Owing to the violation of elementary principles of practical application of phonetic parallelisms, owing to the abuse of generalizations in reconstruction of stems and their reverse application, owing to the extension of semantic complexes and their reverse application, and lastly owing to the liberty in dealing with the material (Section 47), the chance of finding common words is very great..It is now evident how A. Sauvageot could find as many as 214 cases in which over 170 cases contain Tungus parallels [145].
Many particular cases of A. Sauvageot's methodology can be seen from the cases analysed. We may now summarize what has seen observed. In the Tungus parallels A. Sauvageot did not use all available material and sources, as, for example, I. Zaxarov's and E. Titov's dictionaries and several minor collections of lexic ma erial. In using old published material, he is not critical enough and accepts translations which are sometimes not close to the actual meaning of Tungus words. From a series of known meanings, he often makes a selection of meanings suitable for his purpose, and even violates the meaning of words given by the authors. He extends semantic complexes into the Tungus language where they are unknown. He postulates that the Tungus initial glottal and dental are preserved from the hypothetic pra-language. He takes for granted that the Tungus language has lost the initial p. He admits alternations of sounds in Tungus which cannot be justified by the facts. In some cases he does not pay attention to the presence of consonants in the Tungus stems. In some cases he arbitrarily cuts words into supposed stems and suffixes without being familiar with the Tungus etymology and morphology. In most of his cases he does not trouble himself with the task of finding whether the words found in the Tungus vocabularies are Tungus words or recent «loan-words» from other languages. He does not pay attention to the fact of the spreading of Mongol influence over Manchu and Northern Tungus dialects [146]. He takes no notice as to the position of the Goldi in the group of Tungus ethnical units [147]. He considers words found in dialects as belonging to the language where they have been recorded. He does not question himself as to the cultural complexes in which the linguistical complexes might or might not exist.
The enumeration of peculiar conditions of work which have been created by A. Sauvageot may be still extended over that above mentioned, but it will be perhaps too much for what I want to show, namely, why the finding of parallels is not satisfactory, and why after the analysis there are only two words left which are undoubtedly common to all languages here discussed. The «ensemble» of facts about which A. Sauvageot is speaking in reference to the Tungus language is a product of an artificial selection of badly chosen facts [148].
Now another question may be asked, — What was the reason of carrying out such a work, which, after a slight touch from the critics, goes to pieces and even compromises parallels which are not perhaps wrong at all? Yet how does it happen that such experienced linguists, as Z. Gombocz and J. Nlmeth [149], have omitted the most evident blunders and A. Meillet has omitted to notice the crying contradiction between the theories maintained by himself and the theoretical deficiency of A. Sauvageot's work? A. Meillet shows a great cautiousness in the matter of etymologies, semasiology, and phonetics when the question is about the Indo-European languages, but in the case of Tungus parallels A. Sauvageot allows himself such liberty that the dangerous effects may be foreseen by any general linguist who even might know no Tungus.
These questions may be answered by the supposition as
to the existence of a general cause for such an attitude on the part of the
persons responsible for the final form of the work published by A. Sauvageot, —
they wanted to prove a certain proposition, and this desire was so great that
it put secondary the problem of choice of technical ways for reaching the goal.
145. It ought to be pointed out that under these conditions it would be possible to bring still more — this is a question of leisure — but they would be useless as a new support of A. Sauvageot's position.
146. He frequently quotes a dialect (Dr., Castr.) which soon after having been recorded lost its Tungus lexic complex and was substituted by a Mongol dialect. Indeed, in the time of M. Castren, a great part of the words was already substituted by the Mongol elements without speaking of the ethnographical complex, which naturally, together with the terms, was already borrowed from the Buriats.
147. E.g., in 1926 already represented by me (cf. «Northern Tnngus Migrations. Goldi and Their Ethnical Affinities,» op. cit.).
148. As to other facts forming the «ensemble» of facts, namely, vocalic harmony, morphological similarities, etc., with the present advanced study into these phenomena, they cannot be considered as a proof of common origin of the Ural-Altaic languages which is also felt by A. Sauvageot.
149. I have already had occasion of dealing with the etymology of saman proposed by J. Nemeth.