Is really Tartessian a Celtic language? (newly updated)

When Koch's first book appeared in 2009 (the second edition came in 2013), launching his theory Tartessian was a Celtic language, I was enthusiastic. However, by the time he published his second book (2011) I had reached to the conclusion his interpretations were contrived as well as inconsistent.

A major difficulty regarding the Tartessian corpus is the semisyllabic SW script (where vowels are redundantly noted) hasn't yet been fully deciphered and thus the exact value of some signs is still unknown or problematic. For example, Koch assigns the sign H (Phoenician het, Greek epsilon) to the Proto-Celtic voicess bilabial fricative [ɸ] (almost universally lost in historical Celtic languages) coming from IE *p, as in e.g. Hatªaneatᵉe. However, the correspondence between Tartessian Haitᵘura¹ and Iberian baiduŕa², the later seemingly related to baides 'witness', in turn derived from IE *weid- 'to see; to know (as a fact)' (cfr. Celtiberian ueizos), points to the the value /w/ (Greek digamma).

If this is correct and the word is native to Tartessian, it would indicate it's an IE language although certainly a non-Celtic one. A good possibility would be the Paleo-European (after Krahe's Alteuropäische) substrate identified by the Spanish Indo-Europeanist Francisco Villar in the ancient Iberian toponymy, where we find the lexeme *akʷā 'river'³ (see here) used as a suffix (e.g. Turaqua).

Regardless of the actual filiation of Tartessian, it's clear it had contacts with other languages spoken in Western Iberia, namely Gallaecian (Celtic) and Lusitanian (non-Celtic). For example, the segment lokºobºo /lugu-bo/ closely mimicks Gallaecian LVGUBO /Lugubo/, LVCOVBV(S) /Lucoubo/, a dative plural theonym presumably referring to the gods Lug.

____________________________________

¹ Translated by Koch as 'Lady of Baeturia'. See J.T. Koch (2011): The South-Western (SW) Inscriptions and the Tartessos of Archaeology of History, in Tarteso, El emporio del metal.

² The signs for rothics /r/ and /ŕ/ are reversed in the SW (Tartessian) and Southern Iberian scripts with regard to the Levantine one, upon which the usual transcriptions are based.
³ Not 'water' (cfr. Latin aqua) as commonly thought.

Latin aqua 'water' (updated)

Latin aqua 'water' < *akw-ā (f.) is an interesting word with correspondences in Germanic *áxwō 'river' (Gothic ahwa, Old High German aha) as well as Celtiberian akua,  whose meaning is probably 'river' (cfr. tar akuai 'across the river' in Botorrita III¹). Contrarily to what many scholars think, its original meaning isn't 'water' but 'river'² and belongs to the Old European Hydronymy (OEH), a corpus first studied by Hans Krahe and more recently by Francisco Villar and which represents an older form of IE than the historically attested languages³.  

Apparently, *akʷā would be a derivated noun from the adjective *aku- 'quick, fast', found in Latin acupedius 'swift of foot' and accipiter 'hawk', the latter with cognates in Greek ōkýpteros, literally meaning 'swift flyer' (ōkýs 'swift', pterón 'wing'), and Sanskrit āśu-pátvan- 'flying swiftly'⁴. These correspondences suggest Paleo-European *aku- would correspond to late IE *ōḱu- 'quick' > Latin ōcior.

On the other hand, the geminate stop in accipiter would be the result of a sound shift -kʷ- > -kk- which I call Kretschmer's Law and of which are examples in Italoid (e.g. the Lusitanian theonym Iccona) and Celtic (e.g. *sukko- 'pig'). There're also traces of it in Vulgar Latin 'water', as reflected in the Appendix Probi (aqua non acqua).

_________________________________________________
¹ P. de Bernardo Stempel (2007): Water in the Botorrita Bronzes and Other Inscriptions, in Palaeohispanica 7, pp. 55-69.
² Despite so, it has been compared it to other 'water' words, both within the IE family and oustide, even up to the point of building upon it a "global etymology" (Ruhlen).

³ More specifically, its agglutinative morphology would correspond to the "IE I (pre-flexional)" stage proposed by Francisco Rodríguez Adrados. Of course, Vennemann's proposal of OEH being related to Basque must be rejected.

⁴ M. de Vaan (2008): Etymological Dictionary of Latin and the Other Italic Languages, p. 21.

Gaulish *santikā 'ladle; milking vessel' (updated)

Gascon sanja, sansha, santja, sòntja, shansha, sancha, Aragonese sancha and dialectal Catalan sanxa (Cerdanya) designate a milking vessel carved in wood, often made by hollowing out a tree trunk¹. On the other hand, dialectal Basque xantxa (L, Z), xaiñtxa (Z) designates a milking vessel with a long metal handle (kopetxa), and in the latter also 'ladle' (golhare), a meaning which in my opinion would be the original one².

Although the Gaulish origin of this word seems undeniable, a Celtic etymology is more dubious. For example, Matasović proposes a Celtic femenine *sϕanjā corresponding to the masculine *sϕenjo- > Old Irish sine 'teat'³, in turn derived from IE *spen-, which is phonetically unacceptable, among other things because Celtic *sϕ- gives s- in Goidelic but f- in Britonic and probably also in Gaulish. For the same reason, Hubschmid's protoform *sand-ikā from IE *spn̥dh- 'bucket'⁴ can't be accepted, although Celtic *sϕondā would be the origin of Romansh s(u)onna 'bucket', probably a Lepontic loanword.

Following Coromines⁵, I'd propose a Gaulish protoform *santi-kā as a Baltoid loanword corresponding to Baltic *samti-/semti- > Lithuanian sámtis 'spoon, ladle', sémti 'to pump, to scoop (a liquid)', from an IE root *semH- 'to pump, to scoop (a liquid)' also found in Latin sentīna 'bilge; sewer, drainage' and cognate to Altaic *ʃŏ́mo 'to dive; to summerge; to scoop (a liquid)' (EDAL 2193) > Turkic *tʃo:(m)- 'to diver; to swim; to scoop (a liquid); to immerse, to dip' (Turkish čömče 'spoon'⁶). On the other hand, Etruscan śanti (Tabula Capuana) probably designated some kind of vessel.

___________________________________________

¹ G. Rohlfs (1970): Le Gascon. Etudes de phylologie pyrénéenne, § 59. It's likely the same kind of vessel called kaiku in Basque.

² The oldest European metal ladlers are from the Hallstatt culture of the Early Iron Age.

³ R. Matasović (2009): Etymological Dictionary of Proto-Celtic, p. 333.

⁴ J. Hubschmid (1951): Alpenwörter. Ursprungs und vorromanischen Romanischen, p. 61.

⁵ DECLC, p. 667-668.

⁶ J. Hubschmid (1955): Schläuche und Fasser, p. 107, quotes čamča, čumča with a diminutive suffix -ča.

French bourbe 'sludge' (updated)

French bourbe 'sludge' (collective form bourbier) designates a kind of dark, thick mud deposited by stagnant or waste waters and surely derives from Gaulish *borwā (f.), whose supposed meaning 'hot, boiling spring' accroding to standard dictionaries, which link it to Welsh berwi, Breton bervi 'to boil'^1. 

However, in my opinion this etymology is semantically unsatisfactory, and I'd prefer a substrate loanword from Baltic *purwā > Lithuanian pũrva 'smudge, dregs', Latvian pùrvs, purve 'morass, swamp', a word with parallels in Indo-Aryan: Sanskrit púrīṣa- 'crumbling or loose earth, rubbish; feces, excrement, ordure', Sinhalese puraṇa 'fallow or waste land'. 
______________________________

¹ X. Delamarre (2008): Dictionnaire de la langue gauloise, pp. 82-83.

Gaulish *komboro- 'heap, accumulation' (updated)

Basque gonburu (B), bonburu (B) 'excess', usually derived from Latin cumulu- 'heap', is actually loanword from Gallo-Romance *komboro- 'heap, accumulation'¹ (REW 2075). This is the origin of Medieval Latin combrus 'abatis' (an obstacle formed by a row of tree branches pointing to the enemy) and Old French combre 'dam on a river'. Other derivatives, we we've also got French encombrer, Italian ingombrare 'to obstruct, to block' < *in-comb(o)rāre and Spanish escombrar 'to disembarrass' < *ex-comb(o)rāre. 

Although specialists usually give this word a Celtic etymology from Celtic *kom-bero-² >Old Irish commar, Welsh cymer, Breton kemper (hence the toponym Quimper) 'confluence (of rivers)', in my opinion this etymology is semantically unsatisfactory,  so I'd prefer a substrate loanword related to Baltic *kumb(u)r- 'soil elevation, hill' (Lithuanian kumbrī̃s, kum̃bris, Latvian kumburis) and Thracian skumbras (ancient toponyms Skóumbros, Skoũmbro)³. As the source language (either from the Baltic group or a close relative of it), had no differentiated /a, o/ vowels but only /a/, its /u/ was actually realized [ʊ] and therefore was interpreted by Celtic (Gaulish) speakers as being closer to their own /o/⁴.

Working independently of each other, the Catalan linguist Joan Coromines and the Spanish Indo-Europeanist Francisco Villar apparently conflated this Baltoid language with a different Italoid (i.e. akin to Italic) one, called Sorothaptic by Coromines from its purported association with the Urnfield culture, attested in several lead foil inscriptions (written in Latin alphabet) found at the beggining of the 20th century in the thermal station of Amélie-les-Bains/Els Banys d'Arles (Rosselló)⁵. These are votive texts addressed to some water deities (nymphs) referred to as kantas niskas, where kantas presumably derives from IE *k´wen-to- 'holy' and niskas could be related to Basque neska 'unmarried girl'.

________________________________________________________

¹ A meaning proposed by Coromines instead of Meyer-Lübke's original one. 
² X. Delamarre (2008): Dictionnaire de la langue gauloise, p. 122. 
³ I. Duridanov (1969): Die Thrakisch- und Dakisch-Baltischen Sprachbeziehungen, p. 93.
⁴ Villar has an idiosyncratic theory of the development of the 5 vowel system of some IE languages (e.g. Italic, Celtic, Greek) from an older 4 vowel system /i, ε, ɑ, ʊ/ found in e.g. Anatolian, Balto-Slavic, Germanic, as well as in Etruscan. See F. Villar (2000): Indoeuropeos y no indoeuropeos en la Hispania prerromana, pp. 369-378.
⁵ J. Coromines (1976): Els ploms sorotàptics d'Arles, in Entre dos llenguatges (II), pp. 142-216. Lusitanian, a language attested on several inscriptions written in the Latin alphabet, would also belong to this group.