I have taken the instance of the Goldi language as a basis for the discussion of the phenomenon of aspiration and bilabialization for two reasons; namely, the Goldi language shows a series of both aspiration and bilabialization of vowels and the Goldi language is always quoted as one in which the hypothetical *p is preserved. The aspiration and the bilabialization are so common in Goldi that they are observed in the same words; e.g., palga~halga — «the foot»; pemu~ximma — «the lip»; porro~xarrake — «auerhahn»; polo~xolo — «the trembling poplar.» Particularly in the Ussuri Goldi dialect, the aspiration seems to be stabilized and has x instead of p. The stabilization of the aspiration is characteristic of Goldi in general. It may be here noted that in good agreement with other Tungus languages affected by the aspiration, the practice of alteration of s into h and even x is very common in Goldi. On the other hand, the presence of the labio-dental tenuis spirant / and the bilabial media lenis spirant w is more frequent than in any other Northern Tungus dialect besides the occlusive p. These facts point to the meeting of some strong current of words with the initial vowels bilabialized and not altered into p witn another current of alteration of the initial f into p in foreign words. Thus, from the phonetic point of view, this language is a mixed one. I say intentionally «mixed,» for two phonetic currents of different origin are met with which are clear from the conclusions regarding the Goldi, Manchu, and Northern Tungus languages.
The analysis of the vocabulary reveals the same picture. By the side of the typical Northern Tungus elements there is a large number of typical Southern Tungus elements (Manchu and perhaps Nuichen) used parallel with the Northern Tungus words, but already modified according to the type of «aspirating» dialects. Besides these modified Southern Tungus elements there are many «loan-words» which have not yet been modified. Furthermore, this language includes a great number of words which are neither Southern nor Northern Tungus. Some of them, but not all, may be connected with the Gilak words. It ought to be noted that most of these mysterious words designate local phenomena, local animals and plants, special methods of fishing, etc.
The morphology of these dialects is also sometimes mixed with two systems — the Southern and Northern Tungus. In fact, the specimens gathered amongst the Sungari Goldi reveals a parallelism in using, for example, the Manchu suffix -xa (part, perf.) along with the Northern Tungus -che; accumulation of several suffixes for expressing complex relations (complex starters), as, e.g., jatka+-chi+ -du+i, which is typical of the Northern Tungus dialects; the verbal pronominal suffixes are partly preserved; the Northern Tungus suffix -du (locativus and dativus) is used instead of the Southern Tungus -de; wo (accusativus) instead of be, etc.; the pronominal declension is preserved [81]. On the other hand, the Olcha dialect [82] has no such typical Northern Tungus elements and generally morphologically it is much poorer than the Sungari dialect. Some differences are also seen when the Ussuri and Amur dialects are compared with the above-mentioned two other dialects.
The facts given above may suffice for stating that (1) the Goldi language cannot be considered as an entity; (2) it is a group of dialects in various degrees containing Southern and Northern Tungus elements, phonetic, lexic, and morphological; (3) it must be classified as a «mixed» language [83]. The history of the formation of the Goldi as an ethnical group may greatly help us in the understanding of the process of «mixing.» As I have already shown [84], the formation of the Goldi has taken a long time. They were originally a Northern Tungus group, or, even better to say, groups, speaking the Northern Tungus dialects affected by aspiration. In fact, words common for the Goldi, Manchu, and Northern Tungus met with the initial p, f, and 0~h~x denote such elements like «soul» «lip,» «palm of the hand,» «thumb,» «forehead,» «wound,» «trousers,» «hammer,» «to follow,» «downwards,» etc., while the words met with in Goldi with the initial p and in Manchu with the initial f, but unknown in the Northern Tungus, denote either local phenomena (e. g., animals and plants) or special phenomena unknown amongst the Northern Tungus. On the other hand, we meet with words with the initial consonants f, p, w, v in Goldi and Manchu which sometimes correspond with the different Northern Tungus words. Besides these cases, there are a few cases where the initial p of the Goldi corresponds to 0~h~x of the Northern Tungus dialects. This class of facts may be understood as due to the loss of the old Manchu words, or even in some cases to the formation by analogy through the spreading of the bilabial tenuis occlusive over the words with the aspirated initial vowels. This was a secondary phenomenon which might have occurred during the period when Goldi was already tinder a strong Manchu influence. From the above-mentioned facts, the origin of the initial p in Goldi may be better understood as a case of Northern Tungus dialects which have come into contact with the Manchu (also Nuichen) linguistical complex and partly borrowed it, first altering their original phonetic characters (the period when the Manchu-Nuichen f became the p of the Goldi) and afterwards, the change of the old phonetic complex without any essential change of the Manchu words (the period when Manchu f was preserved) [85]. The behaviour of the Goldi towards the Northern Tungus phonetic characters has also changed, for, at the present time, it treats the Northern Tungus phonetic peculiarities as foreign, although the old «aspirating» type is still preserved in many words of a probably early origin. In this respect, it is very interesting to know that they now reproduce very well the Northern Tungus words received without aspiration, although the same words in an «aspirating» dialect must be aspirated. It shows that the influx of the second non-aspirated Northern Tungus element does not meet with the phonetic modifications characteristic of the «aspirating» dialects.
As to the Northern Tungus dialects, they do possess the initial p in rare cases of words for which we have no evidences for showing that these words are loaned from the Manchu, Mongol, and Russian languages, also from some Paleasiatic dialects. These facts seem to show that the Northern Tungus is not hostile to the initial p, as it is hostile, for example, to the corresponding labio-dental tenuis spirant f. On the other hand, the phenomenon of transition from media lenes to tenues is not observed in the Northern Tungus dialects, but in Manchu Sp., which, in so far as the present problem is concerned, particularly affects the initial bilabial media lenis occlusive b, but no tendency as to the transition from spirants to occlusives in bilabials and labio-dentals is observed. The process of transition from media lenes to lenues is observed in Goldi, which makes a very great difference when Goldi is compared with Manchu Sp.
Considering all the facts and conclusions to which we have come, it may be formulated thus: (1) The Goldi language being a mixed one, it is not a typical representative of either the Northern Tungus or the Southern Tungus, and, in addition to this, it is greatly affected by some third elements, which, apart from Mongol, might be various Paleasiatic languages now extinct; (2) the frequent occurrence of the initial p in Goldi is not due to the preservation of this consonant, but to the phonetic reaction on an alien (Southern Tungus) phonetic system characteristic of the Northern Tungus dialects. Therefore the reference to the Goldi language as one preserving the original initial labial (bilabial) tenuis sound is foundless, for it is of a secondary origin.
These inferences are important for our analysis of the problem of the original initial sounds of many Tungus and alien words. Since the process of aspiration and bilabialization is going on in several Tungus dialects forming two distinct phonetic currents of fashion, and since there are no evidences for showing the loss of the tennis bilabial initial consonants, but, just the contrary, reveals their consolidation and spreading, the Tungus languages cannot be taken as evidence of a loss of the above-discussed initial consonants. What has been shown in the previous section and what may be supported by other evidences quoted from the Tungus is that, in a great number of cases, the appearance of the initial consonants — bilabial, labio-dental and glottal, spirant and occlusive, tenues and media lenes—is a relatively recent phenomenon, due to the processes of aspiration and bilabialization, conditioned, perhaps, by the increase of accentuation of other vowels of the words. This conclusion does not imply the hypothesis of the non-existence of the initial p in Tangus, as initial p's are known at the present time and so might have been also in previous periods of formation of the dialects. Yet this conclusion does not imply that all words with the initial f, p, h, x, w, and v must formerly have been words with initial vowels.
81. The above-quoted facts are taken from the material published by W. L. Kotwicz (cf. «Material for the Study of the Tungus Dialects»).
82. According to the material published by P. P. Schmidt, cf. Olcha.
83. The idea of the existence of «mixed languages» has been severely criticized, but the above-quoted facts seem to support it in full sense.
84. Cf. my «Northern Tungus Migrations,» and «Social Organization of the Northern Tungus.»
85. This is a process which is now observed in some Northern Tungus dialects in Manchuria. Some of them are under Manchu influence and some of them are under Mongol influence. Little by little they become familiar with the foreign phonetic system and do not alter the original (foreign) pronunciation.