Relationship Versus Classification
The reason for what has been called in the previous section ‘relativity of proof’, is that the wrong question is being asked. If we ask whether A is similar to B, since everything in the universe is similar to everything else in some way, one never knows how similar something must be to something else (here, of course, one language or language group to another) to draw any conclusions. What is lacking is a standard of comparison, the standard being in these case other languages. But this already brings with it multilateral comparison, degrees of similarity and thus, when carried to its ultimate conclusion, a classification of languages.
Ironically, this point has often been seen, but not carried through with thoroughness or impartiality by opponents of particular hypotheses of classification. An interesting case in point is that of Yukaghir, in regard to the Indo European-Uralic hypothesis. Yukaghir was up to that time, and still is now lumped by many with other languages in Siberia spoken by small populations in a Paleo-Siberian group with the proviso that none of them are provably related to any other.
Paasonen in 1907, in order to refute the hypothesis that Indo-European was related to Uralic, pointed out there were at least as many resemblances between Uralic and Yukaghir as between Uralic and Indo-European and everybody knew that Yukaghir was not related to Uralic. The hidden syllogism of Paasonen’s reasoning might be stated as follows. (1) Yukaghir is not related to Uralic (2) Yukaghir shows at least as many resemblances to Uralic as Indo-European does. (3) Therefore, Indo-European is not related to Uralic. By calling attention to the numerous resemblances of Uralic to Yukaghir, Paasonen inspired Collinder (1940) and others to advance the notion that Uralic was related to Yukaghir. This, of course, does not logically exclude the idea of a relationship between Uralic and Indo-European, only it is a more distant one. For A and B to be related in this broader framework means that they are more similar in essential ways requiring genetic explanation than some other group or groups of languages. Thus, Indo-European and Uralic are more similar to each other than either is to Sino-Tibetan. At this point, we do not posit a relationship between Indo-European and Sino-Tibetan until a still larger picture emerges.
Once we proceed in this way, we end up with a taxonomy, a classification involving degrees of relationship, one in its abstract structure quite similar to that of a biological taxonomy. To make it complete we look for all the languages belonging with A at a particular degree of relationship. Such an entire group we traditionally call a linguistic stock and it corresponds to the biologist’s notion of a taxon.
An essential difference between the approach through relationship as against that through classification is that comparison in the former case might be called decontextualized. We compare only A and B as though no other languages in the world existed. Hence, no standard of comparison exists leading to the difficulties that have just been seen.
