Dear Log,
How (not) to
date languages
not surprised
jhi on 2003-12-16T15:13:19
The "glottochronology" has always smelled a bit funny to me, too... in addition to the problems mentioned in the blog entry (the "constant rate" being quite not so constant between rather close relatives like English and Icelandic, and using only the lexicographic selection as the measure of change), I can think of other problems:
(1) What is a cognate, really? Where does one draw the line? How about false cognates? (words that look like cognates but are not)
(2) Using only one word from each language for one 'thing' is dubious. There maybe be grammar/dialectal/stylistic/taboo reasons for using/not using certain words. The word lists for each language should be generated only by people very, very proficient in that language-- just using a dictionary is unacceptable. The less known the language is, the more important this aspect is.
(3) While the principle of using "core words" of a language, words that are unlikely to change, is a good one, I doubt that it will work that well. For example the names of animals and plants that are not indigenous to the speakers of a language are likely to be borrowed from languages that have words for those-- in which case we are talking of a cultural exchange, not a linguistic "sound change".
In summary, I would say that while "glottochronology" might be useful as a rough estimation tool, using it to make pronouncements like "8700 years ago" is foolish. Could I know the day of the week, too?
