The relationship between different languages In the eyes of observers is defined by their similarity. The latter is defined by the quantity of similar elements, not all of which produce an equal impression of importance. So, for instance, the similar structure of languages may be omitted from one's sight by the fact of the phonetically different sounding starters, and the similarity of the phonetic system may be omitted from one's sight by the same dissimilarity. On the other hand, the similarity of starters, even with the condition of a different «structure» of language and a different phonetic system, may produce a greater impression than the differences. So, in this respect, the personal impression received by the observer, his personal ability of perception of sounds, etc., may be responsible for the intensity of impression of similarity. Yet the variety of impressions also depends upon the ideas of the observer. One may pay attention to the «verbs»; another, to the terms for «numerals»; and a third, to «cultural phenomena»; etc. But in all cases the chief idea is to find «similarity» and «dissimilarity.»
However, an absolutely erroneous inference is often made from the facts of similarities; namely, that similarity indicates a «common origin.» First of all, the similarity may be due to the convergence, as it is, for instance, with the phonetics, where the choice of possible elementary sounds is limited, and where, since the basis of articulation is the same (the variety of articulations is very limited, indeed), the similarity of elementary sounds is quite natural. The same may be stated in reference to the prevailing type of «structure» of language. In a lesser degree it may be referred to the «common» starters. Second, the similarity may be due to the spreading over of a certain territory of certain phonetic fashions, or the «structure of language,» and especially starters. The number of common elements of these origins may cover from zero to the totality of the language. In the last case, two similar languages may produce the impression of being originally the same language, although both of them may arrive at a certain similarity by a gradual accumulation of elements from a third source. Such a common origin may exist, and it does exist, as the transmission of one complex from one to another ethnical unit or from one to another generation.
The classification of languages into groups is simply a method of grouping facts — complexes of language — into larger headings for better memorizing, and it must not imply conclusions as to the «origin» of these languages, for the similarity of languages as ethnographical complexes is not correlated with the origin of the complex as a whole. It is a function, and as such it can have no origin. Indeed, we can and must speak about the origin of a people as a physical body, but we cannot treat it in the same sense as the origin of a language, which is actually a simple metaphor. But what has happened with the classification of languages according to their similarity is that an inference by analogy has been made: since they are similar, they have originated from a pra-language-ancestor.
The elements into which a language is divided may be analyzed and the «origin» of some of the elements may indeed be established. So the whole complex of language may be dissected into its elements and there will always remain some elements which cannot be connected with any of the other existing linguistic complexes. In the process of analysis one may find, in a group of languages, the elements A, which are connected with a known ethnographical and ethnical complex; the elements B, which may be connected with another complex; the elements C, which may be connected with a third complex; and finally there will remain elements D, which can be connected with no one of the existing nor the extinct languages. Will they be the original language? — Not at all. They will be merely unknown elements hypothetically referred to a certain ethnical unit, also hypothetical, and in hypothetical form. These elements D may be of the same various origin as are elements A, B, and C. Here, naturally, limits are put upon us by our knowledge of the «history» of language and the history of its bearers. At which moment shall we refer to the language as connected with a definite ethnical and ethnographical complex? It is an absolutely arbitrary choice, which, practically, is conditioned by other theoretical presumptions.