domenica 22 novembre 2015

Evolution of the Indo-European languages, phentics, cladistics and the scientific method

This video about the evolution of Indo-European languages deserves to be listened. It criticizes a paper on Science, based on the method of clustering analysis, which tries to find the origin of IndoEuropean languages. Although the criticism it presents is 95% (I will come back on this later), I found it rather irritating. The reason of this irriation became clear to me at minute 45.50 where says that the research on science is close to the method of creationsim, since, instead of looking for the theory explaining the facts, it looks for the facts which support the theory. But this is not the method of creationsim, it is the hypothetica-deductive method of science. The method of creationists is "I believe in hypothesi A, let's find the facts that falsify the alternative hypothesis", and the latter is exactly the approach of the video.

Going into more detail, there are two approaches for the reconstruction of the tree of languages. The historical linguistic or traditional approaches compares in detaill single words of the languages, whereas the clustering method compares thousands of words at the same time. The two are obviously complementary, but the supporters of historical linguistics (like the authors of the video) never accepted the numerical methods. Some criticism is nonetheless founded. The two approaches exist also in taxonomy: the numerical method is called phenetic, since it looks for overall similarity, whereas the historical linguistic method is called cladistic, and in the latter method only shared similarities, not overall similarity are considered (Actually cladistic was invested in linguistic earlier than in taxonomy). At minute 24.30 the difference between the two approaches is discussed without the author being aware, obviously, of the existence of the same problems in taxonomy. In fact, if two languages (or two species) are compared on the base of overall similarity and not strictly on shared similarity, they appear more related than they are actually: the overall similarity of dolphins and fishes is large, but they are completely unrelated. Cladistic methods have never been applied in linguistics, to my knowledge, and it could be ftruiful to try.

Concerning the dating of the origin of languages the criticism of the video is that the dates of the tree contradict a few historical facts and that it does not consider movement of people like the latins conquering the Mediterranean. It is sensible, but useless, since the only way to estimated the origin of languages or cluster of languages for which there is no historical record is to assume constant rate of evolution. It is a pire aller, but nothing better exists.

Nessun commento: