In etymological research it is a standard procedure to group words in the languages studied into ‘cognate sets’, i.e., sets of words that are considered historically related to each other by descent from the same form in the common proto-language. Every once in a while, however, one encounters divergent cognate sets with ambiguous boundaries. The kind of situation I have in mind here is when words across related languages show a ‘family resemblance’ in terms phonological and semantical correspondences, but it is difficult to clearly delimit which forms exactly belong to the same cognate set, and whether one is dealing with one or two (or even more than two) distinct etyma. In this post I will discuss one such case: a set of words that can be phonologically traced back to a single Proto-Uralic shape *korpi(-), but with so widely diverged semantics that it is not evident at first sight that we are dealing with one single etymon in the proto-language.
In a paper published over 20 years ago (2002: 15–16) I discussed the etymology of a Samoyed noun with the meanings ‘aurora, northern lights’ and ‘large fire’, and argued for an etymological connection with Saami, Finnic and Mordvin verbs with meanings such as ‘scorch’ and ‘blaze’. I proposed the reconstruction of a PU word-root *korpi(-) on the basis of the following branch-level cognate sets:
- SaaL guorbbat ‘be ravaged by a forest fire (of land); go out gradually (of fire)’, SaaN guorbat ‘lose hair; become barren, overgrazed (of pasture); get burnt (e.g., of clothes in fire)’, SaaI kuorbâđ ‘get burnt’, Sk kuõrbbâd ‘go out slowly (of fire)’ (< PSaa *kuorpe̮-)
- Est kõrbema, Võro kõrbõma, Liv kuorbõ ‘get burnt, scorched, parched; dry up (of soil)’ (< PFi *korpe̮-); Fin korventaa, Ol korvendua, Veps korbeta ‘scorch, parch’ (< PFi *korbe̮-nta-)
- MdE kurvams ‘blaze’ (< PMd *kǝrva-); MdE kurvaźems, kirvaźems, MdM kǝŕväźǝms ‘light up, flare up’ (< PMd *kǝrvaźǝ- ~ *kǝŕvaźǝ-)
- NenF karp° ‘large fire (e.g., when a patch of land is set on fire in order to repel mosquitos)’, NenT xarp°, EnF kabu ‘aurora, northern lights’ (< PSam *kårpǝ̑); NenT xarpǝr- ‘blaze’ (< PSam *kårpǝ̑-r-)
On the other hand, in another later publication (Aikio 2014b: 8–10) I proposed the reconstruction of a homonymous Proto-Uralic noun *korpi ‘woods’ on the basis of the following Finnic, Mordvin and Mansi data:
- Fin korpi (gen. korven), Ol korbi (gen. korven), Veps korb (gen. korben), Est kõrb (gen. kõrve) ‘dense and damp old-growth forest, backwoods, desolate forest land’ (< PFi *korpi : *korpe̮-)
- MdE kuro, MdM kura ‘bush, shrub; bushes, thicket; row of houses along a street, part of a street in a town, part of town’ (< PMd *kurǝ). — The meanings ‘row of houses’, ‘part of town’ must be secondary; they apparently developed by metaphoric extension of “a cluster of trees or bushes” to “a cluster of houses”.
- MsT kʷārp, kārp ‘grove’, KL xōrǝp, KM KU P kōrp, VN VS LL kʷōrp, LU xōrp ‘woods, forest’, So xɔ̄rp-ur ‘dense old-growth forest with several species of trees’ (ur ‘hill, mountain, ridge’) (< PMs *kʷārpǝ)
I did not, however, discuss the possible connections between the apparently homonymous PU word-stems *korpi ‘forest’ and *korpi(-) ‘large fire; burn, blaze’. On the face of it, the meanings of the two words do not seem to have almost anything in common, so this would look like a case of coincidental homonymy. Nevertheless, upon closer inspection of Saami data an etymological connection appears likely. In Saami there are as many as three lexemes of different word classes that show a homonymous stem reflecting PSaa *kuorpe̮(-) < PU *korpi-:
- PSaa *kuorpe̮- (verb) ‘be ravaged by a forest fire (of land); become barren (of land, pasture); lose hair’ (> SaaL guorbbat ‘be ravaged by a forest fire (of land); go out gradually (of fire)’, SaaN guorbat ‘lose hair; become barren, overgrazed (of pasture); get burnt (e.g., of clothes in fire)’, SaaI kuorbâđ ‘get burnt’, SaaSk kuõrbbâd ‘go out slowly (of fire)’)
- PSaa *kuorpe̮ (noun) ‘land ravaged by a forest fire’ (> SaaS goerpe ‘accident’, SaaU guarbba, SaaL guorbba, SaaSk kuõrbb, SaaK kuurrb-jieʹmmn ‘land ravaged by a forest fire’ (jieʹmmn ‘land’). — The semantic development of SaaS goerpe ‘accident’ is unclear; a more archaic meaning is preserved in the derivative goerpesje-laante ‘land ravaged by a forest fire’ (laante ‘land’), the exact cognate of SaaSk kuõrbâš ‘land ravaged by a forest fire; barren pasture’ (< PSaa *kuorpe̮ś).
- PSaa *kuorpe̮ (adj.attr.) : *kuorpe̮-s (adj.pred.) ‘barren (of land); hairless’ (> SaaL guorbba : guorbas ‘barren, having no lichen (of pasture)’, SaaN guorba : guorbbas ‘hairless, having a poor coating of hair (of animal skin); barren (of land)’)
The meanings of many Saami forms are connected with the concept of ‘forest fire’, and this suggests that the reconstructed verbs *korpi ‘forest’ and *korpi- ‘(large) fire; burn, blaze’ could actually represent a single Proto-Uralic etymon. Both the verb and the noun stems are archaic in Saami: the PSaa verb *kuorpe̮- corresponds exactly to Est kõrbema, Võro `kõrbõma, Liv kuorbõ (< PFi *korpe̮-), and the PSaa noun *kuorpe̮ ‘land ravaged by a forest fire’ is evidently the cognate of NenF karp° ‘large fire (e.g., when a patch of land is set on fire in order to repel mosquitos)’ (and the corresponding Tundra Nenets and Enets nouns with the innovative meaning ‘aurora, northern lights’).
PSaa *kuorpe̮ and NenF karp° allow us to reconstruct a PU noun *korpi ‘burnt or burning land (?)’, and the reconstruction of a distinct homonymous noun *korpi ‘forest’ thus turns out to be problematic: it would be most unnatural to assume that Proto-Uralic had two homonymous nouns pertaining to different types of terrain with different vegetation. Instead, we must be dealing with a single etymon, and the question is how the meaning ‘forest’ has developed from earlier ‘burnt land, land ravaged by a forest fire’, or the like.
The most obvious solution would be to assume that the meaning first shifted to ‘complex early seral forest’, i.e. an ecosystem that first develops to occupy a potentially forested area after the previously existing forest has been destroyed by wildfire. Especially the meaning ‘bush, shrub; bushes, thicket’ attested in Mordvin could be very naturally derived from this. However, nowhere in Finnic do the attested reflexes of PFi *korpi refer to any kind of young forest, let alone one that has started to grow after a forest fire. Instead, the noun consistently denotes a quite different type of forest ecosystem: a dense, damp, and often boggy old-growth forest. While this could derive from ‘complex early seral forest’ via multiple steps of semantic change, there is another more attractive possibility. The nouns continuing PFi *korpi also have a strong connotation of the absence of human habitation; for this reason, Fin korpi was even used in old Bible translations in the sense of ‘desert’, a type of ecosystem and landscape for which no native word existed in Finnish. Thus, a meaning ‘backwoods, desolate forest land’ seems to have developed from earlier ‘barren land’ – this meaning is found in the Saami cognate, and even Est kõrb has in many dialects also the meaning ‘barren, sandy soil’, which can be interpreted as a vestige of the more original semantics of the noun. In proof of this, there are also many independent parallels for the etymological connection between nouns for ‘forest’ and ‘barren or desolate land’, including the following:
- Engl heath < OEngl hǣþ ‘heath, wasteland, wilderness’ (< PGerm *haiþī-) ~ Welsh coed ‘forest, wood, trees; shrubs; timber’ (< Proto-Celtic *kaito-)
- Engl (dialectal) wold ‘unforested plain, grassland, moor’ < OEngl weald ‘forest’ ~ Germ Wald ‘forest, woods’, ONo vǫllr ‘field, flat ground, meadow’ (< PGerm *walþu-z)
- SaaL vuovdde, SaaN vuovdi, SaaI vyevdi ‘forest, woods’, SaaSk vuʹvdd ‘forest, woods; area’ < PSaa *vuovtē < Pre-PSaa *awta < PGerm *auþa- (> ONo auðr ‘empty, void, desolate’) (Aikio 2006: 12)
- SaaL miehttse, SaaN meahcci, SaaI mecci ‘wilderness, wilds, desolate land’ < Fin metsä ‘forest, woods’