As I mention in the "About" page, I'm currently working on a reconstruction of Proto-Jê, the common ancestor of the languages grouped under the Jê family (numbering around ten). Although similarities between the members of the family are generally easy to spot, a few cases prove a little more challenging (both phonologically and semantically).
One of the recently reconstructed Proto-Jê words, *prãm 'hunger', is inherited in Southern Jê (Kaingáng and Xokléng languages; the latter is now commonly referred to as Laklãnõ) with the meaning 'summer' (see Wiesemann's dictionary, for Kaingáng, and Bublitz's 1994 master's thesis, p. 47, for Laklãnõ). Although it is not hard to think of a plausible semantic connection between both concepts, some sort of independent confirmation is always welcome. In well-studied families, such as Indo-European, seemingly unlikely semantic connections can ultimately be proven to be true thanks to the existence of written documentation (in addition to the fact that the phonological correspondences are better established, etc.). In lesser-known families or stocks (such as Jê and Macro-Jê), such corroboration is much harder to come by. One has to look somewhere else for corroborating evidence.
Sometimes, one or more related languages provide the "missing (semantic) link". For instance, the likely cognate of the Proto-Jê word *kra 'offspring' in Karajá is ra 'nephew', a possibility that went unnoticed by Davis (1968, 'Some Macro-Jê relationships,' IJAL). The semantic connection between both is far from obscure. But, if one is still unsure, the fact that the cognate in Xavánte (Central Jê) means both "offspring" and "nephew" makes the plausibility of the hypothesis even more obvious. (By the way, the likely Karajá cognate for Proto-Jê *prãm is rǝma 'hunger', illustrating the same process of cluster simplification seen with ra 'nephew'.)
Sometimes the semantic change under consideration is documented elsewhere (a fact that corroborates the plausibility of the hypothesis). In our case in point ("hunger" > "summer"), for instance, there is in the Mataco family a root that shows a similar semantic scope, in which a season comes to be associated with 'hunger': "invierno, época de hambre" (from Verónica Grondona's handout "Algunos cognados en las lenguas dela familia mataca", presented at the 52nd International Congress of Americanists, Seville, 2006). Other pieces of evidence which may shed light on the possible historical circumstances underlying the semantic change may be found in the ethnographic literature. In our case, the following passage, from Jules Henry's Jungle People (1964, p. 6), on the Xokléng, may be elucidating:
"The Kaingáng [Xokléng] are hungry in winter and early spring. Then the tracks of animals are hard to interpret and the tapir can run far and fast, for it is no longer burdened with its young and the forest is cool. To the Kaingáng the tapir is not only the most important food, it is the very symbol of food. When they have no tapir meat there is very little meat of any kind, for the tapir is most plentiful when the wild fruits and nuts are ripe, and when these are gone the monkeys and birds, the rodents and pigs, the deer and the tapir that have fed on them for months, grow scarce or vanish altogether.[...] Summer, with its warmth, its dryness, and its plenty, brings comfort at last to these people [...]."
The only apparent problem with this theory is how to reconcile the fact that (in the available literature)
Águas de março nem sempre fecham o verão. In rural Brazil (at least Central Brazil, as far as I can tell), 'summer' is the lack of rain; 'winter' is the rainy season. The farmer and the weather person (as well as Tom Jobim!) may be talking about very different things with the "same" words.
Update (October 10, 2007, morning)
In response to a query I sent to the Etnolingüística list, Wilmar D'Angelis (UNICAMP), an expert on Kaingáng language and culture, provided me with very interesting insights into this issue (which prompted the corrections above, in
An afterthought (October 10, 2007, evening)
D'Angelis suggests that the fact that both "hunger" and "year" occur one after the other in Wiesemann's dictionary is in itself a good reason to consider them as (etymologically) connected. I guess it is quite the contrary: the fact that the author lists them in two different entries suggests that she saw them as a case of homonymy, not polysemy. This is one of those cases where the border between both categories is blurred -- and that's exactly what makes the kind of ethnographic data mentioned by Jules Henry (and D'Angelis himself) so important.