The hypothesis of common origin of the Altaic languages, and even the Ural-Altaic languages, dates from the first half of the last century. It has found many defenders and opponents. So the question in the opinion of outsiders is not settled at all. In fact, every one who tries to conciliate contradictory opinions and evidences, and to find a moderate, middle solution, is misled, for in this problem there is only one correct solution, while all other solutions will be proved sooner or later to be erroneous. We do not want to enter into discussion as to which group is wrong and which is right, for, as shown in Part I, the question of such a relationship between the languages cannot even be put to an answer, so that the evidences brought forth to prove or disprove, both limited in number and numerous, cannot be convincing either. So my approach of the problem is essentially different. The present part, devoted to the analysis of positive evidences supporting the idea of existence of a common origin of these languages, does not have in view the disproving of such an existence, but it has in view the showing of the process by which these evidences are produced and which linguistical value may have the parallels found.
I have taken the latest and most complete work by A. Sauvageot, who has used in his work a large number of previous publications dealing with the positive evidences for showing the «genetic» relationship between the Ural-Altaic languages. The controversy between the comparatists is great, indeed. There remain very few parallels which are not contested by one of them. I will not use ihis controversy for discrediting the idea and work — such a discrediting as will be later seen has little interest even from the linguistical point of view [67].
Although the work of A. Sauvageot, from the point of view of technique, as will be later demonstrated also, is not perhaps the best one, yet it has been approved and assisted by such authorities as Z. Gombocz and A. Meillet [68]. This may be considered as guarantee that the ideas developed are not discarded by these authorities and that A. Sauvageot is not alone in his position. Indeed, in principle his position does not differ from that of all recent contributors to the hypothesis of common Ural-Altaic languages, as, for example, P. Pelliot, P. P. Schmidt, and a large group of linguists headed by G. Ramstedt. Yet, historically, this idea began nearly a century ago with the works of M. A. Castren, followed by a brilliant group of linguists of the same century, so that A. Sauvageot's study ought to be regarded as a collective work.
Let us now restore the process of building up this theory in so far as one may see from this and some other works previously published.
A. Sauvageot begins his credo by putting emphasis on the weak points of the theory, which are: (1) the paucity of documents as to the history of languages; (2) the impossibility of basing one's self on the vocalic harmony formerly supposed to be one of the fundamental characteristics of the Ural-Altaic languages; (3) the regularity and the simplicity of the structure of all these languages, which is not characteristic of these languages only; (4) the insufficiency of evidence of pronouns. These objections, however, fall down in the front of the new considerations; namely, (1) these languages are spoken and were spoken in a geographical area sharply defined, (2) the present grouping of these languages is not different from that of the most ancient time, (3) to the closeness in territory corresponds unity of structure of language, so he asserts that these considerations «portent a conclure qu'il y a eu unite de filiation dans le temps. Les ressemblances de structure interne qui caracterisent les langues ouralo-altaiques forment donc une presomption en faveur de leur unite genealogique» (op. cit., p. xxi); and later on, after having reviewed contradictory data as to the vocalic harmony, the latter is stated to be present: «dans la plupart des langues de la Siberie et de l'Europe orientale» and «ceci constitue une presomption des plus graves en faveur de leur parente» (op. cit., p. xxvi). Another reason is the alternation of consonants and vowels definitely established for some languages. (Here he has in view, amongst others, «la loi de Ramstedt,» the loss of consonant p [φ] in Turk, Mongol, and Tungus, op. cit., pp. 3, 4). At last, the only really reliable method of establishing the relationship between these languages, according to him, is the finding of common stems, — more exactly, the consonants for the vowels are not definite, — and the semantic value of words are considered (op. cit., p. xxxvi). A. Sauvageot brings forth the significance of his labour in comparing a great number of words and he pretends «a mettre un peu d'ordre et de cohesion» and to look for «solutions precises,» so he hopes that «son effort produira au moins ce resultat que la question de la parente ouralo-altaique ne fera plus desormais 1'objet d'une polemique purement verbale, comme il arrive depuis plus de soixante dix ans. Les arguments precis que cet ouvrage presente au publique reclament des critiques precises» (op. cit., pp. xxxvii et seq.). After having examined two hundred and fourteen cases of parallels in the conclusion, he supposes that «des recherches ulterieures rendront certainement caduque la majeure partie de ces concordances. N'en subsisterait-il qu'une poignee, cell suffirait a prouver que les langues considerees ne sont pas etrangeres les unes aux autres» (op. cit., p. 139). The «laws» of alternations are shown, but in his eyes they have only a «statistical» meaning. Furthermore, since the usual way of refuting the common origin of words is their consideration as loan-words, he refers to the opinion of P. Pelliot («Les Mots,» op. cit., pp. 255, 261) who discredits Altaic parallels from Chinese by P. P. Schmidt and G. Ramstedt (op. cit., p. 255), and who believes that there is no reason for supposing the loan of words for common terms, as, for example, bouillir. A. Sauvageot supports him by his own remark, «Nous nous refusons a croire qu'un peuple emprunte les mots qui exprimcnt l'ensemble des notions primordiales» (op. cit., p. 141). The definition of «loan» has been given in the Introduction; namely, «Quand un mot se trouve dans la plupart des dialectes ouraliens, turk, mongol, etc.... il est difficile de croire a un emprunt, ou s'il y a eu emprunt, cet emprunt a toutes les chances de s'etre produit a une epoque si ancienne que nous sommes en droit de considerer un mot de ce genre comme un vocable indigene» (op. cit., pp. xxxi, xxxii).
With an evident satisfaction as to the results achieved, the author concludes by a remark concerning his senior colleagues: «Nous exprimerons donc en terminant le regret sincere que nous gardons de ne pas voir cette tache menee a meilleure fin par un de nos eminents confreres en linguistique ouralo-altaique» (op, cit., p. 142).
67. I do not want to miss this occasion for pointing cut that discrediting criticism of each other's work and parallels produces an impression that one's work, lacks a serious method. Why this is so will be better seen when the work of A. Sauvageot is analysed. It is inevitable, for it is the goal which implies the choice and treatment of the material and parallels.
68. I believe that A. Meillet, who is also quoted in the preface by A. Sauvageot, is not directly responsible for the final form of the work here analysed. If one compares the last publications of this very cautious comparatist and general linguist (it is sufficient to remember the cases where his ideas have been quoted in the previous part) with the ideas and methods used by A. Sauvageot, one may see at once that these two authors occupy quite opposite wings of linguistical movement, perhaps it would be better to say different places in the marching column, at the head of which A. Meillet occupies one of the leading positions.