Res: Res: [tied] Catalan Bruixa = witch

Chicago etymologist Joan Coromines.

So there I go. In his bruc (the heather plant) and bruixa (witch) entries -totalling nine double-column small-letter pages!- he expounds his tightly argumented solutions.

Catalan bruixa (Port. bruxa, Sp. bruja, Gascon broucho) would come from Celtic (rather Gallo-Latin) *broiksa / *bru:ksa, with three original (and probably consecutive) meanings: (a) heath(s) (dominated by broom-like plants) [or, more generally, uncultivated lands out of town], (b) woman in the heaths (assumedly in the Devil's company),  and  (c) "pagan" (originally a woman, soon everybody else, living in or visiting "the heaths"), exactly paralleling the three senses of German Heide (senses also approximated by English heath and heathen). The plant in question is the heather (German Heiderkraut), which in Catalan is bruc and also bruguera (French bruyère, Spanish brezo [<OS bruezo], Gascon bròc). The "witch word" (usually associated, remember, with brooms and the broomstick sweep, Latin scopa [>Sp. escoba, Cat. escombra], usually made out of [Cat.] bruc = Erica arborea) is plainly an –s– derivative of the name of the plant, which (Coromines argues) would derive from *broikos (and scattered European variants: bru:co- (attested), *brouco-, *brocio-, and *brocco-), the Gallo-Latin name of the specific plant called heather in English. He then explicitly opposes the (now outdated) suggested links with e.g. *bru:ska (cf. João) and *bru:siare (>Cat. abrusar, It. bruciare, =burn) and speculates a very plausible also-Celtic connection with Catalan brossa = French brousse (= weeds, underbrush, brushwood), a (French) dialectal form of which yields English brush (and brushwood).

As to the Celtic *broik-/*bru: k-, Coromines sees it as a Latin phonetic adaptation of Celtic *vroikos (cf. OIr. froech and [Gaulish goddess's name] Vroica), considering that Latin's most approximate equivalent of the (nonexistent) vr- cluster is br-, so that the substitution would be only natural. Also the -oi-/-u:- vocalism may reasonably be a mere approximation to Latin values.