A Semitic 'horse' Wanderwort

According to standard etymological dictionaries (e.g. Kluge's), High German Pferd 'horse' (formerly 'post horse') is an early loanword from Late Latin paraverēdus 'post horse'. This is derivated from verēdus¹, a Celtic (Gallic) loanword *uerēdos < *uɸo-rēdos (e.g. Cymru gorwydd 'horse'), a deverbal noun from *uɸo-rēd- 'to travel fast', a compound verb from *uɸo- 'under' and *rēd- 'to ride'².

However, IMHO this etymology is phonetically implausible, as it would need a huge haplology (syllable collapse) to explain the German word. Taking into account the High German consonant shift and the Dutch cognate paard, the protoform would be *perd-, which I consider a Wanderwort from Semitic *pVrd/parad^ʕ/z-) 'an equid'.  

A similar root *bVrdʕ/barz/dʕ- 'mule, horse' would be the origin of Late Latin burdus 'mule' and būr(r)icus 'donkey, small horse' (a diminutive form which suggests an assimilated variant *burrus). 
__________________________________________

¹ Interestingly enough, Latin verēdus was borrowed (through Greek béraidos, beredos) into Arabic barīd- 'express, courier' and similar forms in other Semitic languages, thus contributing to the widespread confusion about these roots.

² Delamarre (2008): Dictionnaire de la langue gauloise, p. 314.

A confusion of rhotics

Most personal names and theonyms found in Latin inscriptions from the old region of Aquitaine correspond to an early form of Basque commonly referred as Proto-Basque by Vascologists. However, some of these names have been wrongly interpreted due to an imperfect reading of rhotics: while the flap /r/ <r> is usually written as RH, the trill /ŕ/ <rr> can be either R or RR. Thus reading R as /r/ instead of the actual /ŕ/ is a mistake many Vascologists have made.

For example, the Aquitanian theonym HERAVSCORRITSEHE, usually interpreted as containing a first member herauś corresponding to modern Basque herauts 'boar', but this is inexact. Firstly, this word is actually one of the variants of a word *enauś(i), *inauś(i) whose meaning is 'heat of sow' or 'sow/female boar in heat' and whose modern forms¹ display two alternative results from a former *n². The correct reading is heŕaus, which corresponds to modern Basque errauts 'ash' (a compound of erre 'to burn' and hauts 'dust')³.

Another example is the anthroponym LAVRCO, whose first element⁴ is commonly interpreted as the numeral lau(r) '4' (with a flap /r/), when it's actually lauŕ (with a trill /ŕ/). This corresponds to modern Basque labur 'short'⁵, so in this case the Aquitanian form corresponds to Iberian and not to Proto-Basque⁶. 
__________________________________

¹ herause, eresu, (h)erüsi, heusi, iñaus, iraus, i(h)ausi, irausi, irusi.

² According to our own interpretation, the regular shift -n- > -0- in Basque (as well as in Gascon and Portuguese) is consequence of an adstrate influence, while the comparatively rare -n- > -r- is part of its own Iberian pedigree.

³ This was first proposed by Hugo  Schuchardt.
⁴ Also found in Iberian anthroponyms. The second element is the diminutive suffix -ko.
⁵ The form laur is only found in the Baztanese dialect.
⁶ Iberian has *bu > u while Basque has *bo > o.

A bad purchase

Latin emō 'to buy' is an output of IE *H1em- 'to take, to distribute', a root also found in Celtic and Balto-Slavic. The "laryngeal" *H1 is reconstructed to account for initial i- in Balto-Slavic (Lithuanian imù, Old Church Slavonic imǫ). However, given its scarce attestation, the likelihood of this being a non-native IE root is high.

In his own blog, Glen Gordon proposes this root to be a recent (early 1st millennium BCE) borrowing from a supposed Etrusco-Rhaetic [sic] verb *em 'to take', which apparently appears as -em in Etruscan numerals like ci-em zaθrum '17', literally '3 from 20', much like Latin duo-de-viginti '18', lit. '2 from 20'. 

Unfortunately, Gordon confuses what is clearly a postposition with a verb, and for making his point stronger, he notices two instances of the presumed past form *em-e in the scriptio continua nacemeuruiθalθileniθaliχememesnamertanśinamulu (TLE 366). Unfortunately, he doesn't give us a proper segmentation of the whole inscription.

Given the inadequacy of this proposal, we must find a better one. A Vasco-Caucasian loanword from PNC *HVqqVn- (˜ -m-) 'to take, to snatch', which gives Basque k(h)en-du 'to take away; to ask for a price' (Bengtson) and also Etruscan cen- 'to get, to buy', whose participle cen-u 'got, bought' is found in some inscriptions (Perugia Cippus, Tabula Cortonensis), seems likely.

A note to readers (caveat emptor): don't buy crackpot theories.

Iberian anthroponyms

Even before complete decipherment of the Iberian scripts was achieved by Gómez Moreno in 1922, the structure of the Iberian anthroponyms had been known since the finding of the Ascoli's Bronze Plate near Rome in 1908. This is a list of Iberian soldiers which formed the Turma Salluitana (an auxiliary horse troop which fought to the side of Pompeius Strabo in the Allies' War) and to whom Roman citizenship was granted. The names appear in the form of praenomen (first name) and cognomen (last name): X son of Y.

Although this inscription allowed epigraphists to identify personal names in Iberian texts as soon as they could been read, it's far from being an equivalent of the Rossetta stone¹. Most Iberian anthroponyms are compounds of two members, either two nouns or a noun and an adjective. This structure is similar to the one of Celtic or Germanic anthroponymy, whose speakers were also warfare aristocracies of the Iron Age².

Examples of Iberian anthroponyms:

Baise-bilos, Bilos-baiser 'Solitary Eagle'
Balke-bilos 'Eagle's Eye'
Bando-nius 'Chief's Horse'³
Biu-nius 'Chief's Mare'
Iskeŕ-adin, Adin-iskeŕ 'Old Hand'
Nios-iskeŕ, Iske-nius 'Chief's Hand'
Sakaŕ-iskeŕ 'Big Hand'
Sosin-adin 'Old Bull'
Sosin-bilos 'Bull-Eagle'⁴
Sosin-biuŕ 'Bull-Mare'

The apparent similarity of Iberian anthroponym elements with Basque words has fueled countless amateurs to support the discredited Vasco-Iberist theory, which in its most extreme form equates Iberian with an ancestral form of Basque, so the modern language can be confidently used to translate Iberian  texts⁵. This approach is absolutely unscientific and rejected by serious specialists. The hard truth is that Basque alone is of little help to understand Iberian.

However, with the aid of external comparison and a patient research we have been able to open a breach on what it was an impenetrable wall for many. The first Iberian word whose etymology can be surely stablished is adin 'mature, old' (PNC *=VdʑV 'to grow'⁶ + participle suffix *-nV).

Iberian glossary

adin 'mature, old' ~ Basque adin 'age; judgement'

baiser 'solitary' (Aquitanian baese-)⁷ 

balke 'eye' < PNC *ʡwilʡi 'eye'
bando 'horse'⁸ ~ Basque mando 'mule'⁹

biki 'cow' ~  Basque behi 'cow'¹⁰ 

bilos 'eagle, bird of prey' ~ Latin mīluus 'kite'

bi-o-s 'heart' ~ Basque bi-ho-tz 'heart'¹¹

bi-uŕ 'mare' ~ Basque be-hor 'mare'¹² 

is-keŕ 'hand'¹³ ~ Spanish garra 'claw'

niś 'girl'¹⁴ ~ Basque nes-ka 'girl' (Aquitanian nes-ka-to).

nios, nius 'chief'¹⁵                                                               

śani 'boy' ~ Basque sehi, sein 'boy, servant'¹⁶

sakaŕ 'big' ~ Basque zahar 'old' (formerly 'big')¹⁷

saldu 'horse' ~ Basque zaldi 'horse'¹8

sosin 'bull (Aquitanian soson) ~ Basque zezen 'bull'

tautin 'noble' ~  Aquitanian hauten 'noble', Basque hauta 'election, elite'¹⁹

_________________

¹ Unfortunately, bilingual texts are very scanty and often incomplete due to breaking of the material support (ussualy stone).
² This is a quite different picture from the one commonly found in schoolbooks, which present Iberians as peaceful traders.
³ Found in the Latinized form Mandonius.
⁴ Found in the Latinized form Sosimilos.
⁵ Good examples of Vasco-Iberist crackpots in Spanish literature are Juan Luis Román del Cerro and Jorge Alonso García.
⁶ The native Basque output of this root is hazi 'to grow; seed'.
⁷  A compund from *wa- 'this' < PNC *ʔu (˜ *hu) 'demonstrative pronoun (that)' and  *-idʑV 'self, oneself'.
⁸ Iberian ortography doesn't differentiate between /b/ and /m/.
⁹ Borrowed from Celtic *mandu 'young animal; horse', an Altaic Wanderwort.
¹⁰A West Europe substrate item *bekko < PNC *bHe:mtɬɬɨ (˜ -u,-i) 'deer, mountain goat'. 
¹¹ PNC *jerk’wi 'heart'. 
¹²PNC *q’ɦweɫV:/*q’weɫɦV: 'large female domestic animal (cow, mare)'.
¹³ The first element is from *tsHə 'one'.
¹⁴ PNC *nusA (˜ -o-) 'daughter-in-law'.
¹⁵ PNC *nɨwts(w)A: 'prince, ruler; bride-groom'.
¹⁶ PNC *ts’ænʔV 'new'.
¹⁷ PNC *tʃ’ǝqwV 'big'.
¹⁸ PIE *g^(w)Ald- 'foal, young of an ass', with assibilation (palatalization) of the initial velar. 
¹⁹ Celtic *toutā 'people'.

Knives and ploughs

According to the traditional view, English knife, from Germanic *knība-z 'knife' (with no native IE etymology), is a Wanderwort which spread from Old Norse into Old English and other Germanic languages, as well as Old French quenif, quanif 'pocket knife', (and in diminutive form) Occitan ganivet, Aragonese cañivete and Basque kanibet, ganibet, kanabita, gaiñibeta, ganabeta, kanit, gaminta 'penknife'.

However, the German linguist Theo Vennemann¹ (see my earlier post) reversed the propagation direction, considering Basque kanibet to be the original form, as a compound of kana 'cane' (a loanword from Latin canna) and bedoi 'pruning shears'², a dialectal word of limited extension. But (unknowingly to him) the latter is a loanword from Gaulish uidubion 'hoe' (glossed as Latin vidubium), a compound from uidu- 'wood' and -bion < Celtic *bi-na- 'to hit, strike' < IE *b^hejH-.  This root is reflected in Germanic as *bī́ɵla-, *bīdla- 'knife, sword, ploughsare' > English bill.

My own proposal is that Latin novācula 'razor, knife' (Portughese navalha, Spanish navaja, Catalan navalla, Aragonese novalla) < *kne-wa-tlā, traditionally derived from IE *ksew- 'to rub, to whet'³, is actually a reflex of IE *kneH2- 'to scratch, to scrape'. This etymology can also explain Basque nabas- 'plough' (in compounds)³ < *knā-wa-z and Germanic *knība-z < *knē-wa-z as borrowings from some IE language (possibly Italoid) where IE *w gave b. 

The semantic shift from 'knife' to 'plough' can also be observed in the descendants of Latin culter 'knife' > Occitan coltre, French coutre, Italian coltro 'ploughshare', Basque golde 'plough', the original meaning being transferred to the diminutive cultellum > Galician coitelo, Spanish cuchillo, Catalan coltell 'knife' (the latter form being displaced by ganivet).

Another interesting Basque word is saratu 'to hoe', a loanword from IE *sºrp- 'sickle; to cut' (Latin sar(r)iō 'to hoe (vine)' < *sarpiō.)
______________________
¹ Zur Etymologie der Sippe von engl. knife, franz. canif, bask. kanibet*, in Europa Vasconica, Europa Semitica (2003), pp. 427-452.

² Also Bearnese bedui. See García de Diego, Diccionario etimológico español e hispánico (1985), pp. 1063-1064.
³ Considered by many Indo-Europeanists to be an extension of IE *k´es- 'to cut'.
⁴  There is also the Bearnese hapax naves '(large) knife'.

Some wrong Spanish etymologies

Some Coromines' Spanish etymologies (the ones found in the DRAE are generally nastier) enjoy a greater credit than they actually deserve.

For example, if we look at cerrar 'to close, shut' in his Breve diccionario etimológico de la lengua castellana, we can see he derives it from late Latin *serare < Latin sera 'bolt, lock', in spite that Spanish /c/ (in the Middle Ages a dental affricate [ts] and now an interdental fricative [θ]) can't come from Latin /s/. He denies elsewhere any relationship between this word and Catalan serrar, French serrer 'to tighten', although they fit both phonetically (as [ts] > [s] in these languages) and semantically (the meanings 'closed' and 'tight, dense' are related). The answer is all these words¹ are Pyrenaic (a Vasco-Caucasian substrate language) loanwords from the PNC root *=utɕE(rV) 'thick, fat'.

Another weird etymology is jara 'rockrose (Cistus)', a kind of bush common in the Iberian Peninsula which Coromines derives from Arabic ša^cra 'low bush', where [ʃ] /x/ regularly developed into [χ] /j/. Apparently, he ignores Basque zare 'basket' and (t)xara 'rockrose (wood)', a word of Pyrenaic origin for which Bengtson suggests an etymology from PNC *tʃ’ʃ’wɦeli (˜ tɕ’ɕ’,-ʕ-,-ɫ-) 'a k. of foliage tree'². The palatized form would have been borrowed into Spanish xara, then jara.

_________

¹ There's also Basque (Biscayan) zarri 'dense', zarra-tu 'dense; to close, shut'.
² Related to Paleo-Eurasian *tʃwalV 'willow', a root represented in Paleo-European *sal(i)k- 'willow' (Celtic *salik-, Latin salix, Germanic *salx-) and native IE *welik- 'willow' (Greek helíkē, English willow).

Strawberry fields forever

Starostin reconstructs a Paleo-Eurasian root *NEk’rV 'thorny bush' > Altaic *ńíkrV 'a k. of thorny tree', from which I derive Paleo-European *ag´ºr- (Gaulish agraniō- 'sloe, fruit of the blackthorn', English acorn¹, Latin agresta 'green grape', Greek ágrios 'wild'), with loss of the initial nasal. Its IE native counterpart would be *d^hreg´^h- 'sloetree, blackthorn'² (Celtic *dragenā 'sloe, fruit of the blackthorn'³, Albanian drédhë 'strawberry'). Greek hrá:ks, hró:ks 'grape' < *srāg-, is probably a Thracian loanword from the same root.

Paleo-Eurasian *bVrdzV 'thorn' is the origin of Kartvelian *bardzg-/*burdzg(a) 'thorn, thorny plant'. The sibilant *dz is reflexed as *g´ in IE *b^hrag´- 'strawberry' > Lithuanian brãškē 'garden strawberry', Latin frāgum 'strawberry'⁴ and as *g´^h in *wreH2g´^h- 'thorn' > Celtic *wragi- 'needle', Lithuanian rãžas 'dry stalk, stubble; pronk of fork', Greek hrákhos 'thorn-hedge', hrákhis 'spine, backbone'.

Altaic *núŕi (˜ -e) 'a k. of berry, grape'⁵ is related to Hittite muri- '(bunch) of grapes' and IE *moro- 'blackberry'.
_________________

¹ Indo-Europeanists usually group this with Lithuanian uga 'berry', Old Church Slavic jagoda 'fruit', vin-jaga 'grape', but I prefer to group the Balto-Slavic forms with Latin ūva 'grape', thus reconstructing *H3g^wh-ā 'berry'.
² Found (with simplification of the initial cluster) in Basque lahar, nahar 'blackberry (Rubus fructicosus); buckthorn (Rhamnus cathartica); caltrop (Tribulus terrestris)'.
³ Related to this is *d^hren- 'cornel cherry' (Old High German dirn-baum, Russian derën) ~ *tºrnu- 'thorn'.
⁴ Usually though to be derivated from *srāg-. French fraise, Spanish fresa 'strawberry' is probably an Italoid loanword from the same root (Italoid is a "satem" language with *b^h- > f- like Latin). Basque has araga 'strawberry' and arakatz 'gooseberry'.
⁵ Notice than the supposed Korean cognate *nùrúk 'yeast' presented in the database doesn't correspond to this root but to *ńjuŕge 'a k. of weed' < Paleo-Eurasian *nUŕV 'grain, corn'.

El roble y el haya

El IE regional del NW (no PIE) *perk^w- tiene aparentemente dos formas, una masculina 'roble' (latín quercus 'roble, encina') y otra femenina 'pino' (inglés fir 'abeto'). Se trata de un préstamo (loanword) vasco-caucásico a partir del PNC *χwɨrkkV 'árbol, roble', con *χw > *p.

El PIE *b^heH2g´o- 'haya' (latín fāgus 'haya', griego phēgos 'roble', ruso buz 'olmo') es también de origen vasco-caucásico a partir del PNC *mħoqqwe (˜ -ʕ-) 'roble'.

Efectivamente, tal y como propuso Starostin en un viejo artículo, el PIE contiene un buen número de préstamos vasco-caucásicos¹ de época neolítica. Las correspondencias fonéticas entre éstos y el PNC sugieren una única lengua de origen, que tendría características como:

1) PNC *w- > b- en posición inicial. 
2) PNC *l/*r caen en ciertos grupos consonánticos (clusters).
3) PNC *l  > r intervocálicamente y ante consonantes.
4) Una transformación del sistema vocálico del PNC que implica la delabialización de PNC *o.
_______________
¹ Hay que aclarar que el proto norcaucásico (PNC) reconstruído por Starostin y Nikolayev es una entidad mucho más antigua que lo que aparentemente se podría pensar, y por tanto, más próxima a lo que sería el auténtico proto-vasco-caucásico, aunque Starostin nunca incluyó el euskera ni la hipótesis vasco-caucásica en sus investigaciones.