× 2 ekuzi […u]ppari išpant[i] ‘drinks two […] libates into a bowl’ (KBo 25.51 i 18–19; OH/OS) hardly differs from that of ūppari šipanti (KBo 25.61 Vo 9; OH/OS) cited above. It therefore seems extremely unlikely that the stem išpant- has any different sense synchronically from that of šipant­. Both refer to libating conceived as a single telic act and to the other telic meanings derived from that. By the oldest attested Hittite išpant- survives only as a marginal variant of šipant­. In fact, one may reasonably ask: does the very rare verbal stem išpant- continue a genuine prehistoric present stem at all, or is it merely an analogical creation based on the nominal forms (DUG)išpanduzzi- ‘libation’, DUGišpanduzzi(y)aššar- ‘libation vessel’, and (DUG)išpantuwa- ‘libation vessel’? Of course, if one opts for the latter interpretation, then one must ask in turn what the basis was for the nominal stems, which appear to be deverbative. As to DUGišpanduzzi- (from which DUGišpanduzzi(y)aššar- obviously is further derived), if one looks at the class of Hittite nouns in ­uzzi­, some are indeed undeniably deverbative, formed to synchronically existing verbal stems: e.g., KUŠannanuzzi- ‘(part of a) harness’ < annanu- ‘train’, kuruzzi- ‘cutting tool’ < ku(e)r- ‘cut’. Others, however, appear to be rather deradical, being derived from forms of the respective roots whose existence in pre-Hittite as verbal stems is dubious: e.g., išuzzi- ‘belt, chain’ < *s(e)h2­ ‘bind’ (but all verbal forms are based on iši- < *sh2ei­), tuzzi- ‘camp; army’ < *dh(e)h1­ ‘place’ (whereas the present stem of the verb is dai- with an *­i- suffix).18 The nominal stems DUGišpanduzzi- and DUGišpanduzzi(y)aššarare thus not probative evidence for a genuine pre-Hittite verbal stem išpand­. The stem išpanduzzi- may be a primary derivative from the root *spend­. It is true that (DUG)išpanduwa- is hypostasized from the verbal noun (thus with Carruba 1966: 23, note 35), but precisely in this case there are also a number of spellings as (DUG)šipanduwa- (see CHD Š: 396). In this noun, then, the variant išpanduwa- may be analogical, just as in the other verbal forms. I therefore must conclude that evidence for a pre-Hittite present stem of any kind is less than compelling. A h2e-conjugation present *spónd-ei, *spénd-ti may well have existed, but its existence must be based on other evidence (see Jasanoff 2003: 78 on Greek σπένδω ‘pour, libate’ and Latin spondeō ‘vow’). The fundamentally telic senses of the Hittite verb šipand- are in any case fully compatible with the proposal that it continues a reduplicated h2e-aorist. With due revisions, then, the much maligned derivation suggested by Forssman more than twenty 18 The primary meaning of tuzzi- is ‘camp’, as shown by the derived verb tuzziya- ‘encamp’. One must with Kloekhorst (2008: 908) insist on this etymology of Carruba (1966: 23, note 35). There is no connection with western Indo-European *teutā­.

195

H. Craig Melchert

years ago may be upheld. However, one must not overlook that the functional side of the scenario presented here, following Jasanoff, has implications for Indo-European dialectology that are diametrically opposed to those of Forssman’s original formulation: by the present account Hittite šip(p)and- reflects a PIE reduplicated aorist whose development into an “attained-state” perfect is a common innovation of “Core Indo-European”.

Ilya Yakubovich Philipps-Universität Marburg; [email protected]

Response to C. Melchert *

It is appropriate to begin this response by thanking H. Craig Melchert for submitting the paper under discussion to the Journal of Language Relationship. Given the fact that the main claim of this paper radically contradicts the views expressed earlier by two editors of the journal, Alexei Kassian and Ilya Yakubovich, the publication of this piece in our journal is obviously conducive to resuming the discussion on this controversial topic. I hope that our readers will benefit from comparing different approaches to interpreting Hittite cuneiform spellings. In the first part of the response I will dwell on Melchert’s specific claims pertaining to the Hittite verbal stem špand- ‘to libate’. It is my intention to demonstrate that its analysis offered immediately above is fraught with so many complications and arbitrary assumptions that it cannot be acceptable as a viable hypothesis regardless of the broader considerations that have motivated it. The second part of the response turns to a more general issue of how the Anatolian cuneiform reflects the evolution of consonant clusters in the Hittite language. I have to acknowledge here that Melchert’s new approach is internally consistent and has some advantages over his older views. This prompts me to present an alternative account of how špand- may have evolved within the history of Hittite, which largely accommodates Melchert’s contemporary interpretation of Hittite orthography but strives to avoid the pitfalls of his etymological analysis. * This reply is subject to the usual disclaimers. I am grateful to Alexei Kassian and H. Craig Melchert, whose comments to its first drafts led to the overall improvement of my argumentation, and to Stephen Durnford, who has kindly agreed to improve my style. My work on this piece was conducted within the framework of the project “Digitales philologisch-etymologisches Wörterbuch der altanatolischen Kleinkorpussprachen (RI 1730/7-1)” funded by the Deutsche Forschungsgemeinschaft.

196

1. The readers must first be reminded about the nature of the controversy. The Old Hittite texts display a number of forms that contain the reflexes of the IndoEuropean root *spe/ond- ‘to libate’ (LIV2: 577–578). These forms can be divided into two groups displaying the cuneiform spelling beginning with iš-pa- and ši-pa- respectively. Their distribution in Old Hittite / Old Script texts is illustrated in the Table 1 below, which is taken wholesale from Kassian & Yakubovich 2002: 34. It is easy to see that the the third-person forms of the base verb display the variants beginning with both iš-pa- and ši-pa­, with a preference for the first variant, while the rest of the attested forms show exclusively the spelling iš-pa­. It is worth mentioning that the spelling ši-pa- was generalized for all the finite forms by the Middle Hittite period, but the nominal derivatives išpantuzzi and išpantuzzijaššar retained the spelling iš-pa- throughout the history of Hittite (Yakubovich 2009: 549). The controversy concerns the question whether the forms listed in the Table 1 are ultimately derived from one verbal stem or from two. According to the view of Kassian & Yakubovich 2002, which is also maintained in Yakubovich 2009, the variants iš-pa- and ši-pa- reflect different graphic renderings of the same wordinitial cluster /sp-/, which cannot be unambiguouly represented in cuneiform script. In this we followed a tentative suggestion expressed in Melchert 1994: 31. For Melchert (ibid.), the issue was not fully settled, because he could not think of a plausible reason why the two different graphic conventions were adopted in the instance of the root špand- ‘libate’, but not for rendering the other roots with etymological *sC- clusters, which all consistently adopt the spelling iš-CV­. Kassian and Yakubovich (2002: 34) were bolder in defending the same interpretation, because we thought that we had a solution to this problem. According to the

Response to C. Melchert

Table 1: špand- ‘to libate’ and its derivatives in Old Hittite špand- ‘to libate’ prs. 1 sg. išpantahhi/e: 6× prs. 3 sg. išpā/anti: 8× prs. 3 pl. išpantanzi: 1× prs. 3 sg. išpanzaškizzi: 1×

*šipantahhi/e: not attested šipā/anti: 27× šipantanzi: 7× *šipanzaškizzi: not attested išpantuzzi ‘libation vessel’

nom.-acc. sg. išpantuzzi: 7× dat.-loc. pl. išpantuzziaš: 2×

*šipantuzzi: not attested *šipantuzzias: not attested

išpantuzzijaššar ‘libation vessel’ nom.-acc. sg. išpantuzzijaššar: 11× acc. sg. išpantuzijaššaran: 1× acc. pl. išpantuzzijaššaruš: 1×

hypothesis proposed in Kassian 2000 and elaborated in Yakubovich 2009: 549–549 (with fn. 6), the innovative spelling ši-pa-(a)-an-ti ‘libates’ arose as an instance of graphic disambiguation with iš-pa-an-ti ‘in the night’ and later spread to the other forms belonging to the paradigm of the same verb. The gradual generalization of a spelling pattern from the most frequent form of the paradigm to the rest of it appears straightforward. This solution is cited with approval in Giusfredi 2014: 186–187, who also points out that the disambiguation never spread to the nominal derivatives of špand- ‘to libate’, because they are always accompanied by the determinative DUG ‘vessel’ and thus could not be taken for the derivatives of išpant- ‘night’. At the same time one has to acknowledge that a hypothesis of graphic disambiguation between lexemes in a dead language is normally not amenable to independent verification in view of its irreducible character. It can only be falsified, for example, by demonstrating that the phenomenon is not merely graphic, and / or replaced with a superior account. Quite a different view is entertained in the paper to which I am now responding. It is argued there that only the Old Hittite spellings with iš-pa- reflect the etymological stem *spand­, whereas their counterparts beginning with ši-pa- continue the pre-Hittite reduplicated stem *sispand- < *sipand­. Melchert acknowledges his inability to trace the synchronic difference between the two stems within the paradigm of the finite verb. This prompts him to advance a tentative hypothesis that the variant iš-pa- had originally been restricted to the non-finite forms and only secondarily spread to the finite paradigm in Old Hittite. The reason why the reconstructed stem distribution became skewed in

*šipantuzzijaššar: not attested *šipantuzijaššaran: not attested *šipantuzzijaššaruš: not attested

Old Hittite only to be restored in Middle Hittite remains unclear under such an analysis, even though one must acknowledge that one cannot always predict the direction of analogical change. A more serious flaw of the proposed alternative is that it neither simplifies the account for the spelling šipa-(a)-an-ti ‘libates’ nor increases its value for the theory of writing. Melchert acknowledges that according to Sturtevant’s rule the expected reading of /sipánti/ would be ši-ip-pa-(a)-an-ti, the form that is regular in New Hittite, but rare in Middle Hittite and completely unattested in Old Hittite / Old Script texts. He also concedes that **/sibánti/, the expected reading of OH. ši-pa-(a)-an-ti, cannot be derived from /sipánti/ by known sound laws. Thus Melchert essentially concurs with the observation of Kassian and Yakubovich 2002 that the form ši-pa-(a)-an-ti is graphically irregular. His account for the observed irregularity is, however, different and considerably more generic: Kassian and Yakubovich (2002: 33) and Yakubovich (2009: 547) argue that one cannot interpret the first vowel of the Old Hittite/Old Script spelling ši-pa-an-t/d-° as real, because this could only imply a reading /siband-/, and voicing of the stop in this environment cannot be motivated by any known Hittite sound change. This argument reflects a fundamental methodological fallacy and a profound misunderstanding of how orthographies devised by and for native speakers work. Native speakers know how the words of their language are pronounced and also the grammar that predicts where they will occur, and writing systems (especially those used by a 197

Ilya Yakubovich

small elite) need only give just enough clues for another native speaker reader to successfully identify the word intended. (p. 191) In its application to the Hittite cuneiform, this statement logically implies that Sturtevant’s rule can be randomly violated in each and every case where this does not lead to the confusion of lexemes. Given the far-reaching character of this implication, it is not fully clear to me whether the citation above should be taken literally or perceived as a rhetorical device. At any rate, I stand by the description of Sturtevant’s rule in Hoffner & Melchert 2008: 35, where it is regarded as a consistent pattern. To be sure, it can be violated by occasional simplified spellings, but I am aware of no instances where such violations would be generalized for any frequent form or lexeme. Therefore the exceptional orthography ši-pa-(a)-an-ti remains fully ad hoc under Melchert’s analysis. The final vulnerability of the new hypothesis concerns the way /sipánti/ is derived from the alleged reduplicated formation. Here Melchert begins with the stem *sespo/end- and postulates its subsequent development to *sēpo/end­, which supposedly reflects a universal constraint on the identical segments belonging to the same syllable. No Hittite parallels are, however, cited for such a development, while the forms of the Hittite root še/aš- ‘to sleep’ represent patent counterexamples. The last difficulty is implicitly acknowledged by Melchert (p. 193, fn. 13), but the change *sespo/end- > *sēpo/end- is nevertheless called regular! This is arguably the first occasion in the history of Anatolian studies where optimality-theoretical constraints are invoked not as a metalanguage for the empirically proven sound laws, but rather in order to overrule the available empirical evidence. To illustrate the potential dangers of such a practice it is enough to mention that one of the prominent markedness constraints within the framework of Optimality Theory is the constraint on closed syllables. This constraint came to be top-ranked, for example, in Old Church Slavic, where a number of processes conspired in order to trigger the law of open syllable. Does this suffice to claim that any coda simplification on the morpheme boundary, whether regular or not, can be now licensed for ancient Indo-European languages with reference to the sudden prominence of such a constraint at the point when the respective morphological derivation has taken place? For example, one could use such an assumption in order to argue that Hitt. tēzzi ‘says’ goes back to an earlier *tērzi, a putative singular counterpart of taranzi ‘they say’, while e.g. kuerzi ‘cuts’ reflects a later analogical devel198

opment. I doubt, however, that Melchert or any other mainstream Indo-Europeanist would subscribe to such a radical break with the traditional comparative method. While it is true that reduplications have a particular propensity to periodical renewals due to their iconic character, this has little to do with the assumed change *sespo/end- > *sēpo/end­, which is applied to the preexisting reduplication template according to Melchert’s own analysis. Naturally, if one assumes that the attested Old Hittite forms of špand- ‘to libate’ reflect just one stem, the need for such an irregular development simply disappears. Summing up, I claim that the proposed phonetic interpretation of the alternation between iš-pa- and ši-pain the paradigm of špand- ‘to libate’ is inferior to its graphic interpretation on three independent counts. First, it cannot account for the dynamics of distribution between the two stems. Second, it operates with an ad hoc violation of Sturtevant’s rule. Third, it implies a phonetic scenario that contradicts the known sound laws. The first problem can be regarded as merely complicating the proposed analysis, but problems two and three plainly render it untenable, particularly when taken together. It remains to be seen what the considerations that prompted Melchert to give up his original analysis of the stem ‘to libate’ are. 2. Melchert’s new interpretation of the spelling variation in špand- ‘to libate’ represents a consequence of his second thoughts on the development of initial sC-clusters in the history of Hittite. Melchert’s old view on this topic are tentatively put forward in Melchert 1994: 31–32, while his change of opinion is already clearly expressed in Hoffner & Melchert 2008: 27. Nevertheless, since Melchert proposes a very detailed explication of his new stance, I will generally follow his most recent line of presentation in my further discussion. The development of initial clusters in Hittite was a matter of much controversy in the twentieth century (see references in Melchert 1994: 31, and above p. 187 ff. with ref.). But an important contribution to the debate on the wake of the new millennium consisted of two articles that focus on this precise issue, namely Kavitskaya 2001 and Kassian & Yakubovich 2002. The first paper invokes the theory of syllable structure in order to advocate the view that the spelling iš-CV- for rendering such clusters always reflects phonological reality, thus implicitly taking issue with the stance of Melchert 1994 and anticipating certain assumptions of the present paper by Melchert. Curiously enough, this theoretically informed piece of work is not cited by Melchert above, possibly because Melchert’s own analy-

Response to C. Melchert

sis focuses on the structure of Anatolian cuneiform rather than on cross-linguistic generalizations about syllable structure. The second paper dwells on orthographic issues and argues, following the observations of Melchert 1994, that the spelling iš-CV- for etymological sC-clusters represents a graphic convention. Melchert rejects several claims advanced in Kassian & Yakubovich 2002, naturally grouping some of them together with his own dated views. The logical starting point of Melchert can be formulated as follows. The main graphic indicator for a synchronic consonant cluster is the presence of irregular spelling alternations, such as those characterizing the initial signs of ša/e-me-en-zi ‘withdraws’ or ša/e/ippe/ikkušta‘pin’. In Kassian & Yakubovich 2002, such alternations were taken as instances of schwa insertion followed by schwa-harmony (e.g. [same:ntsi] ~ [seme:ntsi]). This interpretation, however, is not compelling, as pointed in de Vaan 2003: 285 with reference to a similar “harmony” in Mycenaean Greek orthography, which clearly has a graphic explanation.1 Furthermore, the data collected in Kassian & Yakubovich 2002 indicate no statistically significant correlation between the alternations of the ša/e-me-en-zi type and the plene spellings of the type ša-(a)-li-ga ‘touches, defiles’, which are surely indicative of vocalic epenthesis (cf. Kavitskaya 2001: 275, fn. 11). On the methodological level, Kassian & Yakubovich 2002 did use irregular spelling alternations in order to recover consonant clusters in some other instances (e.g. za-aš-ki- / zi-kifor /tske-/, on which see below). Therefore it appears fair to invoke the same principle in the case under discussion. So far the critique of Melchert can be regarded as internally consistent. If ša/e/ippe/ikkušta- and similar alternations reflect scribal uncertainty in dealing with word-initial consonant clusters, then cases like išpant- ‘night’ must reflect something else. Hence the next claim by Melchert: prothesis in iš-CV- clusters is phonetically real. An independent argument in favor of this hypothesis, which is not directly mentioned by Melchert, is the broad agreement between the relevant conventions of

1 To be sure, there is a significant difference between the Hittite and Mycenaean conventions. In Hittite, it is the a-vowel that is usually inserted in writing for rendering the etymological clusters “obstruent+resonant”, except for the cluster *tr­, where e-vowel is inserted (Kassian & Yakubovich 2002: 12–21). At least in some of these cases, the epenthesis is also phonetic, as indicated by occasional plene spellings. In Mycenaean, on the contrary, the “dummy” epenthetic vowel normally replicates the vowel that is pronounced in the relevant syllable, e.g. du-ru- for /dru-/, do-ro- for /dro-/ etc. (Melena 2014: 111–112). Deviations from this practice represent exceptions (Melena 2014: 113).

the Old Assyrian and Hittite cuneiform. Decksen 2007 reviews evidence for the spelling iš-CV- in Anatolian appellatives borrowed into Old Assyrian. Thus Old Ass. išpuruzzinnum (3×) ‘roof batten’ cannot be separated from Hitt. išparuzzi- ‘rafter’, itself possibly a derivative of Hitt. išpar- ‘to spread, strew’. Old Ass. išiulum (1×, perhaps a commodity) may refer to a physical object used for binding rather than a written treaty, but this is hardly a compelling reason to doubt its connection with Hitt. išāi- /išija- ‘to bind’, the base of Hitt. išiul- ‘treaty’. Finally, given that nasals before stops are not reflected in writing in Old Assyrian orthography, Old Ass. išpadalum (3×, a commodity) can be either a derivative of išpant- ‘night’, or perhaps that of the root špand- ‘to libate’, which is treated in this paper.2 Kassian & Yakubovich 2002 and Kloekhorst 2008 concur in reconstructing consonant clusters in the roots under discussion. The root etymologies offered in this paragraph are admittedly speculative, especially given the fact that in two of the three cases we cannot determine the semantics of the nouns involved. But if scholars are right in seeing here Hittite loanwords of Indo-European origin, structural considerations would strongly plead for reconstructing *sC- in išpuruzzinnum, išiulum, and išpadalum. The morphemes išpur­, iši­, and išpad­, all segmentable with a reasonable degree of confidence, are unlikely to reflect Indo-European disyllabic roots beginning with i, which vindicates its status as the prothetic vowel. One may argue that two largely independent cuneiform orhographies were unlikely to adopt the same default device of i-prothesis for rendering word-initial etymological sC- clusters unless there was some phonetic substance behind it. The data above need to be reconciled with the synchronic alternation between word-initial iš-pu- and šupu- in the Old Assyrian transliteration of Hittite personal names, which were adduced in Yakubovich 2009: 546. Melchert (p. 189) treats the cases of Šu-puda-a-šu vs. Iš-pu-da-a-šu, Šu-pu-na-a-šu vs. Iš-pu-naa-šu, and Šu-pu-nu-ma-an vs. Iš-pu-nu-ma-an as recurrent instances of genuine phonetic variation. Although this claim derives a degree of support from the over2 The first interpretation is maintained in CAD (I/J): 257a, where the meaning ‘lodging’ is assigned to the noun under discussion, since it is mentioned together with the donkey food. The editors of the CAD were, however, familiar only with one occurrence of išpadalum, whereas its two additional occurrences apparently tip the scales in favour of its interpretation as an object (Dercksen 2007: 36). Can it be some sort of libation vessel, or alternatively a chamber pot (vase de nuit)? Cf. Luv. (CAELUM.*286.x)sà-pa-tara/i-i-sa ‘libation-priest’ and its discussion in Yakubovich 2009: 555–556 vs. Melchert, p. 191 above.

199

Ilya Yakubovich

whelming lexical distribution of the two variants in later Hittite (see below), the data above demonstrate that in the Colony period we are still dealing with free variation, which in turn strongly suggests that this variation was subphonemic. What it means in practice is that the Hittite speakers of the Colony period targeted the phonemic representation /sp-/, and were possibly even able to render it accurately in thorough pronunciation, but optionally implemented either prothesis or epenthesis in spoken forms, perhaps depending on personal idiolects. The only logical alternative to the proposed solution would be to assume that the Assyrian scribes encountered two different Hittite dialects, which were characterized by phonological prothesis and phonological epenthesis respectively, whereas the later dialect of Hattusa represents a sort of koine that drew upon both of them. In the absence of independent evidence for such dialectal divisions, the hypothesis of free subphonemic variation must be preferred as more economical. In a sense, this is the same kind of logic that prompts Melchert to accept free graphic variation in ša/e/ippe/ikkušta- ‘pin’ and similar cases, as opposed to postulating unattested Hittite dialects. Melchert plausibly hypothesises that the Hittite *spclusters represented an arena where two different strategies of breaking *sC- clusters were in competion with each other. One was the i-prothesis, typical of the “s+stop” clusters, the other one was the u-epenthesis, which characterized clusters “s+labial” (or perhaps only those of them that had /u/ in the first syllable). But if one assumes that both strategies were allophonic in a particular environment in the Colony period, the simplest solution is to assume that they were always allophonic at the same historical period. In other words, the source of Old Assyrian išiulum was phonetically [isxiu:l], or something similar, but phonologically /sxiúl/. Naturally, the Hittite loanwords into Old Assyrian reflect the Akkadian phonotactics and therefore the prothetic vowel must have acquired there the phonological status. They also appear to have generalized i-prothesis before *sp- at the expense of u-epenthesis, if the available occurrences of išpuruzzinnum and išpadallum have enough probative force. So much for the situation in the Colony period (20– 18th centuries BC). Moving to the Old Hittite / Old Script corpus (15th century BC), one can observe the ongoing lexicalization of different processes affecting the etymological *sp- clusters. If one follows Melchert’s new phonetic interpretation, one encounters here numerous instances of stable i-epenthesis, e.g. išpant- ‘night’, stable preservation of the original clus200

ter in ša/e/ipe/ikkušta- ‘pin, needle’, and overwhelming u-epenthesis in šuppištuwara- ‘decorated (vel sim.)’. In phonological terms, this situation can be, in principle, interpreted in two different ways. On the one hand, it is possible to argue that we observe here an emerging orthographic convention, which manifests itself through the selection of one phonetic variant per lexeme merely for purposes of writing. According to such an approach, the treatment of clusters in Old Hittite orthography would not be indicative of the actual evolution of language. On the other hand, the standardization of lexical representations may reflect the development of spoken Hittite, in which case one has to conclude that prothesis and epenthesis were well on the way to acquiring phonological status by the 15th century BC. Since there is no independent evidence for the subphonemic character of these processes in later Hittite, in this reply I will pursue the second solution, which also appears closer to Melchert’s own views. One must, however, stress that the phonological prothesis and epenthesis discussed here do not represent mechanical consequences of universal constraints on syllable structure, contrary to what is asserted in Kavitskaya 2001. On the one hand, the diverse reflexes of the etymological sp-clusters strongly suggest that both phonological processes spread by way of lexical diffusion. On the other hand, as shown in Kassian & Yakubovich 2002, there is evidence for even more complicated initial clusters, which are nonetheless synchronically reflected in Old Hittite orthography. The best example is the verbal stem za-aš-ki- alternating with zi-ik-ki- and zi-ki- in the meaning ‘to put (around)’ (Kassian 2002: 136, cf. Yates 2016: 169 fn. 16), the imperfective derived from dāi-/tiya- ‘to put, place’ which can only represent /tske-/. Furthermore, there is enough morphological evidence to argue that zaškaraiš ‘anus’ and zasai- ‘dream’ synchronically contain the clusters /tsk-/ and /tsx-/ respectively (cf. Kloekhorst 2008: 700, 875, Hoffner & Melchert 2008: 47). One needs a vivid imagination in order to build up a hierarchy of universal syllabic constrains that proscribes, for example, word-initial /sk-/ but accommodates word-initial /tsk-/.3 3 This is not to deny the hypothesis that the universal constraints were quietly at work behind the scene as the evolution of Hittite clusters took its particular course. But one is unlikely to acquire a reputation like that of Sherlock Holmes if one begins with invoking the fallen nature of human beings (or the inherent injustice of capitalism) as a motivation for a particular crime. On a more positive note, it is worth pointing out that the ProtoAnatolian word-initial initial *sC- clusters appear to have received differential treatment not only in Hittite but also in Lu-

Response to C. Melchert

It is under the prism of this observation that one has to approach the development of the verb špand- ‘to libate’ in the history of Hittite. If different strategies of cluster simplification spreading by way of diffusion were competing for the etymological sp-clusters in Old Hittite, it is perfectly possible that none of them had yet been generalized in pronunciation for certain lexical items. This is, in fact, more or less what is argued by Melchert in the instance of šuppištuwara­, which is once attested in the shape iš-piš-du-wa-ra(KUB 42.64 Rev. 2). The only reason that appears to preclude Melchert from extending the same type of explanation to the variation between iš-pa- and ši-pa- in špand- is that the strategy of i-epenthesis appears to be otherwise unattested with the etymological sp-clusters. Nevertheless, i-epenthesis has been claimed for other Hittite clusters involving a combination “s+stop”. Thus Kloekhorst (2008: 808) plausibly argues that /tské/á-/ ‘to put around’ began to develop epenthesis already in Old Hittite, as the spelling variant zi-ik-ki­, to become standard in the later period, would appear to suggest. One also encounters 1sg.prs tar-ši-ik-ki-mi, whose stem reflects the imperfective of tarn(a)- ‘to let (off)’, in the Old Hittite / Old Script corpus (Kassian & Yakubovich 2002: 34). In the later period epenthesis of the same type becomes common in other imperfective forms formed from roots ending in coronal stops, e.g. az-zi-ik-ki- /atsiki-/ from ad- ‘to eat’, ar-ši-ik-ki- /arsiki-/ from arr- ‘to wash’.4 But the stems ending in labial and velar stops implemented a different strategy of attachvian. As Melchert (p. 190–191) justly points out, we have sufficient evidence for Luv. */st/ > /t/, but not for the analogous development in clusters containing velar stops. In fact, Rieken (2010: 657) has plausibly argued that Luv. *sk evolved into [ʃk] in the verb sà-ka-ta-li-sà- [ʃkantaliʃ:a­] ‘to provide with decorations, make shine’. Rieken’s interpretation of the Anatolian hieroglyph as a designated syllabogram for rendering the sound [ʃ] is also conducive to taking Luv. (CAELUM.*286.x)sà-pa-tara/i-i-sa ‘libation-priest’ as [ʃpantaris], or something similar. Note, however, that a different development can be observed in Luv. parri(ya)- ‘to spread’ vs. išpar- ‘to spread, strew’ (Melchert 2014: 504) and Luv. part(a/i)- ‘leg (of animal)’ vs. Hitt. išpart- ‘to jump, escape’ (Oettinger 2015: 271–272). Therefore I continue to believe that Luv. (CAELUM.*286.x)sà-pa-tara/i-i-sa represents a loanword from Hittite. 4 Note, however, that even for the Middle Hittite period one can still confidently reconstruct the (optional) lack of epenthesis between the Hittite verbal roots ending in coronal consonants and the imperfective sk-suffix. Cf. such forms as az-za-ki-tin HBM 17 Rs. 43 (MS), ši-pa-an-za-kán-du KUB 40.56 + KUB 31.88+ Rs III 7, 12 (MS). See Kassian & Yakubovich 2002: 37–38 for additional synchronic evidence from Old Hittite. The claim that “there are examples to show that prehistorically there was epenthesis in all sequences of VC-sé/ó- except those in Vs-sé/ó-” (Melchert 2012: 179) is not illustrated with empirical data and therefore can be disregarded for the time being.

ing the imperfective suffix /-ské/á-/, e.g. ša-an-i-iš-kifrom šan- ‘to seek’, 3pl.prs ap-pí-iš-kán-zi from epp- ‘to seize’ (see Kassian & Yakubovich 2002: 33–37 for more examples). It is remarkable that Darya Kavitskaya, who otherwise frequently argues for direct application of phonological constraints, essentially accepts here a spread by diffusion. She claims that “[a]fter the zikke- form was created, one can hypothesize that the analogical extension of this form to other dental stems took place” (Kavitskaya 2001: 283).5 If the epenthesis in /tské/á-/ could influence the epenthesis in /arské/á-/, there are no reasons to a priori exclude the hypothesis that the same process affected the stem /spánd-/. To be sure, this is a non-trivial claim, because it extends the diffusion of i-epenthesis beyond the morphological domain for which it has been demonstrated, but its additional target is an isolated lexeme. One can, however, point out that the uepenthesis in the etymological sp-clusters likewise appears to be restricted to šuppištuwara- ‘decorated (vel sim.) and šuppištuwara- ‘decoration (vel sim.)’. A possible explanation for the rarity of the two strategies is that the productive process of i-prothesis encroached upon both of them within the domain of word-initial clusters “s+stop”. In the instance of u-epenthesis, the onomastics of the Colony period is conducive to reconstructing its productive character within a limited domain of sp(u)-clusters. It is therefore perfectly possible, although not provable, that certain additional clusters “s+stop” also exhibited optional i-epenthesis before the cuneiform was adapted for writing Hittite.6

5 The most recent brief discussion of i-epenthesis in Hittite imperfectives known to me, namely Yates 2016: 169–170, strives to account for it within the framework of the Optimality Theory. This discussion, however, does not go quite to the heart of the matter, because it fails to refer to the faithfulness constraint(s) that interact with the Sonority Sequencing Principle. In my opinion, of utmost relevance here is the No Breaking constraint, which prohibits splitting the phonological units of the input representation. As already pointed out in Kassian & Yakubovich 2002: 43, albeit in different terms, the difference between the derivations /apskV-/ → [ap:iskV­] and /atskV-/ → [ats:ik:V­] lies in the fact that /ts/ is a Hittite phoneme, whereas /ps/ is not. The derivation /atsk-/ → [ats:ik:­] satisfies both the Sonority Sequencing Principle and No Breaking constraint at the cost of violating a lower-ranking principle “align epenthesis with morpheme boundaries”. Such an explanation may not, however, be applied to the case of /arskV-/ → [arsik:V­] (as opposed to [ar:iskV­]) and similar cases, which must, therefore, be explained as an imitation of /atskV-/ → [ats:ik:V­] and similar cases. Since the process under discussion involves a proportion between the underlying ad phonetic representations, it is more appropriate to define it as diffusion of epenthesis rather than analogy. 6 The change in the phonetic treatment of *sC-clusters finds a typological parallel in the history of Persian. Thus it is usually

201

Ilya Yakubovich

Naturally, we would have to assume that at the point when i-prothesis, i-epenthesis, and u-epenthesis had been in competition with each another, all the three processes had been subphonemic. Now it is possible to compare the predictions of my new hypothesis with those of Melchert. I see the variation between the spellings iš-pa- and ši-pa- in the forms of špand- ‘to libate’ as a vestige of free allophonic alternation, of a kind that I also reconstruct behind spelling variations Iš-pu-da-a-šu and Šu-pu-da-a-šu in Old Assyrian. For Melchert, the forms išpand- and šipand- reflect two different stems, so the opposition between them must be phonological. I submit that šipa-(a)-an-ti and similar spellings provide a straightforward argument for preferring my analysis. The seeming violation of Sturtevant’s rule in this form, dismissed by Melchert as a random phenomenon, indicates that the phonological representation of the root was still /spand-/ in Old Hittite. It probably became /sipand-/ in the Middle Hittite period, after the phonetic variant [ispa:nd­] came out of use in finite forms, although the conservative scribal tradition retained the spelling ši-pa-(a)-an-ti for a while. Eventually, however, it was replaced with the predictable šiip-pa-(a)-an-ti, which again fully conformed to Sturtevant’s rule. The likely sociolinguistic reasons for this orthographic reform were discussed in Yakubovich 2009, and I hope that the assumption of a real phonetic epenthesis can only make this account more credible. Two additional advantages of the proposed account over the reduplication hypothesis of Melchert consist in avoiding synchronic suppletion and irreguassumed that the default strategy in processing the Iranian lexemes was epenthesis, as in Pers. setāre ‘star’, but the recent loanwords undergo prothesis, as in Pers. estudyo ‘studio’ (cf. Windfuhr & Perry 2009: 428). Note, however, that even today some Persian native speakers implement a combination of phonetic prothesis and epenthesis while learning the pronunciation of English clusters “s + stop” (Jabbari 2011: 242, Table 2).

lar dissimilation *sespo/end- > *sēpo/end- (compare the previous section). At the same time, the hypothesis of i-epenthesis comes at a considerable price when compared with the graphic disambiguation hypothesis, which was advocated in Kassian & Yakubovich 2002. Beside the necessity of assuming the arbitrary spread of iepenthesis from [tsike:/a:­] to [sipa:nd­], one has to reckon with the loss of direct motivation for the distribution of graphic variants in the Old Hittite paradigm of špand- ‘to libate’. To be sure, a broad explanatory account still remains possible. If the phonetic process of i-epenthesis were spreading by way of lexical diffusion before the i-prothesis was generalized across the board, one might argue that it initially affected the 3sg form [sipa:ndi] in conformity with the general tendency of diffusional sound changes to target first the most frequent forms [Labov 1994: 483]. The subsequent spread from 3sg to 3pl, but not to 1sg, stays within the pool of trivial analogical patterns. But the assumption of graphic disambiguation between šipa-an-ti ‘to libate’ and iš-pa-an-ti ‘at night’ would have an advantage of immediately restricting its scope to the specific form where it happens to be most frequently observed. On the other hand, the scenario of Kassian & Yakubovich 2002 complicates the account for the New Hittite spelling ši-ip-pa-(a)-an-ti and is rendered more problematic by new suggestive evidence for the phonetic character of i-prothesis, as argued earlier in this section. Summing up, the accounts in terms of graphic disambiguation and phonetic epenthesis remain viable alternatives, the selection between which will ultimately depend on the broader question of what happened to etymological sC-clusters in Hittite. I am now leaning toward the phonetic explanation, but I do not consider the issue fully settled. But whichever of these two solutions one prefers, there is no need to assume that the variants išpā/ant- and šipā/ant- historically reflect two different stems.

References Bernabé Pajares, Alberto. 1973. La geminación de s y sonantes en hetita. Revista española de linguística 3: 415–56. Brugmann, Karl. 1897–1916. Grundriß der Vergleichenden Grammatik der Indogermanischen Sprachen. Strassburg: Trübner. Byrd, Andrew. 2010. The Indo-European syllable. Leiden/Boston: Brill. Carruba, Onofrio. 1966. Das Beschwörungsritual für die Göttin Wišurijanza. Wiesbaden: Harrassowitz. CAD = The Assyrian Dictionary of the Oriental Institute of the University of Chicago. Chicago: The Oriental Institute, 1956–. CHD = The Hittite Dictionary of the Oriental Institute of the University of Chicago. Chicago: The Oriental Institute, 1989–.

202

Response to C. Melchert

Dercksen, Jan G. 2007. On Anatolian Loanwords in Akkadian Texts from Kültepe. ZA 97: 26–46. Eichner, Heiner. 1973. Die Etymologie von heth. mēhur. MSS 31: 53–107. Eichner, Heiner. 1975. Die Vorgeschichte des hethitischen Verbalsystems. In: Helmut Rix (ed.). Flexion und Wortbildung: Aken der V. Fachtagung der Indogermanischen Gesellschaft, Regensburg, 9.–14. September 1973. Wiesbaden: Reichert: 71–103. Forssman, Bernhard. 1994. Zu hethitisch šipand- and išpand­. In Jens E. Rasmussen (ed.). In honorem Holger Pedersen: Kolloquium der Indogermanischen Gesellschaft vom 26. bis 28. März 1993 in Kopenhagen. Wiesbaden: Reichert: 93– 106. Giusfredi, Federico. 2010. Sources for socio-economic history of the Neo-Hittite states. Heidelberg: Winter. Giusfredi, Federico. 2014. I limiti della regolaritàgrafematica: alcuni esempi dall’ anatolico. Kadmos 53(1–2): 185–191. Hajnal, Ivo. 1995. Der lykische Vokalismus. Graz: Leykam. Hoffner, Harry A. Jr., H. Craig Melchert. 2008. A grammar of the Hittite language. Part 1: Reference grammar. Winona Lake: Eisenbrauns. Jabbari, Ali A. 2011. Persian learners’ syllabification of English consonant clusters. International Journal of English Linguistics 1/1: 236–246. Jasanoff, Jay. 2003. Hittite and the Indo-European verb. Oxford/New York: Oxford University Press. Jasanoff, Jay. forthcoming. What happened to the perfect in Hittite? A contribution to the theory of the h2econjugation. In Elisabeth Rieken (ed.). 100 Jahre Entzifferung des Hethitischen. Morphosyntaktische Kategorien in Sprachgeschichte und Forschung. Wiesbaden: Reichert. Kassian, Alexei. 2000. Khettskoje ispant- ~ sipant- ‘zhertvovatj’: k foneticheskoj interpretacii. Colloquia Classica and Indogermanica / Klassicheskaja filologija i indojevropejskoje jazykoznanije II: 37–39. Kassian, Alexei, Ilya Yakubovich. 2002. The reflexes of Indo-European initial clusters in Hittite. In Vitalij Shevoroshkin, Paul Sidwell (eds.). Anatolian languages. Canberra: Association for the History of Language: 10–48. Kavitskaya, Darya. 2001. Hittite vowel epenthesis and the sonority hierarchy. Diachronica 18/2: 267–299. Keydana, Götz. Die Indogermanische Perfektreduplikation. Folia Linguistica Historica 22: 61–116. Kimball, Sara. 1999. Hittite historical phonology. Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck. Kloekhorst, Alwin. 2008. Etymological dictionary of the hittite inherited lexicon. Leiden/Boston: Brill. Kloekhorst, Alwin. 2014. Accent in Hittite: A study in plene spelling, consonant gradation, clitics, and metrics. Wiesbaden: Harrassowitz. Knobloch, Johannes. 1956. Hethitische Etymologien. In: Kurt Schubert et al. (eds.). Vorderasiatische Studien. Festschrift für Viktor Christian zum 70. Geburtstag. Vienna: 66–8. Kronasser, Heinz. 1966. Etymologie der hethitischen Sprache. Band 1. Wiesbaden: Harrassowitz. Kümmel, Martin. 2000. Das Perfekt im Indoiranischen. Wiesbaden: Reichert. Labov, William. 1994. Principles of linguistic change: Internal factors. Oxford: Blackwell. LIV2 = Kümmel, Martin, Helmut Rix (eds.). 2001. Lexikon der indogermanischen Verben. Die Wurzeln und ihre Primärstammbildungen. 2. Auflage. Wiesbaden: Reichert. Melchert, H. Craig. 1984. Studies in Hittite historical phonology. Göttingen: Vandenhoeck and Ruprecht. Melchert, H. Craig. 1994. Anatolian historical phonology. Amsterdam/Atlanta: Rodopi. Melchert, H. Craig. 2004. Hittite nominal stems in ­anzan­. In: Eva Tichy, Dagmar S. Wodtko, Britta Irslinger (eds.). Indogermanisches Nomen. Derivation, Flexion und Ablaut. Bremen: Hempen: 129–39. Melchert, H. Craig. 2012. Hittite “heteroclite” s-stems. In: Adam I. Cooper, Jeremy Rau, Michael Weiss (eds.). Multi nominis grammaticus: Studies in Classical and Indo-European linguistics in honor of Alan J. Nussbaum on the occasion of his sixty-fifth birthday. Boston/Ann Arbor: Beech Stave: 175–84. Melchert, H. Craig. 2014. Hittite išpar- ‘to spread out’ and išparre/a- ‘to kick’. In Piotr Taracha (ed.). Proceedings of the Eighth International Congress of Hittitology. Warsaw: Agade: 499–506. Melena, José L. 2014. Mycenaean Writing. In Yves Duoux, Anna Morpurgo-Davies (eds.). A Companion to Linear B: Mycenaean Texts and their World. Volume 3. Louvain-la-neuve: Peeters: 1–186. Neu, Erich. 1970. Ein althethitisches Gewitterritual. Wiesbaden: Harrassowitz. Neu, Erich. 1996. Das hurritische Epos der Freilassung I. Untersuchungen zu einem hurritisch-hethitischen Textensemble aus attuša. Wiesbaden: Harrassowitz. Neumann, Günter. 1987. Review of Gary M. Beckman, Hittite birth rituals (Wiesbaden: Harrassowitz, 1983). IF 92: 279–83. 203

Ilya Yakubovich

Nussbaum, Alan. 2014. The PIE proprietor and his goods. In: H. Craig Melchert, Elisabeth Rieken, and Thomas Steer (eds.), Munus amicitiae Norbert Oettinger a collegis et amicis dicatum, 228–54. Boston/Ann Arbor: Beech Stave. Oettinger, Norbert. 1979. Die Stammbildung des hethitischen Verbums. Nürnberg: Carl. Oettinger, Norbert. 2015. “Hethitisch” partipartiske- ‘laufen’ (*sperdh­) und mutmutali- ‘Schweinewühlplatz’. Münchener Studien zur Sprachwissenschaft 69/2: 269–279. Otten, Heinrich. 1961. Eine Beschwörung der Unterirdischen aus Boğazköy. ZA 54: 114–57. Otten, Heinrich, and Vladimir Souček. 1969. Ein althethitisches Ritual für das Königpaar. Wiesbaden: Harrassowitz. Poetto, Massimo. 1986. Eteo (URUDU)ša/epikkušta­. Sprache 32: 52–3. Rieken, Elisabeth. 1999. Untersuchungen zur nominalen Stammbildung des Hethitischen. Wiesbaden: Harrassowitz. Rieken, Elisabeth. 2000. Hethitisch šumuma­. HS 113: 171–5. Rieken, Elisabeth. 2002. Ein Lautgesetz und der Obliquusstamm des urindogermanischen Personalpronomens der 1. und 2. Person Plural. In: Matthias Fritz, Susanne Zeilfelder (eds.). Novalis. Festschrift für Günter Neumann zum 80. Geburtstag. Graz: Leykam: 407–16. Rieken, Elisabeth. 2010. Das Zeichen in Hieroglyphen-Luwischen. Aygül Süel (ed.). Acts of the VIIth International Congress of Hittitology. Ankara: Anıt: 651–660. Sandell, Ryan. 2016. Obligatory contour principle effects in Indo-European phonology: Statistical evidence and the morphology-phonology interface. In: Stephanie W. Jamison, H. Craig Melchert, and Brent Vine (eds.). Proceedings of the 26th Annual UCLA Indo-European Conference. Bremen: Hempen: 141–60. Schulze-Thulin, Britta. 2001. Fortsetzer urindogermanischer ­o-éye/o-Kausativa/ Iterativa im Hethitischen. In: Onofrio Carruba and Wolfgang Meid (eds.). Anatolisch und Indogermanisch/Anatolico e indoeuropeo. Innsbruck: Institut für Sprachen und Literaturen der Universität Innsbruck: 381–93. Schumacher, Stefan. 2005. ‘Langvokalische Perfekta’ in indogermanischen Einzelsprachen und ihr grundsprachlicher Hintergrund. In Gerhard Meiser, Olav Hackstein (eds.). Sprachkontakt und Sprachwandel: Akten der XI. Fachtagung der Indogermanischen Gesellschaft, 17–23. September 2000, Halle an der Salle. Wiesbaden: Reichert: 591‒626. Tischler, Johann. 2006. Hethitisches Etymologisches Glossar. Teil II/2. Lfg. 14 S/2. Innsbruck: Institut für Sprachen und Literaturen der Universität Innsbruck. de Vaan, Michiel. 2003. Review of Proceedings of the 12th Annual UCLA Indo-European conference, Martin E. Huld et al. (ed.) [Washington, DC: Institute for the Study of Man]. Amsterdamer Beiträge zur älteren Germanistik 58: 283–288. Watkins, Calvert. 1975. La désignation indo-européenne du ‘tabou’. In Julia Kristeva, Jean-Claude Milner, Nicolas Ruwet (eds.). Langue, discours, société. Pour Émile Benveniste. Paris: Éditions du seuil: 208–14. Weiss, Michael. 2010. Language and ritual in Sabellic Italy: The ritual complex of the third and fourth Tabulae Iguvinae. Leiden/Boston: Brill. Windfuhr, Gernot, John R. Perry. 2009. Persian and Tajik. Gernot Windfuhr (ed.). The Iranian languages. London: Routledge: 416–544. Yakubovich, Ilya. 2009. Anaptyxis in Hitt. *spand- ‘to libate’: One more case of Luvian influence on New Hittite. In N. A. Bondarko, N. N. Kazanski (eds.). Indoevropeyskoe yazykoznanie i klassicheskaya filologiya. XIII. Chteniya pamyati I.M. Tronskogo. St. Petersburg: Nauka: 545–57. Yakubovich, Ilya. 2010a. Review of Hoffner and Melchert 2008. Bibliotheca Orientalis 67: 147–54. Yakubovich, Ilya. 2010b. Sociolinguistics of the Luvian Language. Leiden/Boston: Brill. Yates, Anthony D. 2016. Left but not leftmost? Interactions between epenthesis and ictus assignment in Anatolian. Stephanie D. Jamison et al (eds.). Proceedings of the 26th Annual UCLA Indo-European Conference. Bremen: Hempen: 161–178. Zukoff, Sam. 2014. On the origins of Attic reduplication. In: Stephanie W. Jamison, H. Craig Melchert, Brent Vine (eds.). Proceedings of the 25th UCLA Indo-European Conference. Bremen: Hempen: 257–78. Zukoff, Sam. 2015. Poorly-cued repetition avoidance in Indo-European reduplication. Paper presented 9 January at the 89th Annual Meeting of the Linguistic Society of America, Portland, Oregon. https://www.academia. edu/ 10119026/ (accessed 13 May 2016).

204

Response to C. Melchert

Крейг Мелчерт. Начальный кластер *sp- в хеттском языке и глагол šip(p)and- ‘жертвовать’. Статья посвящена механизму развития из праиндоевропейского источника хеттской формы ši(p)pand- ʽсовершать возлияниеʼ. Эта тема остается достаточно противоречивой ввиду того, что от решения данного вопроса существенно зависит не только реконструкция развития начальных сочетаний вида «свистящий + смычный» в хеттском языке, но и определение статуса глагольной категории «перфекта» в анатолийских языках — были ли формы перфекта (которые в древнейших неанатолийских и.-е. языках выражали значение достижения того или иного состояния) унаследованы и затем утрачены в анатолийских языках, или же их следует считать, в рамках «индо-хеттской» гипотезы, общей инновацией на уровне индоевропейского «ядра»? Попытка вывести форму ši(p)pand- из редуплицированного и.-е. перфекта *s(p)e-spónd- в свое время была справедливо отвергнута по целому ряду формальных и функциональных причин; однако, учитывая достигнутый прогресс в изучении рефлексов и.-е. *sp- в хеттском, а также ряд новейших гипотез относительно фонологической природы редупликации и ее роли в и.-е. глагольной морфологии, мы находим веские основания вновь вернуться к этому вопросу. Ключевые слова: hi-спряжение, индохеттская гипотеза, праиндоевропейский перфект, редупликация.

205

Reports / Хроника Galina Sim Institute of Linguistics of the Russian Academy of Sciences (Moscow); [email protected]

Towards Proto-Niger-Congo: Comparison and Reconstruction, Paris, LLACAN, September 1–3, 2016

The 2nd Niger-Congo Congress was held by the research unit “Languages and cultures of Sub-Saharan Africa” (LLACAN) in Paris on September 1–3, 2016 as a follow-up to the first Congress, also held in Paris four years earlier. Despite the fact that J. Greenberg’s original hypothesis is now more than fifty years old, the genetic reality of the Proto-Niger-Congo phylum continues to have a somewhat hypothetical status, and thus, the main stimulus for organizing the previous Congress was a rather ambitious idea to make a significant advance in the reconstruction of Proto-Niger-Congo within a foreseeable timespan through the combined efforts of Africanists from different parts of the world. At the conclusion of the first Congress it was agreed that such events should be held on a regular basis every three or four years. Another outcome of scientific collaboration during and after the congress would be a collective monograph (provisionally titled “Comparative-historical studies in East Benue-Congo” and edited by John Watters) that will be released in the near future. The 2nd Congress was opened by Mark van de Velde, present director of LLACAN. The initial sessions concerned the Niger-Congo phylum as a whole, and the rest consisted of presentations on individual families within Niger-Congo, more or less grouped together by the degree of genetic affiliation between the languages concerned. Addressed issues included comparative and historical studies as such, as well as typological studies without any major conclusions on diachrony. Historical presentations referred to various language levels: segmental phonology, tonology, grammatical and semantic reconstructions. Regrettably, several planned participants were unable to attend the event, but all submitted abstracts and other materials remain available online at the website of the Congress (http://llacan.vjf.cnrs.fr/nigercongo2/index.html). Generally following the chronological order, first, we will briefly describe the presentations that concentrated on specific issues of the Niger-Congo macrofamily, and then outline the various reports on particular branches of Niger-Congo.

One of the most exciting disputes concerned the possible stem/root structure in Proto-Niger-Congo. Two polemical proceedings that succeeded one another represented drastically different points of view. Roger Blench (McDonald Institute for Archaeological Research / University of Cambridge) developed Kay Williamson’s original suspicion, based on Ijoid data, that the canonic Proto-Niger-Congo root structure was possibly not disyllabic (according to Blench, this point of view may be suspected of a “Bantu-centric” bias), but trisyllabic CVCVCV. These originally triconsonantal roots subsequently underwent erosion in most branches, although Ijoid and Dogon languages (as the first Niger-Congo outliers) still show evidence of the initial state. Another argument in support of this is that the three-syllable root structure provides an explanation for front/back vowel alternation between the Niger-Congo subfamilies. An alternative view was presented by Konstantin Pozdniakov (INALCO — LLACAN). Using statistical tools, he came to the conclusion that the most common Proto-Niger-Congo root structure was monosyllabic. In order to identify possible later derivational strata, calculations were carried out separately for verb and noun stems. Taking into consideration all the meso-level language families, the speaker successively addressed three questions: the nature of the initial and final phonemes and the number of syllables, with the consolidated results leading to suggest a monosyllabic biconsonantal CVC structure. In their joint study Dmitry Idiatov and Mark Van de Velde (LLACAN — CNRS) determined to what extent it is justified to postulate labial-velar stops in Proto-Niger-Congo upon analyzing the lexical frequency of labial-velars in languages of Northern SubSaharan Africa. After putting the frequencies on the map, it was discovered that they constituted two easily recognizable high labial-velar frequency clusters (Coastal West Africa and Central Africa) and perhaps one less prominent cluster (South-Eastern Mali and South-Western Burkina Faso). Taking into account the

Journal of Language Relationship • Вопросы языкового родства • 14/3 (2016) • Pp. 207—210

Reports / Хроника

typological rarity of these consonants, the areas seem likely to be pockets of retention of the original state rather than zones where independent innovation should be suggested. On the other hand, since labialvelar phonemes are atypical for the basic lexicon and tend to be used in the expressive part of the vocabulary, such as ideophones or property words, and also because their distribution is mostly restricted to steminitial position, it was assumed that it was unlikely for Proto-Niger-Congo to have labial-velars. The main trigger for the emergence and spread of these phonemes was C-emphasis prosody, partial manifestation of which is stem-initial consonant lengthening and strengthening. The authors also cautioned against properly unfounded reconstruction of labial-velars in intermediate language units. Larry Hyman (University of California, Berkeley) verified the general assumption about the presence of two level tones in Proto-Niger-Congo. The starting point for the survey was an intention to figure out how far it is possible to extend the regular correspondences between the existing tonal reconstruction of Proto-Bantu and reconstructions of higher level groupings which include Bantu. Only verbal stems were considered and only for those languages where root tone characteristics are not a part of verb morphology (TAM, negation, etc.), so as to deal as far as feasible with relics of inherited lexical tones. A further goal was to find out more about Niger-Congo verb structure by considering the verb extensions and their tone patterns. The major findings are that there is no evidence for reconstructing more than two register tones; verb extensions had their own inherent tone features. A first attempt at a modern phylogenetic classification of Niger-Congo was made by Jean-Marie Hombert (Laboratoire Dynamique du Langage, CNRS — Université Lumière Lyon 2), Rebecca Grollemund and Simon Brandford (Evolutionary Biology Group, University of Reading). The input consisted of 100­item wordlists belonging to the basic vocabulary for a sample of languages from different Niger-Congo families; cognate sets allegedly were identified by manual evaluation. However, data sources and quality were not always clear, and, more importantly, neither was the scoring scheme for cognate detection, making it hard to evaluate the usefulness of the new scheme for further Niger-Congo studies. One of the distinguishing features of Niger-Congo was the existence of a nominal classification system. Several of the presentations were devoted to its traces in descendant families, including a survey conducted by Robert Hepburn-Gray (University at Buffalo, SUNY) 208

that presented an overview of Niger-Congo noun class agreement systems, taking into account a number of parameters of cross-language variation, such as type of agreement targets, or number of distinct positionally conditioned realizations of the class markers; it seems, however, that the modest size of the sample and the employed methodology remain insufficient to make any far-reaching conclusions. Ronald Schaefer (Southern Illinois University) reviewed types of subordinated arguments in Niger-Congo branches, distinguishing between sentence-like, infinitival and truncated complements (nominalisations). Two major tendencies are represented within Niger-Congo: to mark indicative, subjunctive, and conditional clauses differently, or to make no distinction between any of them. Unfortunately, this report suffers from the same problem as the preceding one. An important task for Niger-Congo studies is to precisely delineate the boundaries of the phylum. Thus, for instance, because of limited data sources and a number of typological shifts in their history, the status of the Kadu (Krongo-Kadugli) languages had long remained contentious. Under one hypothesis, they were associated with Niger-Congo (within Kordofanian, or as a different branch), under another, they were included with Nilo-Saharan. An additional benefit of George Starostin’s (Russian State University for the Humanities / Russian Presidential Academy) ongoing reclassification of Greenberg’s Nilo-Saharan phylum was that it supplied new evidence for Kadu’s non-Niger-Congo affiliation. The revised classification was established by means of lexicostatistics combined with traditional comparative-historical method involving step-by-step reconstruction (moving from first-order proto-languages to higher levels of classification) of the most stable subset of Swadesh list items (for detailed description, see G. Starostin 2013). The findings show that the basic lexicon of Kadu languages shows far more significant phonetic similarities with Nilo-Saharan (specifically, Central Sudanic, since Nilo-Saharan as a valid taxon remains even more problematic than Niger-Congo), and that there is no lexical basis whatsoever to group them together with Niger-Congo. Turning now to reports on individual (smaller) language groups and families, Sandro Capo Chichi (Université Paris VII-Paris Diderot / Laboratoire de Linguistique formelle) revised the function of the noun suffix ­i in Proto-Gbe. Previously it was understood as marking compounds, diminutives, nouns similar to underived base, and instruments (Capo 1991). According to Capo Chichi, the affix is one of the complementarily distributed variants along with na-

Towards Proto-Niger-Congo: Comparison and Reconstruction (2016)

sality of terminal vowel, and both of them served as nominalizer for marking loanwords with structures uncommon for Proto-Gbe. The Kwa languages demonstrate great diversity in the design of their class systems from residual to quite elaborate (the Guang and the Ghana-Togo Mountain languages). As the classes are attested in genetically remote Kwa branches and have external parallels in other Niger-Congo families, Maria Konoshenko and Daria Shavarina (Russian State University for the Humanities) came to the conclusion that Proto-Kwa had a fully-fledged class system with canonical NigerCongo class marking on nouns and agreement. They identified the main strategies of noun class attrition, comparing them with the generalizations formulated for other families. Almost the same issue, but from a different angle, was addressed by Ines Fiedler (Humboldt-Universität zu Berlin). She described the Kwa class system in terms of gender (cf. Güldemann 2000) and proposed refinements for the already existing reconstructions of Kwa subgroups. Further related to general classification issues, Tucker Childs (Portland State University) offered motivation for establishing Mel as a group separate from Atlantic proper, and suggested an internal reclassification of the family, indicating major clusters: TemneBaga, Bolom-Kisi and Gola as a single language branch. The rest of the report dealt with the reconstruction of definite markers for various Mel subgroups, whose distribution supports current views on the internal classification of Mel. Guillaume Segerer (LLACAN) raised the issue of universality of the basic vocabulary among languages, building on the case of Joola. Joola (Bak < Northern Atlantic), a cluster of languages spoken in Lower Casamance, Senegal, exhibit minor differences in phonology and morphology that reflect their close relationship; however, low cognacy percentages in lexicostatistical calculations contradict this assumption. The speaker highlighted the most stable and most unstable items in the Swadesh list, comparing them with the ones proposed in S. Starostin 2007 and Holman et al. 2008, and outlined the main lexical replacement strategies, among others, taboo for words used for people’s names after their death. Another talk specifically tackled the Southern subbranch of the Mande family. These languages are generally assumed to have isolating morphology, yet their pronominal systems are rather elaborate, with multiple pronominal series differing morphologically and syntactically. Valentin Vydrin (LLACAN — INALCO) presented scenarios for the proliferation of the pronominal series and his own reconstruction of

the Proto-Southern Mande pronominal system. Despite a series of innovations and complications of the system that took place in recent times, the protosystem itself was rather intricate as well, including, among other things, a clusivity opposition for both dual and plural numbers. Lynell Zogbo (University of the Free State / Abidjan) attempted to determine whether a separate adjective category could be reconstructible for Proto-Kru. Given that the class does not exceed 6–20 items, is not particularly homogenous within itself (some adjectives pattern like verbs, others like nouns) and Western and Eastern Kru differ in models of noun class agreement, she concluded that even though the category did exist, it was modest in scope. Jeffrey Heath (University of Michigan) discussed verbal inflection in Proto-Dogon, particularly the issue of stem-gradation, which enables to draw parallels between Dogon and other Niger-Congo families. A significant contribution to the description of several underexplored Adamawa lects was made in the course of fieldwork trips undertaken in 2012–2014 by the Faculty of Asian and African Studies of St. Petersburg State University. In their joint report Anastasiya Lyahovitch and Alexander Zheltov (St. Petersburg State University) presented the main results of the survey. For the languages under consideration the authors compared their pronominal systems as well as numeral systems and copulas, with further discussion on the implications that this could have for the internal classification of the family. Ulrich Kleinewillinghöfer (JG-Universität Mainz) drew attention to several languages within the Samba-Duru subgroup of Central Adamawa that still preserve the original nominal class system, while most other Adamawa languages have reduced or even lost it; the established tentative reconstructions for Central Adamawa bear clear similarities to the noun system earlier proposed for Gur. On the Benue-Congo side of things, a further report by Bruce Connel (Glendon College, York University) focused on tonological reconstruction in Mambiloid (Northern Bantoid). Modern Mambiloid languages appear to have three to four contrastive tones, yet the absence of regular tonal correspondences in their systems imply that they are rather innovative. The speaker suggested a two-tonal system for ProtoMambiloid and, using individual daughter languages as case samples, illustrated the evolution from twolevel tone systems to polytonal ones. Demola Lewis (University of Ibadan, Nigeria) used several algorithms for automated classification, such as the Automated Similarity Judgement Program and the Sound Correspondent Recognition Program, to 209

Reports / Хроника

North Edoid languages. For source data, she used the Leipzig-Jakarta lists of basic vocabulary, a recent alternative to the Swadesh list (with 62 overlapping items), compiled by Martin Haspelmath and Uri Tadmor in the course of their cross-linguistic investigation of borrowings in the world’s languages (cf. Haspelmath & Tadmor 2009). Oyetayo Bankale (University of Ibadan, Nigeria) proposed a new model of stem-initial consonant (the most stable position within the morpheme) inventory for Proto-West BenueCongo, although his reconstruction would seem to deal more with phonetics rather than phonology. Two presentations concerned the Ekoid languages, a small Southern Bantoid subgroup. The Ekoid group proper is often discussed in conjunction with its closest relative, a single language Mbe. During the first stage of divergence, the common ancestor of the Ekoid languages split into Proto-Ndoe and Proto-BakorEjagham, which in turn subdivided into the Bakor and Ejagham clusters. John Watters (SIL International) presented a reconstruction of the Proto-Ekoid-Mbe noun class system. Noun class prefixes and concord markers of Mbe and all the proto-languages of Ekoid subgroups were matched with those reconstructed for Proto-Bantu; the results revealed that despite several transformations and mergers, Proto-Ekoid-Mbe had a relatively well preserved noun class system, since all Bantu classes (except for 13 and 18) have correspondences in Ekoid-Mbe. Continuing the Ekoid subject, Galina Sim (Institute of Linguistics, Russian Academy of Sciences) proposed a hypothetical segmental inventory for Proto-Ekoid, specifically dwelling on some controversial topics such as the voiceless/voiced obstruent split in Proto-Ndoe, which should rather be regarded as an innovation than an archaism. An overview of the Lower Cross speech forms spread in Akwa Ibom State of Nigeria and their suggested classification was presented by Emmanuel Akaninyene Okon (University of Uyo, Akwa Ibom State, Nigeria). Both talks given by Jacky Maniacky (Royal Museum for Central Africa, Belgium) addressed the issues of semantic reconstruction in Bantu. An automated likelihood-based statistical model was devel-

210

oped by Rebecca Grollemund, Simon Branford, and Mark Pagel (Evolutionary Biology Group, University of Reading). It allows to align cognate sets phonemically and detect sound changes by evaluating probabilities for each node in the graph and choosing the most probable scenario. The model was applied to the Proto-Bantu material and enabled to automatically detect some sound changes that actually took place. Investigations of particular Niger-Congo branches not only lay down the groundwork for prospective Niger-Congo reconstruction but also to a certain extent supply possible scenarios that could have taken place during earlier stages. The final chord took the shape of two general discussions moderated by Konstantin Pozdniakov and Larry Hyman. Following the conference, the first meeting of the newly formed Adamawa languages working group was held with the goal of focusing researchers’ efforts on this least well explored branch of Niger-Congo. Nevertheless, many problems in NigerCongo reconstruction still remain to be resolved, and, hopefully, all the initiatives will in fact result in fruitful cooperation and further progress in our understanding of African prehistory.

References Capo, Hounkpatin B. Christophe. 1991. A comparative phonology of Gbe (Publications in African Languages and Linguistics 14). Berlin / New York: Foris Publications; Garome, Bénin: Labo Gbe. Güldemann, Tom. 2000. Noun categorization systems in NonKhoe lineages of Khoisan. Afrikanische Arbeitspapiere 63: 5–33. Haspelmath, Martin, Uri Tadmor (eds.). 2009. Loanwords in the World’s Languages. A Comparative Handbook. Berlin, Boston: De Gruyter Mouton. Holman, Eric, Søren Wichmann, Cecil H. Brown, Viveka Velupillai, André Müller, Dik Bakker. 2008. Explorations in automated language classification. Folia Linguistica 42(2): 331–354. Starostin, G.S. 2013. Yazyki Afriki. Opyt postroeniya leksikostatisticheskoy klassifikacii. Tom 1: Metodologiya. Koysanskie yazyki. Moscow: Yazyki slavyanskoy kul’tury. Starostin, S.A. 2007. Opredelenie ustoychivosti bazisnoy leksiki. In S.A. Starostin. Trudy po yazykoznaniyu. Moscow: Yazyki slavyanskikh kul’tur: 825–839.

Book Reviews / Рецензии П. В. Башарин Russian State University for the Humanities (Moscow); [email protected]

С. В. Кулланда [Sergei Kullanda]. Скифы: язык и этногенез [The Scyths: language and ethnogenesis]. М.: Университет Дмитрия Пожарского, 2016. 215 с.

Монография С.В. Кулланды подводит некоторый итог циклу работ автора, посвященных вопросам скифского языка. Скифские имена собственные и социальные термины являются важным источником не только для лингвистов, но и для историков. На основе их этимологизации Д.С. Раевский, развивая идеи В.И. Абаева (1949: 242—243) и Э.А. Грантовского (1960: 7—9), построил специфическую концепцию о характере скифского общества и его картине мира. Например, в именах трех легендарных братьев, Липоксая — Арпоксая — Колаксая (соответственно, *ripa ‘горы’ по аналогии с названием Рипейских гор, *āpra ‘глубокий’, *xvar ‘солнце’) он видел отражение представлений о трех мирах: верхнего — небесного, срединного — надземного и нижнего — подземных вод. Ряд этимологизаций, в том числе и приведенная выше, не имели под собой достаточных оснований и носили спекулятивный характер 1 . Однако эти гипотезы легли в основу трактовки ряда предметов материальной культуры, связанных с культурами скифского круга и позднее получили распространение в среде специалистов по скифской археологии. На основании анализа скифского ономастикона в науке до сих пор господствует мнение о чисто гипотетическом разделении скифского и сарматского языков. В отечественной науке данная установка была сформирована работами В.И. Абаева, понимавшего под скифским языком «общее название для всех скифо-сарматских наречий и говоров, которые существовали на территории Причерноморья в период от VIII—VII вв. до н.э. до IV— V вв. н.э.»2 и отказавшегося от разделения скифских и сарматских имен собственных. Схожих установок придерживаются и ведущие западные 1 Раевский 1977. См. их критику в рецензируемой монографии (с. 119—120). 2 Абаев 1949: 147. Этой точки зрения ученый придерживался в течение всей своей жизни.

иранисты (М. Майрхофер и Р. Шмитт). Таким образом, по замечанию автора, скифский в понимании упомянутых ученых являлся «конгломератом разновременных фонетических явлений, характерных для разных подгрупп иранских языков» (с. 5). Археологи часто отождествляют понятия «скифский» и «иранский». К.Т. Витчак впервые предложил фонетические критерии разделения скифского и сарматского. Самым важным дифференцирующим признаком стало развитие в двух языках праиранского *d, отразившегося как l в скифском и как d в сарматском. Рецензируемая монография состоит из трех глав. Первая посвящена предыстории скифов и открывается анализом вопроса прародины индоиранцев и иранцев. Ссылаясь на ряд заимствований в прафинно-угорский язык, демонстрирующий фонетические переходы, характерные для индоиранского, а не для праиранского (отсутствие переходов *s > h, *kš > xš), автор предлагает локализовать прародину индоиранцев в зоне, контактной с тогдашним проживанием финно-угров, не слишком далеко от таёжной зоны. Предположение о позднем переходе *s > *h для иранских, позволяющее видеть в народе, контактировавшем с финноуграми, не индоариев, а иранцев, выдвигавшееся с 60­х годов прошлого века некоторыми известными иранистами, так и не нашло достаточного обоснования (с. 17). С другой стороны, известная гипотеза Е.А. Хелимского о некоем не оставившем потомков индоарийском языке как об источнике этих заимствований, скорее применима к более позднему пласту заимствованной лексики. Контактировали с индоиранцами также носители правосточнокавказского (пранахского) языка и картвелы. Потенциальным заимствованиям из северокавказских языков было посвящено несколько статей С.В. Кулланды3. В рецензируемой моногра3

Кулланда 2012; Kullanda 2014.

Journal of Language Relationship • Вопросы языкового родства • 14/3 (2016) • Pp. 211–216

Book Reviews / Рецензии

фии автор обобщает выводы этих публикаций. Опираясь на нерегулярные передачи ряда рефлексов в индоарийских и иранских языках, он полагает, что говорить о единовременном заимствовании северокавказской лексики не представляется возможным. Большинство сопоставлений представляются вполне удачными, например, ἀκινάκης ‘персидский меч’, этимологии которого до сих пор не предлагалось, из пранах. *ɦāhḳ́i- ‘железо’, *neḳV ‘нож’4. Выведение др.-инд. gandharvá и авест. gaṇdrβa- из потенциального пранахского словосложения *ḳanat ‘мальчик, юноша; молодец, удалец’ + *āri ‘шайка, банда’ не выглядит столь же надежной. Сложность локализации носителей картвельских и восточнокавказских языков в 3—2 тыс. до н.э. не позволяет дать однозначный ответ на вопрос о локализации индоиранцев, хотя напрашивается вывод об их продвижении с раннеиндоевропейской прародины через Кавказ в исторически засвидетельствованные места их обитания. Неясно, как эти данные увязываются с финно-угорскими контактами. С другой стороны, восточнокавказские заимствования продолжаются и отдельно в восточноиранские языки. Напрашивается логичный вывод, что последние могли заимствовать ряд лексем в процессе своего движения через Кавказ на Иранское нагорье и далее в Среднюю Азию. Этот аргумент является еще одним доводом в пользу движения праиранцев через Кавказ. С.В. Кулланда полагает, что восточные иранцы населяли область Нижнего Поволжья. Аргументами для такого предположения служит авестийский гидроним Raŋhā, тождественный Qα̃ Клавдия Птолемея и идентифицирующийся с Волгой, а также упоминание бобров (бобрих), из чьих шкур состоит облачение авестийской богини Ардви Суры Анахиты, между тем как в Аму-Дарье и СырДарье эти животные не обитают. Сепаратные контакты собственно скифского проследить не удается. Финно-пермские иранизмы, которые В.И. Лыткин считал заимствованиями из юго-восточных иранских языков на основании наличия в них озвончения интервокального *­š- в ­ž­, могут демонстрировать собственно финно-пермское озвончение. Вторая глава посвящена скифскому языку и культуре. Анализируя археологические материалы, маркеры продвижения носителей скифской культуры, автор приходит к выводу, что носители скифской материальной культуры пришли из 4 Автор не приводит на страницах монографии их полного списка, отсылая к выше цитируемым статьям.

212

Приаралья, откуда происходит самая значительная часть категорий артефактов. Основную часть второй главы составляет скифский глоссарий (с. 41—96). Анализируя все известные научные гипотезы по каждой конкретной лексеме, автор предлагает ряд новых этимологий. Анализ большего числа разбираемых этимонов до сих пор служит предметом острой полемики5. В конце словника автор реконструирует несколько этимонов, не зафиксированных в письменных памятниках: *gauarga этноним, **θarmi ‘вид дерева’, **malaxa (?) ‘саранча’, *paraθu- ‘топор’, *raθana- ‘ремень, веревка’. Однако malax скорее всего было заимствовано из языка типа согдийского (где оно, правда, не зафиксировано, но где, с другой стороны, слова для саранчи пока не найдено) или бактрийского. Предположение автора, что согдийцы могли использовать графический ламед (l) для передачи звука δ (на том основании, что в перенятом согдийцами арамейском курсиве далет и реш не различались, а ламед в согдийском использовался только в заимствованиях за отсутствием там звука r) (с. 98) лишено достаточных оснований. Гипотеза о существовании согдийского диалекта, где существовал переход δ > l, представляется более обоснованной, тем более, что данный гипотетический согдийский диалект разделяет эту особенность с юговосточными иранскими языками. Даже если предположить, что пласт потенциальных согдийских заимствований в персидском на самом деле пришел из бактрийского, для существования подобного диалектального перехода в согдийском остается достаточно оснований (использование l для обозначения спирантов δ и θ)6. Третья глава содержит характеристику основных фонетических особенностей скифского языка. Вслед за К.Т. Витчаком, С.В. Кулланда трактует все случаи появления скифского l как отражение восточноиранского δ и, соответственно, общеиранского *d. Переход в δ осуществился к VIII в., о чем свидетельствуют, с одной стороны, ассирийская и вавилонская передача самоназвания скифов ašguza, asguza, iškūzaia а с другой, — греческая Σκύθαι < *Skuδa. Известна и передача семитского ḏ через тету в греческих текстах. 5

Разбор примеров см. ниже. Ср., однако, исследование П.Б. Лурье и И.С. Якубовича, где демонстрируется, что все доказательства наличия ламбдаизма в постулируемом согдийском диалекте можно переинтерпретировать, а источником заимствования лексем с l в персидском является не согдийский, а бактрийский (Lurje, Yakubovich forthcoming). 6

С. В. КУЛЛАНДА. Скифы: язык и этногенез (2016)

Этононим Σκόλοτοι является более поздней передачей и отражает переход δ > l (*Skula). С другой стороны, этноним Σκύθαι, согласно С.В. Кулланде, наглядно демонстрирует эволюцию фонемы, обозначаемой тетой: переход из придыхательного в спирант. Из этого, полагает автор, следует, что первыми со скифами познакомились не ионийцы, с которыми имели дело жители Передней Азии, а дорийцы, т.е. население Балканского полуострова (если только этноним не попал к грекам через фракийскую передачу, что представляется более вероятным). Гипотеза о таком переходе снимает проблему объяснения l в греческих передачах скифских имен ввиду преобладавшего в иранских языках ротацизма. С другой стороны, она заставляет пересмотреть ряд старых трактовок, связанных с социолингвистической областью. Например, имя Παιρίσαλος, встречающееся в эпиграфике Крыма и Таманского полуострова, традиционно считается фригийским. С.В. Кулланда предположил, что это скифизированная форма имени боспорского царя Перисада (Παιρισάδης). Автор полагает, что данный переход является общегенетическим для юго-восточных иранских языков. Согласно Д.И. Эдельман, материал языков Гиндукуша, где изменение прослеживается в нуристанских и дардских языках, демонстрирует ареальный характер этого перехода. Однако, согласно ряду исследований (П.О. Шервё, А.И. Коган), это явление в языках Восточного Гиндукуша может объясняться влиянием иранских языков. С.В. Кулланда ссылается на заимствования из юговосточных иранских языков в древнеиндийский: передача топонима Бактрия (авест. Bāxδī­) как Bāxlīka, lipi ‘письмо’ при западноиранском dipi (куда лексема, восходящая к шумерскому, попала через аккадский или эламский). Однако нужно заметить, что индийский материал демонстрирует не юго-восточные иранизмы вообще, а заимствования из бактрийского (de Blois 2013, 269). Второй характерной особенностью скифской фонетики, согласно С.В. Кулланде, является переход начального *xš- в сибилянт, что передавалось сигмой в греческой графике. В отличие от этого, в сарматском *xš- сохранялся и передавался греческой кси. На основании этого С.В. Кулланда отказывается от находящей поддержку у ряда современных ученых гипотезе о сарматской атрибуции племени саев и вслед за Б.Н Граковым видит в них скифов. Любопытно объяснение автором перебоев передачи иранских имен собственных с этимологи-

ческим начальным xš- в ассирийской, эламской и греческой передачах (типа ассир. ká-aš/kaš-ta-ri-tu < Xšaθrita, греч. ξατρáπης (эпиграфич.) < *xšaθrapāпри элам. sa-tar-pa/ba-nu < Xšaθrapāna, греч. σατράπης < *xšaθrapā­). Подобные несоответствия вызваны различием в произношении самих иранцев (особенно на фоне того, что греки в остальных случаях различали иранские xš- и š/s­). Практически все имена, содержащие подобные перебои, отражают особенности мидийской фонетики (θr вместо др.перс. ç). Исключением является первая часть имени, зафиксированного в арамейской графике на ахеменидской булле из Телло šs)mr < xšaça (согласно В. Хинцу). При этом переход *xš- > s не был характерен для западных иранских языков (мидийского и древнеперсидского). Автор объясняет эти передачи отражением скифской фонетики (общеиранское *θr отражается в греческих передачах скифских имен как τρ). Этот процесс мог быть вызван влиянием кочевого скифского элемента на местное мидийское население. Сложность представляет собой объяснение интервокального ­d­, появляющегося в ряде имен. Автор полагает, что в данных этимологиях мы имеем дело с отражением не ­d­, а nt- > d. Ради этого он пересматривает традиционные этимологии: Μαδύες из *mantu- ‘советник, правитель’, при общепринятой этимологии *madu- ‘мед’, Ἀμάδοκοι из *a-mantuka ‘не имеющие правителей’, при общепринятой этимологии *āmādaka ‘сыроядцы’ из āma ‘сырой’ и ad ‘есть’. С другой стороны, по мнению автора, сочетание *nd- сохранялось как в Ἰνδάνθυρος, сопоставляемом с древнеперсидским Vindafarnah- Бехистунской надписи. Последнее передавалось погречески как Ἰνταφέρνης / Ἰνταφρένης. Интересна в этой связи трактовка гидронима Παντικάπης, традиционно объясняемого как ‘Путь (panti) рыбы (kapa)’. Во-первых, ­nt- < *­nθ- могло не переходить в d. Во-вторых, гидронимы не обязательно имели скифское происхождение: ср. ниже о названии Дона Τάναϊς (сами скифы, согласно Плинию Старшему, называли Дон Силисом). Согласно гипотезе автора, *ś дало скифское θ, вопреки общему для всех прочих восточных иранских s. Последний рефлекс отражен в передаче массагетских имен. Массагетскому имени собственному у Геродота Σπαργαπίσης соотвествует скифское Σπαργαπείθης. Следует заметить, что этот переход отражен в именах с элементом πείθης. Фасмер предложил влияние греческого элемента ­πείθης- у Геродота. С.В. Кулланда полагает, что это влияние не могло быть решающим и приводит в пример потенциальные скифизмы в осетинском 213

Book Reviews / Рецензии

типа færæt ‘топор’ < *paraθu < *paraśu, где θ появляется без всякого греческого посредства (при ожидаемом осетинском развитии *ś > s). Этноним Θισαμάται сопоставляется с др.-инд. śiṣ- ‘оставлять; выделять’ и трактуется как ‘отклонившиеся, удалившиеся’. Автор приводит экстралингвистический довод — сообщение Геродота о «скифах, отложившихся от царских скифов», живших к востоку от Меотиды (Азовского моря). Однако фисаматы жили в Причерноморье. Это географическое несоответствие автор объясняет перемещением скифских племен на протяжении III в. до н.э. (упоминание фисаматов встречается в декрете Протогена, предположительно конца этого столетия). Особый случай развития *ś в скифском демонстрирует ряд имен божеств. Во-первых, это имя скифского Аполлона *|οιτόσυρος, вторую часть которого автор предлагает связывать с младоавест. sūr ( l ряд иранистов связывал исключительно с диалектом «царских скифов». Впервые об общескифских процессах заговорил, кажется, только К.Т. Витчак. Далеко не все греческие авторы имели о скифском языке точные представления. Типологически любая передача фонетики иноязычных имен часто грешит неточностями и гиперкоррекциями, даже при условии, что передатчик в некоторой степени владеет данным языком. Это видно по гиперкоррекциям в передаче скифского вокализма в приводимом авторе лексиконе. Картина усложняется, если мы допускаем наличие ряда диалектов в самом скифском. В связи с этим не исключается и спорадическое различное отражение одних и тех же рефлексов. Принятая автором гипотеза о едином скифском языке, а не ряда скифских диалектов, которые могли иметь различные фонетические особенности, порождает необходимость ревизии ряда этимологий. Например (если рассматривать примеры с самого начала глоссария), этноним’Αμάδοκοι традиционно этимологизировался как *āmādaka ‘сыроядцы’ из āma ‘сырой’ и ad ‘есть’ (по В. Томашеку). С.В. Кулланда предлагает возводить лексему к *amantu-ka ‘не имеющие правителей’, от *mantu‘правитель’ или даже видеть в нем не скифский этноним. Возведение ΑΔΟΥΞΗΣ (имя из Фанагории) к *hada-uxšan- ‘богатый быками’ (этимология С.Р. Тохтасьева) также не принимается, т.к. *d должно было перейти в l. В этом этимоне С.В. Кулланда предлагает видеть заимствование из какого-то иранского языка, где не произошло данного перехода. Спорадические колебания рефлексации можно проследить на примере θ. В ранних греческих передачах она отражает δ (в Σκύθαι < *Skuδa), а в большинстве этимонов соответствует θ < *ś (например, Αθύρας из *aśu- ‘быстрый’). Также тета появляется в отражении кластера ­st- (σθ) Βορυσθένης / Βαρυσθένης < *baru-stāna ‘место, [где водится] бобр’ (при этом автор не принимает предлагаемую А. Лома этимологию, допускающую *varu-θana­, *varauš-θana- ‘пойма [реки] Вару’). Наконец, встречается случай соответствия θ- ~ t- в композите Ἰνδάθυρσος. Для данного композита предложен ряд этимологизаций, но ни одна из них не предполагает отражение теты во втором элементе. С.Р. Тохтасьев на этом основании полагает, что имя не восходит к диалекту царских скифов8.

7 Мы упоминаем исключительно датировку материала, приводимого автором.

8 Подробнее о различных гипотезах этимологизации этого имени см. рецензируемую монографию (с. 66—67).

214

С. В. КУЛЛАНДА. Скифы: язык и этногенез (2016)

Отдельная часть главы посвящена вопросу потенциальных скифизмов как в древних иранских языках, так и в различных языках близлежащих регионов. Согласно мнению П. Лекока и А.М. Лубоцкого, к скифскому словарю относится лексема farnah-/xvarnah- (эманация божественного света, символизирующего верховную власть). Общепринятой гипотезой является предположение, что форма xvarnah- первична и являлась общеиранским понятием, при том, что начальное xv- нашло отражение только в авестийской лексеме xvarnah. В последние десятилетия получила развитие гипотеза о первичности формы farnah- и о том, что форма с начальным xv- является результатом заимствования и гиперкоррекции. С.В. Кулланда не признает отнесения farnah- к скифскому словарю на том основании, что переход p > f перед гласным встречается в сарматском и осетинском, но не свойствен скифскому (наряду с раннесарматским!). Автор также не находит достаточных оснований в гипотезе А.М. Луботского о скифском переходе *ti > θi (на основании трех древнеперсидских слов: duvarθi- ‘портик, колоннада’, skauθi-/škauθi- ‘слабый, бедный’, *θigra(ka)- ‘чеснок’) из-за отсутствия надежных примеров для такого перехода. Ярким примером потенциального скифизма в мидийском является этноним Skudra, встречающийся в древнеперсидских надписях по отношению к фракийцам, который мог произойти от самоназвания скифов (*Skuδa). Имеется и другой пример передачи восточноиранского δ сочетанием зубной + r: мидийское *Bāxtrī- (при авест. Bāxδī­). С.В. Кулланда полагает, что r в этих примерах является эпентезой для передачи специфического восточноиранского щелевого δ, а не исконным согласным9. Ниже приведены случаи потенциальных скифских заимствований, постулируемых автором. Лексика текстов авестийского корпуса демонстрирует развитие *ś > θ наряду с ожидаемым s (gaēθu- наряду с gaēsu- ‘курчавый’ (Яшт), aiβi-θūra- ‘могущественный’ наряду с sūra- ‘сильный, могучий’ (Яшт), ana-saxta/anaθaxta ‘тот, чей срок еще не прошел’ (Видевдат) и т.д.); xš > s (sātar ‘властитель’ вместо *xšātar); sr > θr (θraotah- ‘поток’ вместо *sraotah­). Эти дублеты и замены издавна объяснялись как результаты смешения разных диалектов, вероятно, в устоявшихся культовых формулах (С. Викандер). Попытку объяснить их влиянием древнеперсидского С.В. Кулланда не принимает. Главным аргументом является то, что помимо перехода *ś > θ 9 В современной иранистике, впрочем, встречаются и прямо противоположные суждения (de Blois 2013).

иных признаков древнеперсидской фонетики они не обнаруживают. Автор делает предположение о существовании скифских объединений, образовывавших симбиоз с меотами и кобанцами. В доказательство этого он сопоставляет этноним Μαῗται / Μαιῶται / Μαιῆται с авест. maēt/θ-/mit/θ- ‘пребывать, проживать’ в смысле ‘коренные жители’. С другой стороны, автор считает меотской лексему из знаменитого митаннийского трактата Киккули a-ša-an-na ‘беговой круг’, во втором слоге которой отражен не индоарийский рефлекс *h, а его индоиранский вариант *źh. Отсюда следует предположение, что арии могли составлять правящий слой меотского общества. Меотам и синдам С.В. Кулланда определяет особую роль в этногенезе Причерноморских скифов, как арийскому (но не иранскому) народу, проживавшему в Северном Причерноморье. На то, что меоты и синды как население Северо-Западного Кавказа могли принадлежать к самостоятельной ветви арийских языков, по мнению автора, указывает ряд косвенных признаков. Во-первых, это разобранные выше имена скифских божеств, которые демонстрируют нетипичное для скифского развитие *ś >s. Синдов, если возводить их этноним к арийскому *sindhu- ‘река’, что должно дать в иранских *hindu, скифы могли называть индами (что и засвидетельствовано рукописями труда Геродота, хотя издатели исправляют Ἰνδοί на Σινδοί). Симбиоз этих культур со скифской по оценке автора демонстрирует ряд особенностей материальной культуры Причерноморских скифов. С.А. Старостин реконструировал для восточнокавказского лексему *vĕlθi- ’войлок, буркаʼ, считая его заимствованием из какого-то индоиранского языка (ср. авест. varsa ‘волос’, др.-инд. válśa- ‘побег, ветвь’). Фонема *θ восстанавливалась им только для нескольких восточнокавказских праформ, которые не относились к исконной лексике. С.В. Кулланда полагает, что она является маркером заимствованием из скифского. Он отмечает, что θ регулярна для древнеперсидского и скифского. Однако предок древнеперсидского не демонстрирует развития *ś > θ, поскольку древнеперсидская форма названия Ассирии Aθura указывает на то, что это название было заимствовано с š, который уже позже перешел в θ. Данное предположение выглядит весьма гипотетично: даже если полностью принять аргументацию автора по поводу проблематичности древнеперсидского источника, сложно исходить из положения, что переход *ś > θ был характерен только для двух языков и не предположить, что могли быть и иные языки, где происходил подобный переход. 215

Book Reviews / Рецензии

О связи с западнокавказскими, согласно С.В. Кулланде, говорит скифское *sana ‘вино’ из западнокавказского *s(w)ana ‘смородина; вино’. Оно же присутствует в осетинском sæn / sænæ ‘вино’ (В.И. Абаев полагал, что переход был противоположным — из скифского в северокавказские). Лингвистический материал (тем более не вполне надежный, особенно в случае с *vĕlθi­) не является основным решающим фактором, подтвержающим связи скифов с народами северного Кавказа. Однако данные связи подтверждены рядом надежных археологических данных, и факт, что северокавказцы были тесно связаны с культурами скифского круга, не ставится под сомнение. Ряд наблюдений может свидетельствовать в пользу контактов скифов с фракийцами. Ряд фракийских племен (сайи, сатры) имеют иранские наименования. Согласно ряду исследователей, об этих контактах свидетельствует этноним Σκύθαι, который мог попасть к грекам через фракийскую передачу. Упомянутый выше этноним Skudra, потенциальный скифизм в мидийском, встречается в древнеперсидских надписях по отношению к фракийцам и мог произойти от самоназвания скифов. С.В. Кулланда также разделяет гипотезу Г. Хольцера о контактах скифских племен с еще одним, не индоиранским, индоевропейским субстратом, языком, в котором происходило оглушение звонких и озвончение глухих: название Дона Τάναϊς (ПИЕ *dh2n- ‘бежать, течь’), имя божества Ταβιτί (ПИЕ *dh2p- ‘готовить ритуальную пищу’). В приложении автор дает перечень и краткую характеристику основных архаических скифских памятников (всего 16 объектов), содержащих вещи ближневосточного производства. В конце монографии помещен указатель словоформ как скифского, так и прочих языков, затрагиваемых в тексте, что значительно упрощает поиски нужного пассажа. Анализируя скифский языковой материал, исследователь неизбежно сталкивается с рядом альтернативных этимологизаций и трудностью в предпочтении одной из них. Ряд альтернативных скифских этимологий порождает острую полемику в кругах специалистов (см., например, Иванчик 2009; Кулланда 2011а; Он же 2011б). Положитель-

216

ной стороной работы является освещение всех научных теорий, наработанных учеными по каждому этимону. Она вряд ли поставит точку в вопросе этногенеза скифов, но, безусловно, станет ценным подспорьем для всех интересующихся скифским языком. Литература Абаев 1949 — В.И. Абаев. Осетинский язык и фольклор. М.—Л. [V.I. Abaev. Osetinskij jazyk i fol’klor. Moskva—Leningrad, 1949]. Грантовский 1960 — Э.А. Грантовский. Индо-иранские касты у скифов. М. [E.A. Grantovskij. Indo-iranskije kasty u skifov. Moskva, 1960]. Иванчик 2009 — А. И. Иванчик. К вопросу о скифском языке. ВДИ 2: 62—88. [A.I. Ivančik. K voprosu o skifskom jazyke. Vestnik drevnej istorii. 2009, 2: 62—88]. Кулланда 2011а — С.В. Кулланда. Уроки скифского. Вопросы языкового родства 5: 48—68. [S.V. Kullanda. Uroki skifskogo. Voprosy jazykovogo rodstva. 2011, 5: 48—68]. Кулланда 2011б — С.В. Кулланда. Скифы: язык и этнос. Вестник РГГУ. Серия «Востоковедение. Африканистика» 2 (64) / 11: 9—46. [S.V. Kullanda. Skify: jazyk i etnos. Vestnik RGGU. Serija «Vostokovedenije. Afrikanistika» 2 (64) / 11: 9—46]. Кулланда 2012 — С.В. Кулланда. К проблеме лексических контактов северокавказских, индоиранских и классических языков. Индоевропейское языкознание и классическая филология XVI. СПб., 2012: 406—415. [S.V. Kullanda. K probleme leksičeskih kontaktov severokavkazskih, indoiranskih i klassičeskih jazykov. Indoevropeyskoje jazykoznanije i klassičeskaja filologija XVI. St Peterburg: 406—415]. Раевский 1977 — Д.С. Раевский. Очерки идеологии скифосакских племен. Опыт реконструкции скифской мифологии. М. [D.S. Raevskij. Očerki ideologii skifo-sakskih plemen. Opyt rekonstruktsii skiskoj mifologii. Moskva, 1977]. de Blois 2013 — F. de Blois. Bactria, Bāxδī, Balx. Commentationes Iranicae. Сборник к 90­летию В.А. Лившица. Под ред. С.Р. Тохтасьева и П.Б. Лурье. СПб: 268—271. [Commentationes Iranicae. Sbornik k 90­letiyu V.A. Livshica. Pod red. S.R. Tokhtas’eva i P.B. Lur’e. SPb: 268—271.] Kullanda 2014 — S.V. Kullanda. North Caucasian Loanwords in Indo-Iranian and Iranian. Scripta Antiqua. Вопросы древней истории, филологии, искусства и материальной культуры: альманах. Том 3. К юбилею Э.В. Ртвеладзе. М.: 717—725. [Scripta Antiqua. Voprosy drevnej istorii, filologii, iskusstva i material’noj kul’tury: al’manah. Tom 3. K jubileju E.V. Rtveladze. Мoskva: 717—725] Lurje, Yakubovich, forthcoming — P. Lurje, I. Yakubovich. The Myth of Sogdian Lambdacism. In: D. Durkin-Meisterernst (ed.). Memorial Volume in Honour of Werner Sundermann (1935—2012). Berlin, forthcoming.

Maria A. Molina Institute of Linguistics, Russian Academy of Sciences (Moscow); [email protected]

Carlotta Viti (ed.). Perspectives on historical syntax. Studies in Language Companion Series 169. Amsterdam: John Benjamins, 2015. 158 p.

The volume under review is dedicated to various issues of historical syntax and syntactic reconstruction. The book is a collection of contributions resulting from the workshop “Syntactic change and syntactic reconstruction: new perspectives” held at the University of Zurich in September 2012. In terms of linguistic reconstruction syntax has always been less investigated than phonology, lexicon or grammar. In the second half of the 20th century research on diachronic syntax began to take its place in the field of historical linguistic studies, Indo-European as well as historical linguistics in general (inter alia, Lehmann 1974, 1976, 2000; Faarlund 1990; Bauer 1995, 2000; Crespo & García Ramón 1997; Devine & Stephens 1999; Barðdal 2001; Hewson & Bubenik 2006; Luraghi 2010; Barðdal & Eythórsson 2012; Ferraresi & Goldbach 2008; Ferraresi & Lühr 2010; Harris & Campbell 1995; Lightfoot 1979, 1991, 1999, 2002a, 2006; Longobardi 2003; Batllori et al. 2005; Roberts 2007; Jonas et al. 2012). Unfortunately, the usual trend in historical syntax is that the research is conducted under different frameworks, with almost no exchange of data or methods. This results in very little consensus in academic spheres on important theoretical and practical issues concerning syntactic reconstruction. Meanwhile, scholarly interest in syntactic change and reconstruction is growing, since this territory is largely uncharted and could provide researchers with a lot of additional information on language relationship and contacts, as well as historical migrations in the history of civilization. With syntax constituting one of the essential layers of language structure, linguistic reconstruction can hardly be complete without considering syntactic change. The very feasibility of such a reconstruction, though, is still under discussion. The obvious reason is that basic syntactic units are freely generated and not memorized, and vary in many more ways than phonemes, morphemes or lexemes. Therefore, they are hard to compare in different languages; it seems impossible to determine with any certainty which construction is etymologically older, and to reconstruct the previous stages according to the principles of the comparative method, which imply regular correspondences between linguistic units.

The book under review contributes to the understanding of historical syntax as a discipline of comparative historical linguistics. Carlotta Viti (University of Zurich) opens the discussion with general notions on the nature of historical syntax in the article “Historical syntax: problems, materials, methods, hypotheses”. Historical syntax is presented here as an emerging field of comparative linguistics; mechanisms of syntactic change and feasibility of syntactic reconstruction are discussed, as well as the general relevance of the volume for current studies in historical syntax. The author summarizes her introduction with a representative (but not exactly comprehensive!) list of references on previous research in historical syntax, about 8 pages in length. The bibliography, however, somehow lacks in fastidiousness: the names of A. Meillet, J. Wackernagel and K. Brugmann, M. Swadesh, C. Watkins and W. Lehmann, W. Labov, P. Kiparsky, J. Roberts and A. Garrett, whose influence on historical syntax and syntax theory cannot be overestimated, go along with dubious works on time depth in historical linguistics, such as Gray & Atkinson (2003), Renfrew et al. (2000), Longobardi & Guardiano (2009). The former two use phylogenetic methods to measure distances between cognate words, with ambitious, but not always reliable conclusions on prehistoric migrations; the latter focuses on building genealogical trees based on a list of syntactic parameters (see detailed discussion in Molina 2016). The aim of the introductory paper, though, is not to discuss the quality of the research, but to introduce the general problems of the field and present the most prominent perspectives of its development. The chapter on syntactic change opens with a paper by Ekkehard König (Free University of Berlin & University of Freiburg), “Manner deixis as source of grammatical markers in Indo-European languages”. The main part of this paper focuses on the quite neglected aspect of relative demonstratives and their role in the process of grammaticalization, resulting in the development of new grammatical categories. The author discusses the well-known change from exophoric to anaphoric and cataphoric meaning, and also gives examples for cases of propositional anaphors,

Journal of Language Relationship • Вопросы языкового родства • 14/3 (2016) • Pp. 217—225

Book Reviews / Рецензии

developed from deictic words, as well as comparative markers, adverbial connectives, quotative, exclamative and approximative markers. He specifically looks into the semantic categories of ‘manner’, ‘quality’ and ‘degree’ (< Proto-Germanic *swa). In particular, he shows that Germanic languages lack a clear differentiation between the three categories mentioned above (Germanic and Romance languages are the ones in focus here). König claims to be using the comparative method for his study, taking a theoretical approach rather than a descriptive one. The general syntactic processes are reconstructed “on the basis of comparative evidence, synchronic observations of possible forms, and patterns of polysemy” as well as on the basis of theoretical information known about grammaticalization of demonstratives — the macro-processes of grammaticalization, observable in a variety of languages. The deictic particles of ‘manner’, ‘quality’, ‘degree’ are taken as a starting point for all further processes concerning other demonstratives. By means of semantic analysis the author demonstrates on the examples of modern languages (English, German, Italian), as well as on ancient and protolanguage material (Latin, Old German), the possible ways of semantic change from exophoric to anaphoric meaning, from anaphoric to connective, from cataphoric to quotative (the latter is typologically supported with data from African languages, via Güldemann 2008). One of the major construction types is the meaning change from an endophoric determiner to a comparative marker with further development into a relative marker (see also Haspelmath 2012; a detailed discussion of manner deictics in comparative constructions across languages is given in König 2013). The last change seems to have almost no support from linguistic material — still, König shows some Old Saxon / Old Low German examples (p. 54) that support this process, providing a perfect candidate for the reanalysis of comparative markers as relative markers: (1) sulike gesidos so Such

companions

he im selbo

gecos

as he himself

chose

‘Such companions as/that he chose for himself…’ (Heliand text, 9th century, cf. Brandner & Bräuning 2013:138)

The main result of the paper is that, as suggested by the material, “demonstratives of manner, of quality and of degree are a highly relevant source for processes of grammaticalization”. However, the author stresses the preliminary character of this idea, and points out that it demands further research. Frans Plank (University of Constance) begins his paper “Time for change” with a reference to physics, 218

astronomy and geology, lamenting that historical linguistics cannot offer the same exact timing for its milestone events, as is given for the Big Bang or the time of Earth emerging, in order to model the evolution of typological diversity in languages. He puts forward the idea that the time needed for a change (‘time-stability’) should be a direct object of study. The basic measuring unit for such a study might be one generation, or one instance of acquisition of a language. Plank claims that the absolute dating of changes in syntax is possible quite deep in time, if relative chronologies of changes are traced (p. 66). He also suggests that for some items of lexicon the time for change goes slower than for others, being incredulous at the idea of glottochronology: “The glottochronological constant has been so decisively discredited, and the identification of cognates has proved so formidable or indeed impossible a task without an in-depth expertise in the histories of the languages concerned, that one can only marvel at the recent surge of neoglottochronological enthusiasm and its gullible reception in high-profile science journals and the general press” (p. 70). It should be remarked that, as far as “neo”-glottochronology is concerned, Plank seems to only be acquainted with the well-publicized works of the Gray & Atkinson group (mentioned above with respect to the article of Carlotta Viti), whose methods were certainly let down by the poor quality of input lexicon material and, consequently, even poorer output results of dating. In fact, Plank confesses this in his own words: “Only one characteristic recent paper shall be mentioned, owing to its exceptional misproportion between rhetorical flourish and phylogenetic sophistication on the one hand and historical linguistic substance on the other: Greenhill, Atkinson, Meade & Gray 2010”. Different modern approaches to glottochronology, such as represented, e.g., in the project “The Global Lexicostatistical Database” (G. Starostin 2011– 2016), are not taken into account by Plank in his paper. As for his own ideas on time change, Plank suggests that there is a theoretical minimum for an elementary syntactic change, which takes three generations: — individuals innovate; — variation appears in the speech community; — whole speech community follows the innovators. According to the author, the loss of dual number in Attic Greek took precisely that minimum time, while in other languages this process took much longer: Old English is just one example, with over 600 years, or 25+ generations, for the change to take effect. Different changes, therefore, demand different spans of time. Plank advocates the idea that there is a list of possible

Carlotta VITI (ed.). Perspectives on historical syntax (2015)

parameters for change, such as abruptness/gradualness, simple or complex innovation, social diffusion of the society, and so on (7 categories suggested in the paper): “Change should be rapid, reaching completion within the minimum span of three generations, if all is easy: simple actuation; abrupt transition; Neo-grammarian mode of implementation; elementary change; discernible, high-profile difference; decisive individuals; small, homogeneous, well-connected community”. One particular case is investigated, namely, the grammaticalization of the local adposition ‘at’ from the noun ‘dwelling, home’. The aim of this case study is to determine the length of time for this change, and to compare its pace between several languages where it has occurred. Relevant instances are French chez ‘at’ from Late Latin casa/chiés; Swedish, Danish, Norwegian hos ‘at’ from Old Norse hus; Icelandic and Faroese hjá ‘at, next to, by, with; of’ from Old Norse hión ‘family, household’; and late Pāli gē ‘at; of’ from Prakritic Indo-Aryan geha. The author shows that this change took approximately the same time (about 400 years = approximately 16 cycles of acquisition) to be completed. The second part of the book, given over to issues of syntactic reconstruction, opens with a paper contributed by Thomas Smitherman (University of Bergen), called “Reconstructing non-canonical argument structure for Proto-Indo-European: methodological questions and progress”. The paper discusses methodological issues that have arisen over the investigation into the likelihood that oblique subject constructions in Indo-European languages are inherited from ProtoIndo-European. A four-year project, Indo-European Case and Argument Structure in a Typological Perspective (IECASTP, led by Jóhanna Barðdal, University of Bergen, in 2008–2012), had attempted to apply the comparative method to syntax, which allows Smitherman to discuss the difficulties encountered by the researchers. As a starting point, he assumes that a syntactic reconstruction may be less reliable compared to a lexical one, but the reason for that is an extra layer of complexity — it should be based on a thorough reconstruction of phonetics, phonology, morphology (with complete understanding of allomorphy), formal and semantic aspects of lexicon. There are certain formal approaches to description of syntactic constructions, and a syntactic reconstruction of a language might look like an inventory of its possible constructions. IECASTP attempted to provide an example of how formal representations might work for PIE syntactic reconstructions (see Barðdal & Smitherman 2013). These representations include reconstruction of predi-

cates (as heads), all separate word forms, cases, semantic roles of arguments and argument structure of the predicate, which constitute a kind of construction grammar. This grammar can, indeed, be used as formal means to compare syntactic units. As described in the paper, the approach is rather close to dependency grammar, which has been actively used in treebanks, including ones for ancient languages (see below on Dag Haug and the PROIEL project). It certainly helps to enforce uniformity and provide an instrument to make comparisons on syntactic level; still, this does not necessarily mean that a certain syntactic construction in Latin has the exact same meaning as, say, in Hittite, which marks the weak point of this approach. The author does not, however, insist on generalizations on the current level of historical syntactic studies: theoretically-determined interpretations, according to Smitherman, should only be attempted “after the empirical data have been examined, after comparisons between languages have been conducted”. The project has succeeded in gathering lists for predicates with argument structures, in which oblique subjects appear, from the oldest languages of IndoEuropean branches: Old Icelandic, Old High German, Middle High German, Gothic, and Old Russian; Latin, Ancient Greek (Homeric to Early Koiné), Old Church Slavonic, Old English, and Old Swedish, Sanskrit and Hittite (partially). For each predicate a PIE etymology was drawn where possible (phonetic reconstruction is based on laryngeal theory, under the assumption of three laryngeals and no vowel-initial morphemes). If a predicate is supposed to be an early borrowing into one IE branch from another (like some German borrowings into Common Slavic), it is not counted on the level of Indo-European etymological comparison. Study of argument roles for the predicate involves analysis of the semantics of affixes and preverbs. Some verbal affixes are assumed to have aspectual values (like ­ske- in Hittite); preverbs in some IE languages might evolve from postpositions or deictic adverbs, which could determine the case of arguments. Semantic transfers in verbs are also checked. IECASTP guidelines identify common semantic correlations as being linked to a single PIE predicate: e.g., burn — be angry or suffer an uncontrollable sensation; bend/twist — be confused/be in pain; eat/consume — be overcome; be light/heavy — have it easy/difficult, etc. The preliminary results are as follows. Roughly 200 cognate sets in 2+ branches, and 90 sets in 3+ branches (Baltic and Slavic are not counted separately) have been analysed for the etymology of verbs and their polysemy, case frames and distribution of oblique subjects. The working hypothesis is that “late PIE had 219

Book Reviews / Рецензии

a contained, probably unproductive or barely productive realm of semantic alignment within a generally Nominative-Accusative language”. According to the author, this ensues from patterns of use of the argument structure to accommodate polysemy, recurrent throughout many branches. Basic methodological problems concern early borrowings between branches; areal contacts, with similar argument structures on some cognate predicates; comparison of non-cognate like word classes, e.g. comparing Latin deponent forms (with *­r), GræcoAryan (*­oi), Slavic or Old Norse neo-formations involving the reflexive pronoun. Another important problem is whether to reconstruct sememes or forms — if several forms have the same or almost the same meaning, should they be considered separately or not? Smitherman does not give any clear answer in his article. He suggests that focused diachronic frequency studies should be conducted in the case of specific sememes for oblique subject construction in IndoEuropean. A controlled test should be invented that could play the role of a Swadesh-type wordlist for syntactic constructions, before we could claim with any certainty what semantic alignment there was in PIE. Summing up, he argues that a reconstruction of oblique subject constructions for PIE is possible, though the use of the Comparative Method for syntax certainly needs further review and refinement. The next paper of the book under review is “An approach to syntactic reconstruction” by Ilja A. Seržant. It is primarily devoted to the methodological discussion of how to reconstruct syntactic patterns. The author distinguishes between two types of inquiries into diachronic syntax: stage reconstruction and etymological reconstruction (p. 117). He focuses on the second one and argues for a methodology based on the principles of the Comparative Method, where all factors other than inheritance should be excluded by the reconstruction process: “Typologically quirky, idiosyncratic features are better reconstructable than typologically ordinary ones”. Seržant applies his method to the development of the independent partitive genitive (IPG) from Proto-Indo-European into Baltic and Russian, and finally into North Russian dialects, to show that this feature was indeed inherited from PIE and how it changed from PIE. The method crucially relies on typologically idiosyncratic properties of every pattern to be reconstructed. For example, morphological properties, as regards their phonetic/phonological realization, are typologically idiosyncratic. The more idiosyncratic properties are found to correlate across comparanda, the higher is the probability of the reconstruction. 220

Since (syntactic) categories never remain the same through time, syntactic reconstruction deals rather with clusters of properties that mutate through time: certain properties may persist while others may drastically change or get lost and new ones can be acquired. The author emphasizes that “superficially similar constructions may in fact have quite divergent underlying syntactic structures at different developmental stages”. A grammatical category, therefore, is treated as a list or as a cluster of properties, with each subgroup analyzed separately. There are four types of profiles for the analysis — lexical, semantic, morphological and syntactic ones: “Profiles of the category can be established in the course of synchronic analyses at every particular stage where data are available”. The reconstruction of the morphological and lexical profiles on a proto-stage can be carried out by means of the Comparative Method. However, “the degree of probability” depends on “the number of idiosyncratic properties” inherited from the respective proto-language on the basis of the Comparative Method (Ivanov 1965: 185). It is only the Comparative Method, applied correctly, that helps to get rid of borrowings in the morphological profile, excluding typologically dominant correlations or correlations that are due to language contact. The syntactic and semantic profiles have to be explored for typologically quirky properties in order to individualize the reconstructed pattern against the typological background and thus claim sufficient probability. The following ranking of profiles represents their relevance for determining etymologically cognate categories across related languages (ranked from most to least crucial): morphological profile > lexical profile > syntactic profile > semantic profile

Speaking about the IPG and the changes it underwent from PIE to Baltic and East Slavic, Ilja Seržant discusses first the morphological and lexical profiles of the construction. The inheritance of morphology and lexicon from PIE to Baltic/Slavic languages was thoroughly proven in previous studies. The difference between the genitive in Baltic/Slavic and in PIE is seen by Seržant, particularly, as loss of all morphological difference between ablative and genitive throughout the singular in the former, “while the latter still distinguishes these cases for one specific NP type, namely, the o-stems”. On the lexical level, it is important that there are reconstructible lexemes that occurred in the construction. Derivational means that are part of the lexeme should not be glossed over,

Carlotta VITI (ed.). Perspectives on historical syntax (2015)

“because different morphological derivations, especially with verbs, may be linked to distinct syntactic patterns, e.g. causatives vs. simplices or denominal vs. deverbal predicates are known to trigger distinct syntactic patterns”. The semantic profile is described on p. 134 as a list of possibly inherited functional properties for which values are drawn for PIE and Baltic/Slavic. The comparison exhibits a number of particular changes in the partitive genitive though its development: — ability to quantify over the host constituent or over the whole clause, — sensitiveness to adverbs quantifying the situation (VP), — sensitiveness to verb-prefixal quantifiers, — invoking the meaning of a temporality (‘for some period of time’) with transfer verbs, — ‘one’ as a possible value of the implicit (head) quantifier, — combination with verb negation, — interaction with aspectuality, — decreased referentiality, — discursive backgroundedness, — gradual loss of the partitive function; prevalence of the pseudo-partitive function, — partitivity constraint, — partial loss of the differential object marking. The syntactic profile consists of five properties: selection restrictions on NPs marked by the IPG (e.g. mass vs. count nouns), selection restrictions on verbs with subject IPG (e.g. existential vs. unergative), verbal agreement with subject IPG, coordination with otherwise case-marked NPs, positional restrictions. Thus, morphological and lexical profiles provide a relatively high probability for the assumption of etymological relationship between the IPG of Baltic and Slavic languages and the same structure in PIE, reconstructed on the basis of ancient IE languages such as Sanskrit, Avestan and Ancient Greek. After the analysis of syntactic properties, the author argues that the IPG in Baltic and Slavic (Russian) languages may be analyzed as a syntactically independent partitive genitive, governed by an implicit pronoun, for which the term pro is used, thus assuming the existence of an implicit head for this construction. The author reminds us that there is no restriction on syntactic position for the IPG in ancient IE languages, which is another reason why he argues for the implicit pronoun assuming case and position in the clause. The same concerns singular/plural of verbs in the partitive constructions — it is the pro which assumes number and person and becomes visible due to its ability to be the controller in

the subject position triggering verbal agreement. He argues that this implicit pronoun (zero head) in Baltic and Russian became even less visible in the morphosyntax and, comparing with PIE, retained only a weak ability to coordinate with accusatives and, partly, nominatives (triggering the default third singular neuter/non-agreeing form). Finally, in those instances where some North Russian varieties allow for the agreement ad formam with the IPG subject, the implicit pronoun may be considered to be lost entirely and the former dependent genitive NP acquires direct access to verbal agreement. The general development of the IPG, thus, can be summarized as: explicit head (dependent partitive genitive) —> “pro” (PIE/ ancient IE languages) —> “PRO” (Baltic/Russian) —> null (some North Russian subdialects). The next paper, “Anatolian syntax: inheritance and innovation”, was contributed to the volume by Annette Teffeteller. It is dedicated to three interrelated topics in the syntax of Anatolian languages: the issue of argument structure, the putative split-ergativity, and the development of subject clitic pronouns. Actually, Annette Teffeteller is reproducing here her own talk at the VIIIth International Congress of Hittitology in Warsaw in 2008, where it was received with relatively little enthusiasm; the problems encountered there remain largely unsolved in the paper, which seriously restricts its usefulness for future discussion on historical syntax. In two areas concerned with subject reference, Anatolian languages display unique syntactic features. First, there is a third-person enclitic ‘subject’ pronoun, marked for gender, common and neuter, restricted to a particular class of verbs (intransitives only, predominantly statives). Second, there is a suffix used with neuter nouns when they occur in correlation with the subject of a transitive verb. Both are topics with a long history of discussion, and for both there is still no consensus as to their origins. Unfortunately, the author adds no new information to the discussion. Her analysis of Anatolian data in the paper is largely restricted to Hittite material, and the examples are mostly not Teffeteller’s own, but have been taken from other works, such as Melchert 2011. She laments that the most prominent syntacticians working in the field of Anatolian languages tend to use generative syntactic theory (see, for example, Hoffner & Melchert 2008:406; Sideltsev 2011), where subject pronouns are treated as null subject. Teffeteller suggests using another framework for Hittite, borrowed from the research on North American languages, according to which personal endings of verbs might be considered as verbal subject markers, i.e. incorporated pronomi221

Book Reviews / Рецензии

nal elements (Jelinek 1984). Nouns, NPs, and independent pronouns in this case are considered as adjuncts, regardless of whether they are found outside or inside the clause. The verb thus constitutes a complete minimal clause in itself. However, the author gives no clear reasons why this concept should be used instead of the more widespread concept of null subject. Teffeteller argues that “the absence of grammatical agreement markers (null subject) is a typologically rare phenomenon”, with reference to Siewierska 1999, and that the generative framework only works for modern European languages. This sounds rather strange, since generative syntax has been tested many times on languages outside of the Indo-European family, and, actually, is now widely used in typological research on the world’s languages. As for the problem of ‘ergativity’ in Hittite, this is an old discussion, in which the specific Hittite ‘ergative’ suffix ­ant- (added to neuter nouns if they are subjects) is sometimes viewed as derivational, and sometimes as inflectional, with a special ‘ergative’ case in the Hittite noun declension paradigm (see recently, inter alia, Melchert 2011, Yakubovich 2011, Goedegebuure 2013). Annette Teffeteller traces this discussion in detail, listing all the arguments pro and contra, and may be safely referred to as a source for the most recent references on the question. In the chapter “Historical syntax and corpus linguistics” the most prominent projects of annotated corpora for historical languages are represented. The opening paper is by Dag Haug from the University of Oslo, the leader of PROIEL, a unique public on-line resource for syntactically annotated corpora of ancient languages, built in the framework of universal dependencies (UD). He argues in the paper for the advantages of using parsed corpora (treebanks) for research in historical linguistics. One important example is basic word order. Raw statistical data on word order in Ancient Greek differs between researchers (see p. 189 for figures on word order in Luke/Acts, according to various authors). The author points out an important question of historical syntax: if we cannot agree even on the raw facts, how can we settle such questions as what (if any) basic word order there was in Ancient Greek, or to what extent it was influenced by Semitic? Another thing is that the results of the research should also be replicable by other scholars, and it is only the corpus approach that could help us achieve this. Initially, PROIEL had developed a parsed corpus of the Greek New Testament as well as several of its early translations into other languages (Haug & Jøhndal 2008; Haug et al. 2009). The paper in question 222

focuses on the description of the oldest part of the project. However, it should be mentioned here that, as of now, the project not only contains data from Ancient Greek (New Testament, Historia Lausiaca, Herodotus: Histories, Sphrantzes: Chronicles), but also from Church Slavonic (Codex Marianus, Codex Suprasliensis, Codex Zographensis), Classical Armenian (New Testament, Koriwn), Gothic (The Gothic Bible), and includes a list of sources in Latin, Old English, Old French, Old Norse, Old Russian, Portuguese and Spanish. The author argues that “a treebank does not in itself define the actual assumptions of research based on it, but it defines the set of possible assumptions that a researcher can make using it”. There are several ways to avoid pre-assumption. Phrase structure based corpora, such as the Penn Treebank (actually, the family of corpora from the Linguistic Data Consorium at the University of Penn), use a much flatter phrase structure than any practitioners of theoretical phrase structure grammars assume, thereby avoiding many contentious decisions. The other option, which was chosen in the PROIEL corpus, is to use a dependencybased analysis, where grammatical relations, such as subject, object, and adverbial, are taken as primitive. Being on the team of linguists and programmers that work on the standards of Universal Dependencies, Dag Haug could have hardly made a different choice. Unfortunately, syntactic annotation in the UD scheme treats the syntax of the world’s languages as if no language-specific features existed in the first place (annotation of language-specific relations as subtags of existing universal tags does not help much). For example, one problematic issue with UD is clitics, with their specific syntax.1 Another problem is the very absence of any assumption in treebanks — in fact, this framework gives the researcher no proper explanation of syntax. However, in terms of pure data PROIEL, with its standards of merely building an improved instrument for search, so far remains the best, if not the only, means of applying statistical methods to the material of early IE languages. The paper of Prof. Dr. Rosemarie Lühr from Humboldt-University of Berlin (“Traces of discourse configurationality in older Indo-European languages?”) concerns the relationship between information structure and syntax on the material of Old Indian, Ancient

1 Joakim Nivre, University of Uppsala, another member of the UD team, informed me in a pers. comm. during his lecture on Universal Dependencies in Moscow, Yandex campus, 20 April 2016, that he had no proper guidelines for annotating, for instance, Hittite subject enclitics.

Carlotta VITI (ed.). Perspectives on historical syntax (2015)

Greek and Hittite, languages with the documented discourse-configurational word order topic — focus — verb. There are many deviations from this word order in the early IE languages. As in the case of a preceding predicate noun in Old Indian, the sequence topic — focus can be inverted. Similarly, a shifting topic may appear at the end of a clause if the first/initial position is taken by a contrastive focus. The positions of contrastive focus are shown in the paper on examples from Ancient Greek and Hittite. Regrettably, Prof. Lühr does not take into account the works of Petra Goedegebuure (2013, 2014), who specifically discussed the types of foci in Hittite and their positions in the clause. This shortcoming of the article under review might be partially explained by its being presented for the first time in September 2012. Still, at the time when the reviewed volume was edited, the abovementioned papers of Goedegebuure, specifically the one concerning focused noun phrases (2013), had already been published and should have been known to Prof. Lühr. Actually, she presents here the preliminary results of her own project in Humboldt-University on word order corpus research in early IE languages. The project took around eight years, was finished in 2015 and has been discussed in several workshops of 2015 and 2016. One presented result is that the marked word order OSV, closely connected with the position of focus, in Hittite appears in 50% of all involved material.2 Unfortunately, the corpus itself is not in the public domain, and there is no way for an independent check. Our own Hittite material (letters and instructions), when subjected to corpus-based analysis of the distribution of OSV, demonstrates rather low values — around 15% (Molina 2015). The author summarizes the paper with the notion that the position of information-structural entities in the old IE languages is inherited from PIE, and that only Greek has demonstrated in the study an innovation specified as “the postverbal new-information focus position”, triggered by the verb moving into the middle position. As has already been said, all the details leading to this assertion should be independently double-checked on corpus material, which strongly demands historical corpora made for the research to be opened for the public. The chapter concerning corpus research for historical languages is continued with the paper “Studying word order changes in Latin: some methodological remarks”, contributed by Lieven Danckaert (Ghent

2 This information was made public at the Workshop “The precursors of Indo-European: The Indo-Hittite and Indo-Uralic hypotheses”, Leiden University, June 2015, and was later discussed personally with Prof. Lühr.

University). He argues that “a linear string of Latin words can correspond to more than one syntactic structure”, and offers a detailed case study on the often discussed OV/VO alternation in the history of Latin (from 1st c. BC to 6th c. AD). The author postulates multiple positions for the object in the Latin clause and demonstrates that “the objects in three different positions are all to be interpreted in a different way”. Two hypotheses are formulated for languages that display variable OV and VO ordering: one, that the choice between possible word orders is influenced by the variety of different usage-based factors, such as information structure, weight and complexity of the object, but the factor itself should not have much influence on the syntactic position of direct objects. Second, quantitative results that emerge from a study that only takes into account syntactically non-ambiguous environments provide a more accurate characterization of the syntactic changes that took place during the evolution from Latin towards the (early) Romance languages. The case study presented in the paper took into account at least 20 clauses with an auxiliary, with a transitive nonfinite verb and an overt direct object for each period and source, “in order to be sure that the calculation of the average values of VO and OV is based on sufficient amount of tokens and thus provides a reliable estimate”. Surprisingly, the statistical data demonstrated that no statistically significant rise of VO could be spotted, in strong contrast with what is commonly assumed. Anna Bonifazi’s article is titled “Problematizing syndetic coordination: Ancient Greek ‘and’ from Homer to Aristophanes”. It discusses the interpretation of three particles, te, kaí and dé, which function as coordinators with the general meaning ‘and’ in Archaic and Classical Greek. Bonifazi focuses on discourse phenomena that cause syntactic distinctions between these particles. She demonstrates that multiple words with the meaning ‘and’ reflect a specific communicative need: for example, te may pragmatically imply shared knowledge, or may indicate a certain genre, while kaí between two conjuncts may be used to indicate a conceptual unity. Overall, summarizes Bonifazi, the range of usage for te, kaí, and dé encompasses a continuum between connective and adverbial functions. The last paper in this chapter concerns epigraphic corpora: “What role for inscriptions in the study of syntax and syntactic change in the old Indo-European languages?”, by Francesca Dell’Oro. She regrets that scholars who deal with syntactic problems, especially of a theoretical nature “tend to dismiss inscriptional records of early IE languages as being ‘not useful’ or ‘too difficult to investigate’”, — although this assertion would seem unfair if one takes into account cer223

Book Reviews / Рецензии

tain cases of thorough investigation of epigraphic sources by Indo-Europeanists, it is indeed true that syntactic research rarely looks into inscriptions for input data. The paper provides a lot of examples from Greek epigraphics, including ones containing syntactic errors that help to reconstruct certain syntactic patterns, but also contains a series of rather commonplace assertions, such as “it is not easy to investigate problems of syntactic change or syntactic reconstruction on the basis of epigraphic material”. The final chapter, concerning questions of historical syntax and linguistic contact, contains two papers based on non-Indo-European material. One of them is a case study of Guinea creole languages (“The Gulf of Guinea creoles: a case-study of syntactic reconstruction”) by Tjerk Hagemeijer; the other is “Syntactic diversity and change in Austroasiatic languages” by Mathias Jenny. Upon first sight, both seem to concern issues that are only tangentially related to the main focus of the volume, but in fact they offer a wider understanding of the problems of syntactic reconstruction, discussing material that usually remains untouched by mainstream researchers in the respective fields. The first article shows that creoles may constitute fertile ground with respect to the reconstruction of syntax, given a high degree of structural identity between sister languages and the fact that many shared syntactic properties, such as discontinuous sentence negation, must have been inherited and diffused from the protolanguage. The second one looks into the syntactic diversity of Austroasiatic languages, where historical data are available only for a small number of units, and seeks possible explanations for the development of this diversity. Two main factors seem to trigger syntactic change, namely, reanalysis and contact influence from neighboring languages, and insights drawn from languages with lengthy recorded histories could help to understand the development of languages with no historical data.

References Barðdal, Jóhanna. 2001. Case in Icelandic: A Synchronic, Diachronic and Comparative Approach. Ph.D. dissertation, University of Lund. Barðdal, Jóhanna, Thórhallur Eythórsson. 2012. Reconstructing syntax: Construction grammar and the comparative method. In: H. C. Boas, I. Sag (eds.). Sign-Based Construction Grammar. Stanford CA: CSLI: 257–308. Barðdal, Jóhanna, Thomas Smitherman. 2013. The quest for cognates: A reconstruction of oblique subject constructions in Proto-Indo-European. Language Dynamics and Change 3(1): 28– 67.

224

Batllori, Montserrat, Maria-Lluïsa Hernanz, Carmen Picallo, Francesc Roca. 2005. Grammaticalization and Parametric Variation. Oxford: OUP. Bauer, Brigitte. 1995. The Emergence and Development of SVO Patterning in Latin and French: Diachronic and Psycholinguistic Perspectives. Oxford: OUP. Bauer, Brigitte. 2000. Archaic Syntax in Indo-European: The Spread of Transitivity in Latin and French. Berlin / New York: Mouton de Gruyter. Brandner, Ellen, Iris Bräuning. 2013. Relative wo in Alemannic: Only a complementizer? Linguistische Berichte 234: 131–169. Crespo, Emilio, José Luis García Ramón (eds.). 1997. Berthold Delbrück y la sintaxis indoeuropea hoy. Actas del Coloquio de la Indogermanische Gesellschaft. Madrid, 21–24 de septiembre de 1994. Wiesbaden: Reichert. Crisma, Paola, Giuseppe Longobardi (eds.). 2009. Historical Syntax and Linguistic Theory. Oxford: OUP. Devine, Andrew, Laurence Stephens. 1999. Discontinuous Syntax. Hyperbaton in Greek. Oxford: OUP. Goedegebuure, Petra. 2013. Hittite Noun Phrases in Focus. In: S. W. Jamison, H. Craig Melchert, B. Vine (eds.). Proceedings of the 24th Annual UCLA Indo-European Conference. Bremen: Hempen: 27–45. Goedegebuure, Petra. 2014. The Hittite Demonstratives. Studies in Deixis, Topics and Focus. Wiesbaden: Harrassowitz. Gray, Russell, Quentin Atkinson. 2003. Language-tree divergence times support the Anatolian theory of Indo-European origin. Nature 426: 435–439. Greenhill, Simon J., Quentin D. Atkinson, Andrew Meade, Russell D. Gray. 2010. The shape and tempo of language evolution. Proceedings of the Royal Society B: Biological Sciences 277: 2443–2450. Güldemann, Tom. 2008. Quotative Indexes in African Languages: A Synchronic and Diachronic Survey. Berlin: Mouton de Gruyter. Faarlund, Jan. 1990. Syntactic Change: Towards a Theory of Historical Syntax. Berlin: Mouton de Gruyter. Ferraresi, Gisella, Maria Goldbach (eds.). 2008. Principles of Syntactic Reconstruction (Current Issues in Linguistic Theory 302). Amsterdam: John Benjamins. Ferraresi, Gisella, Rosemarie Lühr (eds.). 2010. Diachronic Studies on Information Structure. Language Acquisition and Change. Berlin: Mouton de Gruyter. Harris, Alice, Lyle Campbell. 1995. Historical Syntax in Crosslinguistic Perspective. Cambridge: CUP. Haspelmath, Martin. 2012. Equative constructions in a worldwide perspective. Paper given at the conference “Expressions of Similarity from an Africanist and Typological Perspective”, Villejuif, July 2012. Haug, Dag, Marius L. Jøhndal. 2008. Creating a parallel treebank of the old Indo-European bible translations. In: Caroline Sporleder, Kiril Ribarov (eds.). Proceedings of the Second Workshop on Language Technology for Cultural Heritage Data (LaTeCH 2008): 27–34. Haug, Dag, Marius L. Jøhndal, Hanne Eckhoff, Eirik Welo, Mari Hertzenberg, Angelika Müth. 2009. Computational and linguistic issues in designing a syntactically annotated parallel corpus of Indo-European languages. Traitement Automatique des Langues 50: 17–45. Hewson, John, Vit Bubenik. 2006. From Case to Adposition: The Development of Configurational Syntax in Indo-European Languages (Current Issues in Linguistic Theory 280). Amsterdam: John Benjamins.

Carlotta VITI (ed.). Perspectives on historical syntax (2015)

Hoffner, Harry, Craig Melchert. 2008. A Grammar of the Hittite Language. Winona Lake IN: Eisenbrauns. Jelinek, Eloise. 1984. Empty categories, case, and configurationality. Natural Language and Linguistic Theory 2: 39–76. Jonas, Diane, John Whitman, Andrew Garrett (eds.). 2012. Grammatical Change. Origins, Natures, Outcomes. Oxford: OUP. König, Ekkehard. 2013. The deictic identification of similarity. In: Y. Treis, M. Vanhove (eds.). Similative and Equative Constructions: A Cross-linguistic Perspective. Amsterdam: John Benjamins. Ivanov, Vjacheslav V. 1965. Obshcheindoevropejskaja, praslavjanskaja i anatolijskaja jazykovye sistemy (sravnitel’no-tipologicheskie ocherki). Moscow: Nauka. Lehmann, Winfred. 1974. Proto-Indo-European Syntax. Austin TX: University of Texas Press. Lehmann, Winfred. 1976. From topic to subject in IndoEuropean. In: Charles N. Li (ed.). Subject and Topic. New York NY: Academic Press: 447–456. Lehmann, Winfred. 2000. Pre-Indo-European (Journal of IndoEuropean Studies Monograph Series 41). Washington DC: Institute for the Study of Man. Lightfoot, David. 1979. Principles of Diachronic Syntax. Cambridge: CUP. Lightfoot, David. 1991. How to Set Parameters: Arguments from Language Change. Cambridge MA: MIT Press. Lightfoot, David. 1999. The development of language: acquisition, change, and evolution. Oxford: Blackwell. Lightfoot, David (ed.). 2002a. Syntactic effects of morphological change. Oxford: OUP. Lightfoot, David. 2006. How new languages emerge. Cambridge: CUP. Longobardi, Giuseppe. 2003. On parameters and parameter theory. In: E. Stark, U. Wandrusza (eds.). Syntaxtheorien: Modelle, Methoden, Motive. Tübingen: Narr: 273–290. Longobardi, Giuseppe, Cristina Guardiano. 2009. Evidence for syntax as a signal of historical relatedness. Lingua 119: 1679– 1706.

Luraghi, Silvia. 2010. The rise (and possible downfall) of configurationality. In: Silvia Luraghi & Vit Bubnik (eds). A Companion to Historical Linguistics. London: Continuum: 212–229. Melchert, Craig. 2011. The problem of the ergative case in Hittite. In: M. Fruyt, M. Mazoyer, D. Pardee (eds.). Grammatical Case in the Languages of the Middle East and Europe: Actes du colloque international “Variations, concurrence et évolution des cas dans divers domaines linguistiques”. Paris 2–4 April 2007. Chicago IL: Oriental Institute of the University of Chicago: 161–167. Molina M.A. 2015. Markirovannyy poryadok slov OSV v khettskom yazyke, ego funkcii i regulyarnost’ [Marked word order OSV in Hittite, its functions and regularity]. Indoevropeyskoye yazykoznanie i klassicheskaya filologiya 19. SaintPetersburg: Nauka: 655–663. Molina M.A. 2016. Review of: L. Kulikov, N. Lavidas (eds.). Proto-Indo-European syntax and its development (2015). Voprosy yazykoznaniya 5: 141–146. Renfrew, Colin, April McMahon, Robert Trask (eds.). 2000. Time Depth in Historical Linguistics. Cambridge: McDonald Institute for Archaeological Research. Roberts, Ian. 2007. Diachronic Syntax. Oxford: OUP. Sideltsev, Andrey V. 2011. Clitic doubling: A new syntactic category in Hittite. Altorientalische Forschungen 38(1): 81–91. Siewierska, Anna. 1999. From anaphoric pronoun to grammatical agreement marker: Why objects don’t make it. Folia Linguistica 33(2): 225–251. Greville G. Corbett (ed.). Special issue “Agreement”. Starostin G.S. (ed.). 2011–2015. The Global Lexicostatistical Database. Moscow/Santa Fe: Center for Comparative Studies at the Russian State University for the Humanities; Santa Fe Institute. Available: http://starling.rinet.ru/new100 Yakubovich, Ilya. 2011. Privative ergativity in Hittite. Handout for the presentation at the conference “Historical-Comparative Linguistics in the 21st Century”, Pavia, Italy, 22–25 September 2011.

225

Anton I. Kogan Institute of Oriental Studies of the Russian Academy of Sciences (Russia, Moscow); [email protected]

Genealogical classification of New Indo-Aryan languages and lexicostatistics Genetic relations among Indo-Aryan languages are still unclear. Existing classifications are often intuitive and do not rest upon rigorous criteria. In the present article an attempt is made to create a classification of New Indo-Aryan languages, based on up-to-date lexicostatistical data. The comparative analysis of the resulting genealogical tree and traditional classifications allows the author to draw conclusions about the most probable genealogy of the Indo-Aryan languages.

Keywords: Indo-Aryan languages, language classification, lexicostatistics, glottochronology.

The Indo-Aryan group is one of the few groups of Indo-European languages, if not the only one, for which no classification based on rigorous genetic criteria has been suggested thus far. The cause of such a situation is neither lack of data, nor even the low level of its historical interpretation, but rather the existence of certain prejudices which are widespread among Indologists. One of them is the belief that real genetic relations between the Indo-Aryan languages cannot be clarified because these languages form a dialect continuum. Such an argument can hardly seem convincing to a comparative linguist, since dialect continuum is by no means a unique phenomenon: it is characteristic of many regions, including those where IndoEuropean languages are spoken, e.g. the Slavic and Romance-speaking areas. Since genealogical classifications of Slavic and Romance languages do exist, there is no reason to believe that the taxonomy of Indo-Aryan languages cannot be established. However, this scholarly pessimism does have some grounds. Certain approaches and methods used nowadays in Indological historical linguistics have proven to be inefficient, and without changes in research methodology, significant progress in the genetic classification of the Indo-Aryan languages is hardly possible. This issue will be discussed at some length below. In the past Indologists had more than once attempted to classify the languages that they studied. In the late 19th and 20th centuries several alternative classifications were suggested. Since some of them still appear in modern Indological publications, I feel it necessary to dwell on them here. The scholar who seems to have been the first one to approach this problem is Rudolf Hoernle. He hypothesized that the Aryans migrated to the Indian subcontinent in two successive waves. The migrants belonging to these waves spoke two different dialects of Old Indo-Aryan, which he called Magadhi and Sauraseni (Hoernle 1880).1 Magadhi, according to him, was the common ancestor of modern languages spoken in the South and East of the IndoAryan speaking area, i.e. of Marathi, Konkani, Bengali, Oriya, and Bihari dialects,2 whereas Sauraseni was considered to constitute the protolanguage for the forms of speech current in 1 2

These names should not be confused with the identical names of literary Prakrits (Masica 1991: 447). Hoernle preferred to group these dialects under the name of Eastern Hindi.

Journal of Language Relationship • Вопросы языкового родства • 14/4 (2016) • Pp. 227—258 • © The authors, 2016

Anton I. Kogan

Figure 1. Genealogical classification of Indo-Aryan languages according to G. Grierson (adapted from Masica 1991: 449).

the North and West, i.e. for Nepali, Garhwali, Kumauni, Gujarati, Sindhi, Punjabi including Multani, and Western Hindi including Rajasthani.3 Hoernle was of the opinion that the speakers of “Magadhi” once occupied the entire North India but were later pushed back by “Sauraseni” speakers, and it is for this reason that the languages of the Northwest still possess certain vestigial features that are common with the southeastern languages. Hoernle’s idea of a two-wave migration was taken over and further developed by George A. Grierson — a British scholar, the main author of the famous “Linguistic Survey of India”. His classification of Indo-Aryan languages included two main subbranches, which he called Inner and Outer, although they did not precisely correspond to Hoernle’s Sauraseni and Magadhi respectively. The main point of divergence between Grierson’s and Hoernle’s models was the position of the languages spoken in the Northwest of the subcontinent, namely Sindhi and the dialects of Western Punjab including Multani.4 Grierson preferred to include these forms of speech into the Outer subbranch, i.e. to group them together with Marathi, Konkani and the languages of Eastern India (Grierson 1927). Grierson’s arguments in favor of this point of view will be discussed separately below. Moreover, alongside the main subbranches embracing the bulk of the Indo-Aryan languages, Grierson’s classificatory scheme also featured a third one: it was called Mediate and included certain Hindi dialects, the most important of which is Awadhi.5 These dialects, according to Grierson, possessed both “Inner” and “Outer” features. Grierson’s classification is reproduced on Figure 1 in the form of a genealogical tree. 3

To this group Hoernle also added Pashto and Kashmiri, which he reckoned among the Indo-Aryan languages. 4 These dialects, spoken in the vast area covering northern and western parts of the present-day Pakistani province of Punjab, were grouped by Grierson under the name of Lahnda. This term is still rather popular among the Indologists. 5 The other two are Bagheli and Chhattisgarhi. 228

Genealogical classification of New Indo-Aryan languages and lexicostatistics

As we can see, both Inner and Outer languages are subdivided into further groups, i.e. Pahari,6 Central, Eastern, Southern and Northwestern. Grierson gave no clear-cut reasons for the postulation of these groups or the existence of such entities as “Lahnda”, “Western Hindi” or “Rajasthani”. The two main subbranches, however, were established by him on the basis of features which he thought to be diagnostic for classification. The most significant of them are: 1) retention of MIA s (< OIA s, ś, ṣ) in the Inner languages vs. change of this sibilant into other phonemes in the Outer subbranch; 2) loss of the final short vowels in the Inner subbranch vs. their preservation in the Outer languages; 3) the use of the suffix ­i- to form verbal perfect stems in the Inner languages vs. the formation of such stems in the Outer languages with the suffix ­l­; 4) the analytic typology of the Inner languages vs. the synthetic character of the Outer sub-branch. Upon close examination, none of the above arguments can be considered valid. The loss of final short vowels in a number of Indo-Aryan languages took place in the New Indo-Aryan period and thus has nothing to do with dialectal differences in Old Indo-Aryan. The same holds true for the formation of perfect stems. The use of the suffix ­l- for such a purpose is a comparatively recent phenomenon; moreover, it is not characteristic of all the Outer languages and is found in some Inner ones (e.g. in Gujarati). The development of Old Indo-Aryan sibilants was totally different in the East and Northwest of the subcontinent. In the East all the three sibilant phonemes have merged into one, i.e. ś. This reflex is already attested in the Magadhi Prakrit and found, e.g., in present day Bengali. In almost all the other Indo-Aryan languages, including Northwestern ones, the sibilants merged into s,7 sometimes with subsequent phonetic changes in certain positions. This means that reconstructing something like a “Common Outer” or “Proto-Outer” development of the Old Indian sibilant system is simply out of the question. As for the typological argument, it is generally accepted among comparative linguists that such arguments are not relevant for genealogical classification. In addition, Grierson’s statement that all the Outer languages belong to the same synthetic morphological type is not fully correct. In reality, Eastern Indo-Aryan languages possess agglutinative morphology of secondary origin, which developed from the older analytic system as a result of the transformation of function words (e.g. postpositions) into affixes, whereas Northwestern languages are mainly analytic, although sometimes they preserve a few vestiges of old inflection. Grierson’s arguments were analyzed by the Indian historical linguist and philologist Suniti Kumar Chatterji. In the introduction to his renowned work “Origin and development of the Bengali language” (Chatterji 1926), he managed to convincingly show their invalidity, as well as the incorrectness of the Inner-Outer model. As an alternative, he suggested his own classification, which is in certain respects similar to that of Grierson but without the Inner and Outer sub-branches as separate taxa.8 The reality of the Mediate subbranch was also denied. All the dialects that Grierson classified as Mediate were included by Chatterji into the Eastern 6

The Pahari group includes Indo-Aryan languages spoken in the sub-Himalayan region stretching from Nepal in the Southeast to the southern areas of the Indian state of Jammu and Kashmir in the Northwest. 7 Exceptions include several Pahari languages and Romany, which distinguish between two sibilants, i.e. s and ś. The latter reflects both ś and ṣ of Old Indo-Aryan. 8 Chatterji’s classificatory scheme was published in his abovementioned book as a part of the table illustrating the development of Aryan speech in India, see Chatterji 1926: 6. 229

Anton I. Kogan

group. Northwestern, Pahari,9 Southern, and Eastern groups were considered on the same taxonomic level. Moreover, Chatterji postulated two more subbranches, namely, Southwestern (including Gujarati and Rajasthani) and a subbranch consisting of Sinhalese and Maldivian (Dhivehi).10 The list of languages in some groups was somewhat different from the one offered by Grierson. Thus, both Punjabi and Romany were classified with the Northwestern group. The Central subbranch, called Midland by Chatterji, included only the forms of speech traditionally grouped together under the name of Western Hindi, i.e. standard Hindi and Urdu (with Khariboli as their common dialectal basis), Haryanvi, Braj, Kannauji and Bundeli. While working out this classification, Chatterji largely based himself on intuition. Only in relatively rare cases certain historical-phonological isoglosses were taken into account. Among the cited innovations one can mention the merger of the Old Indian sibilants into ś in the East or the simplification of the Middle Indo-Aryan geminates followed by compensatory lengthening of the preceding vowels in the majority of New Indo-Aryan languages vs. the absence of such a process in the Northwest, i.e. in Sindhi, Lahnda and Punjabi. Both Grierson’s and Chatterji’s schemes can even nowadays be occasionally found in Indological linguistic literature. They coexist with several alternative classifications, which were suggested during the last five decades. The latter, however, differ only slightly from Chatterji’s11 and, likewise, remain mainly intuitive. The 20th century saw great progress in the study of Indo-Aryan historical phonology. In addition to the already mentioned monograph on Bengali by Chatterji, Jules Bloch’s book on Marathi (Bloch 1920) and R.L.Turner’s works on Gujarati, Sindhi, Romany and Nepali (Turner 1921a; 1921b 1924; 1926; 1931) significantly extended our knowledge of sound change in a number of New Indo-Aryan languages. This research prompted some scholars to suggest a genealogical tree based on historical phonological isoglosses. Soon, however, it became clear that this task is extremely difficult, if not impossible. Isoglosses are sometimes easily detectable, but usually they cannot be brought together into bundles that would be peculiar for a particular language or language group. For example, the above-mentioned compensatory lengthening of short vowels before simplified geminates is characteristic of such languages as Hindi-Urdu, Gujarati, Marathi, Bengali and many others, but not of Punjabi, Lahnda or Sindhi. This fact makes it tempting to postulate two groups, one of which would include the three last idioms and the other would include all the rest. According to such a classification, Hindi-Urdu must be declared a language more closely related to Bengali and Gujarati than to Punjabi. But if we take another feature, e.g. the development of the Old Indic sibilants, the picture will be quite different. Bengali, where, as noted above, these phonemes have merged into ś, would form a separate group together with Assamese, for which “ś-reflexation” can be traced historically, whereas Hindi-Urdu will find itself closer to Punjabi, Lahnda, Sindhi and Gujarati, where the reflex is s. Gujarati, however, is not affected by another isogloss, which is common for many Indo-Aryan languages, namely the lenition of intervocalic ­m- into ­v- (sometimes with a further change into  and nasalization of the preceding vowel). In this respect, Hindi-Urdu shows close affinity to Marathi, Sindhi, Punjabi, Nepali, and Romany but not to Gujarati (cf. Hindi-Urdu gv, Marathi gāv, Sindhi gu, Punjabi girāũ, Nepali gāũ, Romany gav, but Gujarati gām ‘village’ < OIA grāma­). This complex situation, where different historical-phonological isoglosses are in “conflict” with each other, was tackled by Colin Masica in his book (Masica 1991). He analyzed the geographical distribution of six historical phonological features, i.e. compensatory lengthening, 19

For Grierson’s Pahari languages, Chatterji prefers the term “North” or “Himalayan”. The latter two languages were ignored in Grierson’s scheme. 11 These classifications are cited, e.g. by Colin Masica in Masica 1991: 454–456. 10

230

Genealogical classification of New Indo-Aryan languages and lexicostatistics

Figure 2. The distribution of some basic New Indo-Aryan phonological isogosses (adapted from Masica 1991: 459).

merger of the three OIA sibilants, OIA kṣ > (c)ch (> s), cerebralization of intervocalic MIA ­l­, voicing of voiceless stops after nasals, retention of OIA initial v- (elsewhere changed to b). The results of his analysis were presented in the form of the scheme reproduced on Figure 2. As can be plainly seen, the above data make it hardly possible to postulate a group of languages (i.e. a sub-group of Indo-Aryan) on the basis of more than one isogloss that would be peculiar only for this group, as opposed to all the rest. There can be no doubt that such a state of things is indeed the result of intensive and longlasting language contact, which in South Asia was and is still often facilitated by absence of natural barriers. The gradual spread of certain contact-induced phonological features can be traced back to Middle Indo-Aryan. For example, OIA intervocalic stops (other than cerebrals) were dropped in the Maharashtri Prakrit, while still being preserved in Śauraseni. But later on the disappearance of stops in the intervocalic position also affected the Śauraseni area, i.e. the central part of the Indo-Gangetic Plain. It is attested already in Apabhraṃśa, a late Middle Indo-Aryan literary language formed in this very region, and is characteristic of almost all the local New Indo-Aryan forms of speech. The merger of OIA sibilants in many areas of the Indian subcontinent likewise dates back to the Middle Indo-Aryan period. Thus, in the language of Aśoka’s inscriptions from most parts of India no distinction was made between the reflexes of OIA s, ś and ṣ. This threefold contrast was fully retained only in the inscriptions from the Northwest, which means that in Aśoka’s time, i.e. in the 3rd century BC, the isogloss in question did not affect this area. Later, however, the sibilants did merge in the Northwest as well. In the New Indo-Aryan dialects spoken there nowadays the results of this process can be clearly seen (cf. Lahnda sap(p) ‘snake’ < OIA sarpa­; sad(d) ‘call, shout’ < OIA śabda- ‘articulate sound, noise’; solã ‘16’ < OIA ṣōḍaśa; ghāh ‘grass’ < OIA ghāsa- ‘food, pasture grass’; dāh ‘10’ < OIA daśa; nō̃h ‘son’s wife’ < OIA suṣā­). Such a situation can be best represented within the framework of Schmidt’s wave model. For this reason, many scholars believe that the latter model is the most preferable, if not the 231

Anton I. Kogan

only possible, for the Indo-Aryan group, whereas the tree model is not applicable to it. This pessimism was perhaps most vividly expressed by C. Masica in his previously mentioned book: “We might therefore be well-advised to give up as vain the quest for a final and “correct” NIA historical taxonomy, which no amount of tinkering can achieve, and concentrate instead on working out the history of various features, letting such feature-specific historical groupings emerge as they may, with their overall non-coincidence as testimonial to the complexity of the situation” (Masica 1991: 460). Masica’s practical suggestion was that scholars should confine themselves to drawing isoglosses on the map and postulating zones that they demarcate. Since such zones usually overlap, the term “overlapping genetic zones” was accepted for them. Such a term, however, is patently self-contradictory. ‘Overlapping’ implies that there must be a language or languages belonging to more than one zone, which can never be the case with genetic subdivisions. Moreover, it is well known that areal and genealogical groupings do not always coincide, and for this reason, the use of the term ‘zone’ in the genetic classification does not seem to be correct. Nevertheless, the proposed term became very popular among Indologists, and has even penetrated into some encyclopaedic editions. It is justified to state that genealogical classification of the Indo-Aryan languages has presently been substituted by the areal one, and that the main reason for this paradigm change is the failure to classify the languages in question on the basis of phonological innovations. It should be noted in this connection that phonological isoglosses, as they are traditionally postulated, are not always unquestionable. Sometimes certain sound changes, which are considered to be common for many languages, in reality coincide only partially. This seems to hold true, e.g., for the simplification of the MIA geminates with compensatory lengthening of the preceding short vowels. In quite a number of Indo-Aryan dialects this development takes place in all words and positions, whereas in standard Hindi-Urdu it apparently affects mainly monosyllabic words. If the number of syllables is more than one, geminates are often retained and the vowel remains short. This hypothetical rule is most likely to be valid when the MIA vowel is a: pakkā ‘ripe, mature’ < MIA pakka- < OIA pakva­, cf. Nepali pāko, Bengali paka (a < ā), Romany pako (a < ā), Gujarati pākũ ‘ripe’; makkhī ‘fly’ < MIA makkhiā- < OIA makṣikā­, cf. Nepali, Kumauni mākho, Assamese mākhi, Gujarati mākhī, Romany makh, maki; saccā ‘true’ < MIA sacca< OIA satya- ‘truth’, cf. Nepali, Kumauni sco, Bengali śãca, Awadhi scu, Marwari sāco, Gujarati sācũ, Romany čačo; acchā ‘good’ < MIA accha- ‘clear, transparent, pure, clean’ < OIA accha‘clear, transparent’, cf. Oriya āchā, Kumauni ācho ‘good’, Gujarati āchũ ‘thin, elegant’; pattā ‘leaf’ < OIA pattra­, MIA patta­, cf. Kumauni pātī ‘leaves, letter’, Nepali pāto ‘page, blade of a knife’, Bengali pata ‘leaf, blade’, Awadhi pātā, Gujarati pātũ ‘leaf’; patthar ‘stone’ < MIA patthara- id. < OIA prastara- ‘anything strewn; flat surface; rock, stone’, cf. Awadhi, Kumauni pāthar, Bengali, Assamese path r, Konkani phāttaru ‘stone’, Marathi pāthar ‘flat stone’; apnā ‘one’s own’ < MIA appaṇaya- < OIA *ātmanaka-,12 cf. Kumauni āpṇo, Nepali āphnu, Bengali ap n, Gujarati āpṇũ. Note that new Indo-Aryan disyllables ending in a vowel most probably reflect Old Indian trisyllabic bases enlarged with the suffix ­k-,13 i.e. pakvaka­, satyaka­, acchaka­, patraka- etc. Of particular interest are those cases where we find two cognates in Hindi-Urdu, one of which is monosyllabic and the other is di- or trisyllabic. Such cognates always show a divergent phonetic development: sāt ‘7’ < MIA satta < OIA sapta vs. sattā ‘aggregate of 7; seven in 12

See Turner 1966: 51. Cf. the above-cited words for ‘flyʼ, for which the prototype with the suffix ­k- (makṣikā­) is attested already in Old Indian. On enlarged noun bases in ­k- see in general (Bloch 1965: 111, 163–165). 13

232

Genealogical classification of New Indo-Aryan languages and lexicostatistics

cards’ < MIA sattaya- < NIA saptaka­; hāth ‘hand’ < MIA hattha- < OIA hasta- vs. hatthā ‘handle’ < OIA hastaka­; lāj ‘shame’ < MIA, OIA lajjā- vs. nilajā ‘shameless’ < MIA nilajja- < OIA nirlajja­; kām ‘work, act’ < MIA kamma- < OIA karman- vs. nikammā ‘idle; useless, good-for-nothing’ < MIA *nikkamma-14 < OIA niṣkarman- ‘inactive’. The phenomenon described here certainly needs further study, because there are some unexplained counterexamples (cf., e.g. māthā ‘forehead’ < MIA mattha(ya)- ‘head’ < OIA masta(ka)- ‘head, skull’15), but what can be stated with certainty is that the issue of common historical phonological isoglosses is much more complicated than it might seem upon first sight. Establishing such isoglosses for the Indo-Aryan group is possible only after a detailed and in-depth analysis of all available data. Since such analysis has not always been properly conducted, it is still premature to say that we know the full picture. This, however, can hardly affect the conclusion that intensive language contact has sometimes made it almost impossible to distinguish between phonological innovations common for a genetic subgroup and contact-driven sound changes. It means that Indo-Aryan languages should not be classified based only (or even mainly) on historical phonology, as was frequently done in the past. But what kind of linguistic data should we then use as criteria for classification? It is tempting to turn to morphology, but, as C. Masica points out in his book, “…morphological criteria conflict just as much as phonological criteria” (Masica 1991: 460). Since syntax is even more prone to radical restructuring due to foreign influence and is, moreover, very similar throughout Indo-Aryan, syntactic data can hardly help us to clarify genetic relations within the Indo-Aryan group. The only domain of language that can provide us with relevant information for genealogical classification appears to be the lexicon. To the best of my knowledge, no scholar has so far seriously attempted to create a genealogical tree of Indo-Aryan languages based on lexical isoglosses. Such a state of affairs arguably results from insufficient attention paid by many Indologists to the lexical level in general and basic vocabulary in particular. This is rather unfortunate, because the in-depth study of this part of the lexicon actually helps to solve a variety of problems of Indological comparative linguistics, including those of historical phonology, because it is well known that in basic vocabulary the number of loanwords is always limited and genuine phonetic development always predominates. The latter fact also suggests the possibility of using the lexicostatistical method for classifying the Indo-Aryan languages. The 100­item Swadesh wordlist has on many occasions been successfully used as a sample of basic vocabulary. The greater part of its items is cross-linguistically stable, and cases of phonological change that are characteristic of borrowings from closely related languages are therefore always in a minority within this set. Since the historical phonology of many IndoAryan languages has now been studied in sufficient detail, such cases must often be easily detectable, even if one factors in the limitations of our knowledge stated above. It means that the problem of unidentified loanwords in Indo-Aryan wordlists is hardly crucial, and the resulting genealogical tree is unlikely to differ to a great extent from the real picture. Consequently, in the present article we present an attempt of genealogical classification based on lexicostatistics. The Indo-Aryan lexicostatistical database, prepared by myself,16 14

Cf. Pali nikkamma­, Prakrit ṇikkamma- ‘unoccupied’ (Turner 1966: 422). It should also be noted that the compensatory lengthening of MIA i and u appears not to be confined to monosyllables: sīdhā ‘straightforward’ < MIA, OIA siddha- ‘perfected’; sūkhā ‘dry’ < MIA sukkha- < OIA śuṣka­. 16 I wish to thank Anastasiya Krylova and Eugenia Renkovskaya (Russian State University for the Humanities, Moscow) for their help in the preparation of the database and Ilya S. Yakubovich (Moscow State University) for providing me with dictionaries of several New Indo-Aryan languages. 15

233

Anton I. Kogan

consists of Swadesh lists for 35 languages, namely Hindi-Urdu, Dakhini,17 Punjabi, Pothohari, Hindko, Gojri, Dogri, Lahnda (Multani), Sindhi, Kutchi, Rajasthani (Marwari), Gujarati, Marathi, Konkani, Bengali, Assamese, Oriya, Nepali, Sinhalese, Maldivian (Dhivehi), Kotgarhi, Himachali, Kului, Mandeali,18 Kumauni, Garhwali, Awadhi, Braj, Mewati,19 Wagdi,20 Banjari, Maithili, Parya, Domaaki (Dumaki), and Romany. Unfortunately, it turned out to be impossible to include Old and Middle Indo-Aryan wordlists in the database, because they are either not securely datable (e.g., Vedic and Pali wordlists) or contain too many lacunae (e.g., wordlists of Aśokan Prakrits). Moreover, certain Middle Indo-Aryan languages, such as literary Prakrits and literary Apabhraṃśas, are to a great extent artificial constructs and do not fully reflect spoken dialects of their time. The wordlists are given in a special appendix after the main text of the article. Before proceeding to the results of the lexicostatistical calculations, it seems necessary to make some remarks concerning synonyms and loanwords. Since in many cases we know little or nothing either about semantic nuances or frequency of a particular Indo-Aryan word on the list, it is sometimes impossible to determine the main synonym. In such a situation, we suggest that the best solution to the problem of synonymy is apparently to include no more than two synonyms on the list in the case when each of them has cognates in other Indo-Aryan languages. If only one of the synonyms finds etymological parallels within the group, it is technically considered as the main one. Likewise, in those cases where both an inherited word and a loanword are attested for the same Swadesh meaning, only the former is included in the database (since addition or omission of the latter will be irrelevant for the lexicostatistical results anyway). Loanwords on the lists are for the most part easily identifiable. Usually they are of either Persian or Sanskrit origin. The latter group embraces not only tatsamas (borrowings from Sanskrit, preserved more or less unchanged in modern languages), but also the so-called semitatsama or ardhatatsama words, i.e. early Sanskrit loans which have undergone certain phonetic changes (e.g., Punjabi purakh ‘man’ < Skr. puruṣa­; Dogri, Himachali, Mandeali, Awadhi barkhā ‘rain’ < Skr. varṣā­). Dravidian loanwords are found chiefly in Sinhalese and Konkani wordlists.21 The noun poṭ ‘belly’ seems to be a borrowing from Dravidian in Marathi and Konkani (cf. Proto-Dravidian *poṭ and its reflexes in different Dravidian languages given in (Burrow, Emeneau 1961: 397–398)). Phonetically similar words for ʽbellyʼ in many other Indo-Aryan languages (Hindi-Urdu, Punjabi, Parya, Gujarati, Bengali peṭ, Assamese pet, Oriya peṭa, Garhwali pyaṭ, Romany perr) were connected by R.L.Turner with OIA peṭa- ‘basket’ (Turner 1966: 475). Although such a semantic development, typologically quite possible,22 could in principle take place spontaneously in Indo-Aryan, it cannot be ruled out either that this change was “catalyzed” by the influence of the abovementioned Dravidian noun. 17 Dakhini is frequently considered a regional form of Urdu. Actually, it is a group of closely related dialects spoken by the Muslim population of the Deccan plateau in Central and South India, chiefly in the Telangana, Karnataka and Maharashtra states. Its speakers are mostly descendants of immigrants from North India and the Middle East. The lexical material used in my database belongs to the dialect spoken in Northern Karnataka. 18 Kotgarhi, Himachali, Kului, and Mandeali are spoken in Western Himalaya, mainly in the present-day Himachal Pradesh and Uttarakhand states of India. They are traditionally included in the Pahari subgroup. 19 Traditionally classified as a dialect of Rajasthani. 20 A dialect of Bhili. 21 Sinhalese seems to have been influenced by Dravidian since very early times. In Konkani the Dravidian lexical stratum is mainly the result of contact between this language and its Southern neighbor Kannada. The Swadesh list for Konkani contains 4 Kannada loanwords, viz. moḍ ‘cloud’, tanthe ‘egg’, urūṭ ‘round’ and bāl ‘tail’. 22 As an approximate parallel cf. English chest, meaning both ‘box’ and ‘thorax’.

234

Genealogical classification of New Indo-Aryan languages and lexicostatistics

The languages that show the largest number of loans are Domaaki and Romany. In Domaaki there are 27 loanwords belonging to the Swadesh list. This percentage is no doubt abnormally high, but nevertheless quite explicable for a language on the verge of extinction, whose 300 speakers are all bi- or trilingual. The donor language for 19 loans is Shina,23 the majority language and lingua franca of the area where Domaaki is spoken. The remaining 8 borrowings are of Burushaski origin.24 The Romany Swadesh list25 contains 19 loanwords, which are borrowed from different sources, namely Dardic (parno ‘white’26), Burushaski (cgno ‘small’27), Iranian (por ‘feather’, čehran ‘star’), Armenian (morči ‘skin’), Kartvelian (kišay ‘sand’28), Greek (kokalo ‘bone’, drom ‘road’), Slavic (zeleno ‘green’, koreno ‘root’, pliv- ‘swim’), Romanian (skarča ‘bark’, unjiya ‘nail’, nuvero ‘cloud’, lungo ‘long’, mnt’a ‘mountain’, rotato ‘round’, semnca ‘seed’, galbeno ‘yellow’). The main source of etymologies is R. L. Turner’s comparative dictionary (Turner 1966). Domaaki and Parya etymologies are also taken from Buddruss 1984 and Oranskiy 1977 respectively. The results of lexicostatistical calculations are given in Table 1. The genealogical tree constructed by the StarLing system on the basis of the above data is reproduced on Figure 3. As one can see, the classification represented by this tree differs from earlier classificatory schemes in quite a number of points. Below I list those differences which are, in my opinion, the most important ones. 1. According to the above classification, Indo-Aryan is subdivided into two main subbranches — one including Sinhalese and Dhivehi (Maldivian), and the other consisting of all the other New Indo-Aryan languages. The most proper names for these subgroups would be “Insular” and “Continental”. The split of Proto-Indo-Aryan dates back to the close of the 2nd millennium B.C. 2. The Continental subgroup includes an outlying branch embracing Marathi and Konkani. 3. Romany turns out to form a common subgroup with Hindi-Urdu, Punjabi and dialects of the sub-Himalayan region, traditionally classified as Pahari (Nepali, Garhwali, Kumauni, Himachali, Kului, Mandeali and Kotgarhi). The closest relative of Romany is Domaaki, as was first hypothesized by D. L. R. Lorimer (1939). The split of “Proto-Hindi-Pahari-Romany” dates 23

In order to save space, I will not list all the Shina loanwords here, especially since their identification usually presents no difficulty. The only doubtful case is Domaaki šuno ‘dog’, which, according to Georg Buddruss, should not be considered a borrowing from Shina because of its irregular inflection (Buddruss 1984: 14). Buddruss’s argument, however, does not seem convincing. The reflexes of Proto-Indo-Iranian *śan-/śun­, being widespread in Dardic, are very rare in New Indo-Aryan and almost never used there as the main word for dog, except in a few West Pahari languages, spoken adjacently to the Dardic-speaking area (cf., e.g., Siraji šunā ‘dog’). This fact suggests a high probability of borrowing from Dardic into Indo-Aryan. The immediate source for the above-cited Domaaki word may have been some older form of Shina šũ ‘dog’. 24 These are: burin ‘cloud’, tigon ‘egg’, čhumo ‘fish’, duwal- ‘fly’, jutiqam ‘green’, čhiā ‘mountain’, thop ‘night’, ono ‘seed’. 25 The Romany material reflects the Kalderash dialect, spoken in Romania and Moldova. 26 Cf. Tirahi parana, Maiyã panar, Kashmiri pron (< *paranu). Turner’s comparison of the Romany word with OIA pāṇḍu- (Turner 1966: 454) is doubtful, because it implies irregular phonetic development. The regular reflex of pāṇḍu- in Romany would have been *panrro (cf. punrro ‘leg, foot’ < OIA piṇḍa- ‘calf of leg’). 27 On Burushaski loanwords in Romany see Berger 1959. 28 Cf. Georgian, Laz kviša ‘sand’. The change kv > k is regular for genuine Romany words (cf. kerel ‘cooks’ < OIA kvaṭhati). It may imply that during a certain period of time the cluster kv was proscribed in the language. If the borrowing of the Kartvelian word for ‘sand’ dates back to this period, the loss of v in the initial consonantal group is quite explicable. 235

95

79

84

86

87

85

80

89

84

73

73

73

77

59

53

81

90

88

91

74

93

90

77

64

77

87

83

87

83

78

91

73

81

97

83

87

91

87

85

81

90

86

67

70

69

78

53

47

82

92

88

91

72

92

90

71

60

74

92

85

89

83

77

97

78

79

PNJ

PTH

HNK

GJR

DGR

LHD

SND

RAJ

GUJ

MAR

BNG

ASS

NEP

SNG

MAL

KOT

HIM

KUL

MND

ORY

AWD

KUM

ROM

KNK

DUM

BRJ

GRH

PRY

MAI

KCH

MEW

WGD

BNJ

96

96

DKH

HND

80

75

94

81

82

93

82

88

72

67

71

89

91

70

91

85

92

80

50

57

76

72

70

68

84

92

84

90

93

93

91

89

95

97

67

65

80

76

76

77

73

78

67

60

68

75

78

68

84

79

86

75

45

50

72

63

66

69

73

80

77

82

86

93

89

89

79

83

70

68

82

78

77

83

77

79

71

60

71

78

80

67

85

81

85

77

48

54

70

67

69

67

76

81

82

87

88

95

89

91

84

87

75

74

86

76

75

86

78

82

69

66

70

82

85

69

91

84

90

79

45

52

74

67

69

69

81

84

81

83

91

95

93

93

86

91

74

70

86

79

81

83

76

81

77

66

69

80

87

68

92

86

91

79

45

55

77

67

68

67

79

84

82

87

91

88

86

93

87

87

69

68

83

86

78

80

73

77

68

62

68

78

83

66

85

79

85

76

46

50

69

67

67

63

74

80

84

87

83

87

82

90

85

85

73

67

81

91

74

74

69

77

67

71

61

75

78

67

78

75

75

70

50

53

67

67

68

71

79

83

84

82

81

82

77

84

80

81

82

76

91

85

82

81

79

92

75

70

70

84

90

75

86

86

86

78

50

57

76

70

70

75

90

83

80

84

84

81

80

92

89

90

HND DKH PNJ PTH HNK GJR DGR LHD SND RAJ

82

77

88

82

79

76

76

83

67

72

62

78

84

73

85

82

81

74

49

54

75

70

69

78

90

79

74

79

81

76

73

84

84

86

67

64

69

73

63

60

57

63

65

82

57

64

70

63

69

64

64

62

51

53

64

59

62

78

75

71

63

67

69

67

69

68

73

67

58

60

73

72

79

66

69

70

64

65

59

71

74

87

72

69

69

70

52

56

66

88

62

69

70

68

67

68

69

69

66

70

73

70

63

60

71

71

74

68

71

67

67

66

63

73

76

87

73

73

72

72

55

57

71

88

59

70

70

67

67

67

67

67

63

72

73

69

72

68

80

71

78

71

81

72

73

64

73

86

85

73

82

83

78

80

46

53

71

66

64

75

76

67

69

77

74

70

72

76

77

78

46

47

55

54

59

58

50

53

59

51

53

56

56

57

55

54

53

52

75

53

57

56

53

54

57

53

50

55

52

54

50

57

59

53

42

43

50

49

51

50

44

48

55

49

53

49

49

53

48

45

47

48

75

46

55

52

51

49

50

50

46

45

45

48

45

50

53

47

69

65

82

74

80

78

80

79

77

57

73

85

85

75

88

84

85

48

52

80

72

70

62

74

78

70

76

79

79

77

75

80

81

82

74

74

87

77

82

85

83

86

71

59

72

85

91

73

96

94

85

47

53

78

72

69

64

81

86

75

85

91

90

85

86

92

90

92

74

69

87

76

77

82

86

81

73

61

71

86

87

73

96

94

84

45

54

83

73

69

64

82

86

75

79

86

84

81

79

85

88

88

77

73

90

77

82

86

83

84

77

65

72

86

88

74

96

96

88

48

55

82

73

72

69

85

86

78

85

92

91

85

84

91

91

91

63

68

78

72

80

70

72

72

72

63

65

75

78

75

74

74

76

53

57

74

86

86

64

74

76

68

67

69

70

68

69

72

75

73

82

76

91

79

88

83

87

88

79

67

78

95

77

88

87

91

85

49

56

85

76

74

70

84

90

78

83

87

85

80

78

91

93

92

75

71

91

76

86

83

90

84

78

61

77

95

74

86

86

85

85

49

56

86

73

71

64

78

84

75

78

80

82

78

75

89

90

90

60

61

74

64

68

72

75

70

82

54

77

78

65

72

71

72

73

53

53

73

63

59

57

62

70

61

68

69

70

71

68

71

77

71

66

63

67

75

59

59

53

59

62

54

61

67

63

65

61

58

56

48

49

64

66

63

80

72

72

71

62

66

64

59

58

67

64

60

67

61

81

67

76

77

72

72

60

82

78

79

72

77

73

71

77

55

59

73

67

64

65

67

73

67

68

77

69

71

67

72

77

74

78

74

93

78

82

82

80

72

59

70

84

88

72

84

81

85

78

47

52

72

67

69

62

83

91

77

77

81

81

78

77

88

87

92

69

66

83

68

79

78

80

72

53

75

90

87

72

83

86

82

79

43

49

81

71

67

56

76

78

69

73

76

77

76

71

82

83

85

74

76

84

76

75

78

82

77

59

72

83

83

70

86

82

85

78

50

58

71

68

66

60

76

81

74

80

83

86

83

77

93

87

89

70

67

82

78

75

79

82

76

59

68

86

88

80

82

77

81

79

49

58

78

74

78

62

79

81

74

78

81

74

76

75

82

83

83

75

70

78

78

76

68

78

67

75

64

76

79

72

77

76

76

73

48

53

71

71

71

72

82

86

91

86

79

75

77

75

81

78

77

78

77

77

82

84

82

92

81

66

74

91

91

76

90

87

87

82

50

55

80

71

73

69

88

91

81

83

86

86

82

80

94

91

97

74

78

70

67

76

66

74

61

63

61

71

76

68

73

69

73

64

42

45

68

60

59

63

77

75

67

68

70

73

67

64

75

73

78

74

78

75

70

74

69

78

67

66

60

75

82

63

77

74

74

69

42

46

72

63

58

67

82

81

73

69

74

75

70

67

80

81

79

GUJ MAR BNG ASS NEP SNG MAL KOT HIM KUL MND ORY AWD KUM ROM KNK DUM BRJ GRH PRY MAI KCH MEW WGD BNJ

Table 1. The lexicostatistical matrix of Indo-Aryan languages.

-1.25

-1.07

-1.00

-0.75

-0.47

-0.50

-0.21

-0.25

0.06

0.00

0.36

0.32

0.25

0.18

0.13

0.47

0.47

0.52

0.47

0.47

0.50

0.62

0.76

0.73

0.75

0.84

0.84

0.96

1.02

1.00

1.22

1.22

1.22

1.30

1.25

1.39

1.50

1.75 SNG M AL M AR KNK PTH HNK GJR DGR LHD PRY WGD BNJ SND KCH M AI GUJ BRJ RAJ M EW ORY BNG ASS ROM DUM KOT NEP GRH KUL HIM M ND DKH HND PNJ AWD KUM

2.00

Figure 3. The genealogical tree of Indo-Aryan languages. Figures in the nodes of the tree denote dates of separation in millennia A.D. (positive numbers) or B.C. (negative numbers), and are the results of glottochronological calculations performed in the StarLing system.

Anton I. Kogan

back to the 1st century A.D., and that of “Proto-Romany-Domaaki” to the close of the 5th century A.D. These figures, however, seem to be preliminary. The Swadesh lists of both Romany and Domaaki contain a number of unetymologized words. In the future, when etymologies of such words are established, the percentage of cognates may increase, and the resulting datings may appear to be somewhat younger. It is also worth noting that neither of the two abovementioned dates should be automatically declared the date of the Gypsy exodus from India. The latter, no doubt, could date back to a later period than the linguistic split. Such a possibility is suggested by the fact that most subbranches of Continental Indo-Aryan diverged (sometimes nearly two millennia back) without mass migration of the speakers outside the subcontinent. 4. Forms of speech traditionally classified as Western Hindi do not actually form a single subgroup. Braj shows close relationship with Marwari and Mewati dialects spoken in Rajasthan, and somewhat more remote with Gujarati. Standard Hindi-Urdu and Dakhini are most closely related to Punjabi.29 As for the Eastern Hindi dialects, their only representative in the database, i.e. Awadhi, is the closest relative of Kumauni. 5. Contrary to the traditional view, there is no reason to suggest a Rajasthani origin for a number of Indo-Aryan languages spoken outside Rajasthan. Thus Gojri,30 classified by Grierson as a form of speech close to the Mewati dialect of Rajasthani, actually does not belong to the same group with the latter, but rather shows a close affinity to Hindko. The Banjari language,31 which was also usually considered as a variety of Rajasthani, actually occupies a somewhat independent position within one of the sub-branches of the Continental languages. 6. The above-stated close affinity of Kumauni to Awadhi implies that the Pahari group in the traditional sense does not exist as a genetic subdivision. The West Pahari languages except Kotgarhi (i.e., Himachali, Kului and Mandeali) do, however, form a single subgroup. The split of their ancestral language must have taken place very recently. On the other hand, our classification does not differ from the earlier ones in those instances where the existence of subgroups is obvious or can be postulated on the basis of early linguistic evidence, as is, e.g., the case of the Eastern subgroup consisting of Oriya, Bengali and Assamese. It should be emphasized again that the above classificatory scheme is preliminary and thus remains open to further amendments and improvements. It does not pretend to answer each and every question concerning New Indo-Aryan taxonomy. In a number of cases, it raises intriguing problems for further research. Among such problems, the rather close affinity of Maithili to Braj, Rajasthani dialects and Gujarati, as well as the somewhat isolated position of Parya and Garhwali appear to be particularly noteworthy.32 We hope that the present article will be instrumental in stimulating scholarly interest in these (and related) issues of IndoAryan comparative linguistics.

29

The pair Hindi-Punjabi shows 97% matches, the highest percentage in the whole database. The language of Gujjars, a Muslim nomadic and semi-nomadic ethnic group dispersed in mountainous areas from Afghan Hindu Kush in the Northwest to the Indian state of Uttarakhand in the Southeast. 31 Also called Lamani and Lambadi, spoken by a semi-nomadic community of Banjaras scattered all over Central and Western India. 32 The Garhwali Swadesh list contains a significant number of unetymologized words. This fact may partly account for the relative isolation of the Garhwali language on the genealogical tree shown above. An abnormally low percentage of matches between Maldivian and Continental Indo-Aryan languages may have the same reason. 30

238

Genealogical classification of New Indo-Aryan languages and lexicostatistics

Appendix: Swadesh wordlists The data are based on the following sources: HND, PNJ, SND, GUJ, MAR, BNG, SNG — Kogan 2005; DKH — Sibghatulla, Zamin 2000; PTH — a native speaker; HNK — Sultān Sukūn 2002; GJR — Awan 2000; DGR — Gosvāmī 2000; LHD — Saleem, Shah 2005 and Kogan 2005; RAJ — Suthar, Gahlot 1995 and Mukherji et al. 2011; ASS — Neog, Goswami 1987; NEP — Schmidt 1994; MAL — Abdulla, O’Shea 2005; KOT — Hendriksen 1976; HIM — native speakers; KUL — Mahapatra, Padmanabha, Ranganatha 1980; MND — native speakers and Mahapatra, Padmanabha, Ranganatha 1980; ORY — native speakers and Praharaj 1931–1940; AWD — Samīr 1955; KUM, GRH — native speakers and Grireson 1916; ROM — Boretzky 1994; KNK — Thali 1999–2001; DUM — Buddruss 1984 and Lorimer 1939; BRJ, MEW, WGD — Mukherji et al. 2011; PRY — Oranskiy 1977; MAI — Thakur, Jha 2012; KCH — Rohra 1965; BNJ — Ramesh 2010. Figures in brackets after words refer to numbers of etymologies in the Indo-Aryan etymological database. They also reflect cognacy, words descending from the same OIA root having equal numbers. Negative numbers are assigned to loanwords and lacunae.

1. ALL: HND sab (1), DKH sab (1), PNJ sabh (1), PTH sāre (1), HNK sāre (1), GJR sārā (1), DGR sab (1), LHD sārā (1), SND sabhi (1), RAJ sagḷau (270), GUJ sahu (1), MAR sagḷā (270), BNG šbay (1), ASS xb (1), NEP sabai (1), SNG siyallō (270), MAL emmehā (174), KOT sar (1), HIM sāre (1), KUL sbh (1), MND sabh (1), ORY sabu (1), AWD sab (1), KUM sap (1), ROM savorre (1), KNK sagḷo (270), DUM buṭā (–1), BRJ sab (1), GRH sabbi (1), PRY sare (1), MAI sab (1), KCH sab (1), MEW sabe (1), WGD sabu (1), BNJ sāri (1) 1a. ALL: SND samūro (174), SNG muḷu (381) MAL hurihā (522) AWD sagar (270) 2. ASHES: HND rākh (3), DKH rā (3), PNJ suāh (111), PTH suhāgā (111) HNK chāī (152), GJR bhass (489), DGR kheh (440), LHD chāī (152), SND rakh (3), RAJ khe (440), GUJ rākh (3), MAR rākh (3), BNG chai (152), ASS sai (152), NEP kharāni (140), SNG aḷu (382), MAL aḷi (382) KOT kh (440), HIM suāh (111), KUL ? (–1), MND swāh (111), ORY pāuñša (567), AWD rākhī (3), KUM saji (488), ROM vušar (140), KNK goboru (606), DUM čhor (140), BRJ rākh (3), GRH chāru (140), PRY čhar (140), MAI chār (140) KCH vānī (653) MEW rakhi (3), WGD pāsī (567), BNJ rāk (3) 2a. ASHES: GJR suhāgo (111), DGR bhass (489), LHD suā (111), SND chāī (152), RAJ rākh (3), KOT chār (140), MND bhās (489), ORY chāra (140), AWD chār (140), KUM chār (140) 3. BARK: HND chāl (4), DKH chilṭā (4), PNJ chill (4), PTH chilaṛ (4), HNK chillaṛ (4), GJR chilṛo (4), DGR chilkā (4), LHD chil (4), SND chalu (4), RAJ chilkau (4), GUJ chāl (4), MAR sāl (4), BNG chal (4), ASS bakli (328), NEP bokro (328), SNG potta (383), MAL thoši (523), KOT chāl (4), HIM chillekaṛ (4), KUL khol (552), MND sāṭū (561), ORY chāli (4), AWD chāl (4), KUM bakkhal (328), ROM skarča (–1), KNK sāl (4), DUM ? (–2), BRJ chāl (4), GRH bakkal (328), PRY pōst (–1), MAI chāl (4), KCH chall (4), MEW chāl (4), WGD sal (4), BNJ chāmbḍi (667) 3a. BARK: ASS sal (4), ORY bakkala (328), MEW bokalā (328) 4. BELLY: HND peṭ (678), DKH peṭ (678), PNJ peṭ (678), PTH ḍhiḍ (113), HNK ḍhiḍ (113), GJR ḍhiḍ (113), DGR ḍhiḍḍ (113), LHD ḍhiḍh (113), SND peṭu (678), RAJ peṭ (678), GUJ peṭ (678), MAR poṭ (–1), BNG peṭ (678), ASS pet (678), NEP peṭ (678), SNG baḍa (329), MAL ban’du (329), KOT peṭ (678), HIM peṭ (678), KUL pēṭ (678), MND pēṭ (678), ORY peṭa (678), AWD pēṭ (678), 239

Anton I. Kogan

KUM peṭ (678), ROM perr (678), KNK poṭ (–1), DUM peṭ (678), BRJ peṭ (678), GRH pyaṭ (678), PRY peṭ (678), MAI udar (–1), KCH peṭ (678), MEW peṭ (678), WGD peṭ (678), BNJ peṭ (678) 4a. BELLY: PNJ ḍhiḍḍ (113), GJR peṭ (678) 5. BIG: HND baṛā (5), DKH baḍā (5), PNJ vaḍḍā (5), PTH baṛā (5), HNK baḍḍā (5), GJR baṛo (5), DGR baḍḍā (5), LHD vaḍḍā (5), SND vaḍ’o (5), RAJ baḍau (5), GUJ moṭo (248), MAR moṭhā (248), BNG bṛ (5), ASS br (5), NEP ṭhūlo (358), SNG loku (385), MAL boḍu (669), KOT bḍḍ (5), HIM baṛā (5), KUL bḍḍa (5), MND bḍḍa (5), ORY baḍḍa (5), AWD baṛā (5), KUM baṛo (5), ROM baro (5), KNK v(h)od (5), DUM baḍā (5), BRJ bāṛo (5), GRH baṛū (5), PRY boṛo (5), MAI baṛkā (5), KCH vaḍo (5), MEW baṛo (5), WGD moṭo (248), BNJ moṭo (248) 5a. BIG: RAJ moṭau (248), NEP baṛo (5) 6. BIRD: HND ciṛiyā (6), DKH pakherū (7), PNJ pakkherū (7), PTH pakherū (7), HNK pakhṛ (7), GJR pakherū (7), DGR paĩchi (7), LHD pakkhī (7), SND pakhī (7), RAJ pãkherū (7), GUJ pãkhī (7), MAR pakṣī (–2), BNG pakhi (7), ASS srai (6), NEP carā (6), SNG kurullā (–1), MAL dūni (524), KOT cṛku (6), HIM panchī (7), KUL cīḍu (6), MND pnchi (7), ORY caḍhei (6), AWD ciraī (6), KUM callo (6), ROM čirikli (6), KNK pākko (7), DUM čai (6), BRJ paksī (–1), GRH cakhul (652), PRY kukuṛi (644), MAI ciṛai (6) KCH pakhi (7), MEW ciṛiā (6), WGD huṛo (660), BNJ gargol (668) 6a. BIRD: HND pãkhī (7), DKH caṛī (6), DGR ciṛī (6), RAJ ciṛī (6), ASS pnkhi (7), KOT panchi (7), KUL pnchi (7), MND cīḍu (6), AWD panchī (7), GRH pacche (7), MAI panchī (7), MEW pakherū (7) 7. BITE: HND kāṭ- (8), DKH kāṭ- (8), PNJ kaṭṭ- (8), PTH laṛ- (153), HNK laṛ- (153), GJR khā(490), DGR baḍḍh- (493), LHD paṭ- (501), SND caku pāi- (176), RAJ karaṛ (249), GUJ karaḍ(249), MAR cāv- (271), BNG kamṛa- (296), ASS kamr- (296), NEP ṭok- (114), SNG hapa- (386), MAL daiganna- (550), KOT ṭuk- (114), HIM dānd mār- (550), KUL kāṭ- (8), MND kāṭ- (8), ORY kāmuṛ- (296), AWD ? (–1), KUM kāṭ- (8), ROM dandar- (550), KNK cāb- (271), DUM an- (–3), BRJ kaṭ- (8), GRH kāṭ- (8), PRY kaṭ- (8), MAI bhamhor- (676), KCH cak- (176), MEW kāṭ- (8), WGD kaṭ- (8), BNJ cāb- (271) 7a. BITE: HNK cak mār- (176), GJR laṛ- (153) SND ḍ’āṛh- (177), MAR ḍsā ghe- (177), SNG vik(387), HIM khā- (490), MND khā- (490), PRY čak par- (176) KCH viṛ- (654), MEW cabā- (271), BNJ kāṭ- (8) 8. BLACK: HND kālā (9), DKH kālā (9), PNJ kālā (9), PTH kālā (9), HNK kālā (9), GJR kālo (9), DGR kālā (9), LHD kāḷā (9), SND kāro (9), RAJ kāḷau (9), GUJ kāḷo(9), MAR kāḷā (9), BNG kalo (9), ASS kla (9), NEP kālo (9), SNG kaḷu (9), MAL kaḷu (9), KOT kaḷ (9), HIM kāḷā (9), KUL kāḷa (9), MND kāḷā (9), ORY kalā (9), AWD kariā (9), KUM kālo (9), ROM kalo (9), KNK kāḷe (9), DUM kālā (9), BRJ kāṛo (9), GRH kāḷu (9), PRY kalo (9), MAI kāri (9), KCH kāro (9), MEW kālo (9), WGD kāḷo (9), BNJ kāḷo (9) 9. BLOOD: HND lahū (10), DKH lhau (10), PNJ lahū (10), PTH lahū (10), HNK rat (115), GJR rat (115) DGR lahu (10), LHD ratt (115) SND ratu (115) RAJ lohī (10), GUJ lohī (10), MAR rakt (–3), BNG rkt (–1), ASS tez (330), NEP ragat (–1), SNG lē (10), MAL ley (10), KOT lou (10), HIM laū (10), KUL lōhu (10), MND lahū (10), ORY rakta (–1), AWD rakat (–2), KUM lwe (10), ROM rat (115), KNK ragat (–2), DUM rot (115), BRJ lohī (10), GRH loi (10), PRY law (10), MAI lohū (10), KCH ratt (115), MEW khūn (–1), WGD khõn (–1), BNJ loi (10) 9a. BLOOD: PNJ ratt (115), PTH rat (115), HNK lahū (10), DGR ratt (115), KCH lui (10) 240

Genealogical classification of New Indo-Aryan languages and lexicostatistics

10. BONE: HND haḍḍī (11), DKH haḍ (11), PNJ haḍḍī (11), PTH haḍī (11), HNK haḍḍī (11), GJR haḍḍī (11), DGR haḍḍī (11), LHD haḍḍī (11), SND haḍ’o (11), RAJ hāḍ (11), GUJ hāḍkũ (11), MAR hāḍ (11), BNG haṛ (11), ASS haṛ (11), NEP hāḍ (11), SNG äṭaya (388), MAL kaši (526), KOT haḍḍ (11), HIM āḍ (11), KUL hāḍka (11), MND haḍḍī (11), ORY hāḍa (11), AWD hāṛ (11), KUM hāṛ (11), ROM kokalo (–2), KNK hāḍ (11), DUM hoṭ (11), BRJ haḍḍī (11), GRH hāḍ (11), PRY haḍi (11), MAI haḍḍī (11), KCH haḍo (11), MEW haḍḍī (11), WGD haḍko (11), BNJ haḍkā (11) 11. BREAST: HND chātī (12), DKH chātī (12), PNJ chātī (12), PTH chātī (12), HNK mammā (478), GJR chātī (12), DGR chātī (12), LHD bubbā (154), SND chātī (12), RAJ chātī (12), GUJ chātī (12), MAR chātī (–4), BNG buk (297), ASS buku (297), NEP chātī (12), SNG ḷaya (116), MAL buk (297), KOT cui (141), HIM chātī (12), KUL chāttī (12), MND hīk (116), ORY chātī (12), AWD chātī (12), KUM chātī (12), ROM kolin (592), KNK hardhen (116), DUM titiro (–4), BRJ chātī (12), GRH ? (–1), PRY hik (116), MAI chātī (12), KCH thaṇ (655), MEW chātī (12), WGD sātī (12), BNJ chātī (12) 11a. BREAST: PNJ hikk (116), DGR hikkṛū (116), SND urahu (178), RAJ bobau (154), MAL uramathi (178), KUL hik (116), MND chātī (12), ORY buku (297) 12. BURN TR.: HND jalā- (13), DKH jāl- (13), PNJ bāl- (13), PTH sāṛ- (117), HNK sāṛ- (117), GJR sāṛ- (117), DGR bāl- (13), LHD bāl- (13), SND sāṛ- (117), RAJ phk- (505), GUJ bāḷ- (13), MAR jāḷ- (13), BNG poṛa- (298), ASS por- (298), NEP bāl- (13), SNG dava- (217), MAL ada(217), KOT dzaḷ- (13), HIM bāḷ- (13), KUL phukh- (505), MND phukh- (505), ORY poṛi- (298), AWD jarā- (13), KUM jal- (13), ROM phabar- (593), KNK jaḷayi- (13), DUM däi- (–5), BRJ jāl(13), GRH jagau- (653), PRY jal- (13), MAI jār- (13), KCH bār- (13), MEW jalā- (13), WGD jal(13), BNJ bāḷ- (13) 12a. BURN TR.: PNJ sāṛ- (117), bāḷ- (13), GJR jāl- (13), DGR sāṛ- (117), LHD sāṛ- (117), SND b’ār- (13), RAJ jaḷā- (13), BNG jala- (13), NEP ḍaṛhāu- (217), SNG pulussa- (390), KOT d- (217), MND baḷā- (13), AWD ḍaṛhiā- (217), PRY sar baga- (117) 13. NAIL: HND nā un (–1), DKH nh (14), PNJ nahũ (14), PTH n (14), HNK nh (14), GJR nah (14), DGR na (14), LHD n (14), SND nãhũ (14), RAJ noriyau (14), GUJ nakh (14), MAR nakh (14), BNG nkh (14), ASS nkh (14), NEP naŋ (14), SNG niyapotta (14), MAL niyafathi (14), KOT nš (14), HIM nkh (14), KUL nhš (14), MND nh (14), ORY nakha (14), AWD nah (14), KUM naŋ (14), ROM unjiya (–3), KNK naŋkuṭ (14), DUM naurā (–6), BRJ nh (14), GRH nang (14), PRY nuk (–2), MAI nah (14), KCH nõ (14), MEW nũh (14), WGD nkh (14), BNJ nak (14) 14. CLOUD: HND bādal (15), DKH abhāl (465), PNJ baddal (15), PTH baddal (15), HNK jhaṛ (155), GJR jhaṛ (155), DGR badal (15), LHD baddal (15), SND kakaru (179), RAJ bādḷau (15), GUJ vādlũ (15), MAR ḍhag (272), BNG barid (–2), ASS badl (15), NEP bādal (15), SNG valākula (391), MAL vilā (391), KOT badḷ (15), HIM baddal (15), KUL bādḷ (15), MND baddaḷ (15), ORY megha (–2), AWD bādar (15), KUM bādal (15), ROM nuvero (–4) KNK moḍ (–3), DUM burin (–7), BRJ badalore (15), GRH bādal (15), PRY badal (15), MAI bādari (15), KCH vḍar (15), MEW bādar (15), WGD wādlo (15), BNJ abāḷ (465) 14a. CLOUD: PTH jhaṛ (155) 15. COLD: HND ṭhaṇḍā (458), DKH thaṇḍā (458), PNJ ṭhaṇḍā (458), PTH ṭhaṇḍā (458), HNK ṭhanḍā (458), GJR ṭhaṇḍo (458), DGR ṭhaṇḍā (458), LHD ṭhaḍḍhā (458), SND ṭhadho (458), RAJ 241

Anton I. Kogan

ṭhaṇḍau (458), GUJ thaṇḍũ (458), MAR thaṇḍ (458), BNG ṭhaṇḍa (458), ASS thanda (458), NEP ciso (332), SNG sisil (392), MAL fini (528), KOT šill (218), HIM ṭhaṇḍā (458), KUL ṭhṇḍa (458), MND ṭhaṇḍa (458), ORY thaṇḍā (458), AWD ṭhaṇḍh (458), KUM ṭhaṇḍo (458), ROM šilalo (218), KNK thaṇḍ (458), DUM šildā (218), BRJ siāro (218), GRH ṭhaṇḍo (458), PRY ya

(–3), MAI ṭhāṛ (458), KCH thadho (458), MEW siro (218), WGD ṭaḍu (458), BNJ siḷo (218) 15a COLD: KNK šeḷo (218) MEW ṭhanḍho (458) 16. COME: HND ā- (16), DKH ā(v)- (16), PNJ ā- (16), PTH as- (181) HNK (16), GJR ā- (16), DGR au(16), LHD āv- (16), SND ac- (180) RAJ āv- (16), GUJ av- (16), MAR ye- (180) BNG aš- (301), ASS ah(301), NEP āu- (16), SNG e- (180), MAL ai- (180) KOT ā- (16), HIM ā- (16), KUL āw- (16), MND āu(16), ORY ās- (301), AWD ā- (16), KUM ū- (16), ROM av- (16), KNK yo- (180), DUM (16), BRJ ā(16), GRH ā- (16), PRY a- (16), MAI ā- (16), KCH ac- (180), MEW (16), WGD (16), BNJ ā- (16) 16a. COME: SND āy- (181), MAR ālā- (181), BNG ail- (181), NEP āy- (181), KOT ach- (181), KUL ye- (180), ORY ail- (181), KNK āyil- (181), KCH āy- (181) 17. DIE: HND mar- (17), DKH mar- (17), PNJ mar- (17), PTH mar- (17), HNK mar- (17), GJR mar- (17), DGR mar- (17), LHD mar- (17), SND mar- (17), RAJ mar- (17), GUJ mar- (17), MAR mar- (17), BNG mr- (17), ASS mr- (17), NEP mar- (17), SNG märe- (17), MAL maru- (17), KOT mr- (17), HIM mar- (17), KUL mr- (17), MND mar- (17), ORY mar- (17), AWD mar- (17), KUM mar- (17), ROM mer- (17), KNK mar- (17), DUM mar- (17), BRJ mar- (17), GRH mar- (17), PRY mar- (17), MAI mar- (17), KCH mar- (17), MEW mar- (17), WGD mar- (17), BNJ mar- (17) 18. DOG: HND kuttā (18), DKH kuttā (18), PNJ kuttā (18), PTH kuttā (18), HNK kuttā (18), GJR kuto (18), DGR kuttā (18), LHD kuttā (18), SND kutto (18), RAJ kutau (18), GUJ kutro (18), MAR kutrā (18), BNG kukur (302), ASS kukur (302), NEP kukur (302), SNG ballā (394), MAL baḷu (394), KOT kukkr (302), HIM kuttā (18), KUL kutta (18), MND kutta (18), ORY kukkura (302), AWD kūkur (302), KUM kukur (302), ROM žukel (595), KNK kutro (18), DUM šuno (–8), BRJ kuttā (18), GRH kukur (302), PRY kuto (18), MAI kukkur (302), KCH kauto (18), MEW kutto (18), WGD kutro (18), BNJ kutrā (18) 18a. DOG: MAL kutthā (18) 19. DRINK: HND pī- (19), DKH pī(v)- (19), PNJ pī- (19), PTH pī- (19), HNK pī- (19), GJR pī(19), DGR pī- (19), LHD pīv- (19), SND pī- (19), RAJ pī- (19), GUJ pi- (19), MAR pi- (19), BNG kha- (303), ASS pi- (19), NEP piu- (19), SNG bo- (19), MAL bo- (19), KOT pī- (19), HIM pī- (19), KUL pī- (19), MND pī- (19), ORY pī- (19), AWD pi- (19), KUM pī- (19), ROM pi- (19), KNK pī(19), DUM pi- (19), BRJ pī- (19), GRH pya- (19), PRY pi- (19), MAI pī- (19), KCH pī- (19), MEW pī- (19), WGD pī- (19), BNJ pi- (19) 20. DRY: HND sukhā (20), DKH sukkā (20), PNJ sukkā (20), PTH sukkā (20), HNK sukkā (20), GJR suko (20), DGR sukkā (20), LHD sukkā (20), SND sukalu (20), RAJ sūkhau (20), GUJ sukũ (20), MAR sukhā (20), BNG šukno (20), ASS xukan (20), NEP sukkho (20), SNG viyaḷi (456), MAL hiki (20), KOT šukkh (20), HIM sukk(20), KUL ? (–2), MND sukhā (20), ORY sukhila (20), AWD sukhā(20), KUM sukko (20), ROM šuko (20), KNK sukk(h)o (20), DUM šukhā (20), BRJ sukho (20), GRH sukhyũ (20), PRY suke (20), MAI sukkhā (20), KCH saukko (20), MEW sūkhā (20), WGD suko (20), BNJ suko (20) 21. EAR: HND kān (21), DKH kān (21), PNJ kann (21), PTH kan (21), HNK kaṇ (21), GJR kaṇ (21), DGR kann (21), LHD kan (21), SND kanu (21), RAJ kān (21), GUJ kān (21), MAR kān (21), 242

Genealogical classification of New Indo-Aryan languages and lexicostatistics

BNG kaṇ (21), ASS kaṇ (21), NEP kān (21), SNG kana (21), MAL kanfaiy (21), KOT kān (21), HIM kān (21), KUL kn (21), MND kān (21), ORY kāṇa (21), AWD kān (21), KUM kān (21), ROM kan (21), KNK kānu (21), DUM koṇ (21), BRJ kān (21), GRH kān (21), PRY kan (21), MAI kān (21), KCH kann (21), MEW kān (21), WGD kān (21), BNJ kān (21) 22. EARTH: HND dhartī (22), DKH dharat (22), PNJ dhartī (22), PTH miṭṭī (157), HNK dhartī (22), GJR dhartī (22), DGR dhartī (22), LHD bhoẽ (118), SND dhartī (22), RAJ dhartī (22), GUJ dhartī (22), MAR bhuī (118), BNG maṭi (157), ASS (157), NEP dhartī (22), SNG pasa (396), MAL bin (118), KOT bhuĩ (118), HIM dhartī (22), KUL dhrti (22), MND dhartī (22), ORY māṭi (157), AWD dhartī (22), KUM māṭo (157), ROM phuv (118), KNK bhyi (118), DUM čhor (623), BRJ dhartī (22), GRH dharti (22), PRY zimin (–4), MAI māṭi (157), KCH miṭṭī (157), MEW bhūm (118), WGD dhartī (22), BNJ dharti (22) 22a. EARTH: DKH bh (118), PNJ bhõ (118), GJR miṭī (157), MAR mātī (157), NEP māṭo (157), SNG poḷova (397), KOT maṭṭi (157), MND bhūy (118), AWD bhuĩ (118), KUM bh (118), KNK mātī (157), MEW dharatī (22) 23. EAT: HND khā- (23), DKH khā- (23), PNJ khā- (23), PTH khā- (23), HNK khā- (23), GJR khā- (23), DGR khā- (23), LHD khā- (23), SND khā- (23), RAJ khāv- (23), GUJ khā- (23), MAR khā- (23), BNG kha- (23), ASS kha- (23), NEP khā- (23), SNG ka- (23), MAL ka- (23), KOT khā(23), HIM khā- (23), KUL khā- (23), MND khā- (23), ORY khā- (23), AWD khā- (23), KUM khā(23), ROM xa- (23), KNK kha- (23), DUM khā- (23), BRJ khā- (23), GRH khā- (23), PRY kha(23), MAI khā- (23), KCH khāy- (23), MEW khā- (23), WGD khāy- (23), BNJ khā- (23) 24. EGG: HND aṇḍā (24), DKH aṇḍ(r)ā (24), PNJ āṇḍā (24), PTH aṇḍā (24), HNK aṭṛā (24), GJR ānṭro (24), DGR āṇḍā (24), LHD ān (24), SND āno (24), RAJ aṇḍau (24), GUJ iṇḍũ (24), MAR aṇḍẽ (24), BNG ḍim (304), ASS ḍima (304), NEP aṇḍā (24), SNG aṇḍa (24), MAL bis (398), KOT pinne (442), HIM aṇḍā (24), KUL aṇḍa (24), MND aṇḍa (24), ORY ḍim (304), AWD aṇḍā (24), KUM aṇḍa (24), ROM anrro (24), KNK tanthe (–4), DUM tigon (–9), BRJ āṇḍā (24), GRH aṇḍru (24), PRY anḍō (24), MAI ḍīm (304), KCH īnõ (24), MEW aṇḍā (24), WGD ẽḍā (24), BNJ iṇḍā (24) 25. EYE: HND kh (25), DKH ankhī (25), PNJ akkh (25), PTH akh (25), HNK akh (25), GJR akh (25), DGR akkh (25), LHD akkh (25), SND akhi (25), RAJ kh (25), GUJ kh (25), MAR ḍoḷā (274), BNG cokh (305), ASS sku (305), NEP ãkhā (25), SNG äsa (25), MAL lō (529), KOT akkh (25), HIM ākh (25), KUL hkhi (25), MND hākh (25), ORY ākhi (25), AWD khi (25), KUM ākho (25), ROM yak(h) (25), KNK doḷo (274), DUM ač (25), BRJ k (25), GRH kho (25), PRY ank (25), MAI kh (25), KCH akh (25), MEW kh (25), WGD kh (25), BNJ āngki (25) 26. FAT N.: HND carbī (–2), DKH carbī (–1), PNJ carbī (–1), PTH thendā (474), HNK mẽj (158), GJR carbī (–1), DGR mĩj (158), LHD carbī (–1), SND carbī (–1), RAJ cgtau (220), GUJ carbī (–1), MAR carbī (–5), BNG corbi (–3), ASS srbi (–1), NEP boso (361), SNG tela (399), MAL ? (–1), KOT cikṭ (220), HIM carbī (–1), KUL ? (–3), MND miñj (158), ORY carbī (–3), AWD carbī (–3), KUM baso (361), ROM čiken (220), KNK vos (361), DUM mao (–10), BRJ carbī (–2), GRH bãvāḷi (654), PRY lant (645), MAI carbī (–2), KCH carbī (–1), MEW carbī (–2), WGD carbī (–2), BNJ maḷāi (670) 26a. FAT N.: KOT mīndz (158) 27. FEATHER: HND pãkh (26), DKH pãkh (26), PNJ pãkh (26), PTH par (–1), HNK kham (119), GJR khamm (119), DGR phãgh (26), LHD par (–2), SND khambh (119), RAJ pkh (26), GUJ pchũ (250), MAR pākh (26), BNG palk (306), ASS pakhi (26), NEP pvkh (26), SNG 243

Anton I. Kogan

pihāṭuva (250), MAL dūnifaiy (250), KOT pankh (26), HIM pãkh (26), KUL pankh (26), MND pānkh (26), ORY pālak (306), AWD paŋkh (26), KUM pankh (26), ROM por (–5), KNK pāk (26), DUM ? (–11), BRJ pākh (26), GRH ? (–2), PRY par (–5), MAI picch (250), KCH ? (–2), MEW pankhī (26), WGD pakh (26), BNJ pangkoḍā (26) 27a. FEATHER: PNJ khãbh (119), GJR pakkh (26), DGR khambh (119), SND pakkhu (26), GUJ pkh (26), MAR pīs (250) 28. FIRE: HND āg (27), DKH ag (27), PNJ agg (27), PTH ag (27), HNK ag (27), GJR ag (27), DGR agg (27), LHD bhāh (679), SND bāhi (679), RAJ āg (27), GUJ āg (27), MAR āg (27), BNG agun (–4), ASS zui (334), NEP āgo (27), SNG ginna (–2), MAL alifān (531), KOT āg (27), HIM āg (27), KUL g (27), MND āg (27), ORY ni (673), AWD āgi (27), KUM āgo (27), ROM yag (27), KNK ujjo (334), DUM ak (27), BRJ āge (27), GRH āg (27), PRY ak (27), MAI āgi (27), KCH āg (–3), MEW āgi (27), WGD āgi (27), BNJ angār (671) 29. FISH: HND machlī (28), DKH macchī (28), PNJ macchī (28), PTH machlī (28), HNK macchī (28), GJR machī (28), DGR macchī (28), LHD machī (28), SND machī (28), RAJ māchḷī (28), GUJ māchlī (28), MAR māsā (28), BNG mach (28), ASS mas (28), NEP māchā (28), SNG māluvā (–3), MAL mas (28), KOT macchi (28), HIM macchī (28), KUL machḷi (28), MND macchī (28), ORY māccha (28), AWD macharī (28), KUM mācho (28), ROM mašo (28), KNK māsoḷi (28), DUM čhumo (–12), BRJ macchī (28), GRH māchu (28), PRY mačhi (28), MAI māch (28), KCH machi (28), MEW macchī (28), WGD machlī (28), BNJ mācli (28) 30. FLY V.: HND uṛ- (29), DKH uṛ- (29), PNJ uḍḍ- (29), PTH uḍar- (29), HNK uḍ(ar)- (29), GJR uḍ- (29), DGR uḍḍ- (29), LHD uḍḍ- (29), SND uḍ’ir- (29), RAJ uḍ- (29), GUJ uḍ- (29), MAR uḍ(29), BNG uṛe ja- (29), ASS ur- (29), NEP uḍ- (29), SNG piyāba- (401), MAL uduh- (532), KOT ṛu- (29), HIM uḍ- (29), KUL uḍḍ- (29), MND uḍ- (29), ORY uṛ- (29), AWD uṛ- (29), KUM uṛ(29), ROM (h)urr- (29), KNK uḍ- (29), DUM duwal- (–13), BRJ uṛ- (29), GRH uṛ- (29), PRY uḍ(29), MAI uṛ- (29), KCH auḍ- (29), MEW uṛ- (29), WGD uḍ- (29), BNJ waḍ- (29) 31. FOOT: HND pair (30), DKH pāõ (30), PNJ pair (30), PTH pair (30), HNK pair (30), GJR pair (30), DGR pair (30), LHD pēr (30), SND per (30), RAJ pag (30), GUJ pag (30), MAR pāy (30), BNG pa (30), ASS pau (30), NEP pāu (30), SNG paya (30), MAL faithila (30), KOT pau (30), HIM pair (30), KUL pr (30), MND pair (30), ORY pā (30), AWD pag (30), KUM khuṭṭo (587), ROM punrro (596), KNK pāy (30), DUM po (30), BRJ pair (30), GRH khuṭṭu (587), PRY per (30), MAI pa (30), KCH pag (30), MEW pēr (30), WGD pog (30), BNJ pak (30) 32. FULL: HND bharā (31), DKH pūrbhar (31), PNJ bhariyā (31), PTH pūrā (32), HNK pūrā (32), GJR pūro (32), DGR pūrā (32), LHD bhareā (31), SND bhariyalu (31), RAJ pūr (32), GUJ pūrũ (32), MAR bharlelẽ (31), BNG bhra (31), ASS pur (32), NEP bharī (31), SNG piri (32), MAL furifai (32), KOT bhr (31), HIM bhareādā (31), KUL bhru (31), MND bharyā (31), ORY pūra (32), AWD purhar (32), KUM purro (32), ROM pherdo (31), KNK bhartī (31), DUM ? (– 14), BRJ puriyo (32), GRH bhary (31), PRY par (31), MAI bharit (31), KCH sajo (656), MEW pūro (32), WGD bhareo (31), BNJ bharo (31) 32a. FULL: HND pūrā (32), DKH pūrā (32), PNJ pūrā (32), PTH bhariyā (31), HNK bhariyā (31), GJR bhariyo (31), DGR bharā (31), LHD pūrā (32), SND pūro (32), RAJ bharpūr (31), GUJ bharpūr (31), MAR purā (32), BNG pura (32), ASS bhra (31), KOT pur (32), HIM pūrā (32), MND pūra (32), ORY bhari (31), AWD bharā (31), KUM bhariyo (31), GRH pūru (32), KCH bhariyo (31) 244

Genealogical classification of New Indo-Aryan languages and lexicostatistics

33. GIVE: HND de- (33), DKH de(v)- (33), PNJ de- (33), PTH de- (33), HNK de- (33), GJR de(33), DGR de- (33), LHD ḍev- (33), SND ḍ’i- (33), RAJ de- (33), GUJ de- (33), MAR de- (33), BNG de- (33), ASS di- (33), NEP di- (33), SNG de- (33), MAL din- (33), KOT de- (33), HIM de(33), KUL de- (33), MND de- (33), ORY de- (33), AWD de- (33), KUM di- (33), ROM de- (33), KNK di- (33), DUM de- (33), BRJ dew- (33), GRH di- (33), PRY de- (33), MAI de- (33), KCH ḍe(33), MEW de- (33), WGD de- (33), BNJ de- (33) 34. GOOD: HND acchā (34), DKH acchā (34), PNJ hachā (34), PTH acchā (34), HNK hacchā (34), GJR acho (34), DGR caŋgā (120), LHD cangā (120), SND cango (120), RAJ acchau (34), GUJ bhalo (184), MAR cglā (120), BNG bhal (184), ASS bhal (184), NEP ramro (362), SNG hoñda (402), MAL rangaḷu (533), KOT bhl (184), HIM acchā (34), KUL bhala (184), MND kharā (554), ORY bhala (184), AWD nīk (466), KUM niko (466), ROM lašo (34), KNK cg (120), DUM šonā (624), BRJ ācho (34), GRH accho (34), PRY hačo (34), MAI nik (466), KCH cangho (120), MEW bhalo (184), WGD haue (661), BNJ āco (34) 34a. GOOD: DKH nīkā (466), PNJ cangā (120), PTH cangā (120), HNK cangā (120), GJR caŋgo (120), DGR kharā (494), RAJ bhalau (184), NEP bhalo (184), SNG manā (403), KUL khra (554), MND bhallā (184), AWD acchā (34), KUM ācho (34), GRH bhalu (184) 35. GREEN: HND harā (35), DKH haryā (35), PNJ har(i)ā (35), PTH harā (35), HNK sāvā (160), GJR nīlo (253), DGR saillā (495), LHD sāvā (160), SND sāo (160), RAJ līlau (253), GUJ lilo (253), MAR hirvā (35), BNG šbuj (–5), ASS xeuj (–2), NEP hariyo (35), SNG palāvan (404), MAL fehikulaige (534), KOT hr (35), HIM arā (35), KUL hra (35), MND saillā (495), ORY sabujā (– 4), AWD hariar (35), KUM hariyo (35), ROM zeleno (–6), KNK pācve (612), DUM jutiqam (– 15), BRJ haro (35), GRH haryũ (35), PRY sōz (–6), MAI sabuj (–3), KCH sāo (160), MEW haro (35), WGD haro (35), BNJ haro (35) 35a. GREEN: GJR hariyo (35), RAJ harau (35), MND harā (35) 36. HAIR: HND bāl (36), DKH bāl (36), PNJ vāl (36), PTH bāl (36), HNK bāl (36), GJR bāl (36), DGR kes (185), LHD vāl (36), SND vār (36), RAJ bāḷ (36), GUJ vāḷ (36), MAR kẽs (185), BNG cul (308), ASS suli (308), NEP bāl (36), SNG kespata (185), MAL isthaši (535), KOT bāḷ (36), HIM bāl (36), KUL šrāḷ (36), MND bāḷh (36), ORY cuḷi (308), AWD bār (36), KUM bāl (36), ROM bal (36), KNK kes (185), DUM jāṭ (–16), BRJ bāl (36), GRH bāl (36), PRY bal (36), MAI kes (185), KCH vār (36), MEW bāl (36), WGD wāl (36), BNJ laṭṭā (672) 36a. HAIR: DKH kes (185), SND kes (185), BNG keš (185), KUL cōḍha (308), MND kes (185) 37. HAND: HND hāth (37), DKH hāt (37), PNJ hatth (37), PTH hath (37), HNK hath (37), GJR hath (37), DGR hatth (37), LHD hatth (37), SND hathu (37), RAJ hāth (37), GUJ hāth (37), MAR hāt (37), BNG hat (37), ASS hat (37), NEP hāt (37), SNG ata (37), MAL aiy (37), KOT hatth (37), HIM āth (37), KUL hth (37), MND hāth (37), ORY hāta (37), AWD hāth (37), KUM hāth (37), ROM vast (37), KNK hāt (37), DUM hat (37), BRJ hāt (37), GRH hāt(h) (37), PRY hat (37), MAI hāth (37), KCH hath (37), MEW hāt (37), WGD at (37), BNJ hāt (37) 38. HEAD: HND sir (38), DKH sir (38), PNJ sir (38), PTH sir (38), HNK sir (38), GJR sir (38), DGR sir (38), LHD sir (38), SND siru (38), RAJ māthau (186), GUJ māthũ (186), MAR ḍokẽ (254), BNG matha (186), ASS mur (338), NEP munṭo (338), SNG hisa (38), MAL is (38), KOT mūṇḍ (338), HIM mūṇḍ (338), KUL sir (38), MND sir (38), ORY muṇḍa (338), AWD mṛ (338), KUM muṇḍo (338), ROM šero (38), KNK matte (186), DUM šuṭo (38), BRJ mātho (186), GRH muṇḍ (38), PRY sar (38), MAI māth (186), KCH ḍogo (254), MEW māth (186), WGD monḍ (338), BNJ māto (338) 245

Anton I. Kogan

38a. HEAD: SND matho (186), GUJ ḍokũ (254), BNG šir (38), ASS xir (38), NEP sir (38), KOT šīr (38), HIM sir (38), MND mūṇḍ (338), AWD kapār (581), KUM sir (38), GRH sir (38), MAI mūṛ (338), KCH matho (186) MEW sir (38) 39. HEAR: HND sun- (39), DKH sun- (39), PNJ suṇ- (39), PTH suṇ- (39), HNK suṇ- (39), GJR suṇ- (39), DGR sun- (39), LHD suṇ- (39), SND b’udh- (187), RAJ suṇ- (39), GUJ sāmbhaḷ- (255), MAR aik- (275), BNG šon- (39), ASS xun- (39), NEP sun- (39), SNG asa- (39), MAL iv- (39), KOT šuṇ- (39), HIM suṇ- (39), KUL šuṇ- (39), MND suṇ- (39), ORY šuṇ- (39), AWD sun- (39), KUM suṇ- (39), ROM šun- (39), KNK aik- (275), DUM paruj- (–17), BRJ sun- (39), GRH suṇ(39), PRY sun- (39), MAI sun- (39), KCH soṇ- (39), MEW suṇ- (39), WGD āyak- (275), BNJ sāmaḷ- (255) 40. HEART: HND dil (–3), DKH hiyā (161), PNJ dil (–2), PTH kalejā (143), HNK dil (–2), GJR kāḷjo (143), DGR dil (–1), LHD h (161), SND ho (161), RAJ hīyau (161), GUJ dil (–2), MAR hRday (–6), BNG rhitpiṇḍ (–6), ASS hiya (161), NEP muṭu (365), SNG ḷaya (161), MAL hīy (161), KOT hi (161), HIM kāḷjā (143), KUL dil (–4), MND dil (–1), ORY hrudaya (–5), AWD dil (–4), KUM hiyo (161), ROM ilo (161), KNK kāḷij (143), DUM ya (161), BRJ hiruday (–3), GRH jikuṛu (655), PRY hik (161), MAI hia (161), KCH hīyo (161), MEW hiye (161), WGD dil (–3), BNJ dal (–1) 40a. HEART: RAJ kāḷjau (143), ASS kliza (143), PRY kilijo (143), MEW kālejā (143) 41. HORN: HND sīŋg (40), DKH sing (40), PNJ siŋg (40), PTH sing (40), HNK siŋg (40), GJR siŋg (40), DGR siŋg (40), LHD siŋg (40), SND siŋg (40), RAJ sg (40), GUJ siŋgḍũ (40), MAR šiŋg (40), BNG šiŋg (40), ASS xiŋg (40), NEP sīŋg (40), SNG (h)anga (40), MAL daḷu (537), KOT šīŋg (40), HIM siŋg (40), KUL sīngh (40), MND sīng (40), ORY šinga (40), AWD sīŋi (40), KUM siŋg (40), ROM šing (40), KNK šīng (40), DUM iŋ (40), BRJ sīng (40), GRH siŋ (40), PRY ša

(–7), MAI sg (40), KCH singh (40), MEW sĩh (40), WGD hengṛo (40), BNJ singg (40) 42. I: HND maĩ (41), DKH maĩ (41), PNJ maĩ (41), PTH maĩ (41), HNK mẽ (41), GJR h (41), DGR aũ (41), LHD maĩ (41), SND āũ (41), RAJ h (41), GUJ hũ (41), MAR mī (41), BNG ami (309), ASS mi (41), NEP ma (41), SNG mama (41), MAL ma (41), KOT m (41), HIM aũ (41), KUL hw (41), MND hu (41), ORY mui (41), AWD mah (41), KUM m (41), ROM me (41), KNK āv (41), DUM u (41), BRJ me (41), GRH mi (41), PRY me (41), MAI ham (309), KCH ũ (41), MEW mũ (41), WGD hu (41), BNJ ma (41) 43. KILL: HND mār- (42), DKH mār- (42), PNJ mār- (42), PTH mār- (42), HNK mār- (42), GJR mār- (42), DGR mār- (42), LHD mār- (42), SND mār- (42), RAJ mār- (42), GUJ mār- (42), MAR ṭhār mār- (42), BNG mere phel- (42), ASS mr- (42), NEP mār- (42), SNG mar- (42), MAL mara(42), KOT mar- (42), HIM mār- (42), KUL mār- (42), MND mār- (42), ORY mār- (42), AWD mār- (42), KUM mār- (42), ROM mudar- (42), KNK mār- (42), DUM mār- (42), BRJ mār- (42), GRH mār- (42), PRY mar- (42), MAI mār- (42), KCH mār- (42), MEW mār- (42), WGD mār(42), BNJ mār nāk- (42) 44. KNEE: HND ghuṭnā (43), DKH guṛgā (122), PNJ goḍā (122), PTH goḍā (122), HNK jannū (144), GJR jan (144), DGR goḍḍā (122), LHD goḍā (122), SND goḍ’o (122), RAJ goḍau (122), GUJ ghṭaṇ (122), MAR guḍhā (122), BNG hãṭu (225), ASS ãṭhu (225), NEP ghuṇḍo (43), SNG daṇa (144), MAL kakū (538), KOT dzaṇu (144), HIM jaṇṇū (144), KUL jānnhu (144), MND jānnhu (144), ORY ṭhu (225), AWD ? (–5), KUM ghũṛo (43), ROM čang (597), KNK dimbi 246

Genealogical classification of New Indo-Aryan languages and lexicostatistics

(613), DUM kuṭā (–18), BRJ ? (–4), GRH ghũḍu (43), PRY janu (144), MAI ṭhehun (636), KCH gūḍo (122), MEW ghuṭṇo (43), WGD ? (–4), BNJ goḍo (122) 44a. KNEE: HNK guḍḍar (122), DGR jāhnnū (144), MND goḍḍā (122) 45. KNOW: HND jān- (44), DKH jān- (44), PNJ jāṇ- (44), PTH jāṇ- (44), HNK jāṇ- (44), GJR jāṇ(44), DGR jān- (44), LHD jāṇ- (44), SND jān- (44), RAJ jāṇ- (44), GUJ jān- (44), MAR jāṇ- (44), BNG jan- (44), ASS zan- (44), NEP jān- (44), SNG dan- (44), MAL danna- (44), KOT dzaṇ- (44), HIM jāṇ- (44), KUL jāṇ- (44), MND jāṇ- (44), ORY jāṇ- (44), AWD jān- (44), KUM jāṛ- (44), ROM žan- (44), KNK jāṇ- (44), DUM ? (–19), BRJ jān- (44), GRH jaṇ- (44), PRY samj- (662), MAI jān- (44), KCH jāṇ- (44), MEW jāṇ- (44), WGD jan- (44), BNJ jaṇ- (44) 46. LEAF: HND pattā (45), DKH pattā (45), PNJ pattar (45), PTH pattā (45), HNK pattar (45), GJR patar (45), DGR pattar (45), LHD pātar (45), SND panu (189), RAJ pattau (45), GUJ pdṛũ (189), MAR pān (189), BNG pata (45), ASS pat (45), NEP pāt (45), SNG pata (45), MAL faiy (45), KOT pāč (45), HIM pattā (45), KUL pttr (45), MND pattar (45), ORY patra (–6), AWD pātā (45), KUM pāt (45), ROM patrin (45), KNK pān (189), DUM poṭ (–20), BRJ pattī (45), GRH pathala (45), PRY barg (–8), MAI pāt (45), KCH pann (189), MEW pattā (45), WGD patto (45), BNJ pālo (570) 46a. LEAF: RAJ pānṛau (189), GUJ pataru (45), KNK pāllo (570) 47. LIE: HND leṭ- (46), DKH paṛ- (47), PNJ leṭ- (46), PTH leṭ- (46), HNK lamā ho- (479), GJR lam ho- (479), DGR leṭ- (46), LHD leṭ- (46), SND leṭ- (46), RAJ loṭ- (508), GUJ leṭ- (46), MAR nij- (277), BNG šue thak- (310), ASS pari thk- (47), NEP leṭi rah- (46), SNG hansivī hiṭin- (407), MAL doguhed- (408), KOT pṛ- (47), HIM leṭ- (46), KUL leṭ- (46), MND lambā pai- (47), ORY šo- (310), AWD paũṛh- (582), KUM paṛ- (47), ROM pašlo- (278), KNK āḍ poḍ- (47) DUM ek- (– 21), BRJ leṭ- (46), GRH paṛ- (47), PRY paṛ- (47), MAI leṭ- (46), KCH leṭ- (46), MEW leṭ- (46), WGD so- (310), BNJ āḍo paḍ- (47) 47a. LIE: HND paṛ- (47), PNJ lamme pai- (47), HNK pe- (47), GJR pe- (47), LHD pē- (47), SND pav- (47), RAJ paṛ- (47), GUJ paḍ- (47), MAR pasarlelẽ as- (278), NEP sut- (310), SNG digāvī hiṭin- (408), MAL oiy- (310), KOT leṭ- (46), KUL lmm pw- (47), MND leṭ- (46), MEW so- (310) 48. LIVER: HND kalejā (48), DKH kalejā (48), PNJ kalejā (48), PTH jigar (–2), HNK jigrā (–3), GJR kāḷj (48), DGR kalejā (48), LHD kalejī (48), SND jero (190), RAJ kāḷjau (48), GUJ kalejā (48), MAR yakRt (–7), BNG jkrit (–7), ASS zkrit (–3), NEP kalejo (48), SNG akmāva (410), MAL mey (539), KOT kaldz (–1), HIM kāḷjā (48), KUL kālja (48), MND kāḷjā (48), ORY kalijā (48), AWD karej (48), KUM phyātto (588), ROM buko (496), KNK kāḷij (48), DUM kaulā (625), BRJ haṇḍe (631), GRH ? (–3), PRY kiliji (48), MAI yakRt (–4), KCH jigar (–4), MEW gurdā (–3), WGD kaljā (48), BNJ koljo (48) 49. LONG: HND lambā (49), DKH lāmbā (49), PNJ lammā (49), PTH lammā (49), HNK lammā (49), GJR lam (49), DGR lammā (49), LHD lambā (49), SND ḍigho (191), RAJ lambau (49), GUJ lāmbũ (49), MAR lb (49), BNG lmba (49), ASS dighl (191), NEP lamo (49), SNG dik (191), MAL digu (191), KOT lamb (49), HIM lambā (49), KUL lmma (49), MND lambā (49), ORY lambā (49), AWD lambā (49), KUM lambo (49), ROM lungo (–7), KNK dīg (191), DUM igā (–22), BRJ lambo (49), GRH lambu (49), PRY lamo (49), MAI lambā (49), KCH lamo (49), MEW lambo (49), WGD lambo (49), BNJ lāmbo (49) 49a. LONG: SND lambo (49), ASS lmba (49) 247

Anton I. Kogan

50. LOUSE: HND j (50), DKH j (50), PNJ j (50), PTH jū (50), HNK j (50), GJR j (50), DGR j (50), LHD ju (50), SND ja (50), RAJ j (50), GUJ jū (50), MAR ū (50), BNG ukuṇ (311), ASS okṇi (311), NEP jumrā (50), SNG ukuṇā (311), MAL ukunu (311), KOT jū (50), HIM j (50), KUL jũ (50), MND j (50), ORY ukuṇi (311), AWD ju (50), KUM jũo (50), ROM žuv (50), KNK ū (50), DUM jūya (50), BRJ jūl (50), GRH j (50), PRY jũ (50), MAI j (50), KCH j (50), MEW j (50), WGD j (50), BNJ j (50) 51. MAN: HND mard (–4), DKH purus (123), PNJ purakh (–3), PTH jaṇā (162), HNK jaṇā (162), GJR janū (162), DGR mardmāhnū (192), LHD murs (123), SND māṇuhū (192), RAJ mṭī (509), GUJ bhāyḍo (256), MAR puruṣ (–8), BNG manuš (192), ASS munih (192), NEP lognemanche (192), SNG pirimiyā (411), MAL firihen (123), KOT mrd (–2), HIM lok (480), KUL mrd (–5), MND mard (–2), ORY puruṣa (–7), AWD mansedhū (192), KUM meĩs (192), ROM murš (123), KNK manišu (192), DUM maniš (192), BRJ ādmī (–5), GRH ādmi (–4), PRY anmi (– 9), MAI purukh (–5), KCH māṛū (192), MEW māṇas (192), WGD admī (–5), BNJ māṭi (509) 51a. MAN: BNJ māṇas (192) 52. MANY: HND bahut (51), DKH bhaut (51), PNJ bahut (51), PTH bah (51),HNK muc (481), GJR muc (481), DGR matā (497), LHD bahũ (51), SND ghaṇā (193), RAJ ghaṇau (193), GUJ ghano (193), MAR phār (280), BNG bhu (51), ASS bhut (51), NEP dherai (367), SNG bohō (51), MAL baivaru (51), KOT bto (51), HIM bot (51), KUL bhu (51), MND bahut (51), ORY bahuta (51), AWD bahut (51), KUM bhaut (51), ROM but (51), KNK anek (–5), DUM buṭ (51), BRJ bhaut (51), GRH bahut (51), PRY bot (51), MAI bahut (51), KCH baurā (51), MEW bhot (51), WGD ghanā (193), BNJ ghaṇo (193) 52a. MANY: HNK bah (51), GUJ bahu (51), MAL gina (193), MND mate (497), KCH ghaṇo (193) 53. MEAT: HND gošt (–5), DKH mās (124), PNJ mās (124), PTH mās (124), HNK mās (124), GJR mās (124), DGR mās (124), LHD gošt (–3), SND māsu (124), RAJ ms (124), GUJ ms (124), MAR ms (124), BNG mãkš (124), ASS mngh (124), NEP māsu (124), SNG mas (124), MAL mas (124), KOT mas (124), HIM mās (124), KUL mās (124), MND mās (124), ORY māẽsa (124), AWD māsu (124), KUM māsu (124), ROM mas (124), KNK mās (124), DUM mos (124), BRJ mās (124), GRH sikār (–5), PRY gōš (–10), MAI ms (124), KCH mas (124), MEW ms (124), WGD sīsī (662), BNJ mās (124) 54. MOON: HND cd (52), DKH cd (52), PNJ cann (52), PTH can (52), HNK can (52), GJR can (52), DGR cann (52), LHD candar (52), SND caṇḍu (52), RAJ candau (52), GUJ cāndo (52), MAR cd (52), BNG cand (52), ASS zon (340), NEP jūn (340), SNG hañda (52), MAL handu (52), KOT dzūṇ (340), HIM candramā (–2), KUL joth (340), MND candarmā (–3), ORY candramā (–8), AWD candā (52), KUM candā (52), ROM šonuto (340), KNK candru (52), DUM čon (52), BRJ candā (52), GRH jūn (340), PRY čan (52), MAI candā (52), KCH candhar (52), MEW cd (52), WGD san (52), BNJ cāndā (52) 54a. MOON: KUM jūn (340) 55. MOUNTAIN: HND pahāṛ (53), DKH p(ah)āṛ (53), PNJ pahāṛ (53), PTH pahāṛī (53), HNK pahāṛ (53), GJR pahāṛ (53), DGR phāṛ (53), LHD pahāṛ (53), SND pahaṛu (53), RAJ pahāṛ (53), GUJ pahāḍ (53), MAR pahāḍ (53), BNG pahaṛ (53), ASS pahar (53), NEP pahāḍ (53), SNG kanda (–4), MAL farubada (–2), KOT pāṛ (53), HIM pāṛ (53), KUL pahāḍ (53), MND pāṛ (53), ORY parbata (–9), AWD pahāṛ (53), KUM pahāṛ (53), ROM m!nt’a (–8), KNK parvatu (–6), 248

Genealogical classification of New Indo-Aryan languages and lexicostatistics

DUM čhiā (–23), BRJ pahāṛ (53), GRH pāṛ (53), PRY paṛ (53), MAI parvat (–6), KCH pahāṛī (53), MEW pahāṛ (53), WGD mãro (663), BNJ pāḍ (53) 56. MOUTH: HND mũh (54), DKH m (54), PNJ mh (54), PTH mh (54), HNK mh (54), GJR mũ (54), DGR m (54), LHD mũh (54), SND mũhũ (54), RAJ mũh (54), GUJ mhõ (54), MAR tõḍ (281), BNG mukh (54), ASS mukh (54), NEP mukh (54), SNG muva (54), MAL anga (540), KOT m (54), HIM mũ (54), KUL mūh (54), MND m (54), ORY muh (54), AWD muh (54), KUM mukh (54), ROM muy (54), KNK tõḍ (281), DUM kašā (626), BRJ mũhṭe (54), GRH gicco (657), PRY mo (54), MAI muh (54), KCH mõ (54), MEW mũh (54), WGD munḍo (54), BNJ muṇḍo (54) 57. NAME: HND nām (55), DKH nāõ (55), PNJ n (55), PTH n (55), HNK n (55), GJR n (55), DGR n (55), LHD n (55), SND nālo (55), RAJ nv (55), GUJ nām (55), MAR nv (55), BNG nam (55), ASS nam (55), NEP nāũ (55), SNG nama (55), MAL nan (55), KOT naũ (55), HIM nāũ (55), KUL n (55), MND n (55), ORY nāma (55), AWD nu (55), KUM nou (55), ROM nav (55), KNK nāv (55), DUM nom (55), BRJ nām (55), GRH nau (55), PRY na (55), MAI nām (55), KCH nālo (55), MEW nw (55), WGD nām (55), BNJ nām (55) 58. NECK: HND galā (56), DKH galā (56), PNJ galā (56), PTH dhauṇ (125), HNK gaḷā (56), GJR gaḷ (56), DGR dhaun (125), LHD giccī (196), SND g’icī (196), RAJ nāṛ (514), GUJ ḍokī (258), MAR mān (229), BNG gla (56), ASS gl (56), NEP galām (56), SNG gela (312), MAL karu (413), KOT kṛi (228), HIM kyāṛī (228), KUL kyāḍi (228), MND kaiṛī (228), ORY beka (674), AWD gaṭaī (197), KUM ghṭī (312), ROM korr (599), KNK gaḷo (56), DUM koṭā (599), BRJ nāri (514), GRH dhauṇ (125), PRY gardan (–11), MAI dhonā (125), KCH giccī (196), MEW galo (56), WGD gābṛo (664), BNJ gaḷā (56) 58a. NECK: PNJ dhauṇ (125), HNK gāṭā (197), GJR gāṭo (197), DGR gāṭā (197), SND g’āṭo (197), RAJ dhūṇ (125), GUJ galcī (56), MAR gaḷā (56), BNG ghaṛ (312), NEP ghṭī (312), SNG kara (413), KOT gḷ (56), KUL gḷa (56), MND g"ḷ (56), AWD gar (56), KUM galo (56), GRH gaḷā (56), MAI ghāṛ (312), KCH niṛī (514), BNJ naḷḍi (514) 59. NEW: HND nayā (57), DKH navā (57), PNJ nav (57), PTH nav (57), HNK navv (57), GJR naw (57), DGR nam (57), LHD nav (57), SND naõ (57),RAJ navau (57), GUJ navũ (57), MAR navā (57), BNG nb (57), ASS n (57), NEP nay (57), SNG nava (57), MAL au (541), KOT nuw (57), HIM navā (57), KUL nwwã (57), MND nav (57), ORY nuā (57), AWD naykā (57), KUM nayyo (57), ROM nevo (57), KNK nave (57), DUM namā (57), BRJ nayo (57), GRH nay (57), PRY navo (57), MAI nab (57), KCH na (57), MEW nawā (57), WGD navo (57), BNJ nawo (57) 59a. NEW: SNG alut (541) 60. NIGHT: HND rāt (58), DKH rāt (58), PNJ rāt (58), PTH rāt (58), HNK rāt (58), GJR rāt (58), DGR rāt (58), LHD rāt (58), SND rāti (58), RAJ rāt (58), GUJ rāt (58), MAR rāt (58), BNG rat (58), ASS rati (58), NEP rāt (58), SNG räya (58), MAL reygandu (58), KOT rāč (58), HIM rattī (58), KUL rāt (58), MND rāt (58), ORY rāti (58), AWD rāti (58), KUM rāt (58), ROM rat (58), KNK rātī (58), DUM thop (–24), BRJ rāt (58), GRH rāt (58), PRY rat (58), MAI rāti (58), KCH rāt (58), MEW rāt (58), WGD rat (58), BNJ rāt (58) 61. NOSE: HND nāk (59), DKH nāk (59), PNJ nakk (59), PTH nak (59), HNK nak (59), GJR nak (59), DGR nakk (59), LHD nakk (59), SND naku (59), RAJ nāk (59), GUJ nāk (59), MAR nāk (59), BNG nak (59), ASS nak (59), NEP nāk (59), SNG nahaya (59), MAL neyfaiy (59), KOT nāk 249

Anton I. Kogan

(59), HIM nāk (59), KUL nāk (59), MND nāk (59), ORY nāka (59), AWD nāki (59), KUM nāk (59), ROM nak (59), KNK nk (59), DUM nok (59), BRJ nāk (59), GRH nāk (59), PRY nak (59), MAI nāk (59), KCH nakk (59), MEW nāk (59), WGD nakoṛo (59), BNJ nāk (59) 62. NOT: HND na (60), DKH nakko (60), PNJ nā (60), PTH na (60), HNK nā (60), GJR na (60), DGR n (60), LHD na (60), SND na (60), RAJ nā (60), GUJ nā (60), MAR na (60), BNG na (60), ASS n (60), NEP na (60), SNG nǟ (60), MAL nu (60), KOT na (60), HIM na (60), KUL ny (60), MND na (60), ORY n (60), AWD nāh (60), KUM n (60), ROM na (60), KNK nhãyī (60), DUM ni (60), BRJ nāye (60), GRH na (60), PRY na (60), MAI nahi (60), KCH na (60), MEW ni (60), WGD nā (60), BNJ ni (60) 63. ONE: HND ek (61), DKH yek (61), PNJ ik (61), PTH hik (61), HNK hik (61), GJR ek (61), DGR ikk (61), LHD ek (61), SND hiku (61), RAJ (h)ik (61), GUJ ek (61), MAR ek (61), BNG ek (61), ASS ek (61), NEP ek (61), SNG eka (61), MAL ekeh (61), KOT ēk (61), HIM ek (61), KUL yek (61), MND ek (61), ORY eka (61), AWD yak (61), KUM ek (61), ROM (y)ek (61), KNK ek (61), DUM ek (61), BRJ ek (61), GRH ek (61), PRY yek (61), MAI ek (61), KCH hikṛo (61), MEW ek (61), WGD ek (61), BNJ ek (61) 64. PERSON: HND janā (163), DKH janā (163), PNJ jaṇā (163), PTH ādmī (–3), HNK lok (461), GJR jaṇū (163), DGR māhnū (126), LHD bandā (–4), SND jaṇo (163), RAJ jaṇau (163), GUJ māṇas (126), MAR māṇūs (126), BNG manuš (126), ASS manuh (126), NEP mānis (126), 126 SNG minihā (126), MAL mēhā (126), KOT dzṇ (163), HIM māṇū (126), KUL mhāṇu (126), MND māṇu (126), ORY maṇiša (126), AWD manaī (126), KUM jaṇ (163), ROM manuš (126), KNK manišu (126), DUM maniš (126), BRJ mānus (126), GRH maṇāḷi (126), PRY jaṇo (163), MAI manukh (–7), KCH māṛū (126), MEW ādamī (–4), WGD lak (461), BNJ jaṇā (163) 64a. PERSON: DKH mānūs (126), DGR janā (163), RAJ māṇas (126), BNG lok (461), KOT maṇčh (126), MND jaṇā (163), KUM meĩs (126), GRH jaṇ (163), BNJ māṇas (126) 65. RAIN: HND mẽh (62), DKH barsāt (127), PNJ mh (62), PTH baddal (127), HNK baddal (127), GJR badal (127), DGR barkhā (–2), LHD mh (62), SND mhu (62), RAJ meh (62), GUJ varsād (127), MAR pāūs (127), BNG badl (127), ASS brkhuṇ (127), NEP jharī (369), SNG vässa (127), MAL vārey (127), KOT pāṇī (447), HIM barkhā (–3), KUL gāš (663), MND barkhā (–4), ORY barša (–10), AWD barkhā (–4), KUM me (62), ROM beršind (127), KNK pāvsu (127), DUM orp (661), BRJ meg (62), GRH barkhā (–6), PRY mi (62), MAI barkhā (–8), KCH meh (62), MEW meh (62), WGD mehā (62), BNJ pāṇi (447) 65a. RAIN: DKH mhyūn (62), HNK mh (62), GJR mh (62), SND vas (127), KCH varsād (127) 66. RED: HND lāl (63), DKH lāl (63), PNJ lāl (63), PTH sūā (475), HNK rattā (128), GJR rato (128), DGR lāl (63), LHD rattā (128), SND lālu (63), RAJ rātau (128), GUJ lāl (63), MAR tbḍā (282), BNG lal (63), ASS rŋga (128), NEP rāto (128), SNG ratu (128), MAL raiy (128), KOT ratt (128), HIM lāl (63), KUL lāl (63), MND lāl (63), ORY lāl (63), AWD lāl (63), KUM lāl (63), ROM lolo (63), KNK tãbḍe (282), DUM loyā (63), BRJ lāl (63), GRH lāl (63), PRY lal (63), MAI lāl (63), KCH ratto (128), MEW lāl (63), WGD lāl (63), BNJ rātḍo (128) 66a. RED: DKH ratlā (128), PNJ rattā (128), HNK lāl (63), GJR lāl (63), LHD lāl (63), SND rato (128), RAJ lāl (63), GUJ rātũ (128), ASS lal (63), KOT lāl (63), MAI rāt (128) 67. ROAD: HND saṛak (64), DKH bāṭ (164), PNJ saṛak (64), PTH saṛak (64), HNK šiṛṛak (64), GJR saṛak (64), DGR batt (164), LHD vāṭ (164), SND saṛak (64), RAJ saṛak (64), GUJ saḍak (64), 250

Genealogical classification of New Indo-Aryan languages and lexicostatistics

MAR saḍak (64), BNG pth (–8), ASS baṭ (164), NEP bāṭo (164), SNG maga (415), MAL magu (415), KOT bāt (164), HIM bāṭ (164), KUL bŏtt (164), MND peyṇḍa (448), ORY šaṛaka (64), AWD bāṭ (164), KUM bāṭo (164), ROM drom (–9), KNK mārgu (–7), DUM panā (–25), BRJ saṛak (64), GRH bāṭu (164), PRY ra (–12), MAI path (–9), KCH vāṭ (164), MEW saṛak (64), WGD wāṭ (164), BNJ wāṭ (164) 67a. ROAD: KOT pṇḍ (448) 68. ROOT: HND jaṛ (65), DKH jaṛ (65), PNJ jaṛ (65), PTH jaṛ (65), HNK jaṛh (65), GJR jaṛ (65), DGR jaṛh (65), LHD jaṛṛ (65), SND jaṛu (65), RAJ jaṛ (65), GUJ mūḷ (259), MAR mūḷ (259), BNG šikṛ (313), ASS xipa (343), NEP jarā (65), SNG mula (259), MAL mō (259), KOT dzṛh (65), HIM jaṛ (65), KUL jlāḍa (65), MND jaḍ (65), ORY mūḷa (259), AWD jari (65), KUM jaṛ (65), ROM koreno (–10), KNK mūḷ (259), DUM moli (–26), BRJ jaṛ (65), GRH jaṛ (65), PRY rešo (–13), MAI jaṛi (65), KCH ? (–5), MEW jaṛ (65), WGD mūr (259), BNJ jaḍ (65) 68a. ROOT: GUJ jaḍ (65), MEW mūl (259) 69. ROUND: HND gol (66), DKH gol (66), PNJ gol (66), PTH gol (66), HNK gol (66), GJR goḷ (66), DGR gol (66), LHD gol (66), SND golu (66), RAJ goḷ (66), GUJ gol (66), MAR gol (66), BNG gol (66), ASS gol (66), NEP bāṭulo (283), SNG vaṭa (283), MAL vah (283), KOT gōḷ (66), HIM gol (66), KUL gōl (66), MND gōl (66), ORY gola (66), AWD gol (66), KUM golo (66), ROM rotato (–11), KNK urūṭ (–8), DUM ? (–27), BRJ gol (66), GRH gulgaṇḍo (66), PRY ? (–14), MAI gol (66), KCH gird (–6), MEW ? (–5), WGD ? (–6), BNJ gol (66) 69a. ROUND: MAR vāṭoḷā (283) 70. SAND: HND ret (67), DKH bālū (68), PNJ ret (67), PTH ret (67), HNK rt (67), GJR ret (67), DGR retā (67), LHD ret (67), SND retī (67), RAJ bāḷū (68), GUJ retī (67), MAR retī (67), BNG bali (68), ASS bali (68), NEP ret (67), SNG väli (67), MAL veli (68), KOT baḷu (68), HIM ballū (68), KUL rēt (67), MND bāllu (68), ORY bāli (68), AWD bārū (68), KUM balwā (68), ROM kišay (–12), KNK rev (67), DUM bāli (68), BRJ bālū (68), GRH bālo (68), PRY ? (–15), MAI bālū (68), KCH ? (–7), MEW bālū (68), WGD ret (67), BNJ retu (67) 70a. SAND: HND bālū (68), PNJ bālū (68), SND vārī (68), RAJ ret (67), GUJ vālu (68), MAR vāḷū (68), NEP bāluvā (68), HIM ret (67), MND ret (67), AWD ret (67), MEW ret (67) 71. SAY: HND kah- (69), DKH ka- (69), PNJ kah- (69), PTH ākh- (468), HNK ā - (468), GJR keh- (69), DGR ākh- (468), LHD ākh- (468), SND al- (166), RAJ kah- (69), GUJ kah- (69), MAR mhaṇ- (231), BNG kh- (69), ASS k- (69), NEP bhan- (231), SNG kiya- (69), MAL buna- (231), KOT kiu- (69), HIM bol- (70), KUL bōl-(70), MND bōl- (70), ORY kah- (69), AWD kah- (69), KUM kū- (69), ROM phen- (231), KNK mhāṇ- (231), DUM mun- (231), BRJ keh- (69), GRH bval- (70), PRY ke- (69), MAI kah- (69), KCH co- (657), MEW kah- (69), WGD kah- (69), BNJ ke(69) 71a. SAY: HND bol- (70), DKH bol- (70), PNJ ākh- (468), PTH bol- (70), HNK bol- (70), GJR bol(70), DGR bol- (70), LHD bol- (70), SND b’ol- (70), RAJ bol- (70), GUJ bol- (70), MAR bol- (70), BNG bl- (70), ASS bol- (70), NEP bol- (70), MAL kiya- (69), KOT bol- (70), ORY bol- (70), AWD bol- (70), KUM bol- (70), MAI bol- (70), KCH bol- (70) 72. SEE: HND dekh- (71), DKH dekh- (71), PNJ dekh- (71), PTH tak- (476), HNK de - (71), GJR dekh- (71), DGR dikkh- (71), LHD ḍekh- (71), SND ḍ’is- (129), RAJ dekh- (71), GUJ dekh- (71), MAR pāh- (198), BNG dekh- (71), ASS dekh- (71), NEP dekh- (71), SNG daki- (71), MAL duš(129), KOT dekh- (71), HIM dekh- (71), KUL dēkh- (71), MND dekh- (71), ORY dekh- (71), 251

Anton I. Kogan

AWD dekh- (71), KUM dekh- (71), ROM dikh- (71), KNK paḷe- (617), DUM dekh- (71), BRJ dek- (71), GRH dekh- (71), PRY dekh- (71), MAI dekh- (71), KCH ḍis- (129), MEW dekh- (71), WGD dekh- (71), BNJ dek- (71) 72a. SEE: PNJ diṭh- (129), HNK diṭṭh- (129), GJR ḍiṭṭh- (129), DGR ḍiṭṭh- (129), LHD ḍiṭṭh(129), SND pas- (198), RAJ dīṭh- (129), GUJ dīṭh- (129), MAL deke- (71), KNK dīs- (129), BNJ diṭ- (129) 73. SEED: HND bīj (72), DKH bj (72), PNJ b (72), PTH b (72), HNK b (72), GJR bīj (72), DGR bī (72), LHD bīj (72), SND b’iju (72), RAJ bīj (72), GUJ bī (72), MAR b (72), BNG bij (72), ASS biz (72), NEP bīũ (72), SNG bījaya (–5), MAL oh (543), KOT bīdz (72), HIM bīj (72), KUL bejja (72), MND bīu (72), ORY bīja (72), AWD biyā (72), KUM byū (72), ROM sem!nca (–13), KNK bī (72), DUM %ono (–28), BRJ bīj (72), GRH bīj (72), PRY ? (–16), MAI bīj (72), KCH bijj (72), MEW bihan (72), WGD ? (–7), BNJ bijā (72) 74. SIT: HND baiṭh- (73), DKH baiṭ(h)- (73), PNJ baiṭh- (73), PTH beh- (73), HNK baiṭh- (73), GJR bais- (73), DGR beh- (73), LHD beh- (73), SND vih- (73), RAJ baiṭh- (73), GUJ bes- (73), MAR bas- (73), BNG bš- (73), ASS bh- (73), NEP basi rah- (73), SNG iñdagan hiṭi- (418), MAL inn- (418), KOT bēš- (73), HIM baiṭh- (73), KUL bšš- (73), MND beyṭh- (73), ORY baš- (73), AWD baiṭh- (73), KUM beiṭh- (73), ROM beš- (73), KNK bais- (73), DUM beš- (73), BRJ beṭh(73), GRH baiṭh- (73), PRY beṭ- (73), MAI baiṭh- (73), KCH ve- (73), MEW bēṭh- (73), WGD way- (73), BNJ bes- (73) 75. SKIN: HND camṛī (74), DKH camṛī (74), PNJ camṛī (74), PTH camṛī (74), HNK camṛā (74), GJR camṛī (74), DGR camm (74), LHD camṛī (74), SND camṛī (74), RAJ khāḷ (75), GUJ camḍī (74), MAR camḍī (74), BNG camṛa (74), ASS sal (348), NEP chālā (348), SNG hama (74), MAL han (74), KOT camṛi (74), HIM camṛī (74), KUL camḍi (74), MND khāl (75), ORY camaṛā (74), AWD khāl (75), KUM khāl (75), ROM morči (–14), KNK cām (74), DUM čom (74), BRJ cam (74), GRH camṛī (74), PRY pōst (–17), MAI camṛā (74), KCH camṛī (74), MEW khāl (75), WGD samḍī (74), BNJ khālḍi (75) 75a. SKIN: HND khāl (75), DKH khararī (75), PNJ khall (75), DGR khall (75), LHD khal (75), SND khal (75), RAJ cāmṛī (74), GUJ khāl (75), MAR kātḍī (284), KUL khōlḍi (75), MND cāmṛī (74), AWD camṛā (74), KUM cām (74), KNK sālī (348), MAI khāl (75) 76. SLEEP: HND so- (76), DKH so- (76), PNJ so- (76), PTH seh- (76), HNK sẽ- (76), GJR so- (76), DGR sau- (76), LHD summ- (76), SND sumh- (76), RAJ sūv- (76), GUJ su- (76), MAR nij- (285), BNG ghuma- (314), ASS xo- (76), NEP sut- (76), SNG nid- (199), MAL nid- (199), KOT sut- (76), HIM sut- (76), KUL so- (76), MND sau- (76), ORY šo- (76), AWD so- (76), KUM nīn kar- (199), ROM so- (76), KNK nīd- (199), DUM so- (76), BRJ sow- (76), GRH sī- (76), PRY so- (76), MAI sut- (76), KCH sum- (76), MEW so- (76), WGD so- (76), BNJ so- (76) 76a. SLEEP: SND ninḍ’ kar- (199), NEP nidau- (199), KUM sit- (76) 77. SMALL: HND choṭā (77), DKH lhorā (470), PNJ choṭā (77), PTH nikkā (130), HNK nikkā (130), GJR nikko (130), DGR lauhkā (470), LHD nikkā (130), SND naṇḍho (200), RAJ nain (261), GUJ nānũ (261), MAR lahān (261), BNG choṭ (77), ASS xru (349), NEP sāno (261), SNG kuḍā (420), MAL kuḍa (420), KOT choṭṭ (77), HIM nikkājā (130), KUL choṭṭa (77), MND halkā (565), ORY choṭa (77), AWD choṭ (77), KUM nāno (261), ROM c!gno (–15), KNK sān (261), DUM čunā (–29), BRJ choṭṭo (77), GRH chvaṭṭu (77), PRY nuko (130), MAI choṭ (77), KCH niṇḍho (200), MEW choṭo (77), WGD nenḍho (200), BNJ nānkyā (261) 252

Genealogical classification of New Indo-Aryan languages and lexicostatistics

77a. SMALL: DKH choṭā (77), PNJ nikkā (130), HNK choṭā (77), GJR choṭo (77), DGR nikkā (130), LHD choṭā (77), SND choṭo (77), RAJ choṭau (77), HIM choṭā (77), AWD nanhakā (261), KUM choṭo (77), MAI nānh (261) 78. SMOKE: HND dhu (78), DKH dhuv (78), PNJ dhū (78), PTH dhū (78), HNK dhūv (78), GJR dhū (78), DGR dh (78), LHD dh (78), SND dũhõ (78), RAJ dhũāṛau (78), GUJ dhumāḍo (78), MAR dhūr (78), BNG dhõva (78), ASS dhõva (78), NEP dhuv (78), SNG duma (78), MAL dun (78), KOT dh (78), HIM dhū (78), KUL dhũwã (78), MND dh (78), ORY dhū (78), AWD dhuw (78), KUM dhuo (78), ROM thuv (78), KNK dhuv(o)ru (78), DUM dum (78), BRJ dhũo (78), GRH dhuā (78), PRY tõva (78), MAI dhu (78), KCH dhūo (78), MEW dhū (78), WGD dū (78), BNJ ? (–2) 79. STAND: HND khaṛā ho- (79), DKH khaṛ- (79), PNJ khalo- (79), PTH khal- (79), HNK khal(79), GJR khaḷ- (79), DGR khaṛo- (79), LHD khaṛ- (79), SND bīh- (201), RAJ ūbh- (262), GUJ ubhũ rah- (262), MAR khaḍā as- (79), BNG dãṛa- (315), ASS thiyoi thk- (350), NEP ubhi- (262), SNG nägiṭa hiṭi- (350), MAL theduve hur- (350), KOT khṛ- (79), HIM khaṛ- (79), KUL khḍ(79), MND khaṛ- (79), ORY ṭhiā he- (350), AWD ṭhāṛh ho- (516), KUM ṭhaṛ- (516), ROM terd(516), KNK ubbī- (262), DUM cak hut- (–30), BRJ ṭhāṛe ho- (514), GRH ṭhāḍu ho- (514), PRY kalo ho- (79), MAI ṭhāṛh bhe- (514), KCH aubhū- (262), MEW khaṛo ho- (79), WGD ub- (262), BNJ hubar- (262) 79a. STAND: HNK uṭh- (350), RAJ ṭhāḍau hov- (516), GUJ khaḍũ rah- (79), MAR ubhā rāh(262) 80. STAR: HND tārā (80), DKH tārā (80), PNJ tārā (80), PTH tārā (80), HNK tārā (80), GJR tāro (80), DGR tārā (80), LHD tārā (80), SND tāro (80), RAJ tārau (80), GUJ tāro (80), MAR tārā (80), BNG tara (80), ASS tra (80), NEP tārā (80), SNG taruva (80), MAL thari (80), KOT tar (80), HIM tārā (80), KUL tāra (80), MND tārā (80), ORY tārā (80), AWD taraī (80), KUM tāro (80), ROM čehran (–16), KNK nakṣatr (–9), DUM tāro (80), BRJ tāre (80), GRH gaiṇu (658), PRY taro (80), MAI tārā (80), KCH tāro (80), MEW tārai (80), WGD tāro (80), BNJ tārā (80) 81. STONE: HND patthar (81), DKH pāthar (81), PNJ patthar (81), PTH baṭṭā (167), HNK baṭā (167), GJR baṭo (167), DGR baṭṭā (167), LHD patthar (81), SND pattharu (81), RAJ bhāṭau (167), GUJ patthar (81), MAR dagaḍ (286), BNG pathr (81), ASS xil (351), NEP ḍhungo (371), SNG gala (–6), MAL hila (351), KOT ḍhol (451), HIM pāthar (81), KUL pāthr (81), MND pāthar (81), ORY pathara (81), AWD pathrā (81), KUM ḍhuŋo (371), ROM barr (167), KNK patthoru (81), DUM boṭ (167), BRJ patthar (81), GRH ḍhũgu (371), PRY patar (81), MAI pāhan (677), KCH pathrī (81), MEW pāthar (81), WGD patthar (81), BNJ bhāṭā (167) 81a. STONE: HNK gaṭṭā (483), GJR pathar (81), DGR patthar (81), RAJ pātharau (81), ASS pathr (81), KUM pāthar (81), MEW bhaṭṭo (167), BNJ pattar (81) 82. SUN: HND sūraj (82), DKH sūrīj (82), PNJ sūraj (82), PTH dīh (168), HNK d (168), GJR d (168), DGR sūraj (82), LHD sijjh (82), SND siju (82), RAJ sūraj (82), GUJ sūraj (82), MAR sūrya (–9), BNG šurj (–9), ASS beli (352), NEP ghām (236), SNG ira (82), MAL iru (82), KOT dhṛ (168), HIM sūraj (82), KUL surj (82), MND sūraj (82), ORY sūrya (–11), AWD suruj (82), KUM sūraj (82), ROM kham (236), KNK sūryu (–10), DUM to (628), BRJ sūraj (82), GRH ghām (236), PRY tup (646), MAI surūj (82), KCH sūraj (82), MEW sūraj (82), WGD dāṛobāwsi (168), BNJ dāḍo (168) 82a. SUN: ASS xuruz (82), MND dhaiṛā (168), GRH sūraj (82), BNJ surjyā (82) 253

Anton I. Kogan

83. SWIM: HND tair- (83), DKH tir- (83), PNJ tar- (83), PTH tar- (83), HNK tar- (83), GJR tar(83), DGR tar- (83), LHD tar- (83), SND tar- (83), RAJ tir- (83), GUJ tar- (83), MAR poh- (287), BNG šãtr de- (83), ASS xãtor- (83), NEP tar- (83), SNG pīna- (421), MAL fath- (545), KOT tr(83), HIM tar- (83), KUL tr- (83), MND tar- (83), ORY pahãr- (83), AWD pãvar- (83), KUM teir(83), ROM pliv- (–17), KNK paũ- (287), DUM ? (–31), BRJ ter- (83), GRH tar- (83), PRY ? (–18), MAI tair- (83), KCH tar- (83), MEW tar- (83), WGD põhow- (287), BNJ tar- (83) 83a. SWIM: MAR tar- (83) 84. TAIL: HND pch (84), DKH dum (–2), PNJ pucch (84), PTH pucchaṛ (84), HNK pūchaṛ (84), GJR pūchaṛ (84), DGR pucch (84), LHD pucchaṛ (84), SND puch (84), RAJ pchṛau (84), GUJ puchḍũ (84), MAR šẽpūṭ (288), BNG langul (317), ASS negur (317), NEP pucchar (84), SNG naguṭa (317), MAL nagō (317), KOT pundzhṛ (84), HIM pūnch (84), KUL phunjiṭ (84), MND pch (84), ORY languḷa (317), AWD pūchi (84), KUM punch (84), ROM pori (601), KNK bāl (–11), DUM čipoỵ (288), BRJ pc (84), GRH puchaṛu (84), PRY dum (–19), MAI pūch (84), KCH pucch (84), MEW põch (84), WGD pochṛī (84), BNJ puncḍi (84) 84a. TAIL: MAI lgaṛi (317) 85. THAT: HND vah (85), DKH (v)o (85), PNJ o (85), PTH oh (85), HNK o (85), GJR vo (85), DGR oh (85), LHD o (85), SND hū (85), RAJ (v)o (85), GUJ te (203), MAR to (203), BNG o (85), ASS xi (353), NEP u (85), SNG ō(ka) (85), MAL e (289), KOT s (353), HIM se (353), KUL s (353), MND sē (353), ORY se (353), AWD u (85), KUM u (85), ROM (k)odo (85), KNK theṇ (203), DUM hei (289), BRJ bū (85), GRH vu (85), PRY u (85), MAI ū (85), KCH hū (85), MEW wo (85), WGD o (85), BNJ u (85) 85a. THAT: DKH ti- (203), SND ta (203), RAJ tikau (203), MAR jo (289), BNG ta- (203), ASS teõ (203), NEP tyo (203), SNG ē(ka) (289), HIM vo (85), KUL te- (203), MND te- (203), ORY tāhā (203), MAI soi (353), KCH ta (203) 86. THIS: HND yah (86), DKH (y)e (86), PNJ e (86), PTH eh (86), HNK e (86), GJR yo (86), DGR eh (86), LHD e (86), SND hī (86), RAJ yau (86), GUJ ā (265), MAR hā (265), BNG e (86), ASS i (86), NEP yo (86), SNG mē(ka) (86), MAL mi (86), KOT j (86), HIM yeh (86), KUL y (86), MND e (86), ORY ehā (86), AWD yai (86), KUM yo (86), ROM kado (602), KNK he(ṇ) (86), DUM tahei (86), BRJ ī (86), GRH yū (86), PRY ya (86), MAI ī (86), KCH hī (86), MEW i (86), WGD to (665), BNJ i (86) 86a. THIS: ROM ­le- (86) 87. THOU: HND tū (87), DKH t (87), PNJ t (87), PTH t (87), HNK t (87), GJR t (87), DGR t (87), LHD t (87), SND t (87), RAJ t (87), GUJ tũ (87), MAR t (87), BNG tumi (87), ASS ti (87), 87 NEP tã (87), SNG tō (87), MAL thiya (87), KOT tū (87), HIM tū (87), KUL tū (87), MND t(87), ORY tume (87), AWD t (87), KUM tẽ (87), ROM tu (87), KNK tu (87), DUM tu (87), BRJ tu (87), GRH tū (87), PRY tu (87), MAI t (87), KCH t (87), MEW tu (87), WGD tu (87), BNJ tũ (87) 88. TONGUE: HND jībh (88), DKH jīb(h) (88), PNJ jībh (88), PTH jīv (88), HNK jībh (88), GJR jībh (88), DGR jibbh (88), LHD jibbh (88), SND j’ibh (88), RAJ jībh (88), GUJ jībh (88), MAR jībh (88), BNG jib (88), ASS zibha (88), NEP jibro (88), SNG diva (88), MAL dō (88), KOT dzibbh (88), HIM jīb (88), KUL jībh (88), MND jībh (88), ORY jibha (88), AWD jībhi (88), KUM jiboṛo (88), ROM čhib (88), KNK jīb(h) (88), DUM jibā (88), BRJ jībh (88), GRH jīb (88), PRY jip (88), MAI jībh (88), KCH jibbh (88), MEW jībh (88), WGD jab (88), BNJ jībh (88) 254

Genealogical classification of New Indo-Aryan languages and lexicostatistics

89. TOOTH: HND dt (89), DKH dāt (89), PNJ dand (89), PTH dand (89), HNK dand (89), GJR dand (89), DGR dand (89), LHD ḍand (89), SND ḍ’andu (89), RAJ dt (89), GUJ dt (89), MAR dt (89), BNG dãt (89), ASS dãt (89), NEP dt (89), SNG data (89), MAL daiy (89), KOT dānd (89), HIM dānd (89), KUL dnd (89), MND dānd (89), ORY dānta (89), AWD dt (89), KUM dt (89), ROM dand (89), KNK dt(u) (89), DUM don (89), BRJ dto (89), GRH dt (89), PRY dan (89), MAI dt (89), KCH ḍandh (89), MEW dt (89), WGD dt (89), BNJ dānt (89) 90.TREE: HND peṛ (90), DKH rūk (131), PNJ rukkh (131), PTH būṭā (373), HNK būṭā (373), GJR būṭo (373), DGR būhṭā (373), LHD vaṇ (169), SND vaṇu (169), RAJ rkh (131), GUJ jhāḍ (266), MAR jhāḍ (266), BNG gach (318), ASS gs (318), NEP rukh (131), SNG gaha (318), MAL gahā (318), KOT pēṛ (90), HIM ḍāl (373), KUL buṭṭa (373), MND ḍāl (373), ORY gaccha (318), AWD rūkh (131), KUM rukh (131), ROM kašt (603), KNK rūku (131), DUM tom (–32), BRJ peṛ (90), GRH ḍāḷ (373), PRY dara t (–20), MAI gāch (318), KCH vaṇ (169), MEW rukh (131), WGD peḍ (90), BNJ jhāḍ (266) 90a. TREE: DGR rukkh (131), NEP boṭ (373), SNG ruka (131), KOT buṭṭi (373), KUM boṭ (373), KNK jhaḍ (266), KCH jhāṛ (266) 91. TWO: HND do (91), DKH do (91), PNJ do (91), PTH do (91), HNK do (91), GJR do (91), DGR do (91), LHD ḍ (91), SND b’a (91), RAJ be (91), GUJ be (91), MAR don (91), BNG dui (91), ASS dui (91), NEP duī (91), SNG deka (91), MAL dey (91), KOT dui (91), HIM do (91), KUL duy (91), MND do (91), ORY dui (91), AWD dui (91), KUM dwi (91), ROM duy (91), KNK doni (91), DUM dui (91), BRJ dui (91), GRH dvī (91), PRY do (91), MAI dui (91), KCH b_a (91), MEW do (91), WGD be (91), BNJ dī (91) 92. WALK (GO): HND jā- (92), DKH jā(v)- (92), PNJ jā- (92), PTH gas- (93), HNK jul- (484), GJR ga- (93), DGR jā- (92), LHD vanj- (204), SND vañ- (204), RAJ jā- (92), GUJ jā- (92), MAR jā- (92), BNG ja- (92), ASS za- (92), NEP jā- (92), SNG ya- (92), MAL dā- (92), KOT ṛu- (459), HIM jā(92), KUL jā- (92), MND jā- (92), ORY jī- (92), AWD jā- (92), KUM jā- (92), ROM ža- (92), KNK vas- (204), DUM jā- (92), BRJ jā- (92), GRH jā- (92), PRY ja- (92), MAI jā- (92), KCH vin- (204), MEW jā- (92), WGD jā- (92), BNJ jā- (92) 92a. WALK (GO): HND ga- (93), DKH ga- (93), PNJ ge- (93), HNK ga- (93), GJR jā- (92), DGR ge- (93), RAJ ga- (93), MAR gel- (93), BNG gel- (93), ASS gal- (93), NEP ga- (93), SNG gi- (93), MAL g- (93), HIM ga- (93), KUL g- (93), MND ge- (93), ORY gal- (93), KUM ga- (93), ROM gel- (93), DUM gi- (93), BRJ ga- (93), GRH ga- (93), PRY gi- (93), MAI ge- (93), MEW ga- (93), WGD ga- (93), BNJ g- (93) 93. WARM: HND garm (–6), DKH tattā (170), PNJ kosā (133), PTH tattā (170), HNK tattā (170), GJR tato (170), DGR tattā (170), LHD tattā (170), SND koso (133), RAJ ūnau (133), GUJ ūnũ (133), MAR ūn (133), BNG grm (–10), ASS umi (133), NEP tāto (170), SNG uṇu (133), MAL hōnu (133), KOT ntt (170), HIM tattā (170), KUL ttta (170), MND tātā (170), ORY ušuma (– 12), AWD garam (–7), KUM tāto (170), ROM tato (170), KNK hūn (133), DUM tatā (170), BRJ chiṭṭaiyo (632), GRH tātu (170), PRY garm (–21), MAI tāt (170), KCH koso (133), MEW tātā (170), WGD ōno (133), BNJ garam (–3) 93a. WARM: HNK kosā (133), GJR koso (133), SND tato (170), RAJ tātau (170) 94. WATER: HND pānī (94), DKH pānī (94), PNJ pāṇī (94), PTH pāṇī (94), HNK pāṇī (94), GJR pāṇī (94), DGR pānī (94), LHD pāṇī (94), SND pāṇī (94), RAJ pāṇī (94), GUJ pāṇī (94), MAR pāṇī (94), BNG jl (319), ASS pani (94), NEP pānī (94), SNG pän (94), MAL fen (94), KOT pāṇī 255

Anton I. Kogan

(94), HIM pāṇī (94), KUL pāṇi (94), MND pāṇī (94), ORY pāṇi (94), AWD pānī (94), KUM pāṇī (94), ROM pay (94), KNK udak (–12), DUM pāni (94), BRJ pānī (94), GRH pāṇi (94), PRY paṇi (94), MAI pāni (94), KCH pāṇī (94), MEW pānī (94), WGD pāṇī (94), BNJ pāṇi (94) 94a. WATER: ASS zl (319), KCH jar (319) 95. WE: HND ham (95), DKH hame (95), PNJ as (95), PTH ass (95), HNK as (95), GJR ham (95), DGR as (95), LHD assã (95), SND as (95), RAJ mhe (95), GUJ ame (95), MAR āmhī (95), BNG amra (95), ASS ami (95), NEP hāmī (95), SNG api (95), MAL aharumen (95), KOT hamme (95), HIM ase (95), KUL ass (95), MND asẽ (95), ORY āme (95), AWD ham (95), KUM ham (95), ROM ame(n) (95), KNK āmmī (95), DUM ame (95), BRJ hum (95), GRH ham (95), PRY ham (95), MAI ham sab (95), KCH as (95), MEW ham (95), WGD hamu (95), BNJ ham (95) 96. WHAT: HND kyā (96), DKH kyā (96), PNJ kī (96), PTH keh (96), HNK ke (96), GJR ke (96), DGR keh (96), LHD kyā (96), SND kahiṛo (96), RAJ kãī (96), GUJ šũ (96), MAR kāy (96), BNG ki (96), ASS kih (96), NEP ke (96), SNG mokada (96), MAL kēkey (96), KOT k (96), HIM kyā (96), KUL kī (96), MND kyā (96), ORY kana (96), AWD kā (96), KUM kī (96), ROM so (96), KNK kasane (96), DUM kisek (96), BRJ kae (96), GRH kyā (96), PRY ka (96), MAI kī (96), KCH kuro (96), MEW kā (96), WGD kae (96), BNJ ki (96) 97. WHITE: HND ciṭṭā (97), DKH ujlā (98), PNJ ciṭṭā (97), PTH ciṭṭā (97), HNK ciṭṭā (97), GJR ciṭo (97), DGR ciṭṭā (97), LHD ciṭṭā (97), SND accho (205), RAJ dhoḷau (267), GUJ dholũ (267), MAR pḍhrā (290), BNG dhbl (267), ASS bga (134), NEP seto (375), SNG sudu (428), MAL hudu (428), KOT šukl (452), HIM ciṭṭā (97), KUL šētta (375), MND ciṭṭā (97), ORY dhalā (267), AWD ujar (98), KUM seto (375), ROM parno (–18), KNK dhave (267), DUM šukulā (452), BRJ dhulo (267), GRH saphed (–7), PRY bago (134), MAI ujar (98), KCH acho (205), MEW dhawaṛo (267), WGD doḷo (267), BNJ dhoḷo (267) 97a. WHITE: HND ujlā (98), DKH pḍū (290), PNJ baggā (134), HNK gorā (485), SND bago (134), ASS dhwl (267), SNG äli (375), KUM sukilo (452), MAI dhaval (267), KCH dhauro (267) 98. WHO: HND kaun (99), DKH kon (99), PNJ kauṇ (99), PTH kuṇ (99), HNK koṇ (99), GJR koṇ (99), DGR kun (99), LHD koṇ (99), SND keru (99), RAJ kuṇ (99), GUJ koṇ (99), MAR koṇ (99), BNG ke (99), ASS kon (99), NEP ko (99), SNG kavuda (99), MAL kāku (99), KOT kuṇ (99), HIM kuṇ (99), KUL kūṇ (99), MND kūṇ (99), ORY kie (99), AWD ko (99), KUM ko (99), ROM kon (99), KNK koṇ (99), DUM koṇo (99), BRJ kōn (99), GRH ko (99), PRY koṇ (99), MAI kon (99), KCH ker (99), MEW koṇ (99), WGD kun (99), BNJ kuṇ (99) 99. WOMAN: HND aurat (–7), DKH tiyā (135), PNJ tīvī (135), PTH jaṇāṇī (–4), HNK trīmat (135), GJR janī (491), DGR janānī (–3), LHD tremit (135), SND nāri (206), RAJ lugāī (472), GUJ baiḍī (268), MAR bāyko (268), BNG meye (321), ASS tirota (135), NEP svāsnīmanche (244), SNG gäṇī (429), MAL anhenmēhā (244), KOT chēuṛi (453), HIM tīmī (135), KUL beṭṭḍi (557), MND janāṇ (–5), ORY bāilī (268), AWD joy (471), KUM seṇmeis (244), ROM žuvli (471), KNK bāil- (268), DUM joi (471), BRJ lugaī (472), GRH janānī (–8), PRY ran (136), MAI tīa (135), KCH bāyṛī (–8), MEW berbānī (268), WGD buirī (268), BNJ bāi (268) 99a. WOMAN: DKH joī (471), PNJ rann (136), ORY māī (321), AWD tiy (135), ROM manušni (244), KNK manišī (244) 100. YELLOW: HND pīlā (100), DKH pīlā (100), PNJ pīlā (100), PTH pīlā (100), HNK pīlā (100), GJR pīlo (100), DGR pīlā (100), LHD pīlā (100), SND pīlo (100), RAJ pīḷau (100), GUJ pilũ (100), 256

Genealogical classification of New Indo-Aryan languages and lexicostatistics

MAR pĩvḷā (100), BNG hlde (322), ASS haldhiya (322), NEP pahẽlo (100), SNG kaha (431), MAL rēndōkula (546), KOT piũḷ (100), HIM pīlā (100), KUL piwḷa (100), MND piuḷā (100), ORY haḷadiā (322), AWD piyar (100), KUM pīlo (100), ROM galbeno (–19), KNK haḷduve (322), DUM haliẓ̌žā (–33), BRJ pīlo (100), GRH pingḷu (659), PRY zard (–22), MAI pīyar (100), KCH pīro (100), MEW piyaro (100), WGD halojjo (322), BNJ piḷo (100)

Abbreviations for languages and dialects ASS — Assamese AWD — Awadhi BNG — Bengali BNJ — Banjari BRJ — Braj DGR — Dogri DKH — Dakhini DUM — Dumaki (Domaaki) GJR — Gojri GRH — Garhwali GUJ — Gujarati HIM — Himachali HND — Hindi-Urdu

HNK — Hindko KCH — Kutchi KNK — Konkani KOT — Kotgarhi KUL — Kului KUM — Kumauni LHD — Lahnda (Multani) MAI — Maithili MAL — Maldivian (Dhivehi) MAR — Marathi MEW — Mewati MIA — Middle Indo-Aryan MND — Mandeali

NEP — Nepali NIA — New Indo-Aryan OIA — Old Indo-Aryan ORY — Oriya PNJ — Punjabi PTH — Pothohari PRY — Parya RAJ — Rajasthani (Marwari) ROM — Romany SND — Sindhi SNG — Sin(g)halese WGD — Wagdi

References Abdulla, F., M. O’Shea. 2005. English Dhivehi, Dhivehi-English dictionary. Version 1.0. Belconnen. Awan, Anjum Rafiq 2000. Concise Gojri-English dictionary. Rajouri: Anjuman Taraqi Gojri Adab. Berger, H. 1959. Die Burušaski-Lehnwörter in der Zigeunersprache. Indo-Iranian Journal 3: 17–43. Bloch, J. 1920. La formation de la langue marathe. Paris: É. Champion. Bloch, J. 1965. Indo-Aryan: from the Vedas to modern times. Paris: Libr. d’Amérique et d’Orient Adrien-Maisonneuve. Boretzky, Norbert. 1994. Romani. Grammatik des Kalderaš-Dialekts mit Texten und Glossar. Wiesbaden: Harrassowitz Verlag. Buddruss, G. 1984. Domaakí-Nachträge zum Atlas der Dardsprachen. Münchener Studien zur Sprachwissenschaft 43: 9–24. Chatterji, S. K. 1926. The origin and development of the Bengali language. Part 1: Introduction. Phonology. Calcutta: Calcutta University Press. Burrow T., M. B. Emeneau. 1961. A Dravidian etymological dictionary. Oxford: Clarendon Press. Gosvāmī, Om. 2000. Ḍogrī-hindī śabdkoś. Jammū: J&K Academy of Art, Culture and Languages. Grireson, G. A. 1916. Linguistic survey of India. Vol. 9. Part 4: Specimens of the Pahari languages and Gujuri. Calcutta: Superintendent Government Printing. Grierson, G. A. 1927. Linguistic survey of India. Vol. 1. Part 1. Introductory. Calcutta: Government of India, Central Publication Branch. Hendriksen, H. 1976. Himachali studies. Vol 1. Vocabulary. København. Hoernle, A. F. Rudolf. 1880. A comparative grammar of the Gaudian languages, with special reference to the Eastern Hindi, accompanied by a language-map and a table of alphabets. London: Trübner & Co. Kogan, A. I. 2005. Dardskie yazyki. Geneticheskaya kharakteristika. Moskva: Vostochnaya literatura. Lorimer, D. L. R. 1939. The Dumaki language: outlines of the speech of the Doma, or Bericho, of Hunza. Nijmegen: Dekker & Van De Vegt. Mahapatra, B.P., B. Padmanabha, M.R. Ranganatha. 1980. Census of India 1971. Ser. 1. Language monograph. Monogr. No 7. Survey of Mandeali and Kului in Himachal Pradesh. New Delhi. 257

Anton I. Kogan

Masica, Colin P. 1991. The Indo-Aryan languages. Cambridge University Press. Mukherji, Kakali et al. 2011. Linguistic survey of India. Rajasthan. Part 1. Language Division, Office of the Registrar General & Census Commissioner, India. s.l. Neog, Maheswar, Upedranath Goswami. 1987. Chandrakanta abhidhan: a comprehensive dictionary of the Assamese language with etymology and illustrations of words with their meanings in Assamese and English. Guwahati: Gauhati University. Oranskiy, I. M. 1977. Fol’klor i yazyk gissarskikh par’ya (Srednyaya Aziya). Vvedenie, teksty, slovar’. Moskva: Nauka, Glavnaya redakciya vostochnoy literatury. Praharaj, G. C. 1931–1940. Purnnacandra odia bhashakosha. Cuttack: Utkal Sahitya Press. Ramesh, G. V. 2010. Banjara multilingual dictionary with semantic category section and indexes for Telugu, English and Hindi. Hyderabad: Rajiv Vidya Mission (Education Department of Andhra Pradesh State Government). Rohra, S. K. 1965. Descriptive analysis of Kachi (Kacchi). Thesis submitted to the University of Poona for the degree of Doctor of Pholosophy in Linguistics. Poona: Deccan College, Post-graduate and Research Institute. Saleem, Teyyeb, Wazir Shah. 2005. Conversational Saraiki, Urdu, English. Karachi: Institute of Applied Linguistics. Samīr, Rāmājñā Dvivedī. 1955. Avadhī koś. Ilāhābād: Hindustānī Ekeḍemī. Schmidt, Ruth Laila. 1994. A practical dictionary of modern Nepali. Ratna Sagar. Sibghatulla, Mohd., Syed Abu Turaab Khataai Zamin. 2000. Dakhni lughaat (Classical Urdu dictionary). With Dakhni idioms and phrases. Bangalore: Malik Publications. Sukūn, Sultān. 2002. Hindko Urdū luγat. Pišāvar: Gandhārā Hindko Borḍ Pākistān. Suthar, B. L., S. S. Gahlot. 1995. Rajasthani-Hindi-English dictionary. Jodhpur: Rajasthani Sahitya Sansthan. Thakur, Gajendra, Nagendra Kumar Jha, Panjikar Vidyanand Jha. 2012. Videha English-Maithili dictionary. Delhi: Shruti Publication. Thali, Mukesa. 1999–2001. Rajhauns Konkani-English illustrated dictionary. Panaji: Rajahasa. Turner, R. L. 1921a. Gujarati phonology. Journal of the Royal Asiatic Society of Great Britain and Ireland 3: 329–365. Turner, R. L. 1921b. Gujarati phonology. Journal of the Royal Asiatic Society of Great Britain and Ireland 4: 505–544. Turner, R. L. 1924. The Sindhi recursives or voiced stops preceded by glottal closure. Bulletin of the School of Oriental Studies, University of London 3(2): 301–315. Turner, R. L. 1926. The position of Romany in Indo-Aryan. Journal of the Gypsy Lore Society, 3rd Ser. 5(4): 145–188. Turner, R. L. 1931. A comparative and etymological dictionary of the Nepali language. London: K. Paul, Trench, Trübner. Turner, R. L. 1966. A comparative dictionary of Indo-Aryan languages. London: Oxford University Press.

А. И. Коган. Генеалогическая классификация индо-арийских языков и лексикостатистика. Генетические отношения индоарийских языков до сих пор остаются неясными. Существующие классификации зачастую носят интуитивный характер и не основываются на надежных сравнительно-исторических критериях. В статье делается попытка классификации новоиндийских языков по обновленным лексикостатистическим данным. Сопоставление полученного в итоге родословного древа с традиционными классификационными схемами позволило автору сделать ряд выводов относительно наиболее вероятной таксономии языков индоарийской группы. Ключевые слова: индоарийские языки, генеалогическая классификация языков, лексикостатистика, глоттохронология.

258

М. Е. Васильев †, М. Н. Саенко ‡ † Школа

актуальных гуманитарных исследований РАНХиГС (Россия, Москва), [email protected];

‡ Институт

славяноведения РАН (Россия, Москва), [email protected]

К вопросу о точности глоттохронологии: датирование процесса лексических замен по данным романских языков Статья представляет собой первую часть исследования, посвященного проблеме достоверности лингвистических датировок, получаемых с помощью метода глоттохронологии. В предлагаемой работе рассматривается процесс лексических замен, происходящих в базисной лексике одного языка с течением времени. В качестве исходных данных нами использовались 110­словные списки, собранные на материале 54 современных и нескольких исторических романских идиомов. При этом для измерения скорости замен списки современных языков сравнивались со списками классической и архаический латыни, а также старофранцузского и староитальянского. Временна́я дистанция между сопоставляемыми идиомами определялась с помощью трех различных глоттохронологических методов: классического уравнения М. Сводеша, модифицированной формулы С. А. Старостина, а также недавно предложенной потоковой модели. Сравнение полученных результатов позволило сделать ряд важных выводов о характере лексических изменений, адекватности существующих глоттохронологических моделей, а также численно оценить точность и надежность глоттохронологических расчетов при датировании общего процесса замен. Вторую часть исследования планируется посвятить проблеме датирования относительной дивергенции двух родственных языков. Ключевые слова: глоттохронология, лексикостатистика, список Сводеша, романские языки.

Одной из основных аксиом лексикостатистики является равномерность процесса замен базисной лексики, описанная М. В. Араповым и М. М. Херц следующим образом: Доля р слов из О[сновного] С[писка], которые сохранятся (не будут заменены другими словами) на протяжении интервала времени Δt (равного, например, году, столетию или тысячелетию) постоянна (т. е. зависит только от величины выбранного промежутка, но не от того, как он выбран, или слова какого языка рассматриваются). (Арапов, Херц 1974: 22)

Математическим соответствием этой аксиомы является коэффициент сохраняемости, используемый в глоттохронологических формулах вычисления времени. Впервые коэффициент сохраняемости был вычислен М. Сводешем путем сравнения двухсотсловных списков ряда языков с долгой письменной историей в ранней форме их существования и на более позднем этапе развития: табл. 1. Как можно заметить, процент сохранившейся лексики при данной методике подсчета колеблется в интервале 76—85 % слов за тысячелетие (Сводеш 1960а: 47). В дальнейшем, используя модифицированный список базисной лексики, состоящий из 100 слов, Сводеш пересчитывает на его основании коэффициент сохраняемости, предварительно исключив из рассмотрения коптский (так как он может быть непрямым Journal of Language Relationship • Вопросы языкового родства • 14/4 (2016) • Pp. 259—277 • © The authors, 2016

М. Е. Васильев, М. Н. Саенко

Таблица 1. Значения коэффициента сохраняемости, рассчитанные Сводешем на основании двухсотсловных списков (Сводеш 1960а: 34). Ранняя форма

Поздняя форма

Интервал времени

Процент сохранившихся слов на 1000 лет

Среднеегипетский

Коптский

2300 лет

76

Классическая латынь

Румынский

2000 лет

77

Древневерхненемецкий

Немецкий

1100 лет

78

Классический китайский

Северокитайский

1000 лет

79

Латынь Плавта

Французский Мольера

1850 лет

79

Доминика кариб 1650 н.э.

Современная форма

300 лет

80

Классическая латынь

Португальский

2000 лет

82

Койне

Кипрский диалект

[без даты]

83

Койне

Афинский диалект

[без даты]

84

Классическая латынь

Итальянский

2000 лет

85

Древнеанглийский

Английский

1000 лет

85

Латынь Плавта

Ранний новоиспанский 1600 г.

1800 лет

85

Таблица 2. Значения коэффициента сохраняемости, рассчитанные Сводешем на основании стословных списков (Сводеш 1960б: 72). Интервал времени

Процент сохранившихся слов на 1000 лет

Шведский

1020

94,3

Немецкий

1100

89,0

Язык

Английский

1000

76,6

Румынский

2150

76,4

Французский

1850

77,6

Афинский

2070

83,6

Китайский

1000

79,6

потомком среднеегипетского), испанский, итальянский, португальский, каталанский (поскольку уже есть подсчет по родственному ему французскому), кипрский (поскольку есть подсчет по афинскому): табл. 2. Однако уже в 1958 г. появилась статья Дж. Ри, в которой сравниваются стословные списки восьми романских языков и при использовании коэффициента сохраняемости r = 0,85 (максимальный среди вычисленных Сводешем) были получены нелепые даты расхождения, например, испанского и португальского — 370 лет назад, румынского и итальянского — 826 лет назад. Поскольку история романских языков довольно хорошо известна, очевидно, что эти результаты не отвечают действительности. Исходя из этого, Дж. Ри делает вывод о некорректности всей методики глоттохронологии. В ответной статье А. Крубер возражает, что в данном случае нужно не сходу отказываться от глоттохронологии вообще, а попытаться ее доработать. В частности, для романских языков 260

К вопросу о точности глоттохронологии

Таблица 3. Расчеты К. Бергсланда и Х. Фогта (Bergsland, Vogt 1962: 117—125). Язык-1

Язык-2

Временно́е расстояние (100­словник)

Временно́е расстояние (200­ словник)

Временно́е расстояние (215­ словник)

Фактическая временна́я дистанция

Древнескандинавский

Исландский

63

130

194

1000

Древнескандинавский

диалект Гьестал

345

799

901

1000

Древнескандинавский

диалект Санднес

412

861

964

1000

Древнескандинавский

Риксмол

637

930

1000

1000

Древнегрузинский

Грузинский

338

750

861

1500

Грузинский

Мегрельский

1316

1004

1033

1800—1900

Древнеармянский

Армянский

211

437

437

1500

Таблица 4. Значения коэффициента сохраняемости, полученные Старостиным на материале стословных списков (Starostin 2000: 230). Язык

Интервал времени

λ2 (без учета λ1 (с учетом заимствований) заимствований)

Японский

1200 лет

0,11

0,06

Китайский

2600 лет

0,1

0,1

Английский

1300 лет

0,14

0,1

Немецкий

1200 лет

0,08

0,05

Французский

1500 лет

0,09

0,07

Испанский

1500 лет

0,07

0,06

Румынский

1500 лет

0,09

0,06

при введении поправочного коэффициента 2,2 Крубером были получены вполне приемлемые даты (Kroeber 1958). Тем не менее, последующие подсчеты, произведенные по методике Сводеша К. Бергсландом и Х. Фогтом (Bergsland, Vogt 1962) на материале риксмола, исландского, армянского и грузинского языков, снова дали результат, существенно расходящийся с известными значениями: табл. 3. Как видно из табл. 3, кроме единственного исключения, расчетные значения временной дистанции ни в одном случае не совпадают с фактическими. Особенно разительным оказалось расхождение при сравнении исландского и древнескандинавского языков. В середине 80­х годов прошлого века усовершенствованием метода глоттохронологии занялся С. А. Старостин, который улучшил формулу дивергенции языков и выдвинул требование исключать заимствования из списков сопоставляемой лексики, поскольку заимствования являются следствием внешних контактов, а не внутреннего изменения языка: табл. 4. 261

М. Е. Васильев, М. Н. Саенко

Рисунок 1. Географическое распространение идиомов, используемых в исследовании. Номерами на карте обозначены: 1. мегленорумынский; 2. истрорумынский; 3. арумынский; 4. румынский (литературный); 5. далматинский; 6. фриульский (центральный); 7. ладинский (гарденский); 8. ладинский (фассанский); 9. руманшский (сурсельвский); 10. руманшский (сурмиранский); 11. руманшский (нижнеэнгадинский); 12. пьемонтский (Ланцо-Торинезе); 13. пьемонтский (Барбания); 14. пьемонтский (Карманьола); 15. пьемонтский (Турин); 16. пьемонтский (верчельский); 17. ломбардский (Бергамо); 18. ломбардский (Плезио); 19. эмилиано-романьольский (Равенна); 20. эмилиано-романьольский (Феррера); 21. эмилиано-романьольский (Карпи); 22. эмилиано-романьольский (Реджо); 23. лигурийский (Рапалло); 24. лигурийский (Генуя); 25. лигурийский (Стелла); 26. венетский (Венеция); 27. венетский (Примьеро); 28. венетский (Беллуно); 29. тосканский (Гроссето); 30. умбрийский (Фолиньо); 31. неаполитанский; 32. логудорский; 33. кампиданский; 34. сицилийский (Палермо); 35. сицилийский (Мессина); 36. сицилийский (Катания); 37. сицилийский (юго-восточный; объединены данные, полученные от информантов из Рагузы и Агридженто); 38. каталанский (центральный); 39. каталанский (северо-западный); 40. каталанский (Менорка); 41. каталанский (Кастельон-де-ла-Плана); 42. каталанский (Валенсия); 43. каталанский (Манисес); 44. кастильский (Сория); 45. астурийский (центральный); 46. португальский (литературный); 47. галисийский (центральный); 48. окситанский (провансальский); 49. франко-провансальский (савойский); 50. французский (литературный); 51. пикардский (южный); 52. валлонский (южный).

За последние годы в рамках проекта «Глобальная лексикостатистическая база данных» было накоплено большое количество списков базисной лексики, качество которых стоит на более высоком уровне, чем у материала, доступного Сводешу и Старостину. В частности, в 2015—2016 гг. одним из авторов данной статьи (М. Н. Саенко) были собраны аннотированные 110­словные списки базисной лексики для 54 1 романских идиомов (а также 4 списка для староитальянского Данте, старофранцузского Кретьена де Труа, ла1

В дальнейшем будут использоваться 52 идиома, так как списки для руманч грижун и итальянского литературного были исключены в силу искусственного характера первого и чрезвычайной архаичности второго. Полные списки с источниками, таблицами транслитерации и описанием доступны на сайте «Глобальная лексикостатистическая база данных» http://starling.rinet.ru/cgi-bin/main.cgi?root=new100. 262

К вопросу о точности глоттохронологии

тыни Плавта и Апулея). При этом предпочтение отдавалось диалектам и «малым» языкам без строгой литературной нормы, из 54 идиомов лишь 5 являются строго нормированными (румынский, руманч грижун, итальянский, португальский, французский). Географически исследование было ограничено только Старым Светом (см. рис. 1). Данные были получены как из диалектных словарей романских языков, так и от информантов. В связи с необходимостью за короткий срок охватить большое количество материала, работа с носителями велась через Интернет (рассылался опросник, а после его заполнения информантам задавались уточняющие вопросы). Всего было опрошено 76 информантов. Используя материал романских языков в качестве тестового, мы последовательно применяем к нему два наиболее известных глоттохронологических метода: классическую глоттохронологию М. Сводеша, усовершенствованную методику С. А. Старостина, а также недавно предложенную потоковую модель. При этом определим основные цели и задачи нашего исследования следующим образом: 1) Проверить применимость существующих глоттохронологических моделей (М. Сводеша, С. А. Старостина, потоковой модели) для датирования изменений в базисной лексике романских языков; 2) Определить оптимальные параметры моделей, обеспечивающие наилучшее соответствие расчетных значений и исходных лексикостатистических данных. При необходимости произвести калибровку моделей с учетом новых параметров и сравнить результаты, полученные с использованием калиброванных и некалиброванных моделей; 3) Установить объективные и теоретические пределы точности при вычислении лингвистических датировок с применением рассматриваемых моделей. В первой части данной статьи мы рассмотрим процесс замен в лексике одного языка по мере его развития и подробно остановимся на методике вычисления временной дистанции между языком-предком и языком-потомком. Вторая часть работы будет посвящена относительной дивергенции двух родственных языков и проблемам ее датирования.

1. Глоттохронологические модели общего процесса лексических замен Исходные данные для определения скорости изменений в базисной лексике романских языков были получены путем сравнения современных идиомов с ближайшими родственниками их непосредственных предков: архаической латынью Плавта, поздней классической латынью Апулея, староитальянским Данте и старофранцузским Кретьена де Труа, датировки которых можно установить по историческим источникам. По результатам сравнения составлена табл. 5, где для каждой пары (или нескольких пар)2 языков указан процент совпадений3 между соответствующими основными списками, а также временной интервал между датами их фиксации. 2

При сопоставлении нескольких пар языков (строки 3, 4, 10, 12), в соответствующих столбцах приводится минимальное, максимальное и среднее арифметическое значение. Последнее рассчитывается как сумма всех известных процентов совпадений, деленная на количество слагаемых, что в общем случае не совпадает со средним значением между максимальным и минимальным процентом совпадений. Например, в строке 4 для трех пар языков с долями совпадений 86%, 83% и 87% среднее между минимальным и максимальным значением составит (83+87)/2=85, а среднее арифметическое — (83+86+87)/3=85,3. 3 Значения процентов совпадений приводятся по данным табл. 8, полученной путем попарного сравнения всех собранных списков в программе Starling (см. Дополнительные материалы). 263

М. Е. Васильев, М. Н. Саенко

Таблица 5. Лексикостатистические данные о скорости замен в базисной лексике романских языков. №

Сравниваемые языки

Мин. % совп.

Средн. % совп.

Макс. % совп.

Интервал времени, лет

1

Исходное значение (для любого идиома)

—

100

—

0

2

Архаическая латынь (Плавт, 250 г. до н.э.) — поздняя классическая латынь (Апулей, 150 г. н.э.)

—

98

—

400

3

Староитальянский (Данте, 1270 г.) — современные итальянские (тосканский, умбрийский)

85

86,5

88

730

4

Старофранцузский (Кретьен де Труа, 1140 г.) — современные французские (литературный французский, пикардский, валлонский)

83

85,3

87

860

5

Поздняя классическая латынь (Апулей, 150 г.) — старофранцузский (1140 г.)

—

84

—

990

6

Поздняя классическая латынь (150 г.) — староитальянский (1270 г.)

—

85

—

1120

7

Архаическая латынь (250 г. до н.э.) — старофранцузский (1140 г.)

—

78

—

1390

8

Архаическая латынь (250 г. до н.э.) — староитальянский (1270 г.)

—

78

—

1520

9

Поздняя классическая латынь (150 г.) — далматинский (1900 г.)

—

81

—

1750

10

Поздняя классическая латынь (150 г.) — современные романские (52 идиома, 2000 г.)

69

74,6

83

1850

11

Архаическая латынь (250 г. до н.э.) — далматинский (1900 г.)

—

76

—

2150

12

Архаическая латынь (250 г. до н.э.) — современные романские (52 идиома, 2000 г.)

64

69,5

77

2250

Рисунок 2. Изменение доли сохранившейся лексики романских языков в зависимости от времени. Нумерация рядом с точками указывает на соответствующие строки табл. 5. Для точек 3, 4, 10 и 12 показан диапазон разброса долей совпадений и среднеарифметическое значение.

N(t), доля сохранившихся значений

1.1 1

1

2

0.9 0.8 0.7

3 4 6 5 7 8

9 11 10

0.6

12

0.5 0 0.5 1 1.5 2 2.5 3 t, временной интервал между датами фиксации сравниваемых списков, тыс. лет

264

К вопросу о точности глоттохронологии

Для наглядности полученные данные можно представить в виде диаграммы, которая отражает уменьшение доли сохранившейся лексики N(t) с течением времени (t): рис. 2. Приведенная диаграмма свидетельствует о ярко выраженном статистическом характере исходных данных, который проявляется в значительном разбросе процентов совпадений, полученных для точек с одинаковыми или хронологически близкими датировками. В то же время, очевидно, что все (даже наиболее выделяющиеся) значения хорошо сгруппированы вокруг некоторой средней величины на всем рассматриваемом интервале времени, что позволяет говорить о существовании зависимости случайного процесса лексических замен от времени. Для определения характера этой зависимости и ее параметров перейдем к рассмотрению конкретных глоттохронологических моделей: классическому уравнению М. Сводеша, усовершенствованной формуле С. А. Старостина и потоковой модели. 1.1. Глоттохронологическая модель М. Сводеша. Классический метод глоттохронологии, предложенный М. Сводешем в середине XX в., построен по аналогии с методом радиоуглеродного датирования и базируется на четырех основных допущениях (постулатах4): а) наличие в словаре каждого языка некоторого устойчивого подмножества слов — базисной лексики, из которой можно выделить универсальный список значений, обладающий повышенной стабильностью в любом языке5; б) постоянная скорость лексических изменений в основном списке, не зависящая от выбранного языка и временного периода; в) одинаковая стабильность всех элементов основного списка; г) независимость замен в списках языков-потомков после их разделения. В качестве математического аппарата, отражающего содержание постулатов глоттохронологии, была использована формула радиоактивного распада, описывающая процесс замен в базисной лексике в виде экспоненциальной зависимости с коэффициентом сохраняемости λ, определяющим темп замен:

N(t) = e − λ ⋅t . Как уже говорилось, согласно подсчетам Сводеша, которые проводились на разнообразном материале (в том числе романском), за 1000 лет различные языки в среднем сохраняют около 85% основного списка (Swadesh 1952: 456—460), что соответствует коэффициенту λ=0,16 (е–0,16*1=0,852). Подставив данное значение λ в исходное выражение, получаем формулу для датирования процесса замен в лексике одного языка:

N Sw (t) = e −0,16⋅t . Для калибровки формулы Сводеша по новым данным, полученным на основе романских языков, воспользуемся методом наименьших квадратов. Смысл данного метода сводится к поиску таких параметров модели (в данном случае — коэффициента λ), при 4

Эти и другие положения глоттохронологии Сводеша более подробно излагаются в работе Арапов, Херц 1974: 21—25. 5 Наибольшее распространение получил 100­словный список, зачастую называемый также «списком Сводеша». 265

М. Е. Васильев, М. Н. Саенко

Рисунок 3. Соответствие исходной и калиброванной модели М. Сводеша исходным данным: N Sw (t) = e −0,16⋅ t ; N SwC (t) = e −0,16⋅t .

1.1 1

NSw(t)=NSwC(t)

0.9 0.8 0.7

N(t)

0.6 0.5 0 0.5 t, тыс.лет

1

1.5

2

2.5

3

Рисунок 4. Количество пар языков (nλ) с коэффициентом λ, полученным для калиброванной модели Сводеша по исходным данным (табл. 5).

30

0.16

20

nλ

10

0 0.05 λSw

0.1

0.15

0.2

которых суммарная разница (или отклонение) между фактическими и расчетными долями совпадений, вычисленными для каждой из опорных точек, является минимальной. Например, в соответствии с табл. 5 (строка 2), процент совпадений между списками арахаической латыни Плавта и классической латыни Апулея составляет 98 (Nф=0,98) при разделяющем их временном интервале 400 лет (t=0,4). Подставляя это значение времени в формулу Сводеша, получим выражение: N р = e − λ ⋅0, 4 . Теперь, для того чтобы обеспечить наилучшее совпадение фактического (Nф) и расчетного (Nр) значений, необходимо найти такую величину λ, при которой квадрат разности между фактическим и расчет266

К вопросу о точности глоттохронологии

ным значениями (Nф – Nр)2 окажется наименьшим. Таким образом, задача поиска минимального отклонения ε для данной опорной точки будет иметь вид:

ε = (e − λ1 ⋅0, 4 − 0,98) 2 → min . Суммируя отклонения, полученные аналогичным способом для каждой опорной точки, получим общую формулу для расчета оптимального коэффициента λ: ε = ∑ (N р − N ф ) 2 → min, i

где Nр — расчетное значение доли общей лексики, вычисленное по формуле NSw(t) = e–λ⋅t, i — номер опорной точки, а Nф и t — фактические значения доли совпадений и времени, представленные в табл. 5. Вычисления, проведенные с помощью пакета Mathcad, показали, что минимальное отклонение между расчетными и фактическими значениями достигается при коэффициенте сохраняемости равном 0,16 (при этом величина суммарного отклонения составляет ε=0,094). Таким образом, калиброванное значение λ совпало с константой Сводеша, а полученная модель оказалась идентична исходной (см. рис. 3):

N SwC (t) = N Sw (t) = e −0,16⋅t . Как видно на приведенном графике (рис. 3), расчетные значения модели хорошо соответствуют опорным точкам на всем временном интервале (до 2,5 тыс. лет). При этом распределение значений λ, полученное для всех опорных точек по формуле Сводеша, оказалось близким к нормальному распределению6 с математическим ожиданием λSw =0,16 и средним квадратическим отклонением σλ≈0,02 (рис. 4). Воспользуемся теперь другой глоттохронологической моделью, которая была предложена С. А. Старостиным. 1.2. Усовершенствованная глоттохронология С. А. Старостина. Анализируя критику методики Сводеша, С. А. Старостин приходит к выводу о несостоятельности 2­го и 3­го постулатов глоттохронологии и указывает на необходимость их пересмотра. В частности, он приводит следующие аргументы (Starostin 2000: 229—230, 236—237): а) слова в базисной лексике языка со временем устаревают, и чем больше рассматриваемый промежуток времени, тем больше вероятность замены слова в основном списке, а следовательно — тем выше скорость распада. б) слова в основном списке неоднородны и обладают разной стабильностью, поэтому с течением времени общая скорость распада снижается из-за повторных замен наименее устойчивых значений и увеличивающейся доли более устойчивых. Чтобы учесть эти особенности в математическом аппарате глоттохронологии, С. А. Старостин предлагает ввести в формулу Сводеша две поправки, одна из которых отражает замедление процесса замен, связанное с проявлением в списке наиболее стабильной 6 После исключения из рассмотрения выделяющегося значения λ =0,051, (строка 2 табл. 5), найденный 1 при калибровке коэффициент λ=0,16 не изменился.

267

М. Е. Васильев, М. Н. Саенко

Рисунок 5. Сравнение исходной NSt(t) и калиброванной NStC(t) моделей Старостина: N St (t) = e −0, 05⋅ N St ⋅ t 2

N StC (t) = e −0,11⋅ N StC ⋅t .

1.1

2

;

NSt(t)

1 0.9 0.8 0.7

N(t)

0.6

NStC(t)

0.5 0 0.5 t, тыс.лет

1

1.5

2

2.5

3

Рисунок 6. Количество пар языков (nλ) с коэффициентом λ, полученным для калиброванной модели Старостина по исходным данным (табл. 5).

40

0.11

30

20

nλ

10

0 0 λSt

0.1

0.2

0.3

0.4

части лексики (λ=λ·N(t)), а вторая — наоборот, его ускорение, обусловленное устареванием сохранившихся значений (λ=λ·t). В результате новая глоттохронологическая модель принимает следующий вид: 2 N St (t) = e − λ ⋅ N St ⋅t . При этом величина константы λ, согласно С. А. Старостину, должна составлять около 0,05 для большинства языков: 2 N St (t) = e −0, 05⋅ N St ⋅t . 268

К вопросу о точности глоттохронологии

Для калибровки параметров рассмотренной модели по данным романских языков, как и в предыдущем случае, воспользуемся методом наименьших средних квадратов и получим следующие значения коэффициента λ и минимального суммарного отклонения ε: λ=0,108; ε=0,214. Очевидно, что найденный коэффициент λ≈0,11 более чем в два раза отличается от исходного (0,05), что заставляет нас перейти к уточненной модели вида: 2

N StC (t) = e −0,11⋅ N StC ⋅ t . Как показывает рассмотрение графиков (рис. 5), переход к калиброванной модели позволяет добиться гораздо лучшего совпадения расчетных значений с опорными точками в диапазоне от 1,5 до 2,5 тыс. лет (в отличие от исходной формулы, дающей заметно «заглубленные» датировки практически на любом временном интервале). Распределение значений λ, полученное по исходным данным для модели Старостина выглядит менее равномерным, чем в случае с формулой Сводеша, однако также достаточно близко к нормальному: математическое ожидание составляет λSt =0,11, среднее квадратическое отклонение σλ≈0,02 (рис. 6). При этом наличие шести сильно выделяющихся значений, соответствующих λ>0,2 (точки 3, 4 и 5 на рис. 2), не повлияло на результат калибровки и конечный вид модели7. Наряду с описанными выше традиционными глоттохронологическими моделями М. Сводеша и С. А. Старостина рассмотрим также статистический метод, основанный на потоковой интерпретации процесса лексических замен. 1.3. Потоковая глоттохронологическая модель. В основе потоковой модели лежит представление о том, что процесс замен каждого из значений основного списка является потоком случайных событий, которые происходят с малой интенсивностью и не влияют одно на другое. Таким образом, развитие всего основного списка можно описать как сумму нескольких независимых экспоненциальных потоков с различным, но постоянным темпом распада8. Каждая из таких составляющих соответствует группам значений или отдельным значениям в составе основного списка, которые обладают одинаковой или близкой устойчивостью При этом число групп и их коэффициенты стабильности подбираются в зависимости от данных, используемых для калибровки модели. Например, принимая исходное количество составляющих модели равное трем, получаем выражение следующего вида:

N P (t) = c1e − λ1t + c 2e − λ 2 t + c3e − λ 3 t , где с1, с2 и с3 — доля слов в каждой из составляющих основного списка, а λ1, λ2 и λ3 — соответствующие коэффициенты стабильности каждой из них. При этом сумма всех найПосле исключения всех выделяющихся значений из расчетов величина λSt по-прежнему составила 0,11. Подробное теоретическое обоснование предлагаемой модели приводится в статье Васильев, Милитарев 2008: 518—523, а практические примеры ее применения в работах Васильев, Старостин 2013, Васильев, Коган 2014. Можно заметить, что данный подход, как и методика С. А. Старостина, подразумевает отказ от 3­го и 4­го постулатов Сводеша, однако использует при этом совершенно другие исходные посылки. 7 8

269

М. Е. Васильев, М. Н. Саенко

Рисунок 7. Соответствие между полученной потоковой моделью и исходными данными: N P (t) = 0,2 + 0,8 e −0, 21⋅ t .

1.1 1

NP(t)

0.9 0.8 0.7

N(t)

0.6 0.5 0 0.5 t, тыс.лет

1

1.5

2

2.5

3

денных компонент с1+с2+с3 должна равняться единице (т.е. образовывать полный основной список), а все коэффициенты — иметь неотрицательные значения. Так, например, классическую формулу Сводеша можно представить как частный случай потоковой модели, которая содержит всего одну значимую составляющую с1=1 с коэффициентом λ1=0,16. Определим оптимальные параметры потоковой модели по исходным данным (табл. 5 и рис. 2) с помощью метода наименьших средних квадратов. Полученные значения приводятся ниже: с1=0,200; с2= 0,238; с3=0,562; λ1=0,000; λ2= 0,210; λ3=0,210; ε=0,094. Нулевой коэффициент первой составляющей (λ1=0), указывает на высокую устойчивость входящих в нее значений (около 20% основного списка), которые сохраняются в языке с течением времени. В то же время равенство коэффициентов λ2=λ3=0,210 свидетельствует о том, что заданное число слагаемых было избыточным, и модель без потери точности может быть сведена к двум содержательным компонентам: 20% (с1=0,2) — сверхстабильная часть списка, 80% (с2+с3=0,8) — экспоненциально убывающая часть значений с коэффициентом λ2,3= 0,21:

N P (t) = 0,2 + 0,8 e −0, 21⋅t . Представленный график (рис. 7) позволяет убедиться в хорошем соответствии потоковой модели фактическим данным о процессе замен в романских языках, что численно подтверждается незначительной суммарной ошибкой (ε=0,094), полученной в ходе калибровки. Завершив рассмотрение основных глоттохронологических методов и моделей, перейдем к сравнению и анализу полученных результатов. 270

К вопросу о точности глоттохронологии

Рисунок 8. Сравнение исходных и калиброванных моделей глоттохронологии, полученных по данным романских языков: N Sw (t) = e −0,16⋅t — исходная/калиброванная модель Сводеша, 2 N St (t) = e −0,05⋅N St ⋅t — исходная модель Старостина, 2 N StC (t) = e −0,11⋅N StC ⋅t — калиброванная модель Старостина, N P (t) = 0,2 + 0,8 e −0, 21⋅t — калиброванная потоковая модель.

1.1 NStC(t) 1 0.9 NSt(t) NP(t)

0.8 0.7

N(t)

0.6 NSw(t) 0.5 0 0.5 t, тыс.лет

1

1.5

2

2.5

3

Таблица 6. Сравнение исходных и калиброванных моделей. Общий вид и название модели

Вид исходной модели

Вид калиброванной модели

N(t) = e − λ ⋅t

N Sw (t) = e −0,16⋅t

N SwC (t) = e −0,16⋅t

модель Сводеша

N St (t) = e

− λ ⋅N St ⋅t 2

модель Старостина

N P (t) = c1e − λ1t + c 2e − λ 2 t + c3e − λ3t потоковая модель

λSw = 0,16

N St (t) = e

λSwC = 0,16 (εSw = 0,094)

−0 , 05⋅N St ⋅t 2

N StC (t) = e −0,11⋅N StC ⋅t

λSt = 0,05

2

λStC = 0,11 (εSt = 0,214)

N P (t) = 0,2 + 0,8 e

−0 , 21⋅t

с1 = 0,20; с2 = 0,80; λ1 = 0,00; λ2 = 0,21; (εP = 0,094)

1.4. Сравнение полученных моделей и их оценка. Обратимся к сравнительной таблице (табл. 6), которая содержит как исходные, так и калиброванные модели, а также значения соответствующих параметров. Сопоставление столбцов таблицы показывает, что существенные изменения в ходе калибровки моделей по опорным точкам произошли только в формуле Старостина: полученный коэффициент λSt составил 0,11 при исходном значении 0,05. При этом калиброванная модель Сводеша оказалась неизменной с константой λSw=0,16. 271

М. Е. Васильев, М. Н. Саенко

Рисунок 9. Фактический разброс значений N(t), используемых в качестве исходных данных для калибрования моделей.

1.1 1 0.9 0.8 0.7

N(t)

0.6 0.5 0 0.5 t, тыс.лет

1

1.5

2

2.5

3

Если мы сравним графики представленных моделей (рис. 8), то обнаружим, что потоковая модель и модель Сводеша (несмотря на принципиальное отличие используемых подходов), дают практически идентичные датировки на всем временном диапазоне с одинаковой суммарной погрешностью εSw=εP=0,094. При этом, как следует из рисунка, обе эти модели обеспечивают наилучшее совпадение с опорными точками, в то время как применение модели Старостина приводит к существенным неточностям как с исходным коэффициентом (λSt=0,05), так и после его корректировки в соответствии с исходными данными (λStC=0,11). И в том, и в другом случае расчетные датировки в интервале 0,5…1,5 тыс. лет оказываются завышенными по отношению к фактическим. Главной причиной этих расхождений является неравномерный «замедленно-ускоренный» характер распада, вызванный введением ускоряющей и замедляющей поправок. Таким образом, можно сделать вывод, что процесс изменений в базисной лексике одного языка наиболее корректно описывается с помощью экспоненциальной зависимости с постоянным (но не обязательно одинаковым!) темпом замен отдельных значений или частей списка. Убедившись в принципиальной адекватности математического аппарата глоттохронологии для описания процесса словарных замен, мы можем перейти к численной оценке точности и надежности различных моделей при датировании лексической дивергенции. 1.5. Оценка точности глоттохронологических моделей. Представляется очевидным, что точность расчетных датировок в первую очередь обусловлена точностью исходных данных, которые используются для идентификации и калибровки параметров моделей. Следовательно, измерение точности следует начинать с оценки фактического разброса значений, объективно присутствующего в исходных данных в силу их статистического характера. Графически данный разброс можно представить в виде кривых, соединяющих крайние значения опорных точек на всем интервале времени: рис. 9. 272

К вопросу о точности глоттохронологии

Рисунок 10 а, б. Иллюстрация разброса фактических долей совпадений и соответствующих датировок по отношению к расчетным значениям N(t) и времени t, полученным по калиброванной модели N Sw (t) = e −0,16⋅t . 1.1

1.1

1

1

0.9 0.8

2ΔN=0,14 (±7%)

0.7

0.9

0,83

0.8

0,75

0.7

0,69

0.6

N(t)

N(t)

0.6 0.5

0,75

1,85 0 0.5 t, тыс.лет

1

1.5

2

2.5

3

0.5

1,16 0 0.5 t, тыс.лет

1

2,3

1,85 1.5

2

2.5

3

2Δt=1,14 (±570 лет)

Рисунок 11 а, б. Доверительный интервал, рассчитанный для усредненной модели фициентом λ=0,16 и заданной вероятностью 0,7. 1.1

1.2

1

1

NSw(t)

0.9

0.6

0,75

0.7

0.4

0.6

0.2

0.5

NSw(t)

0.8

1,45 1,85 0 0.5 t, тыс.лет

1

1.5

2,09 2

2Δt=0,64

2.5

N(t)

N(t)

0.8

N Sw (t) = e −0,16⋅t с коэф-

3

0

0,5

0 t, тыс.лет

1,68 1,39 1 2

3,7 4,3

1,09

2Δt=0,59

3

4

4,9 5

6

2Δt=1,2

Используя диаграммы на рис. 10 (а и б), можно численно оценить разброс процентов совпадений для выбранного значения времени или наоборот — интервал неопределенности датировки для известного процента совпадений между сравниваемыми языками. Например, для временного отрезка 1850 лет разброс доли совпадающих значений составит около 14% (т.е. в среднем ±7 слов при использовании 100 словных списков). Аналогичным образом для доли совпадений N(t)=0,75 диапазон временной неопределенности составит 1140 лет (±570 лет). Изложенный способ оценки точности глоттохронологических моделей подразумевает наличие достаточно большого объема фактических данных о разбросе процентов 273

М. Е. Васильев, М. Н. Саенко

Рисунок 12. Доверительный интервал, рассчитанный для усредненной модели N Sw (t) = e −0,16⋅t с коэффициентом λ=0,16 и заданной вероятностью 0,95.

1.2 1

NSw(t)

0.8 0.6

0,5

0.4

N(t)

0.2 0

0,79

0 t, тыс.лет

1,98 1

1,39

2

3,12 4,3 3

2Δt=1,19

4

5,49 5

6

2Δt=2,37

Таблица 7. Зависимость величины доверительного интервала от выбранного времени и вероятности. Приведены усредненные значения. В скобках указан процент величины интервала от значения расчетной датировки. Расчетная датировка, лет

1000

2000

3000

4000

5000

6000

Величина доверительного интервала для вероятности p = 0,7

±250 (25%)

±360 (18%)

±470 (16%)

±560 (14%)

±650 (13%)

±750 (13%)

Величина доверительного интервала для вероятности p = 0,95

±500 (50%)

±720 (36%)

±940 (32%)

±1120 (28%)

±1300 (26%)

±1500 (25%)

совпадений сравниваемых списков и соответствующих достоверных датировок — т.е. является эмпирическим. Вместе с тем, на основе описанного выше потокового подхода оценка точности моделей может быть произведена теоретическими методами — благодаря известным статистическим свойствам этого процесса. А именно, если представить общий поток замен в одном списке как сумму потоков замен его значений (каждый из которых является экспоненциальным), то для суммарного потока первых замен по известным формулам можно рассчитать доверительный интервал (Вентцель, Овчаров 1969: 235) значений, полученных с использованием модели. Зная величину доверительного интервала, вычисленную для некоторого процента совпадений, мы можем указать диапазон времени, в который с заданной вероятностью9 укладывается искомая датировка (рис. 11). Например, при известной доле совпадений N(t)=0,75, пользуясь моделью N (t) = e–0,16⋅t, получаем расчетную датировку 1850 лет и доверительный интервал 640 лет, из чего сле9 Например, вероятность р=0,7 указывает на то, что в 70 случаях из 100 фактическая датировка будет находиться в пределах нижней и верхней границы рассчитанного доверительного интервала.

274

К вопросу о точности глоттохронологии

дует, что фактическое значение с вероятностью 0,7 может варьироваться в диапазоне от 1450 лет до 2090 лет (рис. 11а). С увеличением временной дистанции ширина доверительного интервала также будет расти. Так, для значения N(t)=0,5 и той же вероятности (p=0,7) его величина достигает уже 1200 лет (рис. 11б). Однако следует отметить, что в процентном отношении величина доверительного интервала постепенно снижается по мере удревнения датировок: например, от 25% (для t=1000 лет) до 13% (для t > 4 тыс. лет), (см. табл. 7, p=0,7). Это обстоятельство хорошо объясняет трудности глоттохронологического сопоставления списков языков с малыми глубинами расхождений. Численно оценить величину доверительного интервала для разных значений времени и выбранной вероятности можно с помощью табл. 7 и рис.12. Сопоставив рис. 11 с рис. 12 и значениями из табл. 7, можно убедиться, что фактический разброс значений хорошо совпадает с теоретическим доверительным интервалом, установленным для вероятности p=0,95, что позволяет сделать вывод о статистической адекватности используемых моделей, а также их практической пригодности для получения лингвистических датировок. При этом, однако, следует помнить, что все результаты вычислений будут иметь вероятностный характер, т.е. чем выше желаемая надежность датировки, тем больше величина ее неопределенности. На практике это означает, что при датировании лексических процессов с использованием глоттохронологии следует говорить не о конкретной дате, а о диапазоне дат с известной вероятностью. Например: «нами получена датировка 2000±360 лет с вероятностью 70%» или «2000±720 лет с вероятностью 95%». Несмотря на то, что исследование процесса лексических замен, происходящих в списке одного языка с течением времени, имеет большое теоретическое значение и составляет основу любых лексикостатистических методов, его применение на практике весьма ограничено. Действительно, случаи, когда необходимо определить временную дистанцию между языком-предком и его потомком, встречаются в компаративистике довольно редко. Гораздо более распространенной задачей является датирование разделения двух современных языков, обладающих предположительным или установленным генетическим родством. Таким образом, с практической точки зрения было бы интересно провести анализ глоттохронологических моделей, описывающих относительную языковую дивергенцию. Этому анализу, как уже говорилось выше, будет посвящена вторая часть настоящей статьи. Тем не менее, полученные результаты позволяют сформулировать некоторые важные выводы уже по итогам первой части проведенного исследования: 1. Процесс лексических изменений, наблюдаемый в базисной лексике романских языков, наиболее корректно описывается экспоненциальной зависимостью с постоянной скоростью замен отдельных значений или частей списка. Данная зависимость реализована, в частности, в классической модели М. Сводеша, а также, в общем случае, — в потоковой модели. 2. Калибровка параметров глоттохронологических моделей позволяет добиться хорошего соответствия получаемых датировок исходным данным. При этом отдельные фактические значения могут существенно отличаться от расчетных, что свидетельствует, с одной стороны, о ярко выраженной статистической природе и неравномерности лексического процесса, а с другой — о недостаточном количестве опорных точек, используемых при калибровке. 3. Лингвистические датировки, полученные с использованием глоттохронологических методов, имеют вероятностный характер и представляют собой не точную величину, а диапазон (доверительный интервал) значений, к которому с известной вероятностью будет принадлежать искомая дата. 275

М. Е. Васильев, М. Н. Саенко

4. Величина доверительного интервала может быть определена по фактическим исходным данным, а также рассчитана теоретически — на основе установленных статистических свойств процесса лексических замен. При этом она зависит от измеряемого временного отрезка и желаемой надежности расчетных значений. По мере их увеличения доверительный интервал растет, а точность датировки, соответственно, снижается. 5. Повышение надежности лингвистических датировок, получаемых с использованием глоттохронологии, возможно за счет привлечения дополнительных исходных данных из различных языковых групп и семей, что является важной задачей для будущих лексикостатистических исследований.

Дополнительные материалы доступны на: • http://jolr.ru/ Файл MS Excel содержит: • Таблица 8. Проценты совпадений между 110­словными списками романских идиомов.

Литература Арапов, М. В., М. М. Херц. 1974. Математические методы в исторической лингвистике. Москва: Наука. Васильев, М. Е., А. Ю. Милитарев. 2008. Глоттохронология в сравнительно-историческом языкознании. Модели дивергенции языков. Orientalia et Classica: Труды Института восточных культур и античности 19: 509—536. Васильев, М. Е., А. И. Коган. 2013. К вопросу о восточнодардской языковой общности. Journal of Language Relationship 10: 149—177. Васильев, М. Е., Г. С. Старостин. 2014. Лексикостатистическая классификация нубийских языков: к вопросу о нильско-нубийской языковой общности. Journal of Language Relationship 12: 51—72. Вентцель, Е. С., Л. А. Овчаров. 1969. Теория вероятностей. Москва: Наука. Сводеш, М. 1960а. Лексикостатистическое датирование доисторических этнических контактов. Новое в лингвистике 1: 23—52. Сводеш М. 1960б. К вопросу о повышении точности в лексикостатическом датировании. Новое в лингвистике 1: 53—87. References Arapov, M. V., M. M. Herz. 1974. Matematicheskiye metody v istoricheskoy lingvistike. Moskva: Nauka. Bergsland, K., H. Vogt. 1962. On the validity of glottochronology. Current anthropology 3(2): 115—153. Kroeber, A. L. 1958. Romance history and glottochronology. Language 34(4): 454—457. Rea, J. A. 1958. Concerning the Validity of Lexicostatistics. International Journal of American Linguistics 24(2): 145— 150. Starostin, S. 2000. Comparative-historical linguistics and lexicostatistics. In Colin Renfrew et al. (eds.). Time Depth in Historical Linguistics. Cambridge: McDonald Institute for Archaeological Research. Vol. 1: 233—259. Swadesh, M. 1952. Lexico-statistic dating of prehistoric ethnic contacts. Proceedings of the American philosophical society 96(4): 452—463. Swadesh, M. 1960a. Leksikostatisticheskoye datirovaniye doistoricheskikh etnicheskikh kontaktov. Novoye v lingvistike 1: 23—52. Swadesh, M. 1960b. K voprosu o povyshenii tochnosti v leksikostatisticheskom datirovanii. Novoye v lingvistike 1: 53—87. Vasilyev, M. E., A. I. Kogan. 2013. K voprosu o vostochnodardskoy yazykovoy obshchnosti. Journal of Language Relationship 10: 149—177. 276

К вопросу о точности глоттохронологии

Vasilyev, M. E., A. Yu. Militaryov. 2008. Glottokhronologiya v sravnitel’no-istoricheskom yazykoznanii. Modeli divergentsii yazykov. Orientalia et Classica: Trudy Instituta vostochnykh kultur i antichnosti 19: 509—536. Vasilyev, M. E., G. S. Starostin. 2014. Leksikostatisticheskaya klassifikatsiya nubiyskikh yazykov: k voprosu o nil’sko-nubiyskoy yazykovoy obshchnosti. Journal of Language Relationship 12: 51—72. Venzel, E. S., L. A. Ovcharov. 1969. Teoriya veroyatnostey. Moskva: Nauka.

Mikhail E. Vasilyev, Mikhail N. Saenko. How accurate glottochronology can be? Dating the lexical replacement process in the Romance languages. In this paper we discuss the accuracy of glottochronology, a lexicostatistical method used in the dating of linguistic divergence. Our study provides a detailed analysis of the process of lexical replacement in the basic lexicon of one language over the course of time. To measure replacement rates and determine other statistic features of lexical change we use 110­item wordlists, compiled over the past two years for 54 modern and several historically attested Romance languages. Pairwise comparison of modern wordlists with those of Archaic Latin, Late Classical Latin, Old French, and Old Italian allows to obtain several control points suitable for calibration of glottochronological equations. To estimate the time distance between the compared idioms, three different methods have been applied: the classic formula of M. Swadesh, the modified glottochronology of S. Starostin and a recently proposed approach based on simulation of lexical changes of every meaning on the Swadesh list as stationary Poisson processes. Further analysis resulted in several important conclusions concerning the following questions: (a) what are the main characteristics of lexical divergence in one language; (b) which of the existing models maps these characteristics more efficiently; (c) how precise and reliable glottochronological dating can be in general. We plan to follow this research by another study in which the process of relative divergence between two or more languages with the same ancestor will be considered. Keywords: Romance languages, lexicostatistics, glottochronology, Swadesh wordlist.

277

И. С. Якубович Марбургский университет имени Филиппа (Марбург, Германия); Российский государственный гуманитарный университет (Москва, Россия), [email protected]

Славянский чертежник: этимология слав. *čьrtъ ‘черт’ В настоящей работе обсуждается славянское наименование черта, которое реконструируется как *čьrtъ на общеславянском уровне, но не имеет очевидных индоевропейских когнатов. Типологические параллели из других индоевропейских мифологических традиций, в первую очередь армянские и анатолийские данные, приводят автора к заключению, что праславянское *čьrtъ может быть интерпретировано как обозначение неземного чертежника, определяющего человеческие судьбы. Постепенно этот бог судьбы стал восприниматься как ангел смерти, что и привело к его синкретизму с дьяволом — воплощением зла в христианской традиции. Ключевые слова: этимология, черт, славянский, армянский, анатолийский.

1. Введение. Убедительная этимология лексемы с нетривиальным семантическим развитием не должна сводиться к внешним лингвистическим соответствиям1. Следует также представить сценарий того, как могли развиться новые значения, в идеале сопроводив их типологическими параллелями из других языков. Корректное лингвистическое сравнение не обязательно влечет за собой правильное понимание эволюции значения. Для иллюстрации этой идеи я хотел бы обратиться к случаю, когда когнаты были подобраны корректно, но этимология опирается на неубедительную семантическую реконструкцию, и предложить лучшую альтернативу. Хочется надеяться, что данный пример окажется небезынтересным также и для индоевропейской сравнительной мифологии. 2. Славянский ‘черт’. Рефлексами праслав. *čьrtъ являются польское czart ‘черт, дьявол, злой дух’, чешское и словацкое čert ‘черт, злой дух; беда’, верхнелужицкое čert ‘черт, демон’, нижнелужицкое cart ‘черт, дьявол’, русское чëрт, украинское и белорусское чорт ‘черт, дьявол’ (Трубачев 1977: 164). Наличие потомков этой лексемы в южнославянских языках является предметом дискуссии. Проще всего считать словенское čt ‘черт’ ученым заимствованием из других славянских языков (Bezlaj 1977: 89), которое могло распространиться посредством искусственно Первая версия настоящей статьи была представлена на XIV Конгрессе Indogermanische Gesellschaft в Копенгагене (2012 г.). Я благодарен участникам конгресса за комментарии и предложения, послужившие улучшению настоящей статьи. Особую благодарность я выражаю Вацлаву Блажеку (Брно), Александру Лубоцкому (Лейден), Крэгу Мелчерту (Лос Анджелес) и, особенно, Грачу Мартиросяну (Лейден). Алексей Касьян (Москва), Армен Петросян (Ереван) и Крешимир Вукович (Оксфорд) помогли мне с рядом библиографических ссылок. Ответственность за высказанные в данной работе суждения лежит, разумеется, на ее авторе. Работа над финальной версией статьи была поддержана стипендией имени Гумбольдта во время моего пребывания в Марбургском университете. Задержка с публикацией материалов конгресса, а также особый интерес, который данная работа может представлять для русскоязычного читателя, побуждают меня представить ее русскую доработанную версию для публикации в «Вопросах языкового родства». 1

Journal of Language Relationship • Вопросы языкового родства • 14/4 (2016) • Pp. 279—291 • © The authors, 2016

И. С. Якубович

созданного личного имени Črtomir 2. Заимствованное происхождение труднее продемонстрировать для лексемы čt ‘ненависть, вражда’, зафиксированной в словенском языке XIX века (см. Трубачев 1977: 164 и ссылки). Хорватское črt ‘черт’ также встречается в источниках XIX века, и его объяснение через раннее заимствование из словенского не является убедительным ввиду наличия этого корня в местных топонимах (Katičić 2003—2004: 260—263). Таким образом, нельзя исключать гипотезу о том, что рефлекс праславянского *čьrtъ встречался и на Балканах до того, как был там замещен различными эвфемизмами, такими, как vrag. Был ли исконный *čьrtъ утерян в общеюжнославянском или в отдельных южнославянских языках, сравнение между восточно- и западнославянскими языками оправдывает его реконструкцию для праславянского3. Однако в самом праславянском языке он, по-видимому, являлся инновацией по сравнению с более архаичным обозначением злого сверхъестественного существа, сохранившемся в балтийских языках. Литовское vélnias и латышское velns ‘черт’ связаны на синхронном уровне со словами, обозначающими духи мертвых: лит. vlės, лат. velis. Эти лексемы относятся к тому же индоевропейскому корню, что и древнеисландское valr, древнесаконское и древневерхненемецкое wal ‘воин, убитый на поле брани’, откуда, несомненно, происходит и название валькирий, досл. «выбирающих мертвых» (Fraenkel 1962—65: 1218—19; Smoczyński 2007: 731, 732). Несмотря на невозможность провести формальную реконструкцию общего теонима из балтийских и германских когнатов со значением бога — покровителя мертвецов, концепт такого божества вполне вероятно является по происхождению северноиндоевропейским4. В то же время, праслав. *čьrtъ представляется совершенно изолированным в рамках индоевропейского религиозного лексикона5. 3. Формально неприемлемые этимологии Если *čьrtъ представляет собой исключительную славянскую инновацию в области индоевропейской демонологии, весьма вероятно, что его деривационная основа также сохраняется в славянских языках. Ранние гипотезы относительно славянской деривации 2 Период формирования словенского литературного языка совпал с периодом творчества поэта-романтика Франца Прешерна (1800—1849), чье влияние на последующую литературу в Словении трудно переоценить. Его эпическая поэма Krst pri Savici («Крещение при Савице») изображает протагониста с именем Črtomir, формально являющимся дериватом от črt. Языческий воин Чртомир в конце концов принимает христианство под влиянием своей возлюбленной Богомилы. Я глубоко признателен Гашперу Бегушу (Любляна), который подтвердил, что имя Чртомир, повидимому, не использовалось в Словении до Прешерна, несмотря на его популярность в наши дни. 3 Несмотря на то, что рус. чёрт, мн. черти засвидетельствован лишь с XVI или даже начала XVII веков, гипотеза о том, что это существительное было заимствовано из польского czart (Раденкович 2014) представляется неприемлемой по формальным соображением. Чередование e / ё в корне и архаическое множественное число на ­и указывают на то, что данная лексема уже существовала в общевосточнославянском. 4 Несколько иной подход к этимологии балтийского слова для ‘черта’ был предложен Ивановым и Топоровым (1974: 31—74). Российские ученые сравнивают лит. vélnias со славянским теонимом Велес / Волос и ведическим демоном Вала, предполагая, что они представляют собой различные трансформации индоевропейского хтонического божества, главного оппонента бога грозы, а также властелина (или похитителя) скота. Доказательство этой гипотезы поднимает большой пласт вопросов, касающихся реконструкции праиндоевропейского пантеона, рассмотрение которых находится за рамками настоящей статьи. Хотелось бы, однако, подчеркнуть, что гипотеза Иванова и Топорова 1977 года только подкрепляет архаический характер лит. vélnias и лат. velns по сравнение с праслав. *čьrtъ ‘черт’. 5 См. также попытку Дуковой (1984) найти мифологические существа, которые обозначались бы когнатом лексемы *čьrtъ вне славянских языков. Дукова предлагает древнеисл. skrati ‘тролль’ и древневерхненем. scraz ‘фавн’ как возможные соответствия. Это сравнение невозможно, поскольку германские формы требуют реконструкции *scrod° (vel sim.).

280

Славянский чертежник: этимология слав. *čьrtъ ‘черт’

*čьrtъ собраны, например, в словарях Vasmer 1953—58, III: 328 и Bezlaj 1977: 89, но ни одна из них не убедила авторов соответствующих словарей и не стала более вероятной в свете новейших открытий6. В самом деле, две из них имеют очевидные формальные недостатки. Гипотеза о родстве с унаследованным праслав. *čьrnъ ‘черный’ (~ вед. krṣná- ‘id.’) оставляет без объяснения суффикс ­t­, а связь с русским терять невозможна не только ввиду необъяснимой палатализации k > č, но и поскольку примеры данного глагола ограничены восточнославянскими языками. Какие бы семантические соображения ни приводились бы в пользу черта как «черного» или «потерянного», соответствующие этимологии сталкиваются с непреодолимыми формальными препятствиями. То же самое можно сказать и об этимологии *čьrtъ, которой отдали предпочтения Vasmer и Bezlaj (loc. cit.), исходя из допущения, что словообразовательная база славянского существительного наличествует в балтийском. Цитирую Фасмера: «Ursl. *čьrtъ wird als *­to- Part. ‘der Verwünschte’ und für verwandt angesehen mit lit. kyrti ‘böse werden’, į-kỳrti ‘sich ekeln’, ap-kyrti ‘überdrüssig werden’, į-kyrùs ‘lästig, aufdringlich’, kerti ‘mit bösem Blick bezaubern’». Отсутствие славянских когнатов само по себе говорит против данной корневой этимологии. Что, однако, более существенно, ни одна из балтийских форм не является семантически совместимой с Фасмеровской реконструкцией черта как ‘проклятого’. К тому же, все процитированные формы, за исключением į-kỳrti ‘заикаться’, представляют собой стативные образования на *­ē­, не образующие *­to- причастий в праславянском. Таким образом, можно лишь согласиться с оценкой Трубачева (1974—, IV: 165): «Каким образом при этом могло получиться причастие на ­to- со значением ‘проклятый’ … остается для нас загадкой, авторы, кажется, не очень считаются с семантикой и морфологией привлекаемых для сравнения основ»7. Единственная этимология *čьrtъ, упомянутая Фасмером и Безлаем, которую можно считать формально приемлемой, связывает эту лексему с лат. curtus ‘короткий, обрубленный’, вероятно, произведенным от и.-е. *(s)ker ‘обрезать, обрубать’ (Rix 2001: 556—57). С формальной точки зрения, можно предполагать, что мы имеем дело с лексическими когнатами и прямыми рефлексами этимологического причастия *kr-to­. Эпитет черта «куцый», разделяемый рядом славянских традиций, мог бы представлять собой семантическую мотивацию для обсуждаемой этимологии (Черных 1993, II: 384). Против данного решения, однако, говорит тот факт, что единственным славянским глагольным рефлексом и.-е. *(s)ker является диалектное укр. *čru (Rix 2001: 557), а сама лексема *čьrtъ не встречается в функции прилагательного. Заметим, что праслав. *kortъkъ ‘короткий’ произведено от и.-е. *(s)kert ‘резать’, вероятно, вариант и.-е. *(s)ker с расширителем, выделившийся в отдельный корень уже на праиндоевропейском уровне (Rix 2001: 559—60). Если бы *čьrtъ ‘куцый’ > ‘черт’ было внутриславянским развитием, следовало бы ожидать больше когнатов данного корня в славянских языках. 4. Семантически проблематичные этимологии. В последние десятилетия в славистических кругах завоевала популярность другая этимология *čьrtъ. С формальной точки зрения, славянская лексема может представлять

Ср. также современный обзор этимологий черта в статье Березович, Виноградов 2012. Следует отметить, что лит. kerti ‘ворожить’ этимологически не связано с другими литовскими лексемами, приводимыми Фасмером. Это прозрачный когнат лит. keraĩ ‘колдовство’ и менее очевидный когнат лит. kùrti ‘строить, сооружать’ (Smoczyński 2007: 277). Относительно семантического перехода ‘строить’ > ‘колдовать’, см. ниже. Что касается лит. į-kyrùs ‘назойливый, навязчивый’ и родственных лексем, их индоевропейская этимология не установлена, однако они не могут являться когнатами kùrti ‘строить, сооружать’ ввиду различий в значении и в корневом вокализме. 6 7

281

И. С. Якубович

собой отглагольное имя, производное от праслав. *čersti / čьrtǫ ‘чертить, бороздить’, родственное лит. kir̃sti / kertù ‘рубить’. Обсуждаемый балтославянский корень является еще одним рефлексом и.-е. *(s)kert ‘резать, рубить’, упомянутого в предыдущем разделе. Подобное существительное едва ли могло быть унаследовано из праиндоевропейского, где простые тематические именные производные отглагольных корней чаще всего характеризовались о-вокализмом (ср. тип τόμος / τομός в древнегреческом). В истории славянских языков, однако, получили развитие вторичные тематические производные с нулевой ступенью корня. Они, как правило, соответствовали нулевой ступени корневого вокализма производящего глагола. Например, от праслав. *smьrděti / smьrdjǫ ‘смердеть, вонять’ было образовано первичное имя действия *smordъ ‘смрад, вонь’ и вторичное имя деятеля др.-рус. smьrdъ ‘смерд’, досл. «вонючка» (Vasmer 1953—58, II: 671, 676). Другое инновативное образование с нулевой ступенью корня отражено в ст.-слав. tlъkъ ‘толмач’, образованном от и.-е. корня *tlok, сохраненного в лат. loquī ‘говорить’ (Vasmer 1953—58, III: 115) или от праслав. *telkti / tlъkǫ ‘толочь’ (Vaillant 1974: 249)8. Еще чаще инновативные производные с нулевой ступенью корня используются в значении имен действия или объекта, ср., напр., др.-рус. tъlkъ ‘толк’ (Vasmer 1953—58, III: 115). Другим дериватом подобного же рода является сербскохорватское и словенское o-črt ‘контур’, относящееся к тому же корню, что и вышеупомянутое *čersti / čьrtǫ ‘чертить, бороздить’ (Vaillant 1974: 239). Наконец, славянские производные a-основы с нулевой ступенью корня, подобные *čьrta ‘черта’ (Vasmer 1953—58, III: 328) можно рассматривать как дополнительный вариант той же словообразовательной модели. Очевидно, что если от глагольной основы можно было произвести вторичное nomen rei actae с нулевой ступенью корня, от нее также можно было произвести и nomen agentis с нулевым корневым вокализмом. Таким образом, сопоставление между славянским обозначением злого духа и производящего глагола со значением ‘чертить, бороздить’ является формально безупречным. В самом деле, праслав. *čersti / čьrtǫ является единственной распространенной глагольной основой, от которой может быть образовано производное существительное *čьrtъ. Однако, семантика предложенной деривации ни в коей мере не является очевидной и в прошлом служила основанием для различных интерпретаций. Первая такая попытка была предпринята в работе Brückner 1918: 174 и разработана в статье Jacobson 1959: 276 (без ссылки на Брюкнера). Согласно Брюкнеру, *čьrtъ первоначально обозначало ‘колдун’. И Брюкнер, и Якобсон связывают данное существительное с праслав. *čьrta ‘черта’, а также *čara ‘черта’, засвидетельствованном в западнославянских. Якобсон определяет чеш. čára как «a bordeline up to which something is permitted or magically prohibited, e.g. the line which marks the so-called magic circle (where the evil demons retain or lose their power)». Мы увидим, что волшебство, осуществляемое за защитной чертой, релевантно и для моих собственных построений. Но этимологическая связь между *čara и *čьrta представляется проблематичной, посокольку *čara непосредственно связана со ст.-слав. čarъ ‘колдовство, чары’, лит. kerti ‘ворожить’ и другими архаичными производными от и.-е. *ker ‘строить, сооружать’, не сохранившегося или плохо сохранившегося в виде финитного глагола в праславянском. В качестве типологической параллели к подобному семантическому переходу можно привести др.-исл. gørningar ‘дела, деяния, магия, колдовство’, производное от gørva ‘делать’ или скр. krtya- ‘дело, деяние, магия, колдовство’ от kar ‘делать’, в конечном итоге восходящего к тому же и.-е. *ker ‘строить, сооружать’. Таким образом, связь между чеш. čára и черчением, вероятно, вто8 Семантический переход ‘толочь’ > ‘переводить’ обсуждается в работе Vaillant 1974 (loc. cit). В качестве дополнительной типологической параллели ср. рус. молоть чепуху.

282

Славянский чертежник: этимология слав. *čьrtъ ‘черт’

рична, как и связь между *čьrta ‘черта’ и колдовством, хотя, как мы увидим, вторая ассоциация и восходит к праславянской эпохе. Поэтому в той мере, в какой сценарий Брюкнера и Якобсона поддается реконструкции, он представляется основанным на народной этимологии9. Другая реконструкция того, каким образом *čьrtъ может быть связан с *čersti / čьrtǫ ‘чертить, бороздить’, была детально разработана Трубачевым (Трубачев 1974—, IV: 166). Отправной точкой данного сценария является слов. čt ‘раскорчеванное место, межа’. Трубачев полагает, что данная лексема совместима со значением производящего глагола ‘копать’, каковое значение сосуществовало со значеним ‘чертить, бороздить’ или даже ему предшествовало. Если это так, то *čьrtъ мог исходно являться именем деятеля со значением ‘землекоп’, уместным для подземного духа. Особым указанием на склонность черта к земляным работам является распространенный топоним *čьrto-ryjь, досл. ‘прорытый чертом’, употребляемый по отношению к различным оврагам и промоинам (Трубачев 1974—, IV: 163—4). Наконец, в той же работе приводятся другие предполагаемые производные от обсуждаемого корня, а именно, ц.-слав. krъtъ ‘крот’ и лит. kertùkas ‘землеройка’, которые также могли исходно значить ‘землекоп’ (Трубачев 1974—, XIII: 58—9). Оценка данной этимологии напрямую зависит от убедительности реконструкции значения ‘копать’ для *čersti / čьrtǫ в праславянском. Насколько я могу судить, в анализе Трубачева она не подкрепляется вескими аргументами. Прямое сравнением между праслав. *čьrtъ ‘черт’ and *krъtъ ‘крот’ невозможно про формальным причинам, как и признает и сам Трубачев, поскольку первая лексема отражает палатализацию *k > č перед рефлексом слогового *r, тогда как вторую можно возводить только к *krutu/o-10. Спекуляции относительно «единства корня kŭr и ker», подкрепляемые ссылками столетней давности, могут лишь скомпрометировать современное лингвистическое исследование. Словенский ландшафтный термин čt, несомненно, этимологически связан с праслав. *čersti / čьrtǫ, однако, не требует изощренной семантической реконструкции, поскольку его значение тривиально выводится из производящего глагола ‘рубить (деревья)’ или ‘бороздить (межу)’. С другой стороны, хотя демоны и ассоциируются с подземным миром в славянской и во многих других традициях, отсюда отнюдь не следует, что их основной функцией считались земляные работы. Для древних славян было естественным приписывать овраги и промоины (*čьrto-ryjь) дьяволу, а не Богу по причине их уродливого вида и потенциального вреда для сельского хозяйства. Однако славяне также использовали рефлексы *čьrto-polxъ досл. ‘чертов ужас’ для различных неприятных растений и ядовитых грибов (Трубачев 1974—, IV: 163). Суммируя вышеизложенное, гипотеза Трубачева является формально ущербной и семантически избыточной.

19 С чисто формальной точки зрения, *čьrtъ можно реконструировать как страдательное причастие *k!r-to- от того же корня, что и ст.-слав. čarъ ‘колдовство, чары’. Недостатком данной этимологии является, с одной стороны, отсутствие финитных рефлексов и.-е. *k!er в праславянском, а с другой, сложности с возведением обозначения злого духа к страдательному причастию ‘заколдованный’ (скорее, ожидалось бы ‘колдующий’!). Заметим, что *čьrtъ не мог быть произведен от дескриптора ‘находящийся за чертой’, поскольку здесь ожидалось бы префиксальное образование, напр., **za-čьrtъ или **otъ-čьrtъ. 10 Напротив, лит. kertùkas ‘землеройка’, действительно, может являться производным от kir̃sti / kertù ‘рубить’ (Fraenkel 1962—65: 223 sub kar̃stas). Этого, однако, недостаточно для реконструкции исходного значения ‘копать’ для более ранней стадии семантического развития kir̃sti / kertù, поскольку обозначение землеройки вполне могло быть основано на метафоре разрубания земли. Сходным образом, англ. cutter ‘катер’ не предполагает, что исходным значением производящего глагола to cut является ‘плыть’.

283

И. С. Якубович

5. Предпосылки для семантического анализа. Общей особенностью этимологий, обсуждавшихся в предыдущем разделе, было то, что все они исходят из специфических особенностей сверхъестественного существа, известного как черт. Такой подход, разумеется, вполне оправдан, если иметь дело с хорошо сохранившимися религиозными традициями. Его достоинство менее очевидно в применении к славянской религии, которая перестала существовать как таковая с приходом христианства и не оставила за собой первичных письменных источников. Хотя некоторые славянские демоны и сохранили за собой специфические функции после христианизации Восточной Европы, это в меньшей мере относится *čьrtъ’у, который подвергся синкретизации с падшими ангелами христианства и их предводителем Дьяволом / Сатаной. В частности, ни защитная черта, упоминаемая Якобсоном, ни хтонические ассоциации черта, используемые Трубачевым в своей работе, не являются особенностями славянского демона, а, скорее, отражают общие черты средневековой христианской демонологии. Поэтому ни одна из этих особенностей, вероятно, не была связана с *čьrtъ’ом в дохристианской религии славян. По существу, для реконструкции исходной функции *čьrtъ’а существенны лишь два соображения. С одной стороны, поскольку *čьrtъ не является калькой какого-либо известного эпитета Сатаны, данный термин, вероятно, уже употреблялся славянамиязычниками для обозначения сверхъестественного существа. С другой стороны, это имя или титул не имеет очевидных индоевропейских когнатов, относящихся к той же сфере, а его внутренняя форма предполагает реконструкцию производного имени деятеля от производящей основы *čersti / čьrtǫ ‘чертить, бороздить’. Поэтому *čьrtъ может быть реконструирован как «чертежник», исходя из чисто структурных соображений. Для того же, чтобы оценить вероятность данной гипотезы, следует опираться на типологические параллели. Если подобных «чертежников» можно обнаружить в других религиозных традициях с коннотациями, способствующими их демонизации, это может как подкрепить обсуждаемую этимологию черта, так и помочь реконструировать функции *čьrtъ’а в дохристианском славянском пантеоне. 6. Иранские параллели? Одна из потенциальных параллелей засвидетельствована в иранских языках. Иранский глагольный корень *kart ‘резать’, рефлекс того же самого и.-е. *(s)kert, который дал и праслав. *čersti / čьrtǫ, приобрел вторичное терминологическое значение ‘стряпать (творить) злые существа’ в зороастрийской традиции. Обычным творцом в таких случаях является Ахра Манью / Ахриман — воплощение и создатель зла. В авестийском языке сотворение злых существ обычно выражается глаголом kart с префиксом fra- или его производными (добрые акты творения передаются глаголом θßarš с теми же приставками). Например, оппонент героя Фраетаона описывается как «весьма могучий демондрудж, которого состряпал (fraca krṇtat) Ахра Манью против материального мира, ради сокрушения творений Правды» (Ясна 9.8, ср. Pirart 2004: 66—67). Родственный среднеперсидский глагол kirrēn- ‘стряпать (творить) злые существа’ используется без префикса (Расторгуева и Эдельман 2000—, IV: 312)11. Существуют обильные свидетельства доисторических контактов между иранцами и славянами, отразившиеся в заимствовании ряда религиозных терминов, таких как *bogъ 11 Можно осторожно предположить, что иранский корень *kart скрывается в первой части среднеперсидского композита kerdagār (пехлеви ⟨krtk’l⟩, манихейское ⟨kyrdg’r⟩) ‘творец’ (ср. MacKenzie 1971: 49, где данное слово переводится как ‘могучий’).

284

Славянский чертежник: этимология слав. *čьrtъ ‘черт’

‘бог’ and *rajь ‘рай’12. Тем не менее, гипотеза о заимствовании славянами зороастрийского концепта злого творца сталкивается с непреодолимыми трудностями. С одной стороны, нет никаких оснований полагать, что те иранцы, от скифов до аланов, которые находились в контакте со славянами, следовали учению Заратуштры. Вместе с тем, маловероятно, что терминологическое использование иранского корня kart для сотворения злых существ, или даже религиозная доктрина о такого рода деятельности, существовали до развития зороастрийского дуализма. С другой стороны, семантика праслав. *čersti / čьrtǫ не содержит и тени указания на акт творения и не обладает пейоративными коннотациями, тогда как авестийский язык не сохранил производного имени деятеля для злого демиурга. Данная комбинация исторических и лингвистических нестыковок заставляет меня пойти по другому пути в поисках славянского чертежника. 7. Анатолийские чертежницы. Сравнительная информация иного рода доступна из анатолийских текстов бронзового века. В анатолийских языках встречаются генетически родственные теонимы, такие как хеттское dGulses (мн.ч.), лувийское dGUL-za- = /kwanza-/ (ед.ч), и палайской dGulzannikes (мн.ч.), связанные с лувийским глагольным корнем, засвидетельствованным в хеттском guls- и лувийском GUL-za(i)- ‘царапать, писать’, а также в лувийском GUL-zattar ‘чертеж, черновик’ (ср. Kloekhorst 2008: 492—93)13. Многочисленные контексты, в которых хетт. dGulses встречается в клинописных текстах из Хаттусы, позволяют неплохо реконструировать их функции (ср. Haas 1994 и, в особенности, Otten 1971). Данная группа богинь происходит из подземного мира, привлекается к заботе о новорожденных и часто ассоциируется с месопотамской богиней-матерью DINGIR.MA†. Их основной задачей было, однако, записывать (guls­) человеческие судьбы. Вероятно, по этой причине их иногда называли «злыми» (KUB 58.108 i 5), тогда как к неудачнику можно было обратиться с утешением ‘тебя обидели Гулсы’ (KUB 23.85 rev. 6). Кроме того, они определяли время смерти каждого человека. Аккадско-хеттская билингва содержит аккадское выражение a[­na] u4­mi [ši-ma­]ti-ka ‘в день твоей судьбы’, чему в хеттском тексте соответствует GIMan=ma=(t)ta dGulsas UD.KAM-us tianzi ‘когда для тебя наступят дни Гулс’ (KBo 12.70 obv.? 14 ff., ср. Taracha 2000: 186, fn. 89)14. Ряд ученых утверждал, что праслав. *bogъ ‘бог’ не заимствован из праир. *baga- ‘бог’, а является его когнатом (ср. Трубачев 1974—, II: 161—63). Решающий аргумент против исходного генетического родства между этими двумя терминами оказался доступным с открытием закона Винтера. В качестве tertium comparationis в данном случае выступает ведийский теоним Бхага, подтверждающий непридыхательный характер индоиранского *­g­. Поэтому славянский когнат данной лексемы звучал бы как **bagъ с рефлексом компенсаторного удлинения. 13 В соответствии с гипотезой, развитой в работе Waal 2014, теоним dGulses следует читать как dGUL-ses, где GUL — шумерограмма с базовым значением ‘ударять’. Ваал также предлагает сопоставить лувийский глагол, известный как GUL-za(i)- в клинописной передаче, с глаголом REL-za- (= kwa/i-za­) иероглифических надписей и приписать получившемуся глаголу терминологическое значение ‘писать иероглифами’. Моя реакция на данную гипотезу, которую я принимаю для лувийского, но не для хеттского языка, представлена в работе Yakubovich 2013. Для целей настоящей работы достаточно отметить, что принятие или отклонение гипотезы Ваал не отменяет этимологической связи между анатолийскими именами божеств судьбы и глаголом ‘писать’. Между тем, для аргументации настоящего раздела существенна лишь данная этимология. 14 Перевод Тарахи отличается в случае хеттского предложения: ‘когда Гулсы установят твои дни’. Перевод, приведенный выше, любезно предложенный мне К. Мелчертом, предполагает, что сказуемым в данном предложении является tiya-mi ‘ступать’, а не dai / tiya-i ‘ставить’. Данная гипотеза является предпочтительной, поскольку окончания именительного падежа множественного числа на ­as в хеттском являются редкими и нерегулярными, тогда как окончания на ­us широко распространены и даже обобщены в позд12

285

И. С. Якубович

Синхронная связь между именами анатолийских богов судьбы и анатолийским глаголом для письма требует отдельной дискуссии. Хеттская цивилизация в позднем бронзовом веке была знакома с грамотой и даже использовала две различные системы письма: месопотамскую клинопись и туземные анатолийские иероглифы. Тем не менее, нетривиальные фонетические различия между хетт. dGulses, лув. dGUL-za- = /kwanza-/, и пал. dGulzannikes предполагают, что если эти теонимы действительно связаны друг с другом, речь не идет о недавних заимствованиях, а скорее о рефлексах общеанатолийского прототипа или результате древней диффузии. Каким же образом могли пониматься имена божеств судьбы в конце III — начале II тыс. до н.э.? Можно предположить, что праанатол. *guls- (vel sim.) обладал значением ‘чертить, рисовать пиктограммы’, тогда как производный теоним имел значение ‘чертежник’ или ‘чертежница’, представляя тем самым типологическую параллель к праслав. *čьrtъ. Разумеется, невозможно строго доказать, что анатолийским мойрам приписывалась способность влиять на человеческую судьбу посредством черчения или манипуляции пиктограммами. Однако, известно, что сами хетты пользовались подобными средствами, чтобы получить представление о будущем. Популярным методом гадания среди хеттов была клеромантия, известная также как KIN-оракулы в хеттологической литературе. Сущностью данной практики, проинтерпретированной в работе Archi 1974, была манипуляция случайными символами при помощи случайного «медиума», приводящего их к случайной цели. Символы, «медиумы» и цели могли быть как положительными, так и отрицательными. К последней группе принадлежали, например, такие понятия, как Враг, Зло или Великий грех (термин для обозначения смерти царя или царицы). Archi (ibid: 130—131) осторожно предполагает, что процедура гадания заключалась в наблюдении над животными. Определенное животное могло попасть на площадку для гадания через те или иные ворота, тем самым взяв на себя функции «медиума», дотронуться до определенных символов и метафорически привести их к определенной цели, то есть к выходу с площадки для гадания. Можно предположить, что набор пиктограмм использовался для обозначения различных входов и выходов, а также символов на площадке для гадания. Насколько можно судить, подобный тип гадания был крайне редким на Ближнем Востоке за пределами Анатолии. Все вышеизложенное согласуется с гипотезой о том, что идея «предначертания судьбы» предшествовала грамотности на территории Малой Азии. 8. Черт и гадание у славян. Теперь уместно обратиться к предсказанию будущего у славян в дохристианскую эпоху. Ключевая информация по данному вопросу содержится в полемическом трактате О письменах, предположительно написанном в 893 году монахом Храбром, учеником св. Мефодия, но дошедшем до нас в более поздних списках. Он открывается словами: «Прежде славяне не имели книг, а читали и гадали чертами и резами, будучи язычниками»15. Можно оставить в стороне спекуляции о дохристианской грамоте среди славян, поскольку навыки чтения будущего не требуют грамотности в обыденном смысле, как это уже было отмечено в связи с ситуацией в Анатолии. У нас нет точной информации нехеттском для определенных морфологических классов, включая сюда и консонантные основы, к которым принадлежит UD.KAM-us = siwattus. 15 Ср. старославянский текст: Прѣжде ѹбо словѣне не имѣах( книгъ, н( чрътами и рѣзами чьтѣах( и гатаах( погани с(ще. Относительно возможных этимологических связей ст.-слав. rězy, см. Топоров 1974: 12—19. 286

Славянский чертежник: этимология слав. *čьrtъ ‘черт’

о том, как черты и «резы» использовались при гадании. Однако нет оснований подвергать сомнению свидетельство Храбра, который жил на рубеже христианизации славян и поэтому должен был обладать определенными познаниями в их дохристианских традициях. Ц.-слав. črъta ‘черта’, упоминаемая Храбром как инструмент гадания, исторически связана с праслав. *čьrtъ, если его первоначальным значением было «чертежник». Соотношение между этими двумя существительными функционально напоминает соотношение лув. GUL-zattar ‘чертеж, черновик’ и GUL-za ‘божество судьбы’. Развивая данную параллель, можно предположить, что праслав. *čьrtъ также считался божеством судьбы. Подобно тому, как в праанатолийском глагол *guls- использовался для предначертания судьбы, праслав. глагол *čьrtati мог использоваться в том же значении (ср. рус. пред-начертать). Начертание пиктограмм, возможно, ассоциировавшееся с процессом предначертания судьбы у древних анатолийцев и древних славян, могло также использоваться гадателями, желавшими проникнуть в замыслы предначертателей судьбы. В империи Хаттусы подобная гадательная практика приняла облик оракулов KIN, тогда как славянский источник сохранил лишь смутное упоминание о чертах и «резах». С другой стороны, связь между гаданием, чертом и защитной чертой хорошо отражена в русском фольклоре. Несмотря на неодобрение православной церкви, в сельской местности была широко распространена практика гадания из-за защитной черты, в особенности среди девушек, желавших увидеть своего суженого или узнать его имя. Андрей Трофимов (Москва / Марбург) любезно поделился типичным заклинанием, сопровождающим данный тип гадания, которое он записал в Каргопольском районе Архангельской области: Че́рти, по черте. По закругу три чёрта, в кругу нет никого16. Предполагалось, что чертей можно принудить выдать заветную информацию, если ритуал правильно совершался на сакральном пространстве за защитной чертой, однако при нарушении правильной процедуры ритуала черти могли навредить девушкам17. Из-за большого промежутка времени к любым параллелям между современными ритуалами и гаданиями древних славян следует относиться с осторожностью. Существенно, однако, что русский фольклор наделяет черта знанием будущего, что как будто бы не вытекает из других его известных свойств. Вполне вероятно, что данная особенность могла быть унаследована от его языческого предшественника. Для понимания того, как концепт предначертания судьбы мог независимо распространиться в Анатолии бронзового века и в юго-восточной Европе в эпоху раннего средневековья, следует напомнить, что оба региона развивались в тени письменных цивилизаций, соответственно, Месопотамии и Византийской империи. Языковые сообщества, не обладающие навыками письма, а лишь смутными представлениями о данной практике, легко могут ассоциировать его с волшебством. Классическим мотивом в поддержку данного наблюдения является история Беллерофонта, отправленного в Ликию с запечатанной табличкой, содержащей инструкцию по его собственному убийству. Данный эпизод является единственным упоминанием письма в Илиаде (Il. VI: 166—170), причем содержимое таблички определяется там как «злосоветные знаки» (σήματα λυγρὰ). Было бы неудивительно, если бы все письменные тексты воспринимались сходным образом в Данное заклинание было записано в этнолингвистической экспедиции РГГУ в 1995 году. Информантом была М. В. Крапивина (1918 г.р.) 17 Это не единственный пример применения защитной черты в восточнославянской магии. Как любезно напомнил мне М. Живлов, русская идиома очертя голову восходит к жесту древних воинов, проводивших условную защитную черту головы перед тем, как ринуться в бой. 16

287

И. С. Якубович

гомеровской Греции, где микенская слоговая письменность уже вышла из употребления, алфавит финикийского происхождения еще не приобрел популярности, но тем не менее имелись смутные представления об идее письма18. Сходным образом, поверхностное знакомство с идеей передачи письменных сообщений могло побудить как анатолийцев, так и славян приписать им магические свойства, а покровителями данного типа магии стали, соответственно, dGulses и *čьrtъ19. 9. Армянский чертежник. Обсуждаемый сценарий вызывает естественный вопрос о путях демонизации *čьrtъ’а. Можно предположить, что функция ангела смерти, естественная для божества судьбы, могла играть ключевую роль в данном процессе. Подобная функция уже упоминалась в контексте хеттских dGulses, а одна из греческих мойр, Атропос, специализировалась на перерезании нити жизни. Здесь, однако, уместно остановиться еще на одной параллели, которая иллюстрирует пейорацию божества судьбы в контексте христианской культуры. Согласно Расселу (Russell 1987: 305), армянская мифологическая фигура Groł, досл. «писатель» (ср. арм. grel ‘писать’), представляет собой адаптацию западноиранского бога Тира, покровителя писцов и помощника при толковании сновидений. Вероятно также, что он воплотил черты урартского бога d1uṭuini, если последнего действительно можно отождествить с богом судьбы (ср. Petrosyan 2007: 181)20. В средневековом армянском фольклоре Groł обладал четкими функциями: он записывал человеческие дела, забирал души тех, для кого настал час смерти, и передавал их на небо для суда21. Это была почтенная роль, которая привела к его отождествлению с архангелом Гавриилом в армянской придворной поэзии (Russell 1987: 302—303). Однако в современном армянском фольклоре Groł обладает совершенно иными коннотациями. Его боятся и ненавидят как ангела смерти, что и приводит к его частому появлению в проклятиях (ср. Russell 1987: 310). Два современных восточноармянских мягких ругательства, упоминающих Groł > Groγ находят прямые параллели в русском языке: арм. groγ tani ‘черт возьми’ и арм. groγ k4ez het ‘черт с тобой’. Данное сходство, возможно, вызвано контактным схождением 18 Возможным косвенным свидетельством негативных коннотаций индоевропейского корня *pei2- ‘рисовать, писать’ возможно, являются начальные строчки гимна Сапфо, посвященного Афродите, в котором содержится призыв к «бессмертной Афродите, обладающей расписным троном» (Ποικιλόθρονʼ, ἀθάνατʼ Ἀφρόδιτα) не подавлять дух поэтессы. Можно лишь задаваться вопросом, обладало ли греческое прилагательное ποικίλος ‘расписной’ пугающими ассоциациями в лесбийском диалекте древнегреческого. Я благодарен Н. Эттингеру (Эрланген) за данную параллель, предложенную с должной осторожностью. 19 Независимым доводом в пользу пассивного знакомства древних славян с письменностью до Св. Кирилла и Мефодия является германское происхождение общеславянского *buky ‘буква (ед.); книга, документ (мн.)’. Германские формы, обычно приводимые в этой связи, включают готское boka ‘буква (ед.); книга, документ (мн.)’ и древневерхненемецкое buoh ‘буква (ед.); книга (мн.)’. Данные германские существительные являются производными от названия бука, что обычно объясняется в контексте использования буковой коры для древнегерманских рунических и алфавитных надписей (Pronk-Tiethoff 2013: 80—82). Как отметил один из рецензентов, сами германские племена могли наделять руны магическими чертами, ср. др.-исл. rún ‘буква (ед.); магия (мн.)’. 20 Функции d3uṭuini не могут быть восстановлены по контекстам, хотя этимологически можно сравнить урартское абстрактное существительное 5uṭutuhi с позитивным значением (‘счастье, удача’(?)). Основным аргументом в пользу интерпретации 3uṭuini как бога судьбы является сходство его имени с хурритскими теонимами d3utena и d3utellura, которые, в свою очередь, принадлежат к функциональным эквивалентам хетт. dGulses, согласно текстам из Хаттусы (Otten 1971). 21 Армен Петросян (личн. сообщ.) напоминает мне о внутренней форме древнеармянского čakata-gir ‘судьба’, досл. ‘то, что написано на лбу’.

288

Славянский чертежник: этимология слав. *čьrtъ ‘черт’

между армянской и русской эксплетивной идиоматикой, но такой процесс едва ли был бы возможен без предварительного отождествления Groγ’а и черта. Тот же маршрут семантической пейорации можно предположить и для славянского Чертежника. Из всех событий человеческой жизни, предопределяемых судьбой, славяне, вероятно, ассоциировали с чертом самое последнее. Не исключено, что это было результатом синкретизма между чертом и иным божеством, относящимся к миру мертвых, возможным когнатом лит. vélnias22. Тем не менее, роль мифологического ангела смерти едва ли более способствует улучшению репутации, чем роль палача в этом подлунном мире. От нее был один лишь шаг до отождествления с воплощением зла, что и произошло с приходом христианства. Единственное, что осталось от старой функции черта — это, возможно, его связь с гаданием. 10. Заключение. Сравнительная мифология является областью, где строгие доказательства едва ли возможны. Хотелось бы, однако, надеяться, что предлагаемая этимология праслав. *čьrtъ обладает двумя ключевыми преимуществами перед альтернативами, обсуждаемыми в начале статьи. Во-первых, она постулирует производящую основу *čersti / čьrtǫ ‘чертить, бороздить’, которая хорошо засвидетельствована в славянских языках и прозрачный дериват со значением ‘чертежник’, который хорошо согласуется с правилами славянской морфологии. Во-вторых, все постулируемые гипотезы относительно религии славян находят типологические параллели в других индоевропейских традициях. Обозначение бога судьбы как ‘чертежника’ обнаруживает полную параллель в анатолийских языках и приблизительную параллель в армянском, тогда как предполагаемое развитие бога судьбы в демоническую фигуру прямо повторяет эволюцию армянского Groł. Некоторые альтернативные этимологии для праслав. *čьrtъ также можно счесть формально безупречными, другие являются семантически удовлетворительными, но ни одна из них не обладает обоими качествами. Поэтому предлагаемая реконструкция славянского чертежника является наилучшим доступным решением.

Литература Березович, Е. Л., Л. Н. Виноградова. 2012. Черт. Славянские древности, под ред. Н. И. Толстого. Том 5: 519— 527. М.: Международные отношения. Дукова, Уте. 1984. Праслав. *čьrtъ ‘черт, злой дух’ / герм. skrat- ‘лесной дух, черт. Этимология 1982: 61—3. М.: Наука. Черных, П. Я. 1994. Историко-этимологический словарь современного русского языка. 2­е изд. М.: Русский язык. Иванов, В. В., В. Н. Топоров. 1974. Исследования в области славянских древностей: лексические и фразеологические вопросы реконструкции текстов. М.: Наука. Раденкович, Л. 2014. Черт и/или межевой. Наименование беса «черт» у славян. Wiener Slawistisches Jahrbuch (Neue Folge) 2: 152—162. Расторгуева, В. С., Дж. И. Эдельман. 2000—. Этимологический cловарь иранских языков. Т. 1—. М.: Наука. Топоров, В. Н. 1974. Славянские комментарии к нескольким латинским архаизмам. Этимология 1972: 3—19. М.: Наука. Трубачев, О. Н. (ред.). 1974—. Этимологический словарь славянских языков. Т. 1—. М.: Наука. В этой связи уместно упомянуть демонизацию славянского бога Велеса / Волоса в ряде традиций, в особенности в древнечешской (Иванов, Топоров 1974: 66). Можно предположить, что частичный синкретизм черта и Велеса имел место уже в дохристианский период, вне зависимости от того, принимать ли этимологическую связь между праслав. *welsъ и лит. vélnias. 22

289

И. С. Якубович

References Archi, Alfonso. 1974. Il sistema KIN della divinazione hittita. Oriens Antiquus 13: 113–144. Berezovich, E. L., L. N. Vinogradova. 2012. Chert. Slavyanskie drevnosti, pod red. N. I. Tolstogo. Tom 5: 519—527. Mocsow: Mezhdunarodnye otnosheniya. Bezlaj, France. 1977. Etimološki slovar slovenskego jezika. Vol. 1: A–J. Ljubljana: Slovenska akademija znatnosti in umetnosti. Brückner, Aleksander. 1918. Über Etymologien und Etymologisieren. Zeitschrift für vergleichende Sprachforsachung 48 (3/4): 161–229. Chernykh, Pavel J. 1994. Istoriko-etimologicheskij slovar sovremennogo russkogo jazyka. 2nd ed. Moscow: Russkij jazyk. Dukova, Ute. 1984. Praslav. *čьrtъ ‘chert, zloj dukh’/germ. skrat- ‘lesnoj dukh, chert’. Etimologija 1982: 61–3. Moscow: Nauka. Fraenkel, Ernst. 1962–65. Litauisches etymologisches Wörterbuch. 2 vols. Heidelberg: Winter. Haas, Volkert. 1994. Geschichte der hethitischen Religion. Leiden: Brill. Ivanov, Vyacheslav V., Vladimir N. Toporov. 1974. Issledovanija v oblasti slavianskikh drevnostej: leksicheskije i frazeologicheskije voprosy rekonstrukcii tekstov Moscow: Nauka. Jacobson, Roman. 1959. Marginalia to Vasmer’s Russian etymological dictionary (р-я). International Journal of Slavic Linguistics and Poetics 1959 (1/2): 266–277. Katičić, Radoslav. 2003–2004. Uz rub rječnika hrvatskoga kajkavskoga kniževnog jezika. Folia Onomastica Croatica 12–13: 259–274. Kloekhorst, Alwin. 2008. Etymological dictionary of the Hittite inherited lexicon. Leiden: Brill. MacKenzie, David N. 1971. A Concise Pahlavi dictionary. Oxford: Oxford University Press. Otten, Heinrich. 1971. Gulš-Gottheiten. Reallexikon der Assyrioligie 3: 598. Petrosyan, Armen. 2007. State pantheon of Greater Armenia: earliest sources. Aramazd (Armenian Journal of Near Eastern Studies) 2: 174–201 Pirart, Éric. 2004. L’éloge mazdéen d’ivresse : édition, traduction et commentaire du Hōm Stōd. Paris: L’Harmattan. Pronk-Tiethoff, Saskia. 2013. The Germanic Loanwords in Proto-Slavic. Amsterdam: Rodopi. Radenković, Ljubinko. 2014. Chert i/ili mezhevoj. Naimenovanie besa «chert» u slavjan. Wiener Slawistisches Jahrbuch (Neue Folge) 2: 152–162. Rastorgujeva, Vera S. & Dzhoj. I. Edelman. 2000–. Etimologicheskij slovar iranskikh jazykov. Moscow: Nauka. Russel, James A. 1987. Zoroastrianism in Armenia. Harvard University Press. Rix, Helmut et al. (eds.). 2001. Lexicon der indogermanischen Verben. 2nd ed. Wiesbaden: Reichert. Smoczyński, Wojciech. 2007. Słownik etymologiczny języka litiewskego. Vilnius: Vilniaus Universitetas. Taracha, Piotr. 2000. Ersetzen und Entsühnen. Das mittelhethitische Ersatzritual für den Grosskönig Tuthalija (CTH *448.4) und verwandte texte. Leiden: Brill. Toporov, Vladimir N. 1974. Slavianskije kommentarii k neskolkim latinskim arkhaizmam. Etimologija 1972: 3–19. Moscow: Nauka. Trubachev, Oleg N. (ed.). 1974–. Etimologichskij slovar slavianskikh jazykov. 32 vol. Moscow: Nauka. Vaillant, André. 1974. Grammaire comparée des langues slaves. Vol. 4. La formation des noms. Paris: Klincksieck. Vasmer, Max. 1953–1958. Russisches etymologisches Wörterbuch. 3 vols. Heidelberg: Winter. Waal, Willemijn. 2014. Hittite gulš-/GUL-š­, Dgulšeš/DGUL-šeš, Cuneiform Luwian gulzā(i)-/GUL-zā(i)­, Hieroglyphic Luwian REL-za- and the Kuwanšeš-deities. In P. Taracha and M. Kapelus (eds.). Proceedings of the Eighth International Congress of Hittitology, Warsaw, September 2011: 1016–1033. Warsaw: Agade. Yakubovich, Ilya. 2013. The Luwian Deity Kwanza. Aramazd (Armenian Journal of Near Eastern Studies) 8 (1–2): 282–297.

Ilya S. Yakubovich. The Slavic draughtsman: etymology of Slav. *čьrtъ ‘devil’. The paper addresses the origin of the Slavic word for ‘demon, devil’, which can be reconstructed as *čьrtъ but does not have obvious cognates elsewhere in Indo-European. Using typological parallels from other Indo-European religious traditions, notably Anatolian and Armenian, I aim to demonstrate that Proto-Slavic *čьrtъ can be interpreted as a supernatural 290

Славянский чертежник: этимология слав. *čьrtъ ‘черт’

Draughtsman in charge of determining human fate. The original god of destiny probably came to be perceived as the bringer of death and then syncretised with the embodiment of all evil in the Christian tradition. Keywords: etymology, devil, Slavic, Anatolian, Armenian.

291

Mikhail A. Zhivlov Russian State University for the Humanities; School for Advanced Studies in the Humanities, RANEPA (Russia, Moscow); [email protected]

The origin of Khanty retroflex nasal * Proto-Khanty is traditionally reconstructed with a retroflex nasal phoneme *ṇ, whose origin remains disputed. According to one theory, it is directly inherited from Proto-Uralic. The other theory holds that Proto-Uralic *n, usually preserved as *n in Khanty, sporadically yielded *ṇ. We argue that Proto-Khanty *ṇ results from a regularly conditioned sound change. Keywords: Ugric languages, historical phonology, retroflex consonants, Proto-Uralic.

1. Introduction Khanty is unique among branches of the Uralic family in that it has a retroflex nasal phoneme /ṇ/ (often called “cacuminal” in Uralistic literature), distinct from both alveolar /n/ and palatal(ized) /ń/. The Proto-Khanty opposition of alveolar vs. retroflex vs. palatal(ized) nasals is reconstructed on the basis of the following correspondences: PKh

V

Vj.

Vart.

Likr.

Mj.

Trj.

J

Irt.

Ni.

Š

Kaz.

Sy.

O

*n

n

n

n

n

n

n

n

n

n

n

n

n

n

*ṇ

ṇ

ṇ

ṇ

ṇ

ṇ

ṇ

n

n

n

n

ṇ

ṇ

n

*ń

ń

ń

ń

ń

ń

ń

ń

ń

ń

ń

ń

ń

ń

Proto-Khanty also had a similar triple opposition in the lateral series: alveolar *l, retroflex *ḷ and palatal(ized) *. Together with the retroflex affricate * (usually transcribed as *č), inherited from Proto-Uralic, *ṇ and *ḷ form part of a tightly integrated consonant system of Proto-Khanty:

stops

labial

alveolar

p

t

affricates

retroflex

palatal(ized)

k č

ć

n

ṇ

ń

lateral approximants

l

ḷ



trill

r

sibilant fricative

s

lateral fricative



nasals

glides

m

w

velar

j

ŋ

γ

Each of the five places of articulation could also be used to form homorganic clusters of nasals and stops (or affricates): *mp, *nt, *ṇč, *ńć, *ŋk (Honti 1999: 105). Unlike most other clus* I am grateful to Anna Dybo for comments that have helped to improve this paper. Any remaining mistakes are my own responsibility. Journal of Language Relationship • Вопросы языкового родства • 14/4 (2016) • Pp. 293—302 • © The authors, 2016

Mikhail A. Zhivlov

ters, these homorganic clusters could never be broken up by epenthetic schwa (Nikolaeva 2000). Proto-Khanty had a morphophonological rule whereby *ṇ plus *t yielded *ṇč (Honti 1999: 98). Also, proto-Khanty *t and *n, on the one hand, and *č and *ṇ, on the other, could not co-occur within the same stem (Helimski 2002). The existence of a retroflex nasal in Proto-Khanty and the absence of anything similar in other Uralic languages raises the issue of the origin of this phoneme. There is one group of cases where this origin is immediately clear: Proto-Uralic *nč always yields Khanty *ṇč (note that *č is a retroflex affricate both in Khanty and Proto-Uralic), while Proto-Uralic *nt is regularly preserved as Khanty *nt. However, there are many instances of inherited words with Khanty *ṇ in other positions as well, so the question of its origin remains unsolved. From the early 20th century, there were two principal answers to this question in the Uralistic literature. The first was formulated already by Kustaa Fredrik Karjalainen, who had discovered the triple opposition of coronal nasals during his fieldwork on Khanty. According to Karjalainen, this opposition goes back to at least Proto-Finno-Ugric (Karjalainen 1913–1918: 6). He shows that when Khanty has *ṇ and all other Finno-Ugric languages, including Mansi, have n, Hungarian has ny. Thus there are not two, but three series of regular correspondences: Khanty

Hungarian

other Uralic

n

n

n

ṇ

ny

n

ń

ny

ń

Karjalainen also tried to find traces of original *ṇ in other Uralic branches, notably Permic, but in order to do this, he needed to assume that in certain cases original *ṇ yielded Khanty n for no apparent reason (Karjalainen 1913–1918: 28–30). In his “Comparative Grammar of the Uralic Languages” (1960), Björn Collinder, following Karjalainen’s theory, reconstructed both *n and *ṇ (*ñ in his notation) for Proto-Uralic. However, for some reason, he failed to mention that Hungarian has distinct reflexes of these consonants (Collinder 1960: 73, 133–134), thus losing the most powerful argument in favor of his reconstruction. Karjalainen’s theory was also adopted by V. M. Illich-Svitych, who projected the reconstruction of the retroflex nasal back to Proto-Nostratic (Illich-Svitych 1967: 323; 1971: 150), and by E. Helimski (1985: 75). All the abovementioned scholars were following the Neogrammarian paradigm, prohibiting unmotivated splits in historical phonology. The alternative theory of the origin of the retroflex nasal in Khanty stems from scholars working outside the Neogrammarian paradigm. Formulated already by Erkki Itkonen (1957), it was most succinctly summarized by László Honti: The cacuminals (*ṇ *ḷ) are the result of a secondary development within pO: they originally occurred as allophonic variants when adjacent to the (non-distinctively) cacuminal affricate *č, then spread, beginning with affective and descriptive vocabulary, to other positions, where they became phonemic. (Honti 1998: 337)

The secondary origin of Khanty *ṇ was accepted by P. Sammallahti, although his wording is more cautious and does not necessarily suggest sporadic sound change: An additional change in the consonantal system was caused by the ‘split-genesis’ of retroflex /l͔/ and /n͔/. At least /n͔/ was originally a contextual variant of /n/ before /č/, but for /lč/ no reliable etymologies can be found. (Sammallahti 1988: 512) 294

The origin of Khanty retroflex nasal

So far, all scholars either projected the Khanty retroflex nasal back to Proto-Uralic, or were willing to accept an unconditioned phonemic split in the prehistory of Proto-Khanty. As far as we know, nobody has tried to explain the origin of Khanty *ṇ in terms of a conditioned split of reflexes. Below, we will try to show that no sporadic sound changes need to be invoked in order to solve this problem.

2. Rules accounting for Proto-Khanty *ṇ The following rules can be postulated for the development of Proto-Khanty *ṇ from ProtoUralic *n. Rule 1: PU *nč > PKh *ṇč. •

We will not list any examples for this rule here, since it is self-evident and uncontroversial. Rule 2: PU *kVnV > PKh *kVṇ

•

• •

PU *kana- ‘to dig’ > PKh *k ṇ- > V Vj. VK Likr. Trj. kṇ­, Mj. J kn­, DN Kam. KoP Kr. Ts. χen­, Fil. χeń­, Kaz. Sy. χṇ­, O χan- ‘to dig; to scoop’ (DEWOS: 508; UEW: 125, Sammallahti 1988: 545). Cf. PMs *kūn- > TJ kōn­, KU χūn­, P kūn­, So. χūn- ‘to scoop’; Hung hány‘to throw’. The word is ultimately borrowed from Proto-Indo-Iranian *khan- ‘to dig’. PU *kunV ‘belly’ > PKh *kuṇ > V Vj. VK Vart, Likr. Trj. kŏṇ, Mj. J kŏn, DN-Sal. Fil. KoP Kr. Sog. Ts. Ni. Š O χŏn, Kaz. Sy. χŏṇ ‘belly’ (DEWOS: 509–510; UEW: 208). PU *ka/oni ‘on one’s back’ > PKh *kuṇ-čāγ > V Vj. kŏṇčaγ, Trj. kŏṇʿγ, J kŏnγ, DN χŏnča, KoP χŏnč, Kr. χănča, ­, Ni. Š χŏnša, Kaz. Sy. χŏṇša, O χŏns ‘on one’s back’ (DEWOS: 514; UEW: 179). Cf. PMs *kān- > TJ TČ kanāw, KU χonī, P kōnγ, So. χni ‘on one’s back’; Hung hanyatt ‘on one’s back’. One might think that in this word a trivial development *nč > *ṇč has taken place, but in reality what we are dealing with here is the Proto-Khanty morphophonological rule (already mentioned above), according to which, *ṇ plus *t on a morpheme boundary yields *ṇč. The adverb *kuṇ-čāγ ‘on one’s back’ contains the adverbial suffix *­tāγ found also in the following words: PU *kuma- ‘face down; to turn over’ > PKh *kom-tāγ > V Vj. komtaγ, Trj. J kămtγ, DN χŏmta, DT χămta, KoP χămt , χŏmt , Kr. χămta, ­ , Ni. Š Kaz. χŏmta, O χămt ‘prone, facedown’ (DEWOS: 502; UEW 201; Sammallahti 1988: 537); PU *perä ‘behind’ > PKh *pir-tǟγ > V Vj. prtäγ, Trj. J prʿγi, DN KoP Kr. ­prt , Kaz. Sy. părta, O prt ‘backwards’ (DEWOS: 1220; UEW 373; Sammallahti 1988: 553).

•

•

PU *künä- ‘elbow’ > PKh *küṇč- > Ni. Š kŭnš-tŋ, Kaz. kŭṇš-ŋ, O kuns-alŋ ‘elbow’; PKh *küṇč-ŋǟj > V Vj. Likr. kṇʿŋi, VK Vart. Mj. J knŋi, Trj. knʿŋi, DN Fil. Koš. kŏšŋj, Tš. kŏšŋaj, KoP Kr. Ts. kŏŋnj ‘elbow’ (DEWOS: 647; UEW: 158; Sammallahti 1988: 544). Cf. PMs *kün> KU koänγ, So. konl-wl ‘elbow’; Hung könyök ‘elbow’. Although not everything in this etymology is clear (the Proto-Saamic reflex *kerńēlē is especially hard to explain), at least the Mansi form shows the reflex of *­n­. Khanty *č may well result from the aforementioned morphophonological change *t > *č; but the identity and function of the suffix remain unknown. PUg *knk ‘light’ (adj.) > PKh *käṇγ > Mj. k!ṇγ, J knγ, DN KoP Kr. Ts. Ni. Š ken, Kaz. k"ṇ, Sy. keṇ, O kon ‘light’ (adj.) (DEWOS: 648; UEW: 862). Cf. PMs *kinγǟ > KO kiγn, LU kinna ‘light’ (adj.); Hung könnyű ‘light’ (adj.). 295

Mikhail A. Zhivlov

•

•

•

•

PKh *käṇγ-1 > Vj. könγ­, köṇγ­, Trj. k!nγ­, k!ṇγ­, J knγ­, KoP Kr. knj­, Kaz. k"ṇi­, Sy. keṇi- ‘to growl (of bear, dog)’ (DEWOS: 648; UEW: 856). Cf. PMs *kīnγ- > KM kenγ­, So. kēnγ- ‘to growl (of bear)’. Comparison with Hung könyörög- ‘to pray’, accepted in UEW, seems improbable. The word can be onomatopoeic. Still, it shows the expected development of *n. PU *kaŋa-la ‘armpit’ > PKh *kVṇŋ > V kunŋ-pt$, VT kuṇŋ-pt, Vj. kunŋ-ptä, Trj. kunŋpt$, J kunŋ-pt, DN χŏnŋ-pt, DT KoP Kr. χănŋ-pt, Ni. χuŋn-păt, Š χuŋn-ĭtp, Mul. χuŋn-păti, Kaz. χǫŋṇ-păti, χǫṇŋ-păti, Sy. χŏṇŋ-păt, O χŏnm-p't ‘armpit’ (DEWOS: 515– 516; UEW: 178; Sammallahti 1988: 543). Cf. PMs *kanl > TJ TČ kalnā, KU χånl, P kanl, So. χanl ‘armpit’; Hung hón (dial. hóny, hany, hóln) ‘armpit’. Here the Proto-Uralic form apparently had *­ŋ­, but forms in daughter languages suffered various assimilations and metatheses. At least the Khanty word goes back to something like *kanVŋV­. The Mansi word reflects *kanVlV, while the Hungarian form goes back to *kalnV. Exceptions: PKh *kānŋ > V Vj. kanŋ, Trj. J kånŋ, DN KoP Kr χonŋ, Ni. Š Kaz. χnŋ ‘bank (of a river); edge (of a forest, shawl etc.)’ (DEWOS: 514–515; UEW: 124–125). This word is compared in UEW to PMs *kant- > KU ­χant, KM ­kant ‘near, close to’, Hung (dial.) hany ‘swamp’ and Komi-Permyak (dial.) kan ‘side’. These words are not especially close semantically or morphologically. The dubious nature of the comparison is recognized in UEW, where all parts of it are supplied with question marks. The most obvious connection of the Khanty word is with Proto-Selkup *k͔ank ‘bank (of a river)’ (Alatalo 2004: 289). The Selkup word is thought to be a Khanty loan (UEW: 124–125), but the only reason for this is the supposed Finno-Ugric ancestry of the Khanty form. If the Finno-Ugric etymology is erroneous, the direction of the loan can be reversed. The Selkup word, in its turn, has been compared to similar words in Yeniseian languages: Kott hanaŋ ‘shore’ and Pumpokol kónnoŋ ‘mountain’ (Helimski 1982: 249). According to S. Starostin, “Pump. kónnoŋ (despite Helimski КС 249) should be distinguished from Kott. hanaŋ ‘shore’, which — as rightly pointed out by the author — is a Uralism (Selk. qan ŋ, Khant. χonŋ ‘shore’)” (Starostin 2005). The Pumpokol form is compared by Starostin to Ket qaŋńeŋ — plural of qaʔj ‘mountain (wooded)’ < Proto-Yeniseian *qäʔj. It seems that the only reason to treat the Kott form as a Uralism is once again the supposed inherited nature of the Khanty word. If this premise is erroneous, we have two available options. One is to compare Pumpokol kónnoŋ ‘mountain’ with Kott hanaŋ ‘shore’ and abandon the idea of a connection between the former and the Ket plural. Then we can reconstruct Proto-Yeniseian *kanaŋ 2 (consonant correspondences are regular, Starostin 1982: 148, 160, 162; as for vowels, we would expect Pumpokol ⟨a⟩, but it is not quite clear to what extent we can rely on transcription of vowels in 18th century sources). Another option is to accept Starostin’s etymology of the Pumpokol word and take the Kott word to be a loan from Pumpokol (Kott initial h- regularly goes back to *k-, Starostin 1982: 160). In both cases, the Selkup word can be considered a Yeniseian loan that was further transferred to Khanty. Either scenario seems preferable to accepting the traditional Uralic etymology of the Khanty word. PKh *kn- > Trj. J kăn­, DN DT KoP Kr. Ni. Kaz. Sy. O χăn- ‘to stick (to), adhere (to) (intr.); to touch, move’; PKh *kan-t- > V kont­, Vj. kont­, Trj. kŏnt­, Ni. χunt­, Kaz. χǫnt- ‘to stick 1 2

296

Here and below we do not provide PU/PUg reconstructions for exclusively Ob-Ugric words. Reconstructions *knŋ, *kanŋ and *knaŋ are also possible.

The origin of Khanty retroflex nasal

(to), to glue (tr.)’ (DEWOS: 504–505). Cf. PMs *kan- > TČ kan­, KU χån­, P kan­, So. χan- ‘to touch; to hang (intr.); to stick (to) (intr.)’; PM *kan-t- > KM kånt­, P kant- ‘to hang (tr.)’. This is a genuine exception, but the root is exclusively Ob-Ugric. If (as we argue below in section 4) the shift *n > *ṇ has taken place in Proto-Ugric times, a word that was borrowed from an unknown source into Proto-Ob-Ugric need not be subject to this sound law. Rule 3: PU *…kVn(V) > PKh *…γVṇ •

•

•

PU *ikin > PKh *ǟγṇ > V VK äγṇ, Vart. Likr. Trj. γṇ, Mj. J γn, DN Fil. KoP Kr. ŋṇ, Ni. Š aŋn, Kaz. Sy. Pit. aŋṇ, O ŋn ‘chin; lower jaw’ (DEWOS: 43–44; Sammallahti 1988: 541). Cf. PMs *īγn > TČ īn, KU iγn, P jēn, So. ēŋn ‘chin’; Hung íny ‘gum; palate’. PKh *kiγṇ > V Vj. Trj. kγṇ, VT J kγn, DN Kr. Ts. kŋn, Ni. Š Mul. kĭŋn, Kaz. kĭwṇ, Sy. kĭjṇ, O kijn ‘laces, strings (on clothes, shoes); button’ (DEWOS: 605–606). Cf. PMs *kīγn > TJ TČ kīń, KU kiγn, P kīn, So. kēŋn ‘button; string on clothes’. PKh *ćoγṇ > V Vj. VK Vart. Likr. +ăγṇ, Mj. +ăγoṇ, J +ăwn, KoP +ŋn, Kaz. śŭŋn ‘fist; knuckles’ (DEWOS: 1503–1504). Cf. PMs *ćaknī > TČ ćaχnē, KU śåχn, P śaχn, So. śaχni ‘fist’. In the last two cases a PU reconstruction is not possible, since the words are limited to Ob-Ugric. Nevertheless, the environment is the same as in the first case. Rule 4: PU *nVkkV (POU *nVkV) > PKh *ṇVk

•

•

•

• •

PU *nikkä- ‘to stick in, push’ > PKh *ṇik- > VK nk­, Likr. ṇk­, Trj. J nk- ‘to push (smb) lightly’; PKh *ṇikj- > Kaz. ṇăki- ‘to push, nudge’ (DEWOS: 984; UEW: 304–305). Cf. PMs *näk- > KM nǟk- ‘to push’. PKh *ṇēkī > V ṇiγ, VT niγ, Vj. ṇiki, Trj. J niki, DN Kam. KoP Kr. Ts. nek, Ni. năk, Kaz. ṇ"ki ‘trigger (of a trap), peg (for strings in a musical instrument), etc.’ (DEWOS: 986). Cf. PMs *nǟŋk > KM n1χ, P naχ (pl. naŋkt) ‘pintle, pivot’. If the P plural form naŋkt was created by analogy with nouns that have the alternation ­χ / ­ŋk­, the PMs form can be reconstructed as *nǟk, which would agree better with PKh *ṇēk. PKh *ṇ2k- > V nγ­, Vj. ṇγ- (ṇka), VK Vart ṇoγ­, Likr. naγo­, ṇako­, Mj. ṇok, ṇoko­, Trj. ṇok­, J nok­, DN Fil. KoP Kr. Tš. noχ­, Ni. Š nuχ­, Kaz. ṇǫχ­, Sy. ṇuχ­, O noχ- ‘to peck’ (DEWOS: 987–988). Cf. PMs *n ko- > TJ TČ nā͕k­, KU nāχ­, P nēko­, nēk­, So. nāχo- ‘to peck’. PKh *ṇāk- > Ni. nχ­, Kaz. Sy. ṇχ­, O naχ- ‘to limp’ (DEWOS: 988). Cf. PMs *nāk- > KU nōχ­, P nōk­, So. nχ- ‘to limp’. PKh *ṇik > Ni. năk, Kaz. ṇăk ‘joint’ (Ni.), ‘spell (of weather)’ (Kaz.) (DEWOS: 983–984). Cf. PMs *näk > TJ TČ KU P näχ, So. nak ‘joint’. Judging by its distribution in Khanty, the word can be a Mansi loan; but the correspondences are still regular and we list it here for the sake of completeness. Rule 5: POU *…nVγ > PKh *…ṇVγ

Here, in the only example that has a Uralic etymology, final *­γ is a suffix, added in (Ob­)Ugric times. So it makes little sense to formulate the input of the rule in terms of Proto-Uralic reconstruction. • PU *śni ‘tinder, bracket fungus’ > PKh *sǟṇγ > V ­sänγ, V Vj. VK ­säṇγ, Trj. sṇγ, J snγ, DN KoP Koš. Kr. Sog. Ts. sn, KoP snk, Fil. śn, Ni. san, Kaz. Sy. saṇ, O sn ‘bracket fungus’ (DEWOS: 1345; UEW: 494–495; Sammallahti 1988: 548). Cf. PMs *šīnγ > TJ TČ šīnw, KU šēnī, P šēniγ, So. sēniγ ‘bracket fungus’. 297

Mikhail A. Zhivlov

•

•

PKh *ǟṇγ > V Vj. äṇγ, Likr. Mj. Trj. ṇγ, J nγ, DN KoP Kr. n, Ni. Š an, Kaz. Sy. Pit. aṇ, O n ‘cup; plate’ (DEWOS: 114–115). Cf. PMs *ǟnī > TJ TČ ǟńī, KU 1n, P na, So. āni ‘cup, plate, vessel’. PKh *säṇγ > V Vj. VK seṇγ, Likr. Mj. Trj. s'ṇγ, J s'nγ, DN KoP Kr. Ts. Ni. sen, Kaz. s"ṇ, Sy. seṇ, O sen ‘nit; crab louse’ (DEWOS: 1345–1346). Cf. PMs *šǟnγ > TJ TČ šǟnw, KU š1nī, P šniγ, So. sāniγ ‘nit’. Rule 6: PU *mVnV > PKh *mVṇ

•

•

•

•

•

•

PU *muna ‘egg; testicle’ > PKh *maṇ > V Vj. moṇ, Trj. mŏn, J mŏn, DN Fil. KoP Kr. mun, DT măn, Ni. mun, Kaz. mǫṇ, O mon ‘penis; testicle’ (DEWOS: 935; UEW 285–286; Sammallahti 1988: 538). Cf. PMs *man > TJ man, KU mån, P mon, LO mon ‘testicle’; Hung (obs., dial.) mony ‘egg; testicle; penis’. PUg *minV- ‘to tear; to dislocate’ > PKh *mäṇm- > Ni. Š O menm­, Kaz. m"ṇm- ‘to tear off’ (DEWOS: 935–936; UEW: 870–871). Cf. PMs *mänmt- > TJ mänmt­, KU mänmt­, P mänmt­, So. manmt- ‘to tear’; Hung (dial.) ki-mënyül- ‘to be dislocated (of joint)’, kimënyít- ‘to dislocate (of joint)’. Exception: PU *meni- ‘to go’ > PKh *min- > V Vj. VK Sur. Irt. mn­, Ni. Š Kaz. Sy. măn­, O m'n- ‘to go’ (DEWOS: 931–932; UEW: 272; Sammallahti 1988: 538). Cf. PMs *min- > TJ miń­, KU P So. min- ‘to go’; Hung mën- ‘to go’. See section 3 below on the possible cause of this exception. Forms with unexpected PKh *ṇ 3 PU *enä > PKh *äṇǟ > V Vj. VK eṇ$, Vart. Likr. Trj. 'ṇ$, Mj. 'ṇ$, 'n$, J 'n, DN KoP Kr. en ‘thick; big’ (DEWOS: 109–110; UEW: 74–75; Sammallahti 1988: 541). Cf. PMs *jänγ > TJ jinw, KU jäni, P jäniγ, So. janiγ ‘big’. Cf., however, another derivative from the same root: PKh *änm- > V Vj. VK enm­, Likr. Mj. Trj. J 'nm­, Irt. Ni. Š enm­, Kaz. "nm­, Sy. enm­, (rarely) eṇm- ‘to grow’. It is possible that the Mansi word with ­γ directly reflects the Proto-Ob-Ugric form, while in Khanty ­γ was secondarily replaced with another suffix. If so, both the presence of *ṇ in the adjective and its absence in the verb can be explained by Rule 5. PUg *pᴕn$- ‘to fart’ > PKh *pṇ > DN păn, Kr. păn­, Ni. pŏn, Kaz. pŏṇ, O păn ‘a fart’; PKh *paṇγ- > V Vj poṇγ­, Trj. pŏṇγ­, J pŏnγ­, DN KoP Kr. păn­, Ni. punij­, Kaz. pǫṇi- ‘to fart’ (DEWOS: 1169; UEW: 413). Cf. PMs *pnγ > TJ ponχ, KM pånγ, P ponγ, LO ponγ ‘a fart’; Hung fing ‘to fart’. The verb has a regular *ṇ (Rule 5), whereas the noun apparently acquired the retroflex nasal under the analogical influence of the verb. PU *niwa- ‘to remove hair from skin’ > PKh *ṇaw- > Kaz. Sy. ṇw­, O naw- ‘to remove hair from reindeer hide’ (DEWOS: 1024; UEW: 306; Sammallahti 1988: 546). A genuine exception.

3. The problem of secondary ny in Hungarian Hungarian parallels to Khanty words listed above demonstrate the validity of the correspondence between Khanty ṇ and Hungarian ny, discovered already by Karjalainen (1913–1918: 3 We list here only those words that have parallels in Mansi or other Uralic languages. Actually, many other Khanty words without external parallels also follow the rules formulated above.

298

The origin of Khanty retroflex nasal

24–25). Below we will separately list all reliable etymologies with Hungarian ny going back to PU *n. These etymologies are not especially numerous, and it would be hard to formulate the conditions of the change, were it not for the fact that the same conditions are valid for the Khanty change *n > *ṇ. Rules 2, 3, 4 and 6, formulated above for Khanty, work for Hungarian as well. Rule 1 (PU *nč > PKh *ṇč) has no counterpart in Hungarian, since Hungarian has simplified all clusters of nasals and homorganic obstruents. The Hungarian reflex of PU *nč is r (Aikio in press). We do not have relevant examples for a possible Hungarian counterpart of Rule 5 (POU *…nVγ > PKh *…ṇVγ). The remaining rules have their counterparts in Hungarian.

• • • •

Rule 2: PU *kVnV > PKh *kVṇ, Hung h/kVny. PU *kana- ‘to dig’ > Hung hány- ‘to throw’ (UEW: 125, Sammallahti 1988: 545). Cf. PKh *k ṇ- ‘to dig; to scoop’, PMs *kūn- ‘to scoop’. PU *ka/oni ‘on one’s back’ > Hung hanyatt ‘id.’ (UEW: 179). Cf. PKh *kuṇ-čāγ ‘id.’, PMs *kān- ‘id.’. PU *künä- ‘elbow’ > Hung könyök ‘id.’ (UEW: 158; Sammallahti 1988: 544). Cf. PKh *küṇč‘id.’, PMs *kün- ‘id.’. PUg *knγ ‘light’ (adj.) > Hung könnyű ‘id.’ (UEW: 862). Cf. PKh *käṇγ ‘id.’, PMs *kinγǟ ‘id.’.

Rule 3: PU *…kVn(V) > PKh *…γVṇ, Hung …ny. We have only one example for this rule in Hungarian, so it could hardly be formulated without the Khanty parallel. • PU *ikin ‘gums’ > Hung íny ‘gum; palate’ (UEW: 80–81; Sammallahti 1988: 541). Cf. PKh *ǟγṇ ‘chin; lower jaw’, PMs *īγn ‘chin’. Rule 4: PU *nVkkV > PKh *ṇVk, Hung nyVk. Here the Hungarian part of the rule can also be illustrated by one example only. • PU *nokki ‘nape of the neck’ > Hung nyak ‘neck’, PSelk *nuku ‘nape of the neck’ (UEW: 328–329; Alatalo 2004, #1385). This etymology, rejected by Janhunen (1981) and Sammallahti (1988), can be rehabilitated if we compare it to two other cases with the same vowel correspondence: PU *soski- ‘to chew’ > PSelk *tutu- ‘id.’ (UEW: 448–449; Alatalo 2004, #1068); PU *totki ‘tench’ > Hung tat hal ‘id.’, PSelk *tutu ‘crucian carp’ (UEW: 532; Alatalo 2004, #1066).

While the details of development, especially the origin of PSelk *u in the second syllable, are not clear, the regularity of the correspondence is not in doubt.

• •

•

Rule 6: PU *mVnV > PKh *mVṇ, Hung mVny. PU *muna ‘egg; testicle’ > Hung (obs., dial.) mony ‘egg; testicle; penis’ (UEW 285–286; Sammallahti 1988: 538). Cf. PKh *maṇ ‘penis; testicle’, PMs *man ‘testicle’. PUg *minV- ‘to tear; to dislocate’ > Hung (dial.) ki-mënyül- ‘to be dislocated (of joint)’, kimënyít- ‘to dislocate (of joint)’ (UEW: 870–871). Cf. PKh *mäṇm- ‘to tear off’, PMs *mänmt- ‘to tear’. Exception: PU *meni- ‘to go’ > Hung mën- ‘id.’ (UEW: 272; Sammallahti 1988: 538). Cf. PKh *min- ‘id.’, PMs *min- ‘id.’. The exceptional behaviour of this verb may have something to do with its morphophonological peculiarity in Ugric. Helimski (1990: 64–66) has shown that Hungarian reflexes of Proto-Uralic verbs *meni- ‘to go’ and *woli- ‘to be’ build their present stems 299

Mikhail A. Zhivlov

with the same suffix *­ś- that forms present stems of Hungarian monosyllabic verbs of the shape CV-: mëgy- ‘to go’ < *men-ś­, vagy- ‘to be’ < *wol-ś­, ësz- ‘to eat’ < *se-ś- < *sewi-ś­, vësz‘to take’ < *we-ś- < *weγi-ś- etc. We can suppose that these two verbs became (or remained) monosyllabic already in common Ugric times. This can explain the exceptional preservation of *n: the final consonant of a monosyllabic stem was protected from change by the immediately following initial consonants of affixes, after which the unchanged *n was generalized to prevocalic position. It is especially important for the chronology of the change in question that this exception is common to Hungarian and Khanty. One more rule can be tentatively formulated for Hungarian only (Khanty has *n in the words in question):

• •

•

Rule 7: pre-Hungarian *nVl/r > Hung nyVl/r. PU *nüδi ‘handle’ > Hung nyél (acc. nyelet) ‘id.’. Cf. PKh *nül ‘id.’, PMs *näl ‘id.’ (DEWOS: 997–998; UEW: 304; Sammallahti 1988: 538). PUg *närk$ ‘saddle’ > Hung nyerëg (acc. nyerget) ‘id.’. Cf. PKh *n1γr ‘id.’, PMs *nǟγrǟ ‘id.’ (DEWOS: 996; UEW: 874). Judging by irregular vowel correspondences, the word for ‘saddle’, together with other horse-related terms, was borrowed separately by different Ugric languages in Common Ugric times from an unknown source. However, this does not explain the discrepancy in initial consonants. Since the position before ­r- in this word is reminiscent of the position before ­l- in nyél, and these are the only reliable etymologies where secondary ny in Hungarian corresponds to Khanty *n, we prefer to postulate a separate rule for these cases. The rule can be confirmed by one more etymology: PU *nara or *nora ‘spring’ > Hung nyár (acc. nyarat) ‘summer’, PS *nårå ‘snow crust; spring (season)’ (see Janhunen 1977: 98 for Samoyed data; the comparison of Hungarian and Samoyed words was suggested by A. Dybo 2007: 170). Two etymologies of the Hungarian word are discussed in EWUng: 1) an inner-Hungarian semantic development nyár ‘swamp’ > nyár ‘summer’ and 2) borrowing from Proto-Turkic *jār² ‘spring, summer’ (more precisely, from pre-Proto-Turkic *ńār²). The first etymology involves an improbable semantic shift: parallel cases adduced in EWUng demonstrate developments like ‘summer’ > ‘melted water’, ‘a place free from snow’, not vice versa. Borrowing from pre-ProtoTurkic is also highly dubious, since no certain traces of such a layer of loanwords are known (see additional arguments against this etymology in Dybo 2007: 169–170). In the earliest layer of Hungarian loanwords from Turkic languages, Turkic *j- is rendered by Hungarian gy- (Dybo 2007: 29–30).

4. Conclusion We can see that Hungarian shares with Khanty Rules 2, 3, 4 and 6 together with a common exception from Rule 6 — the reflex of PU *meni­. Moreover, despite the fact that most of the rules formulated above involve presence of a velar consonant, PU *nVkV yields *nV(γ/w) both in Khanty and Hungarian: PU *näki- ‘to see’ > PKh *nǖ(w)­, Hung néz­. Of the three possible explanations — completely independent parallel development, diffusion of rules between closely related languages and common inheritance from an intermediate parent language — the first one can be definitely excluded. The rules formulated above are 300

The origin of Khanty retroflex nasal

typologically uncommon, and the existence of a common exception betrays a historical connection between Khanty and Hungarian developments. “Diffusionist” account remains a possibility, but were it true, we would expect more differences between Hungarian and Khanty rules (cf. subtle differences between effects of RUKI-rule in Indo-European languages). The remaining possibility involves a common intermediate node on the Uralic tree. As far as we know, no one has ever suggested a Uralic subgroup that would include Khanty and Hungarian, but exclude Mansi. The development of nasals is hardly sufficient for the postulation of such a subgroup. The only reasonable version of a “genetic” account would involve acceptance of the Ugric node. Then we could formulate Rules 2, 3, 4 and 6 as having taken place between the Proto-Uralic and Proto-Ugric stages. Under this account, Proto-Ugric *ṇ that resulted from these rules became (marginally) phonemic as a result of analogical leveling (PKh *min­, Hung mën- ‘to go’) and, perhaps, common borrowings from unknown languages (PKh *kn­, PMs *kan- ‘to stick’). Later, Proto-Ugric *ṇ was preserved in Proto-Khanty, merged with *n in Proto-Mansi (cf. the fate of *ṇ in a number of Khanty varieties) and yielded ny in Hungarian.

Abbreviations for languages and dialects4 Hung — Hungarian PKh — Proto-Khanty

PMs — Proto-Mansi POU — Proto-Ob-Ugric

PS — Proto-Samoyed PSelk — Proto-Selkup

PU — Proto-Uralic PUg — Proto-Ugric

References4 Aikio, Ante (Luobbal Sámmol Sámmol Ánte). In press. Notes on the development of some consonant clusters in Hungarian. In: Sampsa Holopainen, Janne Saarikivi (eds.). Peri orthotētos etymōn. Uusiutuva uralilainen etymologia (Uralica Helsingiensia). Alatalo, Jarmo 2004. Sölkupisches Wörterbuch aus Aufzeichnungen von Kai Donner, U. T. Sirelius und Jarmo Alatalo. Zusammengestellt und herausgegeben von Jarmo Alatalo (Lexica Societatis Fenno-Ugricae XXX). Helsinki: Finnisch-Ugrische Gesellschaft. Collinder, Björn. 1960. Comparative Grammar of the Uralic Languages. Stockholm: Almqvist & Wiksell. DEWOS = Steinitz, Wolfgang. 1966–1993. Dialektologisches und etymologisches Wörterbuch der ostjakischen Sprache. Lfg. 1–15. Berlin: Akademie Verlag. Dybo, A. V. 2007. Lingvisticheskie kontakty rannikh tyurkov: leksicheskiy fond: pratyurkskiy period. Moskva: Vostochnaya literatura. EWUng = Benkő, Loránd (ed.). 1992–1997. Etymologisches Wörterbuch des Ungarischen. Budapest: Akadémiai Kiadó. Helimski, Eugen. 1982. Keto-Uralica. In Ketskiy sbornik. Antropologiya, etnografiya, mifologiya, lingvistika: 238–251. Leningrad: Nauka. Helimski, Eugen. 1985. Rez.: L. Honti, Geschichte des obugrischen Vokalismus der ersten Silbe. Sovetskoe finnougrovedenie 21(1): 63–76. Helimski, Eugen. 1990. Ugrica: etimologii s istoriko-foneticheskim podtekstom. Nyelvtudományi Közlemények 91: 63–68. Helimski, Eugen. 2002. Eine Regel der Konsonantenkompatibilität im Ostjakischen. In: E. Helimski und A. Widmer (Hg.). Wŭśa wŭśa — Sei gegrüßt! : Beiträge zur Finnougristik zu Ehren von Gert Sauer dargebracht zu seinem siebzigsten Geburtstag (Veröffentlichungen der Societas Uralo-Altaica 57). Wiesbaden: Harrassowitz: 111–116. Honti, László. 1998. ObUgrian. In: D. Abondolo (ed.). The Uralic Languages. London / New York: Routledge. Honti, László. 1999. Az Obi-Ugor konszonantizmus története (Studia uralo-altaica: Supplementum 9). Szeged: Universitas Szegediensis de Attila József Nominata. 4 We use standard abbreviations for Khanty and Mansi varieties, see DEWOS for Khanty and Steinitz 1955 for Mansi.

301

Mikhail A. Zhivlov

Illich-Svitych, V. M. 1967. Materialy k sravnitel’nomu slovaryu nostraticheskikh yazykov (indoevropeyskiy, altayskiy, ural’skiy, dravidskiy, kartvel’skiy, semitokhamitskiy). In Etimologiya 1965 (Materialy i issledovaniya po indoevropeyskim i drugim yazykam): 321–373. Moskva: Nauka. Illich-Svitych, V. M. 1971. Opyt sravneniya nostraticheskikh yazykov (semitokhamitskiy, kartvel’skiy, indoevropeyskiy, ural’skiy, dravidiyskiy, altayskiy). Vvedenie. Sravnitel’nyy slovar’ (b-Ḳ). Pod redakciey i s vstupitel’noy stat’ey V. A. Dybo. Moskva: Nauka. Itkonen, Erkki. 1957. Suomalais-ugrilaisen kantakielen äänne- ja muotorakenteesta. Virittäjä 61/1: 1–22. Janhunen, Juha. 1977. Samojedischer Wortschatz. Gemeinsamojedische Etymologien. (Castrenianumin toimitteita 17.) Helsinki. Janhunen, Juha. 1981. Uralilaisen kantakielen sanastosta. Suomalais-Ugrilaisen Seuran Aikakauskirja / Journal de la Société Finno-Ougrienne 77: 219–274. Karjalainen, K. F. 1913–1918. Beiträge zur geschichte der finnisch-ugrischen dentalen nasale. Suomalais-Ugrilaisen Seuran Aikakauskirja / Journal de la Société Finno-Ougrienne 30(24). Nikolaeva, Irina. 2000. The status of schwa in the phonological description of Northern Ostyak. Finnisch-ugrische Mitteilungen 23: 121–148. Sammallahti, Pekka. 1988. Historical Phonology of the Uralic languages with special reference to Samoyed, Ugric and Permic. In: D. Sinor (ed.). The Uralic Languages. Description, history and foreign influences. Leiden / New York: Brill: 478–554. Starostin, Sergei. 1982. Praeniseyskaya rekonstrukciya i vneshnie svyazi eniseyskikh yazykov. In Ketskiy sbornik. Antropologiya, etnografiya, mifologiya, lingvistika: 144–237. Leningrad: Nauka. Starostin, Sergei. 2005. Yenisseian etymology. Available: http://starling.rinet.ru/ [Accessed 27.11.2016]. Steinitz, Wolfgang. 1955. Geschichte des wogulischen Vokalismus (Finnisch-ugrische Studien II). Berlin: Akademie Verlag. UEW = Rédei, Károly. 1988–1991. Uralisches Etymologisches Wörterbuch. Budapest: Akadémiai Kiadó.

М. А. Живлов. Происхождение хантыйского ретрофлексного носового. Для прахантыйского традиционно восстанавливается ретрофлексный носовой согласный *ṇ, происхождение которого остаётся спорным. Согласно одной теории, он непосредственно унаследован из прауральского. По другой теории, прауральская фонема *n, обычно сохраняющаяся в хантыйском как *n, спорадически давала *ṇ. Мы приводим доводы в пользу того, что прахантыйский ретрофлексный носовой *ṇ является результатом регулярных позиционно обусловленных фонетических изменений. Ключевые слова: угорские языки, историческая фонология, ретрофлексные согласные, прауральский язык.

302