2019, Zeitschrift für Celtische Philologie


46 Pages

This work tackles a number of problems presented by presumably Celtic onomastics attested in Venetic indigenous inscriptions. To this effect, beside etymological analysis proper, a number of questions concerning early loss of /p/ in Celtic, Early Celtic accent, word-initial clusters, word order, patterns of agent noun composition, misperception or unexpected phonemic interpretations of foreign forms by local scribes, and areal phenomena like loss of intervocalic -u̯- and -s- will be addressed.


Celtic and Venetic in Contact:The Dialectal Attribution of the Personal Namesin the Venetic record



Dieser Aufsatz behandelt eine Reihe von Problemen, die in der vermutlich keltischen Onomastik auf venetischen Inschriften auftreten. In diesem Zusammenhang werden hier nebender eigentlichen etymologischen Analyse eine Reihe von Fragen bezüglich des Verlusts von /p/  im Keltischen, des frühkeltischen Akzents, Anlautgrup-pen,  Wortstellung, Schemata  der  Bildung komponierter Nomina agentis, Fehlwahr-nehmungen oder unerwarteter phonemi- scher Interpretationen fremder Formen durch einheimische Schreiber sowie areallinguistischer Phänomene wie des Verlusts von intervokalischem -u ̯- und  -s- behandelt.


1. Introduction¹


As is well known, the Venetic record contains a number of unmistakably Gaulish names, both in the Venetic and the Latin alphabet. In this work,I shall set out to find some common traits that allow to distinguish them from the Italic names and give us a number of clues on the specific Gaulish dialect of the area andthe way in which Gaulish words were adopted by Venetic and written downby Venetic scribes in both alphabets. Conversely, a number of phonetic and morphological traits of the Venetic language which have passed unseen will be isolated and compared to those of Celtic. Additional-ly, I shall enumerate the arguments that allow us to differentiate Gaulish and Venetic names occurring in the Venetic record when the similarity to names attested elsewhere does not  suffice.    Given  the  comparatively  early  age    of  the Venetic inscriptions,  as   well  as the differences between the Latin and the Venetic alphabet, some hitherto over-looked facts about Eastern Gaulish will be brought to light. Finally, I shall try to cement the idea that Gaulish names have been treated in a rather predictable way, that is to say that by virtue of being filtered into the Venetic record they have been stabilised, and as a consequence they show homogeneous renditions.

1 I thank the editor and an anonymous referee for their work and kind advice. The Spanish government has explicitly refused to finance this work.

Abbreviations: PN= Personal Name; DN = Divine Name; RN = River Name; EN = Ethnic Name; FN = Father’s Name (patronymic derivative); GN = gamonym; PlN = Place Name, and VH=Venetia et Histria. Texts conducted in the Latin alphabet are rendered in VERSALITES; texts in the Venetic and other varieties of the Etruscan alphabet are in italics.


8 Blanca María Prósper


This contradicts the possibility that spelling variation in the rendition of Gaulish in Latin epigraphy has anything to do with scribal hesitation or inaccuracy. I shall start from traditional assumptions regarding what the definition of Celtic should be,since I regard some current attempts to make this definition more flexible as unfounded escape routes to bolster aprioristic preferences about the classification of particular languages. In this vein, it will be shown that new, independent arguments can be adduced in defense of the classication of Venetic as an Italic language.


2. The - s-stems of Venetic: Areal loss of -s- in eastern Gaulish and western Venetic and archaic IE compounds

In this section I shall examine several overlooked adjectival compounds of the possessive type in -es-. The showcase example in the area is the PN VESCLEVESIS, VESCLEVESI (gen. and dat., CIL III, 3038, 3058, 10138, Dalmatia) = Skt. vasu-śravas- ‘of good fame, glorious’, which may be Venetic in view of its centum character

Other compounds point in the same direction, but some of them are compatible with a Celtic ascription: thePN VESPADEVS (x2, AE  1957, 129, Borno/Camunni)      may   ultimately   come   from   *h₁u ̯esu-g u ̯heh₂id-es -  ‘having  a  good  shine’  or ‘he whose goods are his dazzle. In contra-distinction to other branches,Celtic tended to eliminate this kind of compound and has simply refashioned it as a thematic formation, probably attested in the PNN BAEDA (Britannia), BAEDIO (dat., Rome), BAEDI (pseudo-gentilic, gen., Ostia, Latium). Related forms are *gu ̯heh₂id-ro - in Hispano-Celtic *bai ̯dro  - (a PlN attested as Baedro, BAEDRONENSES, DE VICO BAEDORO in Baetica), the PN BAEDARI (gen., CIL III, 1585, Dacia) or Gk. φαιδρός ‘bright’, and Greek may have lost the corresponding stem *φαῖδος, only extant in the isolated gloss φαίδει: ὄψει (Hesych). The ascriptionof this set to the Caland system is rounded out by the EN baeduī in Callaecia, from *gu ̯heh₂id-u-.³



2 Cf. KATIČIĆ 1976: 172, 175. *h₁u ̯esu- is held not to be attested in Italic, but GARCIA RAMÓN 2016: 357–58 has made a case for the Oscan festival name Vesul(l)ia - (Iovila Capuana) and the gentilice vesulliaís  (Samnium) going back to a possessive adjective *h₁u ̯esu-h₃no-.

3  The PNN VESGASIONI,  VESGASSIS, VESGASSAE (VH) may also be of Celtic ancestry and go back to a possessive compound *ghasteh₂ → *h₁u ̯esu-ghasth₂-o-  ‘having a good spear’ or ‘whose spear is his goods’.

4 Occasionally corrected as Voltigenei .


Celtic and Venetic in Contact: … 9

The oblique cases and derivatives of Venetic PNN with a no-minative in -es  share an intriguing feature:they lack any trace of a medial sibilant.The extant cases are enogenes (LV  123, Isola Vicentina) .e.nogene.i. (LV  133, Padua), and a dat. sg. vo.l.tigeni (LV  28, Padua), voltigenei  (LV  56, Morlungo, Este, 6 th – 5th C.B.C.), derivative FN (LV 28, Padua), and hugene.s. (LV 199, Calalzo) vs. FUGENIA (LV  109, Este), which are haplological for *fugi/o-genes  and its derivative *-genes-(i)i    ̯o -.⁵


We may add to this list a group of unexplained forms comprised of the GN  (nom. sg., LV 28, Este), the FNN,  u.r.kleiio.i . (dat. sg., LV 68, 82, Este) and the FN VRCLESONI (a derivative in -onii ̯os  in the gen., CIL I 2,2793, Este). The matter is further complicated by the existence of a text reading .e.go u.r.kli .e.getoriio.i . .a.kutiio.i. (LV  60, Este). The PN u.r.kli  has been alternatively held to be thematic, and then inflected for the genitive and referring to the possess or of the monument (LEJEUNE in LV  60),or athematic, and then inflec-ted for the dative, either containing a misspelling for †.u.rkle.i. (UNTERMANN 1961:8) or bearing the locative morph in *- as a result of the ongoing process of syncretism of both cases (cf. ESKA & WALLACE 2002: 269).⁶

'SOLINAS 1996: 159–60 has found a different but rather noncommittal solution to this riddle, elaborating on an idea by A. PROSDOMICI: u.r.kl(e)i  is connected with the FN VRLESCONI, and then it is a dative ‘in connessione con una base -e(s). No cogent explanation can be found in UNTERMANN 1961:168: ‘...Nachname zu einem IN. *Urclesō, der -ōn-Variante eines IN. *urcles(os?), der durch Erweiterung mit einem -s-Element aus urkli hervorgegangen ist’.

I find these accounts only partly satisfactory: if we focus on the visible cognacy and functional similarity of u.r.kli with     u.r.kleiio.i . and ,and resist the temptation of correcting the text, we have to acknowledge that itcan be the genitive of *urkl(i)i    ̯o- or *urklo-, but not of *urklei ̯o-.Conversely, it can be the locative of a stem *urkli-, which is intriguing, since -i-stems are the only ones in which we would not expect to find the first traces of the proposed syncretism, and the apparent cognacy with VRLESCONI is ignored or compromised. All this barely qualifies as an explanation and falls short of accounting for the discrepancy between the -es vis à vis the -s-less forms.

While  this problem  is  traditionally  sidestepped,  mere  substraction of the suffix by analogy with other stems, yielding a skewed paradigm, is most unlikely. Since the starting point must be an -es-stem, we have to explore the possibilitythat inter-vocalic -s- has undergone weakening and eventual effacement and has never been reinserted because it could not be recovered by the speaker on the strength of morphophonemic alternations. By contrast, the known cases of -VsV-, for instance the preterite donasa.n., have to be justified as secondary, analogical restorations favoured by the singular, etc.


5 See the Av. PN buǰi / a-sravah-. On the other hand, these names could be compared to Skt. bhojana- ‘possession’, or the μιθρα-βουζανης (on which see BENVENISTE 1966: 114). Whether Vesuenes  in Val Camonica (rock 50) belongs here and is a continuant of  *u ̯esu-genēs  or not is impossible to ascertain.

6 They believe both u.r.kli  and vo.l.tigeni  (they do not mention the alternative form Voltigenei) to be morphological locatives on their way to ousting -ei  as the dativesingular ending and plausibly take u.r.kli  to correspond to a nominative in -es, but never mention VRLESCONI.


10 Blanca María Prósper


While this is hypothetical with our meager available materials, it could go some steps towards explaining the riddle of u.r.kli and vo.l.tigeni, thus far our best cases of the anomalous dative morph -i.

If the two final syllables have evolved as follows: *-es-ei ̯ > -ez-ei̯ > -eɦ-ei̯ > -ēi ̯  the vowel [e:] resulting from the contraction may have been high enough in that context for the diphthong to have been perceived and / or realised as [ɪi̯] or [i:], which would make the rendition -<i> inescapable in the long run.⁷

In the case of the very archaic voltigenei (Este) we still find -ēi ̯  reflected as <ei>. This is, in sum, a phonetic, not a morphosyntactic solution to the problem.

Whether one accepts this idea or not, the FN VRLESCONI must be late and based on the nominative form, which is reminiscent of the GN in the name LOLLIA VXESINA (Altino, CIL V, 2197).⁸ In my view, it is the only surviving descendant of the ancient compound *h₁u ̯r̥Hu-ḱleu ̯es ‘whose fame is wide-spread’ in western IE, and the match of Skt. uru-śravas-, Gk. εὐρυκλῆς. Its closest Celtic counterpart is contained in the Gaul. DN ροκλοισιαβο ‘the renowned ones’,from Celtic *φro-kloues-(i)i̯o- (RIG I, 65), identical to Gk. προκλέης and Skt. pra-śravas-.⁹

Two Gaulish PNN VERCLOVI (gen., Pannonia, edited as VERCIOVI)¹⁰ might continue *u ̯eru-klou̯es-i(i)o-, which raises questions regarding the (areal?) loss of lenited intervocalic /s/ in the Alpine area.

The pseudo-gentilic CLEVIO (dat., CIL V, 4717, Brescia) and CLEVIAE (dat., CIL V, 1816, Gemona), both in VH, could match the pseudo-gentilic CLEVSIVS (EDCS -08000771, Verona) if they are traced to a primitive adjectival derivativeof the well-known type *ḱléu ̯-os ḱléu-s-o- ‘glorious’. If CLEVIO and CLEVSIVS were identical names, the scribal hesitation in the rendition of original -s- may obey unrecoverable distributional reasons. There are many other casesin which we find the disparate remains of otherwise infrequent sibilant stemsin this area. For instance, the Ven. PN Bagsoniia is of Celtic ancestry judging by its Anlaut,¹¹ and may be directly compared with Skt. bhaksam ‘joy, food’, either from *bheh₂g-s- (under acceptance of Lubotsky’s Law) or directly from *bhag-s- (LIV 65). It consequently goes back to an adjective *bh(e)h₂g-s-o- or *bhag-s-o- ‘well fed, nourished’.


7 The digraph <ii> habitually noted the consonantal variant [i ̯] and would probably have been interpreted by the reader as an abbreviation.

8 It is probably glossed over to look like a Latin cognomen in -inā, as in the Veneto-Latin GNN LOXINA, FREMAISTINAI, OSTINAI (CIL I, 2802, 2807, 2809, Este), and the PN on which it is based must be *ups-es- ‘having height’, directly related to Gk. ὕψος.<x> seems to point to Celtic ancestry.

9 This militates in favour of the attribution of the thematicised adjectives Lat. urvum ‘central piece of the plough’ and O. uruvú  to this etymon (cf. RIX 1995).

10 Found at close quarters, respectively in Carnuntum (CIL III, 14359) and Scarbantia(cf.  HILD 1968:  nr. 399).

11 Montebelluna, cf. CRESCI MARRONE & MARINETTI 2012: 230–31. The pseudo-gentilic VAXONIVS (AE  1966, 146, Porec, Istria, VH) is probably identical.




Celtic and Venetic in Contact: … 11

In addition, the sibilant stem *leu ̯k-os- in Skt. rocas has a regular derivative *leu ̯k-s-o- in OIc. ljóss ‘bright’ and the base of W. lluched ‘lightning’. And it is also present in Venetic underlying the pseudo-gentilic LEXIVS (AE  2012, 556, Este),¹² as well as in the GN LOXINA (LV 116, Este) and the FN lo.g.sii(ai) (LV  33, Este). The last two forms reflect the Celticoutcome of *leuk-s-o- with regional monophthongization [ou ̯] > [o:]. This type also survives in the equally pseudo-gentilic GECS(IS) (AE 1993, 1286, Ptuj, Pannonia Superior),which it is very tempting to derive from a neuter stem *ǵeu ̯-os only surviving in Skt. javas-, Avv.  .zauuah- ‘swihtness, speed’.

Syncope of medial /u/, common to the Celtic and Venetic forms reviewed thus far,  is expected in a polysyllabic form, but the Venetic evolution  -eu ̯e- > -ē- requires further assumptions. In fact, a sequence <eve> is attested in moltevebos (tablet of Este, conducted in the Paduan alphabet) and the PN .e.nokleve.s (, Padua).¹³

But the latter form in all likelihood underlies the PN ENOCLIAE in Altino (dat., CIL V, 2221, indigenous onomastic context).¹⁴

This leads me to theconclusion that at least the Venetic variety spoken in Este underwent loss of -u ̯- between two identical vowels.¹⁵

These forms additionally militate against Lejeune’s idea, according to which Venetic [eu ̯] yields [ou ̯] in Este, since they presuppose a change -kleu ̯e- > -klē-, completed before the fricativization of -u ̯- took place.

While  some details remain obscure, we may hypothesise  that  the apparently  contra-dictory attestations are the product of the different range of spread of twoisoglosses of effacement of intervocalic consonants: While the disappearanceof a lenited, approximant-like -s- seemingly encompasses the whole of Venetia Euganea and possibly some areas of the Alpine regions, leaving Dalmatia apparently untouched, loss of -u ̯- seems to be confined to its westernmost fringe,  epitomised by the   examples from Este showing both changes and consequently constituting extreme cases of reduction. In sum, u.r.kli and vo.l.tigeni are neither genitives nor syncretic locatives, but datives in -ei ̯ which appear to show defective writing because their synchronic stem was respecti-vely klē- (whose genitive would have evolved as -kleu ̯eɦos > -klēɦos > -klēos > ? -kle̯os), and gene- (-geneɦos >? -gene̯os).


12 The IE adjective *leu ̯ko- ‘bright’ in Gk. λευκός, etc. is perhaps attested in the pseudo-gentilic LEVCONIO (dat., CIL V, 4902, Voberna, VH, but is also attested in Noricum, Alpes Cottiae and Aemilia, which casts some doubts on this etymology).


14  There is a P N OENOCLIONI in Pompeii (dat., CIL  IV, 7384), interpreted as the transcription of a Greek name †Οἰνοκλίων, which, given its context (STABVLIO OINOKLIONI BIBIS), is likely to be a nonce formation, and then unrelated to ENOCLIAE.

15 This phenomenon is well described for Latin (see WEISS 2009: 154) and attested in Gaulish, as in NAMETOS fromnau ̯ameto-‘ninth’(cf. LAMBERT 1994:44, and see belowon na.i.śo.i., But cf. the veneticised Gaulish FN kavaron:s  in the Alps (LV  246, Gurina). Este is also more innovative than Padua in other respects, notablyin the loss of the final -i- resulting from final syncope -i ̯o-s > -is


12 Blanca María Prósper


Their analysis was consequently ambiguous when vocalic endings followed.¹⁶ The actual, highly fusional dative forms of these paradigms were *-klēi ̯ and *-genēi ̯. The derivatives *-geneɦ-(i)i ̯o- and *-kle(u ̯)eɦ-(i)i ̯o- resisted longer, probably because the resulting sequence ē.i ̯- was hetero-syllabic, but the later forms ENOCLIAE and FUGELIA leave little room for doubt as to their evolution.

The only stumbling block for this hypothesis would consequently be thelate form ENONI (Belluno, LV  236) if it were a dative of a nasal-stem PN *e / ēnō. In fact, such a name does not exist. It is only grounded on the existence of,itself the product of a hypothetical segmen- tation of a mangled sequence enov[- (LV  191, Calalzo; note that the expected punctuation of thevowel <e>- in Anlaut is lacking). A PN .e.nno  is attested in Calalzo (LV  205),and so is the corresponding FN .e..n.non.s., -ENNONIA, ENNONIOI. This seems to point to ENONI being a thematic gen. sg. misspelled for †ENNONI as more or less tacitly as-sumed,but this cannot be proven and the text shows three cases of consonant doubling. At any rate one inscription reads VANTI ENONIO (CIL I,2795, Este).

As a blind neogrammarian change, the process described above eventually blurred the morphological boundaries, thereby creating the skewed paradigms -genes, dat. -genēi ̯ and -klēs, dat. -klēi ̯. Pace LEJEUNE (LV 100), it is certain oncomparative grounds that all these names are ancient adjectival compounds in -es-, and his descriptive account ‘le passage à la flexion en -i- dès nos premiers documents témoignerait qu’au moins dans nos adjectifs composés la flexionmsigmatique avait cesséd ’être vivante en véte’ tiptoes around the real nature of the  problem:  the oblique  cases are as likely to  contain an alternative  stem  as  the Gk.dative εὐρυκλεῖ or the PN Εὐρυκλεί to belong to a different stem from thatof εὐρυκλῆς.Even assuming that these stems had been whole sale transferred to the -i-inflection for whatever reason (in itself a less likely dumping place for discarded stems than the thematic inflection, especially as far as the type -genēs  is concerned), not only would their dative form end up in -ei̯, but thiswould probably be the last slot upon which the original locative ending -would be expected to encroach, since it was synchronically parsed as a full-grade stem -ei ̯- + zero ending. The original locative would have been eitheridentical,or,according to the neogrammarian reconstruction, an unstable -ēi ̯, which would be likely to be abandoned (and its function predictably assumedby prep. + dative -ei) if competition between the morphs -ei ̯ and -i  had arisen in the consonant stems.

This is not to deny that these forms could be synchronically understood as-members of the fifth declension,reflected in the (dat.-)loc. sg. diie.i. ‘day’¹⁷ or the PN ka.n.te.s., dat. ka.n.te.i. But in the case of -kle(u)ēs or -genēs this pointhas been reached after two crucial phonetic changes have been completed, and not by way of stem substitution.


16 The form  mo.l.tevebo.s.(Tavola d’Este) is suggestive of a pervasive insertion of -e- before the endings beginning by a consonant. Cf. PRÓSPER 2018b for details.

17 Tavola d’Este, see MARINETTI 1998.


Celtic and Venetic in Contact: … 13


In the first case, the changes proceeded sofar as to compromise the identification of the root itself. Since neither rule of consonant loss has bled the input of the other, their relative chronology canhardly be established. This is a synoptic table of the relevant forms:


In conclusion,there is every reason to believe that IE /s/ had a contextually lenited pronunciation in Venetic. The possibility that it also affected Celtic should not be underestimated in view of a number of widespread isoglosses concerning  the  weakening  of  final  (see STIFTER 2012)  and  intervocalic sibilants  (PRÓSPER  2018c). Partial effacement of IE /s/ in northern Italy can be due to an areal process of debuccalization and eventual deletion of /s/ which, in the Venetic case atleast, may be step-by-step described as [z] > [z  ̞] > [ɦ] > ø. We can rephrase this conclusion in a broader perspective by starting from a Proto-Italic (or earlier) change [s] > [z] in intervocalic position and codas preceding a voiced sound (as contended by UNTERMANN 1968). In fact, intervocalic weakening of /s/ is expected in languages with a contrast /s/ vs. /s:/ because the speaker tends to maximise the articulatory distance between both sounds in order to preservethe contrast (cf. MENDEZ DOSUNA 1996). I shall come back to this problem in another work.


14 Blanca María Prósper

The first stages of this shift encompass Messapic. This dialect shows debuccalization of /s/ in initial and intervocalic position (rendered <h>), which at some point is totally effaced with the result that <h > comes to be used as hiatus-filler. It stands in contrast with a tense (not necessarily palatal) /s/ in codas and also in initial and intervocalic position. While the relevant literature is toovast to be reviewed here, it is generally assumed that intervocalic <s>, <ss>,<>, <š> and <šš> are used to note the reflex of *-si ̯ - (or *-ʃi ̯-). But alternative reconstructions for some names that look like agent nouns or past participles open up more promising possibilities: for instance, keošorrihi  (gen. sg., Ceglie, MLM 22 Cae)  is a likely cognate of the Ven. pseudo-gentilic  CAVSORIVS;¹⁸ poššes (nom. sg.), poššonnihi (gen. sg. from -ōn-(i)i ̯o-, Vaste, MLM  1 Bas) can go back to *po(s)-dh h₁-to- (cf. OCS. pozdě  ‘late’ < *pos-dh h₁-o -).

As regards Sabellic, Oscan preserved [z] in intervocalic position, but Umbrian shared with Latin a change [z] > [ɹ] > [r]. In codas, when preceding a sonorant, we find Latin and possibly Venetic [z] > [z. ] > [ɦ] > ø with compensatory lengthening of the preceding vowel, while Oscan and Umbrian still write <s>, which may or may not be reflecting an actual sibilant. This is consistent with the idea that an approximant-like [z̞] was weakened into [ɦ] in intramorphemic and intervocalic position at least and disappeared with compensatory length ending after syncopation in the superlatives (cf. Weiss 2017). We can reconstruct  a process -iz ̞ʊmo- > -iɦmo- > -īmo- that originated in Sabellic, perhaps favouredby the root bearing the stress, and progressively engulfed the northern regions, comprising the Celtic dialect of the Transpadana (probably through bilingualspeakers) and possibly the whole of Venetic (PRÓSPER 2018c). In Latin, however, the outcome -issimus  is irregular, and the sibilant was reinforced somewhere down the line, probably for pragmatic/expressive reasons.¹⁹


18 Cf. CAVSORIO FELICISSIMO, CAVSORIA VALENTINA (CIL V, 150, 584, Pula and Trieste).I trace them back to *keh₂u-dhh₁- ‘to strike’ in Lat. cūdō  and possibly cauda, caudex, causa. The underlying *kaussōr  designates a smith or a magistrate tasked with coin-age (cf. excūsor argentarius).

19 Cf.GUNKEL 2017:the sound pattern produced by regular sound change (pessimus, facillimus) is spread to the rest of the morphological category, aided in this case bythe fact that expressive lengthening is iconic, and thus favoured in comparative andsuperlative forms. 


Celtic and Venetic in Contact: … 15

3. IE  /p/  in the Venetic onomastics 
3.1. The Gaulish and Venetic stop system