16 November 2011

Celtic *dra(:)wā 'ryegrass' (updated)

Celtic *dra(:)wā 'ryegrass' is found in dialectal French droue, druive as well as other similar forms in Gallo-Italic (e.g. Lombard droga). Brittonic languages (Welsh drewg, Breton draok, dreokreflect a suffixed form *drāwā-kā 'darnel (Lolium temulentum)', also found in Late Latin dravoca, continuated in Gallo-Romance *dravocāta > French dragée 'forage'1 (not to be confused with its homonymous 'bonbon' or 'tablet')2

IE cognates of the Celtic word are Germanic *tarwō, found in Middle Dutch tarwe, terwe 'wheat' and English tare 'weed; vetch' (also used in translations of the Bible to name a type of ryegrass, probably darnel), as well as Baltic *dirwā 'field', Slavic *derevna 'arable field, village'. From these forms, Indo-Europeanists reconstruct a protoform *derH2u- 'a k. of wild cereal', which in my opinion belongs to the lexicon of Mesolithic hunter-gatherers, with a semantic shift in some languages due to the adoption of agriculture in the Neolithic, a process called acculturation.

This word has cognates in Altaic: Turkic *dạrɨ-g 'millet', Mongolian *darki ‘brushwood’, Tungus-Manchu *daragan 'quitch; reed, cane', Korean *tār 'reed', pointing to the initial consonant being an ejective *t
1 The French word for 'darnel' is ivraie, from Latin ēbriaca (f.) 'intoxicated'. There's also dialectal Basque libraka (R), with agglutination (proclisis) of the definite article. 
2 X. Delamarre (2008): Dictionnaire de la langue gauloise, p. 147-148.

25 October 2011

Etruscan *netś-, *nethś- 'entrails'


From two bilingual inscriptions we know Etruscan netśvis, netśrac are respectively equivalent to Latin haruspex 'adivinator'1 and haruspīcina 'adivination'. This leaves us with an Etruscan word *netś-, *nethś- 'entrails', surely a loanword from Greek nēdús 'stomach, belly, womb'.

The etymology of the Greek word is itself problematic. I think this could be a Thracian loanword parallel to Lithuanian vidùs 'middle, interior' < IE *(d)ui-dhH1-u-, a compound from *(d)ui- '2' and *dheH1- 'to make, to put' whose Celtic and Germanic reflexes mean 'wood, forest, tree'2. This is also the origin of Latin dīvidō 'to divide' < *dis-widhH1-. 

Etruscan visc- 'center', attested in visc ame ren-s '(this) is the center of the hand', could be explained from an derivated form *(d)ui-dhH1-sk-. This word was in turn borrowed into Latin scus, vīsceris 'entrails.
1 Itself a compound from IE *g´hor- 'guts' and *(s)pek´- 'to look'.
2 This would imply forests were considered as a hinterland, an intermediate region between inhabited settlements. Interestingly, the Baltic reflexes of IE *medh-jo- 'middle' mean 'wood, forest, tree'.

15 October 2011

The Atlantidic substrate (updated version)

The German linguist Theo Vennemann1 is the proponent of a Western Europe substrate languages he calls Atlantidic, an Afrasian language akin to Semitic and which left loanwords in Germanic, for example *krabb- 'crab' (also Greek kárabos), which he relates to Semitic *ʕa-k’rab- 'scorpion'. 

However, this word is rather isolated within Afrasian, with no likely cognates except perhaps (and these by no means sure) in Chadic. But as both crabs and scorpions have large claws, IMHO this is cognate to PIE *ghreb(h)- 'to take, to seize' and possibly also to Kartvelian *k’rab-/*k’rap- (-e-) 'to gather'2.  

Other loanwords I've investigated myself are Celtic *gabro-, Latin caper 'he-goat', Germanic *xafra- 'buck'3 (with no plausible PIE etymology), which can be linked to Arabic ɣafr-, ɣufr- 'young of deer/goat', and also Latin aper, Germanic *ibura-, Balto-Slavic *weper- 'boar'4, which would correspond to Arabic ʕifr-, ʕufr- 'pig, boar; piglet'5. 

These evidence
would suggest that
Atlantidic was spoken in Neolithic Europe and whose linkings with Chadic would date back to the time the Sahara had a monsoonic climate (roughly between 6,000-3,000 BC).
1 An Indo-Europeanist formerly defensor of the glottalic theory. 
2 Wrongly linked by Nostraticists to words meaning 'fruit, harvest'.
Greek kápros 'boar' could have arisen from contamination with the other word.
4 The initial labial arises from assimilation to the following *u.
5 Possibly a specialization from the preceding root.

06 October 2011

Germanic *sajwi- 'sea, lake'

Germanic *sajwi- 'sea, lake' is a word with no PIE etymology1, although a relationship with Kartvelian *zoɣw- 'sea' seems likely. IMHO they could be related to NEC *ts’s’æ:k’wV 'sour, raw', reflecting the fact the sea is a mass of salty water.

By contrast, Starostin links these 'sea' words to Altaic *sjògu 'shallow (place)'2. This is probably also the origin of Germanic *sinkw-an- 'to sink', a word with no PIE etymology3 and which suggests a Mesolithic environment like Doggerland, a former landmass in the North Sea which in the last Ice Age bridged Great Britain to the continent.  

Although Doggerland gradually sunk by rapid post-glacial raising sea levels, actually becoming an island of much more reduced extension, according to recent research its actual disappearence seems to have been caused by a catastrophic event around 6200 BC, the Storegga Slide, a submarine landslide in the Norwegian Sea which triggered a huge tsunami.

Some authors have suggested the echos of this prehistoric catastrophe are the origin to the legend of Atlantis, but according to Plato's description IMHO this would mean the Pillars of Hercules were located at the Strait of Dover rather than at the Strait of Gibraltar3.
1 Nikolayev links it to Baltic *sī̂w-a-, *sīw-iā̃ 'juice'.
2 But IMHO Mongolian *siɣar 'dreg, sediment' doesn't belong here.
3 Worse yet, Greenberg links the Altaic word to IE *seik- 'dry' (Latin siccus).
4 In case of doubt, the British Empire holds one of the pillars at both places.

22 September 2011

PIE *dom- 'household'

PIE *dom- 'household' (e.g. Latin domus) is a noun-root whose genitive form *dem-s- is attested in the compound *dem-s-pot- 'master of the house' (e.g. Greek despótēs). Many Indo-Europeanists (e.g. Mallory-Adams) consider it to be derivated from *dem(h2)- 'to build', a root only attested in Greek (e.g. démō 'to build',  dómos, dõ:ma 'house') and Germanic (e.g. English timber). 

A part of the problem is that, unlike French, English doesn't make a clear distinction between the concepts of 'house' as a building (French maison) and 'household' as a dwelling place (French demeure). This explains why the French linguists Émile Beneviste1 and Pierre Chantraine2 think that PIE *dom- refers to the latter rather than the former.

IMHO, this suggests the original meaning was 'to remain, to live', thus linking it with Proto-Afrasian *dam- 'to live, to last, to sit', which AFAIK has no Eurasiatic correspondences.

In terms of linguistic archaeology, this points to the Neolithic, when (pre-)IE-speaking hunterer-gatherers learned farming techniques and thus became sedentary. By contrast, Afrasian speakers (mostly nomadic pastoralists) had a different lifestyle and didn't develop the household institution.
1 Le vocabulaire des institutions indo-européenes, vol. I, pp. 293-307.
2 Dictionnaire étymologique de la langue grecque, pp. 292-293.

18 September 2011

Latin pariēs 'wall'

Although some Indo-Europanists link it to Proto-Germanic *spar(r)e:n, *sparrio:n 'stake, beam', Latin pariēs 'wall' has no inherited PIE etymology. But IMHO it can be considered as a substrate loanword from PIE *Hwer- 'to close, to cover' (usually with various suffixes), a root from which I'd also derive Romance barra 'bar, barrier' and *berruculu- (Spanish berrojo, Gascon berrolh) 'a wooden or iron bar or bolt placed across gates on the inside'1.

This would be a PIE B root corresponding to Proto-Altaic *t`jù:ru 'to hold, to obstruct' and PIE A *twerH- 'to keep, to hold, to fence', found in Balto-Slavic and also the source of Latin obtūrō 'to block, to stop up' and Romance *at-tūrō 'to stop'. 
1 Which Coromines thinks it was an alteration (possibly by contamination with ferrum 'iron') of Latin verūculum, diminutive form of verūs 'roasting spit' < PIE *gwer(H)-u-.

25 July 2011

Initial labials in Basque

Unlike what we've seen in last post, the native result of Latin and Romance f- in Basque is /p/, lenied to /b/ in the standard variety at word-initial: 

Latin fāgu- 'beech' > romance *fago > Basque pago, bago id. (but also fago)
Latin fīcu- 'fig' > Romance *fīco > Basque p(h)iko, biku id. (but also fiko, fiku)
Germanic *fīlō 'file' > Spanish filosa 'sword, knife' > Basque piruxe 'scissors'
Latin fīlu- 'thread' > Basque p(h)iru, biru id. (but also firu)
Latin fōnte- 'source, spring' > Basque ponte, ponde 'baptismal font' 
Latin fōrma- 'mould' > Basque borma 'wall'
Latin fortia- 'strength' > Romance *fortsa >  Basque bortxa 'force, violence'
Latin fortis 'strong' > Basque portitz, bortitz 'hard, strong'
Romance *freca 'stroke' > Basque p(h)ereka id. (but also fereka)  
Latin fronte- 'forehead' > Basque boronde, boronte id.
Latin furca- 'pitchfork' > Basque urka 'pitchfork, gallows'

Also *w from Latin v- (and in some cases also b-) became /p/:

Latin bēstia- 'animal' > Basque piztia, pisti(a) 'beast, pest'
Latin brāca- 'trousers' > Basque prakak (plural)  'trousers' (but also frakak)
Latin gaudiu- 'joy' > Romance *gots [wots]1 > Basque poz, bo(t)z 'joy, happiness'
Latin vēsīca- 'urine bladder; blister' > Basque pu(t)xika 'urine bladder'
Latin viride- 'green' > Romance verde> Basque p(h)erde, berde id. (but also ferde)
Latin voluntāte- 'will, desire' > Basque borondate, boront(h)ate id.
Latin vulture- > Romance *wutre2 > Basque putre id. (but also futre)

By contrast, a minoritary set of words has a nasal /m/ instead:

Celtic *bakko- 'hook, (curved) stick' > Basque mako 'hook; pitchfork, prong; walking-stick, shepherd's staff', mak(h)ila, mak(h)illa 'stick' (diminutive)
Romance barca 'boat' > Basque marka id.
Late Latin persica- 'peach' > Basque mertxika 'apricot, peach' 
Latin pīca- 'magpie' > Basque mika id.
Latin portu- '(mountain) pass' > Basque mortu, maurtu 'desert' (but also portu)
Latin vāgīna- 'sheath' > Basque magina, magiña 'sheath, pod'
Latin vascellu- 'little glass' > Basque maskelu, maskillo 'little cauldron'

Even more oddly, in some dialectal words the initial labial becomes a dental /d/ (/l/ in the standard variety):

Romance banca 'small bench' > Basque lanka, lanke id.
Romance *pendiolu- 'hanging'3 > Basque dindullu 'earrings'
Latin verberāre 'to whip, to hit, to beat' > Basque dirdira 'sunshine reflection'
1 Spanish gozo < *gotso.
2 Asturian utre &lt.
3 Catalan penjoll.

20 July 2011

Initial pl-,fl- in Basque

As it's well known, Basque abhors "muta cum liquida" clusters and thus we've got boronde, boronte < Latin fronte- and p(h)ereka, fereka 'stroke' from Romance *freca < Latin fricāre 'to rub'. 

There's however a number of words where Latin pl-, fl- have given Basque l-: 

Latin plācet 'it pleases' > Basque laket 'pleasure'
Latin plānu- 'flat' > Basque lau(n), leu(n), legun 'flat, smooth'
Latin plūma- 'feather' > Basque luma id.
Latin flamma- 'flame, fire' > Basque lama 'flame, heat'
Latin flore- 'flower' > Basque lore id.

The evolution fl- > l-1 can also be found in Spanish words such as lacio < *flaccidu- 'flacid' and the personal name Laín < Flavinus. IMHO this could be explained if these labials were realized as a voiceless bilabial fricative [φ] like in Proto-Celtic2, where IE *pl- became *φl- and later l- in Celtic languages. Of course, we should expect the same evolution in genuine Celtic loanwords in Basque3:

Celtic *φlāro- 'floor' > Basque larre 'meadow'
Celtic *φletro- 'hide, leather' > Basque larru, narru 'skin'

IMHO, the realization of /f/ as [φ] (which in fact is the usual pronounciation in most of Northern Spain (also including Basque-speaking areas) would explain the aspiration of Latin f- as h- in Gascon and Spanish4.  

In Basque, this consonant disappears at word-initial, mostly before back vowels, where also /p/was realized as [φ]:

Romance *fīco 'fig' [φiko] > Basque iko id.
Latin fīlu- 'thread' [φilu] > Basque iru id.
Romance *fondo 'bottom' [φondo] > Basque ondo 'side, bottom' 
Romance *fongo 'fungus' [φongo] > Basque onddo 'mushroom'
Latin fōrma- 'mould' [φorma] > Basque horma, orma 'wall'
Latin frāga- 'strawberry' [φ(a)raga] > Basque arraga, araga id.5
Romance pollo 'chicken' [φoλo] > Basque oilo 'hen'
1 However, in most words the cluster fl- became assimilated to pl-, then palatalized in Spanish and Portuguese, as in flamma- > Spanish llama, Portuguese chama. 
2 In Proto-Celtic, PIE *p became a voiceless bilabial fricative *φ (then lost in the daughter Celtic languages) at word-initial and intervocally, but *b in other contexts.
3 In fact, there's ample evidence of Celtic toponymy in the Westernmost part of the Basque country, suggesting the area was previously inhabited by Celtic speakers. 
4 Unlike Gascon, Spanish never aspirates clusters fl-, fr- (see note 1). Also Castilian alone doesn't aspirate f- when followed by /w/ (resulting from diphtongation of Latin short /o/), hence fonte- > fuente. However, this doesn't happen in Cantabrian and Eastern Asturian varieties, where we find /xwénte/ instead.
5 In the variant arraga, [φr] evolved to a trill [rr], which then adquired a prothetic vowel. Compare Gascon ahraga

¿A dónde me llevas?

18 July 2011

Una calle no es una arruga, sino un arroyo

Como es bien sabido, el castellano arroyo es una forma masculina derivada del latín hispánico (Plinio) arrugia 'galería de mina', un conducto artificial por donde circulaba el agua al objeto de lavar el mineral1. Lo que ya no es tan conocido es que una forma emparentada *rūga es el verdadero origen del portugués rua y francés rue 'calle, calleja'2, erróneamente atribuído por el romanista Meyer-Lübke (REW 7426) al latin rūga 'arruga'3.

De la forma diminutiva *rūgula deriva el gascón arrrolha, rolha, garrolha 'arroyo, cavidad, canal', prestado al euskera arroil, erroila 'canal; barranco; cavidad; raya del pelo'. 

En mi opinión, todas estas formas provienen de una lengua vasco-caucásica pero no vascónica a la que llamo cantábrico. La etimología es a partir del PNC *jɦerqqwɨ (˜ -o:) 'ladera; barranco, trinchera', que en euskera nativo da erreka 'barranco, lecho de río o arroyo; arroyo; surco; raya del pelo'. En las lenguas hispánicas encontramos cognatos en portugués rego, asturiano riego, catalán rec, castellano regato 'riachuelo; canal de riego'. 

Aunque algunos linguistas han sugerido para estas palabras una etimología céltica a partir de *φrikā 'surco de arado' (latín porca)4, hay que rechazarla por razones semánticas como hace Corominas. En este sentido, el parecido con el antiguo eslavo eclesiástico rĕka 'río' (cuya ascendencia IE es dudosa) parece algo más que una simple anécdota.
1 Este método era especialmente usado en las minas de oro, hasta el punto que los romanos perforaron una montaña (Montefurado) y desviaron a través de ella el río Sil.
2 Meyer-Lübke cita también las formas ruga (dialectos sud-itálicos), arruga (campidorés) y carruga (lucano). Y en euskera encontramos el vizcaíno arruga 'plaza, mercado'.
3 Coromines acepta esta etimología hasta el punto de afirmar que "vistas desde lejos, las calles de una localidad semejan arrugas". Sin comentarios.
4 Del PIE *perk´- 'surco', que en mi opinión proviene del PNC *ɦwə:rqqe: 'línea de los montes; límite' > euskera hegi.

07 July 2011


Latin colō 'to inhabit, to dwell' (hence in-cola 'inhabitant', in-quil-īnus 'tenant') is derived by Indo-Europeanists from PIE *kwel- 'to turn', assuming the original meaning was 'to turn up soil'. However, this explanation strikes me as a bad one, so I searched for a better etymology.

I see here a PIE A root *kwel- 'to dwell' (homonymous with the above one) cognate to PAltaic *gù:lí 'dwelling, cottage'.  This is in turn related to PNC *qəlV 'house, hut', which gives also Kartvelian *xl- 'to dwell' (hence *sa-xl- 'house' and PIE B *sel- 'dwelling, settlement' (Latin solum, PGermanic *sala-), an instance of Fournet's Law1.
1 See my earlier post for more information.