Of the various Turkic languages, the North-Western types, Karachay, Balkar, and Kumyk, and the South-Western type, Azerbaijani, are near neighbours of the languages of North and South Caucasian stock, all of which are spoken within the congested area of the Caucasus. Summarising the researches of Baron P. K. Uslar, R. von Erckert, N. J. Marr, A. Dirr, and others, Prince N. Troubetzkoy (Trubeckoj) distinguishes three major types of languages in use there, exclusive of Turkic and Indo-European, viz. North-East, North-West, and South Caucasian. The North-East Caucasian, or Checheno-Lezginian type comprises the Chechen, Ingush, and Tsova-Tush (Bats) of the former Chechen-Ingush Autonomous Republic (now the Grozny province) and the Lezginian complex, consisting of Awaro- Andian (twelve languages spoken in North-West Daghestan), Dargwa, Kubachi, Lak (also called Kazikumukh), and Archi in Central Daghestan, and the Samur group, i.e. Lezgin proper, in its two principal dialects Kyuri and Akhti, Agul, Tabassaran, Budukh (Budug), Jek, Rutul (Mykhad), Tsakhur, Khinalugh, which is spoken in a single village, and Udi, which is the language of two villages, in South-East Daghestan. Some investigators, among them Marr and G. Dumézil, prefer to separate the Chechenian from the Lezginian types by treating the latter as more properly East Caucasian, and the former as Central Caucasian. AH these languages are structurally and lexically related among themselves and more distantly to North-West Caucasian, or Abkhazo-Circassian. To the last group belong the more southerly Abkhaz, with its dialects, in the Abkhaz Autonomous Republic (capital Sukhumi), which is federated with Georgia, and the more northerly Adyge, spoken in two forms—Kabardin, in the Kabarda Autonomous Republic (capital Nal'chik), and Circassian (Cherkess, Kyakh), in the Circassian and the Adyge Autonomous Republics (capitals Cherkessk and Maikop respectively). A third type of North-West Caucasian is usually listed, viz. Ubykh, which is now almost extinct along the north-east coast of the Black Sea and survives chiefly in Asia Minor.

South Caucasian (Iverian),the only well-articulated languagestock in the Caucasus, consists of Georgian and its congeners Mingrelian (Megrel) and Laz (Chan), now sometimes covered by the inclusive name Zan, and the rather more divergent Svanetian (Svan). These languages are spoken as varieties of a single structural type in the Georgian Federal Republic (capital Tbilisi, formerly Tiflis).

Immediately flanking Georgian on the north is Ossetic, the language of the North Ossete Autonomous Republic (capital Dzaujikau, formerly Vladikavkaz) and of the South Ossete Autonomous Province (capital Stalinir, formerly Tskhinval) in Georgia, which bestride the central Caucasus range, and flanking it on the south is Armenian, which is now largely concentrated in the Armenian Federal Republic (capital Erevan). Both Ossetic, with over 350,000 speakers, and Armenian, with over two millions, are Indo-European languages : Ossetic is a member of the Iranic branch and cognate with Persian and its dialects Talysh and Tat, which are spoken in Soviet Azerbaijan, as well as with Tajiki and the Pamiri dialects of the federal republic of Tajikistan (capital Stalinabad, formerly Dyushambe); Armenian stands isolated, like the Albanian, Hellenic, and Tokharian branches of Indo-European.

The connection of Ossetic (Os, Ir) with the language of the ancient Alans, a Sarmatian tribe mentioned in the sixth century by Jordanes, may be demonstrated circuitously by adducing the vocabulary it shares with Hungarian. It is well known that the Hungarians and the Alans were medieval neighbours and allies before the former finally migrated to and settled down in Slavonic Pannonia, i.e. the major Danubian plain. And the name ‘Alan’ has been equated with ‘Ārya’ and with the Ossetic ‘Ir’.

Ossetic resembles the geographically eastern types of Iranic, especially Sogdian and its modern descendant Yaghnobi, and displays a Caucasian taste for consonantal groupings, though these are not nearly so complex as in North and South Caucasian because of the presence of alleviating vowels. The chief dialectal, mainly phonetic differences are between West Ossetic (Digor) and East Ossetic (Tagaur), which is numerically the more important of the two and the basis of the literary language. The vowel system of both dialects is of a conservative Indo- European type, with front rounded vowels absent altogether. Length is a ‘suprasegmental’ phoneme and occurs in diphthongs as well as in pure vowels (e.g. ūi/ui and ī/i). The consonantal system contains uvulars (e.g. q) and has absorbed ‘ Caucasian’ ejectives, or ‘abruptives' (p', t', k',ts',tš'). Unlike the other Iranic languages which have mostly simplified their system of declension, Ossetic, as between its dialects, retains eight cases, including an interior and an exterior locative and a sociative (in Tagaur), and ekes these out with a large number of prepositions and postpositions.

As a linguistic mosaic the U.S.S.R. superficially resembles India. In either case we have a profusion of individual languages, several distinct groups or stocks distributed in a broadly similar fashion, a residue of isolated types, and an Indo-European (Aryan) lingua franca, whose lines of expansion run mainly eastwards and southwards. But apart from the ubiquitous Indo- European, none of the other stocks occurs in both countries: North Caucasian, for instance, is as typically Soviet as Dravidian is typically Indian. Bishop Robert Caldwell's correlation of Dravidian with Uralian (his ‘Scythian’) in 1856, still retained by J. L. Wyatt and T. R. Pillai in their revision of the third edition (1913) of his standard Comparative Grammar of the Dravidian or South-Indian Family of Languages, has rarely been taken seriously by competent linguists, and the more recent hypothesis propounded by the Hungarian scholar W. Hevesy, that Uralian is ultimately cognate with Kolarian (‘Mundā-Sprachen’), carries as little conviction as Father W. Schmidt's Austric theory, connecting Austroasiatic (including Kolarian) with Austronesian, which it is designed to refute.

The U.S.S.R. has five language stocks—Uralian, Altaic, North Caucasian, South Caucasian, and Indo-European—each with a genetic basis, and a language complex tentatively called Palaeoasiatic, which comprises related and evidently unrelated languages placed together on geographical grounds and because of a vague general resemblance in sound and structure. The linguistic pattern of the country may be represented geometrically as follows: a central wedge of Indo-European (Russian) divides a northern, Uralian zone from a southern, Altaic zone, and tapers off towards an eastern, Palaeoasiatic rim, while the mountainous isolation of the Caucasus in the south-west harbours two individual and territorially compact stocks—North Caucasian and South Caucasian—and, like the Pamirs farther east, preserves some ancient fragments of Indo-European.

The Palaeoasiatic languages are scattered over the Far Eastern littoral of the U.S.S.R., which reaches across more than twenty degrees of latitude from the intercontinental Bering Strait to Korea and includes the island of Sakhalin. They comprehend a knot of three languages—Chukcha (Luoravetlan), Koryak (Nymylan), and Kamchadal (Itel'men), constituting a single stock and spoken from Chukotka to Kamchatka; Yukagir (Odul), the mother-tongue of fewer than 500 persons in Northern Yakutia; Gilyak (Nivkh), used in North Sakhalin and on the lower Amur; Ainu, spoken in South Sakhalin, which was recently annexed from Japan; and the isolated language of the Yenisei Ostyaks (Kets) in Central Siberia. Some of these languages are separated from one another by intruders of Altaic stock (Yakut and Tungus), and all of them have been variously modified by contact with Russian. Ethnic and linguistic idiosyncrasy have been partly recognised by the Soviet administration, which has subdivided the north-easterly (Anadyr') thrust of the Khabarovsk region (kraj), where the Palaeoasiatic languages are in use, into a Chukcha (northern) and a Koryak (southern) national area (okrug), extending from inner Kamchatka to the Arctic Ocean.

Whether related genealogically or geographically, the Palaeoasiatic languages have common traits, some of which they share with North American (North Amerindian), others with Korean and Japanese. Such considerations have led the Latvian orientalist P. Šmits (Schmidt) to include the last two in his Palaeoasiatic group. These languages, he observes, prefer initial voiceless consonants to voiced ones; final t tends to become r in Yenisei Ostyak, Gilyak, Ainu, and Korean; and there is no f in Chukcha, Koryak, and Yukagir, and perhaps it was missing in the older forms of the Sakhalin vernaculars, Gilyak and Ainu. Like Eskimo and the Indian languages of North America (e.g. Algonkin), though not necessarily to the same extent, the Palaeoasiatic group shows a tendency to various degrees of incorporation, i.e. to crystallise the verb and its associates into a holophrase, or polysynthetic mass (cf. Chuk. gelvulin ‘he got a wild reindeer’ with Greenland Esk. naakisarppaait ‘they used to pity him’). Other peculiarities are: a pattern of vowel harmony, implying a classification of vowels into two contrasted series, the relative scarcity of words with initial voiced consonants, and a system of numeration based on finger and hand.

The transition from one type of Uralian to another is hardly easy, but the transition from Uralian to Altaic still appears to be insuperable. Altaic has three branches—Manchurian (Tungusic), Mongolian, and Turkic (Turanian). The relations between them are even less intimate than those between the three branches of Uralian, and Soviet scholarship prefers to keep them apart. About a thousand common words have been collected, but of these only some two hundred occur in all three branches, and all three possess a considerable stock of Chinese loan-words. Mongolia, or at least the immediate vicinity of the Gobi desert seems to have been the focus of dispersion, and the repeated invasions of China, India, and Europe, radiating from there throughout the Middle Ages, made the more settled peoples of Eurasia only too familiar with the aggressive vitality and the power, if not with the languages of their Altaic-speaking conquerors. To-day the Altaic languages survive in parts of China, Afghanistan, and Europe, but no longer in India.

The study of the comparative grammar of Altaic is rendered difficult by the lack of any information about the origins of the speakers (the history of Central Asia becomes less obscure only after the beginning of Turkic ascendancy in the sixth century); the extreme mobility of the Altaic peoples, who included the Turkic Huns, Avars, Pechenegs, Kipchaks, Cumans (Polovecians, Russ. Polovcy), and Khazars, and the Mongolian Tartars (more properly Tatars); the numerous recorded changes of language (e.g. the Kirgiz, according to Chinese sources, were located between the Ob' and the Yenisei in the first century B.C. and were not originally Turkic); the relatively close resemblance existing between the constituent members of individual branches (e.g. among Turkic languages, only Yakut and Chuvash are markedly aberrant); and the slowness of evolution of recorded languages, viz. Turkic from the eighth, Mongolian from the thirteenth century. The affinities between Turkic and Mongolian are more obvious than those between either and Manchurian. The phonetic features which all three have in common are—vowel harmony at various stages, a double series of velar plosives flowing out of the vowel harmony, the tendency to eschew initial voiced constrictives (continuants), the clarity and stability of vowels and the instability of final n/ŋ, the absence of consonantal grouping initially and finally, the nonexistence of long or geminated consonants, and proneness to open syllables.

Against the Palaeoasiatic zone, which follows the Pacific littoral, two genetic stocks abut in broad, horizontal bands, viz. the Uralian, stretching across tundra and taiga from the Baltic Sea to Yakutia, and the Altaic, stretching across steppe and sand from the Crimea to the relatively recent Tungus (Lamut) lands of the Sea of Okhotsk.

Since the earlier half of the nineteenth century, when Castrén, Hunfalvy, and Böhtlingk accepted W. Schott's ‘Uralo-Altaic’ hypothesis, correlating Uralian and Altaic genetically in virtue of shared vowel harmony, and since the later and more considered views of H. Winkler, who extended the linguistic limits of the complex to include Korean and Japanese, attempts have not been lacking (e.g. A. Sauvageot's) to establish a Uralo-Altaic unity, in spite of O. Donner's insistence on the obvious need for a preliminary detailed study of individual languages. Unfortunately, and notwithstanding the recent valuable work of Soviet scholars, the absence of an adequate comparative grammar of the two stocks, especially of Altaic, makes it impossible to go beyond superficial speculations, and the existing dichotomy of Uralian and Altaic must be allowed to remain. Centuries of geographical proximity have no doubt contributed something to the undeniable, if vague, likeness between them. Vowel harmony (synharmonism) characterises most languages of both stocks; there are widespread traces of the Uralian type of consonantal alternation (Stufenwechsel, Finn, astevaihtelu, Hung, fokváltakozás) discovered by E. N. Setälä the phonetic systems exhibit points of resemblance (e.g. the occurrence of rounded front vowels and the avoidance of initial consonant-groups); the morphological features include agglutination in various stages of development; there are similar personal pronouns; and some investigators have even collected a skeleton common vocabulary of words expressing rudimentary nominal and verbal notions, but excluding the numerals.

The Uralian languages of the U.S.S.R. separate into three branches—Samoyedic, Ugrian, and Finnic (Somian). The essential unity of the Samoyedic dialects or languages with Ugrian and Finnic was confirmed in epitome a generation ago by E. N. Setälä. It is obvious not only in structure but in vocabulary (cf. Yur. Sam. sawa ‘good’, jaha ‘river’, maņ, ‘I’, lb ‘bone’, jileş ‘to live’, tu ‘fire’, haļa ‘fish’, pa ‘tree’, ne ‘woman’, jaBeş ‘to drink’, Tav. Sam. ņim ‘name’, Ost. Sam. pelæek ‘half’, with the corresponding words in Finnish, viz.

A number of languages, including regional ones, are missing from this survey. The languages of the many national minorities, which constitute small colonies in town and countryside (see, 1927), do not strictly enter into our linguistic pattern and, even if admitted, would not alter its characteristic outlines. Among such languages are various Balkan types, Baluchi (Beluji) in Turkestan, the modified Indie of the Gypsies, and marginal languages like Asiatic Eskimo and Aleut, mainly on North-East Siberian islands, Kurdish in Transcaucasia, ‘Moldavian’ (a Rumanian dialect) in the Moldavian Federal Republic, Finnish in the Garelo-Finnish Republic, and Polish. These languages have their focus of characterisation outside the Soviet frontiers along with the two types of Germanic—-German and Yiddish, both of which are used by a considerable body of speakers inside the revised frontiers of the U.S.S.R. and exhibit distinct regional associations, the first with the former German Republic on the lower Volga and the partitioned and annexed East Prussia, the second with White Russia and the Ukraine and the artificially created Birobijan (Jewish Autonomous Province), north of the middle Amur in Eastern Siberia.

For ease of recognition and reference I have yielded to the inertia of tradition and reluctantly abandoned the innovations in nomenclature which I advocated in the epitome of this book ‘The Language Pattern of the U.S.S.R.’ (The Slavonic and East European Review, xxv, 65, London, 1947) and necessarily the Confucian epigraph with which I pointed and emphasised them. My position in this matter remains unchanged. I am still persuaded that, for the sake of clarity, terms like ‘ Somian’ (new to English) for ‘Finnic’, ‘Turanian’ for ‘Turkic’, ‘Iverian’ for ‘South Caucasian’, and, for the sake of brevity, terms like ‘Aryan’ (taken from Sweet by Jespersen) for the awkward and inanimate ‘Indo-European’ are eminently desirable. Everyone acquainted with English books on languages, especially those of Eastern Europe and Northern Asia, will, I think, admit that our linguistic name-giving stands in serious need of revision.

In most cases I mechanically reproduce the Russian name and spelling of a particular language (e.g. ‘Kirgiz’ for ‘Kirghiz’, ‘Uigur’ for ‘Uyghur’, etc.) and leave the name in its radical form, i.e. without the Anglo-Latin ending -ian or the Anglo-Greek -ic.