Here follows a paper outlining one of the areas to be presented in the WAC5 session with this title. We hope that all participants in the session (or related sessions) can read papers posted here in advance of the Congress, so there will be more time for fruitful discussion of new ideas among participants. The following paper is Copyright © 2002 Yuri Berezkin. All Rights Reserved. More such papers are to come. For the WAC5 web site please click here.

     The primary technique for estimating shared inheritance vs. chance resemblance is to observe how commonly a motif which would be classified as belonging to the same type occurs on a world-wide scale, and how randomly distributed such occurrences are. The more commonly such a type occurs, the less value it has as evidence for specific historical connections. Some motifs which occur very widely, and thus may be treated as mostly "noise" in any statistical analysis designed to evaluate evidence, refer to rivalry, in-law problems, greed, power, some kind of figure in the moon (not further specified), etc.

     This is not a dichotomy, yes or no, but rather a matter of degree, and it is made more specific by noting patterns of occurrence geographically, and by reference to areas which are known to share many vs. few motifs. One can even "weight" observed similarities, giving greater weight to those motifs which are of rarest distribution. Such weightings could be used in a mathematical procedure to compute degrees of relationships of peoples in their stock of legends. A figure in the moon holding a water pail is specific enough that it does reveal some historical relationships, where a category requiring simply some figure in the moon does not. A very specific motif occurring at two widely separated places in the world which share almost nothing else specific will have little value as evidence, where the same match between two peoples who are not too distant may be very valuable in reconstructing shared inheritance or borrowing.

     These sorts of patterns can be displayed on maps using real geography. Once some historical relations have been established (family trees of legends, motifs, and cycles of legends; or plausible locations of borrowings, in many ways analogous to corresponding concepts in linguistics), then somewhat more abstract maps can also be used, in which adjacency reflects the degree of shared history at some time in the past, even if not exactly datable. Some such maps generated automatically by a database will be discussed.

     Cycles or clusters of legends sharing common actors or themes can also have revealing distributions, which help to evaluate the significance of the distributions of individual legends or motifs of which they are composed.

     Patterning of distribution of motifs at the margins of the area of that motif's primary distribution can reveal distinctions between borrowing and shared inheritance. Especially ways in which a motif is oddly transformed at such margins may be useful, just as some kinds of oddity in sound correspondences can be used to estimate the likelihood that spoken words are borrowed between languages.

     We can observe correlations with environment, in estimating whether similarities are due to shared inheritance or to other factors. Peoples who live primarily by fishing may have some motifs in common because of that, not because they ever shared parts of their history together. So may those who live in deserts. A bit more subtly, those who live at the polar extremes may share common sensibilities to light and darkness, to cold and warmth, and to seasonal cycles, because of how they live, not because they share parts of their history.

     We can consider some motifs as variants of each other. For example, in cycles of trickster legends, the identity of the trickster may respond to the zoological environment (Fox in Asia and in South America, vs. Coyote in the Southwest of the USA). These may or may not be historically related variants, and the degree to which there is sharing of the complex of many legends in a cycle of legends may influence our judgement here. Raven, with possible variants Loon, Jay and others, is another trickster. It appears that the Fox and Coyote cycles and the Raven cycle have historically distinct origins, that they have come into contact with each other and partially overlapped geographically. Wolverine as trickster is very restricted in the motifs where it occurs. Wily Rabbit is another trickster. Study of disjoint and overlappping distributions yields many additional clues.

     A database in which rows (motifs) and columns (peoples for whom the motifs are recorded) can be permuted and easily regrouped is very useful in exploring whether several motifs have approximately similar distributions across peoples or geography, or whether they are in near-complementary distribution, or numbers of other questions.

     Of course hypotheses of relationship derived from distribution of legend motifs can be compared with hypotheses derived from common genetic descent of languages (family trees), from culture areas, from archaeological areas, etc. As we compare these, we may learn the degree to which different sources of information permit us to probe deeply into history. It is of course possible that peoples retain their legends even when they switch to another language, which might mean evidence from legends can penetrate more deeply than evidence from languages in situations of borrowing and intimate cultural influence. This presentation does not make such a claim, but merely raises the question as one worth investigating.

Email to the Author Dr. Lloyd Anderson, Ecological Linguistics, PO Box 15156, Washington DC USA