CC BY
40
TWO NOTES ON OLD CELTIC MORPHOLOGY1
В статье обсуждаются два сложнейших вопроса древнекельтской морфологии: суффиксальное образование географических названий, созданных с помощью дентальных суффиксов *-ati- и *-et-. Из-за скудости языкового материала невозможно определить происхождение и функцию суффиксов, анализируя только внутренние языковые факты, потому необходимо воспользоваться комбинацией исторического и этимологического методов. Следует отметить, что при таком подходе смешивается образование апеллятивов и имён собственных, а также диахронически унаследованная функция суффиксов и их синхронная системная функция.
Происхождение сложного суффикса *-ati- следует искать в nomina agentis типа *gala-ti- (ср. др.-ирл. taid < *ta-ti-), от которых произошла притяжательная функция суффикса, встречающаяся в многочисленных вторичных десубстантивах: ср. этнонимы с суф. *-ates и эпитеты божеств типа Mars Sinatis*. Из этой вторичной адъективной функции суффикса позднее развились nomina agentis, зафиксированные как продуктивная категория в островном кельтском языке (тип др.-ирл. cetlaid). Кельтский языковой материал свидетельствует о различных функциях суффикса *-et-, в том числе притяжательной (часто в расширенном виде, прежде всего с элементом *-io-). В результате сравнения антропонима Bageto*, а также топонимов Brigetio* и Ворвпторауо^ можно заключить, что речь идет о атематическом дентальном суффиксе *-et-, который произошел от древних гистерокинетических nomina agentis типа др.-гр. apysx- ‘candidus’. Его употребление в функции десубстантивного притяжательного суффикса, безусловно, вторично и могло возникнуть в результате реинтерпретации вторичных десубстантивных nomina agentis типа др.-гр. шяо-т-, т. е. субстантивированных притяжательных прилагательных.
Ключевые слова: континентальный кельтский, топонимия,
словообразование, основы на дентальный согласный.
The state of research into Old Celtic (i.e. especially Gaulish) word-formation has been lamented on several occasions, yet given the nature of Continental Celtic linguistic data, it is no surprise that much concerning its interpretation still remains unclear and difficult. As much of the available material is onomastic, however, one would
1 I wish to thank Professor Natalia Belova for her generous help with the Russian abstract and to Dr. Alexander Falileyev subsequently. The text has been prepared with the input system ZRCola (http://ZRCola.zrc-sazu.si) developed by Dr Peter Weiss, a fellow of the Scientific research centre at the Slovene Academy of Sciences and Arts (http://www.zrc-sazu.si).
in addition need to bear in mind a clear-cut distinction between name-formation and word-formation, the former of which will as a rule be archaic/conservative and display different productivity, and/ or function of the morphemes involved. Needless to say, it is only possible to maintain such a distinction where the two systems are equally well documented so as to be liable to juxtaposition. Admittedly, such situations are sufficiently rare to make the endeavour seem futile, but there is an easy solution to the problem. As long as one does not make generalizations about Old Celtic word-formation on the basis of the data extracted solely from the onomastic material there can be no danger of mixing the two systems. There is a weakness to this approach, however. As we are dealing with a poorly documented linguistic system, the most pragmatic and often the only way to account for a particular structural phenomenon is by way of external comparison (etymology, in other words), hence drawing inferences about the synchronic function of a particular topoformant from its diachronic validity as a clearly word-formational element. As an illustration of the above, two difficult suffixal formations from toponymy will be discussed at some length, viz. *-ati- and *-et-, the first providing an example of a cross-systemically productive morpheme with an identifiable internal history, which can be fruitfully used to regulate the diachronically obtained data, the second allowing for diachronic interpretation only. It remains to be noted that due to the paucity of accessible material one cannot hope to document all four parameters of the suffixes involved, that is their function, valency, synchronic and diachronic productivity, and an overall pattern of areal distribution; furthermore, any of these parameters will be open to reconsideration with regard to any new data that may come to light in the future.
Derivatives in *-ati-
A large number of toponyms formed with this suffix properly belong to the Gallo-Roman period, during which the independent suffix -ate < *-ati enjoyed considerable productivity (cf. Dauzat 1946: 185-192, Negre 1990: 163-165). In order to assess correctly its original function(s), however, only the oldest set of data should be consulted as only such data is above suspicion of systematic corruption and can be assumed to reflect Old Celtic morphology proper. Imposing these limitations, there are a limited number2 of geogra-
2
The attestations and their sources are mainly quoted from AcS I—III, AE (L'Annee epigraphique), DCC, Isaac 2004, and PNRB.
phical names that can be safely considered to have been formed with the help of *-ati-: Bpiouaxn^ (scil. ^i^v) (Ptol. II, 8, 1) < *Bnu-ati-(cf. *Briuatensis, Briuate — Lambert 2000: 162), [G]lanat[e] (AE 1958: 225) < *Glan-ati-,3 Gesocribate (TP 1B1) < *Gaisokrib-ati-, Cambate (IA 386, 5) < *Kamb-ati-, 'Paxiaxov (Ptol. II, 7, 6) < *Rati-ati- (cf. uico Ratiatense, RACIATE)4, and cases such as Nantuaden-ses, Nantuadis (Delamarre 2012: 203, Negre 1990: 127), implying
*Nantu-ati-. It can be assumed without much doubt that these were derived respectively from (in the same alphabetical order) *brma ‘bridge’, *glano- ‘clean’, PN *Gaisokribo- (otherwise unattested as a personal name!), *kambo- (recte *kwmbo-) ‘crooked’, *rati- ‘earthen rampart, earthwork’, and *nantu- ‘valley’5. A significant number of undoubtedly related formations can be further observed in ethnonyms such as the often-quoted Abrincates, Atrebates, Boiates, Bouiates, Briuates, Carnuates, Cornacates, Dexiuates, Galates, Genuates, Hercyniates, Nantuates, and Tolosates, the latter of which has even been formed on the basis of a non-Celtic place-name (TLG: 207—8, LG: 35)6. The form BRIVATIOM, which as a part of an essentially Gaulish inscription (RIG II/1: L—3) = *Brmation is to be interpreted as the native genitive plural form, strongly implies that in the case of other attestation of this particular ethnonym, as undoubtedly also in the case of other ethnic names ending in -es, we are clearly dealing with a Latinized i-stem nom. pl. *-is < *-ei-es
3
Concerning the internal Gaulish development of -i to -e by weakening in absolute auslaut cf. more gl. mare (Gloss. Endl.) < *mori, suiorebe (RIG II/1: L-6) < *suesori-bi, anmanbe (RIG II/2: L-93) < *anman-bi, gandobe (instr. pl.) (RIG II/2: L-662) < *gando-bi etc. (LG: 41, McCone 1996: 112, Uhlich 1997: 295).
4 DCC: 185. Concerning the systematic translation of i-stem neuters singular into thematic neuters such as further witnessed in Apyevxopaxov (vs. KovSaxs) see Lambert 2000: 167.
For the elements in question see DLG2 and DCC, Principal Elements, s.vv.
6 Concerning the EN Gaesati (alongside several variants in -тої, -таї, for which see TLG s.vv.), i.e. rather than **Gaesates, the thematic plural must in all likelihood be considered original, the underlying *Gaesatos (cf. PNN Gesatus, Gesatius, Gesato-rix (OPEL II: 166, KGP: 215) and the recently discovered vocative Гп^атє on an inscription from Albania (Falileyev 2013: 69)), representing a barbatus type derivative *gais-ato- ‘having a spear, armed with a spear’ uel sim., to which compare PNN Giamatus (Delamarre 2007: 104) = og. GAMATI (M 258, Ziegler 1994: 181-182) to Gaul. *giiamo-/OI gam < *gamo- ‘winter’. Consider also LN BavvaTia (Ptol. II, 3, 8) formed from an underlying *bann-ato- (to *banna/*benna ‘top, summit, peak’) etc.
— — — 7 8
(*Brmatis < *Brmateies) , cf. vixioppoysi^ for the actual *Nitio-brogis < *-eies (to Gaul. *brogi- ‘place, land’)9 — a supposition that may be further supported by such forms as the Ptolemaic nom. pl. AxpsP&xioi, with a gen. pl. in -irov, and the existence of a purely artificial consonantal form AxpsPa^ (Ptol. II, 9, 4; Strabo 4, 3, 5; Cass. Dio 40, 43, 1) (AcS I: 269). Admittedly, a number of these ethnonyms may be secondary and can be attributed to one of the specializations of the suffix’s original function(s) (such formations, however, would be difficult to separate from the primary core responsible for the spread of the ethnic-name suffix, a good candidate perhaps being the EN Boiates, which appears beside the underived name of the Boii), moreover there is a problem of clearly non-Celtic ethnic names that seem to display the same suffix (consider, for example, Pannonian tribes such as the Oseriates or the Ara-biates etc.), but the fact that most of these ethnonyms are straightforward possessive derivatives from simple appellatives (^ topo-lexemes) or pre-existing geographical names — Abrincates ^ *abrink-ati- (to *abrinka or LN Abrinka)10, Briuates ^ *briu-ati- (to
*briua ‘bridge’ or LN *Briua)11, Carnuates ^ *karnu-ati- (to ^ ^ ^ 12 *karnu- ‘horn’), Nantuates ^ *nantu-ati- (to *nantu- ‘valley’) ,
Teutates ^ *tout-ati- (to *touta ‘people’), Cornacates ^ LN
*Kornakon, Genuates ^ LN *Gen(o)ua, Hercyniates ^ LN
13
*[P]erkunia , Tolosates ^ LN Tolosa — makes it indubitable that we are dealing with a genuine Celtic derivational pattern. It seems reasonable to assume, then, that the original function of *-ati- was desubstantival rather than deadjectival, which makes LNN such as Glanate and Cambate, insofar as these do not yet represent the later Gallo-Roman type, distort a coherent picture. That is unless, of course, the former name is really based on an already concretized rather than purely adjectival meaning of *glano- (cf. y^avsiKaPo < *Glan-iko- ‘belonging to Glanum’) and the latter is traced back to Gaul. *kambo- ‘riverbend’ (an underlying LN *Kambo- would of course be an equally likely possibility) not *kambo- ‘crooked’ = OI
7 LG: 59—60, cf. Irslinger 2002: 195.
8 With <ei^> = -is, cf. yXaveiKaPo (RIG I: G—64) for *Glan-iko-, or the usual Greek rendering of *-rixs as -pei^.
9 See LG: 59, DCC: 170
10 On this presumably Celtic element see De Bernardo Stempel 2005: 84.
11 Also attested in the gen. pl. as BRIVATIOM (RIG II/1: L—3).
12 Pace NWAI: 376, who presupposes Nantuates < *Nanto-at-es.
13 I.e. to *herku- < *perku- ‘oak’ (concerning the preservation of *p as h before front vowels see Oa^H^eeB 2006).
camb, for the substantivized meaning of which there is some convincing proof14.
Moving further to other onomastic subsystems, *-ati- is again encountered as a fairly productive means of formation of theonyms (largely native epithets of Romanized deities) such as Baginatiae* (Delamarre 2007: 37), Baginatis* (CIL XII/2383), Dunatis* (CIL XIII/2532), Sinatis* and Toutatis* (CIL III/5320), all of which seem to be (substantivized) adjectival formations, i.e. *bagin-ati-, *dun-ati-, *sin-ati-, *tout-ati- (to *baglno- ‘faginus’ (?) (*bago- ‘beech’),
*duno- ‘fort’, *sino- ‘?’ 15, and *touta ‘people’), comparable in form and function with Gaul. va^auoaxu; (RIG I: G—153) ‘Nemausensis’, which serves as an adjectival modifier of xouxiou^ [recte xouxio^j and seems fully interchangeable with similar possessive desubstantives in *-iko-, cf. ^axpsPo va^auoiKaPo (RIG I: G—203) ‘Matronis Nemausensibus’. Compare in this respect a clearly deonomastic divine epithet Excingiorigiatis* (AE 1935: 29), which finds a perfect structural match in Dubnocartiacos* (AE 1980: 641—644) ‘belonging to *Dubnokaratos’ (or *Dubnokara(n)tos, both of which are equally likely), not least for the fact that they both evince a structural peculiarity of deanthroponomastic adjectival derivatives in that they avoid being built directly from the underlying anthroponym (and are thus in effect deadjectival): *Exs-kingio=rIg-i-ati- ^ *E/s-
kingio=rig-io-, *Dubno=karat-i-ako- ^ *Dubno=karat-io-.
In Insular Celtic, as is well known, there is no trace of the adjectival function so clearly observable in Continetal Celtic material, its place having been occupied by a fairly productive variant *-i-atis16, employed in the formation of desubstantival agent nouns, e.g. OI cetlaid = OW centhliat < *kantl-iiatis ‘singer’ ^ *kantlom ‘song’ etc.17 This use is usually believed to be absent from Gaulish with the sole exception of aruerii'atin (RIG II/2: L—1002), which however is normally interpreted as an epithet, be it desubstantival or derived
14 Cf. *Cambodunum et sim. (DLG2, s.v. cambo-).
15 KPP: 53 considers OI sin gl. cruind, muince (DIL s.v.) to *sH2ei- (LIV2: 544), to which further compare Ved. siman- ‘Scheitel’.
16 The Insular Celtic conglomerate *-i-ati- (in effect *-iiati-) most likely spread from cases where the suffix was added to i-stems (cf. McCone 1995: 7), to which compare the identical process of resegmentation in the case of another productive agentive suffix *-(i)i-amon- (Stuber 1998: 147)), or io-stems, a case in point here definitely being Gaul. Boiates < *Boi-ati- to
*boios ‘fighter’ < *boi-o-.
17 See a more detailed discussion in NWAI: 374—380. On agent nouns in general see McCone 1995: 7.
1 R
from a place-name , and is as such superficially similar to the above cases. An equally problematic case is Gaul. *longatis ‘nautonier’ (Delamarre 2004: 122), hypothesised on the basis of a couple of arguably Celtic toponyms such as Longatico (IA 129, 1), and possibly *Longaticia (Delamarre loc. cit., 2013: 19), but again, even if such a noun did actually exist in Gaulish, as is likely, it would be directly comparable to the type recognized in Toutatis &c.
There was thus within Celtic a prominent group of primary desubstantival possessives most readily observable in toponymy (cf. Bpiouaxn^ ‘(place) belonging to a bridge’), which could, when reference was made to living entities, effortlessly acquire agentival function by way of superficial substantivization (cf. lat. socius ‘associate’, got. hai'rdeis ‘shepherd’, both originally deriving from possessive adjectives) and start off as a productive category in Insular Celtic, possibly leaving a few traces of the latter development in Gaulish. The key question is, however, whether the precedence of the adjectival function of *-ati- can really be maintained in the light of suffix’s genesis.
As is well known, the shape of the suffix was obtained by straightforward reanalysis of formations from roots with a final laryngeal (cf. NWAI: 377, Zair 2012: 196), i.e. < *-d-ti- < *-H3-ti-. The latter has its most natural starting point in a peripheral group of agent nouns with limited productivity, which sprung up from action nouns of the type seen in sti. stuti- ‘praise’ (itself probably structurally enlarged by the element *-i-, cf. *noku-t-(i-), Ved. sam=i-t-(i-) ‘meeting’), cf. OI taid ‘thief’ < *teH2-t-i- (originally ‘stealing’)19, OI flaith ‘ruler, prince’ (alongside the more elementary meaning ‘ruling, rule’) < *ulH3-t-i- etc. (IdgGr II/1: §318, McCone1995: 6, NWAI: 376 ff.). Alternatively, *-ti- could be traced back to a group of adjectival agent nouns of the type seen in Gr. nXfoq *‘swimming, swimmer’ < *pleH3-t- (LIV2: 485), OInd. rit-‘flowing’ < *H3riH-t-, OI nia ‘warrior’ < *neiH1-t- etc. 20 ,
18 Cf. for example Lambert 1996: 53 (=LG: 152), DLG2 s.v. arueriatis.
19 Cf. the argument put forward by the author (Jezikoslovni zapiski 16/2 (2010), 165—173) to trace back the likes of OI tathae (gen. sg. to taid) <
*tatantos to a participial derivative *teH2-t-eH2-nt- > ta-t-ant- of a denominative *teH2-t-eH2- (Gr. x^xn ‘stealing, thieving’), based on a structurally yet unenlarged stem.
20 See IdgGr II/1: §313, Risch 1974: §72, Rieken 1999: 85—8, NWAI: 171 ff., Irslinger 2002: 45—6. To all appearances, this type of agent nouns goes back to decompositional simplicia arising from secondary exocentric (re)interpretations of tatpurusas with verbal abstracts as second members,
subsequently remodelled by analogy with secondary agent nouns in
*-ti- (cf. nWAI: 376—377 for a similar line of thought). Be that as it may, leaving aside the larger part of Old Celtic ethnic names in *-ati-, which indicate a clear desubstantival origin (Nantuates ^ *nantu-ati- etc.), there are at least two derivatives which may in fact conceal a deverbative origin. In the case of Atrebates, i.e. *Atrebatis, it would of course be possible to start from the substantival *ad-treba ‘dwelling-place’ uel sim. (cf. OI treb, W tref ‘id.’ < *treb-a), in which case this would be yet another case of secondary derivation fully parallel to e.g. *Brm-ati- ^ *briua, but it would be equally possible to derive it from the denominative verb *ad=treb-a- (cf. OI ad-treba ‘inhabits’), a deverbative nomen agentis *ad=treb-a-ti- thus signifying a ‘dweller, inhabitant’ uel sim. If the derivational history of the name Atrebates is de facto much too ambiguous to afford reliable conclusions, rakaxnq for *Galatis (see above) < *Gala-ti- to
*gelH-ti-21 most clearly cannot be convincingly argued to be desub-stantival, nor can it be explained away as a secondary creation after the fashion of other ethnic names in *-ati-,22 thus providing us with a valuable piece of internal evidence relating to the origin and spread of Celtic *-ati- in all its subsequent functional and semantic shifts.
Derivatives in *-et-
Formations involving this suffix form a heterogeneous group. Instantly recognizable are a set of deverbatives in *-eto- such as Gaul. vs^nTov (RIG I: G—153) < *nem-eto- (= OI nemed gl. sacellum), Gaul. geneta ‘(young) girl’ (DLG2: 176) < *genH1-eto-, to which compare Gr. ysvsx^, Lat. genitus (Vine 1998: 56 ff.), and to all appearances also *kal-eto- (Joseph 1982: 40) ‘hard’ > OI calad ‘id.’, PN Caletus23, EN Caletes (if thematic, cf. var. -i, -oi (AcS I: 696)), in which case it is to be clearly separated from the purely agentival OE hwle, -ep- ‘hero, champion’ < *kal-et- (!).
There is further a coherent group of the derivatives in *-et-mostly supplied by onomastic data (this however is not significant as the imbalance is no doubt purely a matter of attestation), which
cf. Ved. °stut- ‘praising X’ ^ ‘praising X, who praises X’, °drs- ‘watching X’ ^ ‘(the one) watching X’ etc.
21 *gelH- (LIV : 185—186, Schumacher 2004: 324—326). Unproblematically via *gela-ti- by Joseph’s Law.
22 Irslinger 2002: 202 contra (in my view without any convincing justification) Schumacher op. cit., p. 326.
23 Beside *k( )al-et-io- as in Caletius (OPEL II: 23) = OI Cailte , cf. gen. sg. Og. CALLITI (M 81).
represent a productive pattern of agent noun formation, vestigially known from Insular Celtic as well, cf. OI rig, -ed ‘fore-arm’< *H3rg-et-, cing, -ed ‘warrior’ (Irslinger 2002: 58—59). It is widely accepted that these go back to PIE hysterokinetic agent nouns of the type apy^, -ex- ‘shining’ < *H2rg-et-,24 which is indeed likely given the invariably full-grade structure of the suffix and the overall agentive semantics of such nouns. The latter becomes particularly apparent in case of functionally overlapping derivatives such as Mogetius (OPEL III: 85) < *-et- beside Mogons* (Delamarre 2007: 135) < *-ont-25 (to
*magh-) or, possibly, Loucetius beside OI loichet (ntr.!) ‘lightning-flash’ < *leuk-nt26. That most of these nouns represent uninherited formations based on an internally-driven pattern with secondary productivity (as opposed to OI rig, for example)27 is evident from the consistency with which they are built from full-grade roots:
*dem(H2)-et-, *derk-et-, *k/ghengh-et-, *H3erg-et-, *leuk-et-, *magh-et-, *nem-et-, *segh-et-, * uenH1-et- 28 . This more or less comprehensive list of nomina agentis has been based on An^xai (Ptol. II, 3, 12), Dercetius* (CIL II/5809)29, gobedbi (instr. pl., RIG
24 IdgGr II/1: §313Y, Rieken 1999: 86—9, Irslinger 2002: 45—47. No_t to be sought in Irish (Early Goidelic) Apyrca (Ptol. II, 2, 2) = /Argeda/ < *Arganta (Pokorny 1950: 130, Sims-Williams 2000: 6, idem 2007: 329), cf. CIb. PN arkanta, pace De Bernardo Stempel (2000: 103), who fails to offer convincing counter-arguments in favour of an otherwise thoroughly unjustifiable reconstruction tArg-et-a.
25 See De Bernardo Stempel 2003: 43, 53—54.
26 Vs. W lluched < *leuk-s-et-o- (IdgGr II/1: § 411).
27 Cf. Rieken (1999: 88), who traces the innovation back to Indo-European. But much of the material adduced in support of this claim belongs to the hysterokinetic type proper (type OI fili, Gr. Ke^c;, -^t- etc.). There exists, however, a clear divide in Celtic between such inherited e-grade deverbatives that preserve the original lenghtened grade in the suffix and the non-ablauting type as represented by *king-et-. Also worth noting in this respect is Irslinger's suggestion (2002: 68) that the proterodynamic and hysterodynamic patterns here clearly influenced each other and that, ultimately, «die Zuordnung des kel.[tischen] Materials zum einen oder anderen ist nicht moglich».
28 I do not include here the puzzling OI tocad ‘fate, luck’< *tonk-eto- (= W tynged, cf. PN Tongeta, Og. TOGITTACC (M 172)) and OW ocet ‘harrow’
< *H2ok-Vta (cf. OWED: 124, NIL: 229—300, Anreiter (perhaps too optimistically) 2001: 162), not only because they obviously represent thematic(ized) derivatives as opposed to athematic agent nouns in -et-, but mainly due to their perplexing o-grade, which must be left unaccounted for.
29 The exact interpretation of this hapax is difficult (cf. DCC: 113). As an oronym it can hardly have carried the agentive meaning, although perhaps it could originally have been detheonymic. However, it is also doubtful whether the underlying derivative is indeed deverbal rather than formed
II/1: L—13) < *gobet-bi (cf. Hamp 1988a: 54), Cinget-o-rix (KGP: 171—172) beside Cingetius (CIL XIII/3707) (cf. Cingetissa (RIB:
30
630), the corresponding feminine hypocoristic) , Leucetius/Loucetius (Delamarre 2007: 119—120), Mogetius (see above), Nemetes (AcS II: 708 ff.), Orget-o-rix (KGP: 252—253) Orgetius (CIL III/11803), Segeta (Delamarre 2007: 163), Uenetes (AcS III: 160 ff.). Extensions by the element -io- are most straightforwardly explained as purely structural enlargements (i.e. by analogy with the dominant class of compound and simplex agent nouns in *-io-, cf. Russell 1984: 37 ff., McCone 1995: 5 ff.); compare the similar situation in Curmi-sagius (AE 1939: 260) < *°sag-io-, with its corresponding simplex (see Uhlich 2002: 418) Sagius (AcS II: 1288) occurring alongside the originally athematic *sag- as attested in EN Tecto-sages.31 Also here perhaps belongs Gaul. *trag-et-io- ‘foot’ (reconstructed on the basis of Late Gaulish treide gl. pede (Gloss. Endl.)32, cf. Briotreide if for *Briuo-tragetion (DLG : 299)) beside OI traig ‘foot’ < *tragh-et- (Irslinger 2002: 59—60), with a straightforward semantic specialization of the more primary ‘Laufer’ (Rieken 1999: 89).
Much more problematic, however, are a group of desubstantival toponyms, which at face value seem to employ the same suffix:
Brigetio* (IA passim, see DCC: 80—81), cf. Bpiyaixiov (Ptol. II, 15, 3)
Bodetia33 (IA 294, 1)
BopPnxo^ayo<; (Ptol. II, 9, 9)
Cunetio* (AI 486, 5)
(?) Dercetio* (CIL II/5809)
Duretia* (= Duretie (TP 1B2))
TaPp^xa (sc. ft^n) (Strabo 7, 1, 5)
Lutetia34
from *derkom ‘eye; look’ (cf. OI derc), to which a number of other names clearly belong, cf. Condercos, Derceia, Inderca etc. (KGP: 179, 192, Stuber 2005: 78).
The suffixless variant *kingo- is irrelevant in this context as it is originally limited to verbal-governing compounds such as *Exs=king-o-(OPEL II: 130) < *°king-o- (cf. var. Excingius (AE 1973: 362), reinforced with the augmentative agentive suffix *-io-), to which compare the structurally identical Andegenus (Delamarre 2007: 21) or Nemetogena (CIL XIII/603) < *° gen-o-.
31 To *sag- ‘look for, search’ (Schumacher 2004: 555).
32 Although a simple thematic *trageton, albeit isolated, is not entirely excluded, cf. prenne gl. arborem grandem (Gloss. Endl.) for original
*prennon ‘tree’ (see DLG2 s.v. prenno-).
33 Beside Bodecia (see DCC: 77).
Saletio* (IA 354, 6)
TaZyaixiov (Ptol. II, 12, 3)
OmKsxia (Strabo 5, 1, 8; Ptol. III, 1, 26)
Of these, Lutetia is far too uncertain to be further included in the discussion35, and the interpretation of Dercetio* remains difficult (see above). OmKsxia, on the other hand, if relevant at all (cf. var. OrnKsvxia as early as Strabo, and ultimately also the modern-day Vicenza < *Vikentia), can only be conceived as a second-grade derivative from a (purely hypothetic) *uik-et- ‘fighter’ uel sim., a (transitive) agentival meaning being inappropriate for a place-name 36 . However, in view of the limited Celtic imprint on the onomastic landscape of Cisalpine Gaul37, the name itself does not have to be interpreted as a Celtic coinage at all (the same goes for the equally problematic Bodetia), a circumstance which perhaps receives additional support from comparison with such clearly non-Celtic toponyms as ’Ercsxiov (Gl. Krahe/Meid 1967: 177), including the fact that we would probably expect the underlying agent noun to have been a full-grade derivative.
The remaining cases may be confidently traced back to *briga ‘hill’ (or *brig- ‘id.’),38 *boruo- ‘water-spring’, *gabro- ‘goat’,*kun-
34
For various differing attestations see DCC: 154.
35 The attestations disagree wildly, cf. older variants of the name such as Lutecia (Caes., Gall. passim) beside Ae/ouKoxeKia in Ptolemy and Strabo (for a detailed account of these forms see DCC: 154). The latter is usually thought to derive from *luk-ot- ‘mouse’ = OI luch etc. (see Irslinger 2002: 61—62), to which Lutecia could perhaps represent an allegro form, but then the question of how to account for the suffix remains. *-eKia, however, cannot find a suitable explanation within Celtic and must surely be due to scribal corruption (note a similar oscillation, though on a significantly smaller scale, between t and c in Bodetia ~ Bodecia). Lambert (2005: 246) suggests *-ik-ia, derived from an underlying DN/PN *Lukot-iko-, but in light of the above, a desubstantival base *lukot-et- is perhaps a not unlikely alternative.
36 With the notable exception of hydronyms. Cf. De Bernardo Stempel (2000: 94), who proposed to translate the name as ‘the Fighting town’.
7 Contrary to the impression maintained by De Bernardo Stempel 2000: 92—96, 105—106.
38 In favour of the apparent coexistence of these two homonymic variants in Gaulish (as opposed to Insular Celtic and Hispano-Celtic, which either point in the direction of a root-noun (OI bri, CIb. nertobis) or its thematized variant (W bre etc.), but note the peculiar case of Old British BRIGE as if for athematic *Brig-e (Hamp 1991/92: 9, contrasting PNRB: 277—278)) consider at least AXioPpi^ (Pt. III, 10, 5) < *Alio-brixs (Falileyev 2007: 4— 5); the numerous compounds with *Brigo- as their first element are, of course, ultimately ambiguous.
‘dog’, *sal- ‘salt’ (?),*tasgo- ‘badger’ (?) ,which at face value seem to have been further derived with the help of *-etio- or a simple thematic *-eto-. On closer inspection, however, it appears that apart from TaPp^xa and the compounded form in Boppnxo-^ayot; (to *magos ‘plain’ uel sim.) the exact shape of the suffix in question was
*-etion-, i.e. *-eti-on-, enlarged by the nasal individualizing suffix *-on- of the type nicely observable in EN Senones, literally ‘The old’, cf. Germanic Burgundiones < *bhrg -nt-i-on- (Bichlmeier 2009/10) etc. The corresponding Ptolemaic forms ending in what appears to be a simple thematic -iov need not present a complication since nasal stems are as a rule only exceptionally recorded as such in Greek sources. The underlying *brig-etio-40, *kun-etio-, *sal-etio-, *tasg-etio- are thus immediately comparable to structurally identical anthroponyms such as Og. CuNITTI (M 149) < *Kun-etio- (Ziegler 1994: 162), Cetetiu* (CIL III/3861, KPP: 263) < *Kait-eti-on-(fem.!), Oxetius (CIL XVI/123, KPP: 239) < *Oxs-etio- (to *oxsi-‘?’)41, and Tasgetios (RIG IV: M—159, OPEL IV: 109) < *Tasg-etio-, the isolated compositional form BopPnxo° perhaps finding its exact parallel in *Trougetomaros if the latter is to be correctly restored for the actual Troucetimaros (CIL III/3367 etc., KPP: 119, KGP: 282), cf. Troucetissa (CIL III/14349, 8, KPP: 207)42. The crucial form for the present discussion, however, seems to be Bageto* (KPP: 219) <
*Bag-et-on-, which exhibits a plain suffix *-et-, unenlarged by *-io-,
39 See DLG2, PNPG (Celtic Elements), and DCC (principal Elements) s.vv.
40 The latter formation received an unnecessarily complicated treatment by Hamp 1990: 59—62, who suggested that we see in it a remodelled deverbative.
41 Gl. DLG2, s. v. oxso-, oxi- ‘boeuf’, ‘haut’, both weak.
42 Cf. Mogetimarus (Delamarre 2007: 134) and Orcetirix (KGP: 252) for
*Moget-o- and *Orget-o- with -i- by vowel harmony (on this phenomenon and composition vowels in general see most recently Sims-Williams 2013, forthcoming). It is conspicuous, though, that all attestations of the Trouceti-names are limited to Noricum and Pannonia, as is, for instance, the peculiar hapax Mogitmarus (CIL III/3325, KPP: 111) with uncertain restoration, and that it may not be at all coincidental that one feels rather uneasy attributing the problematic -i- in Mogetimarus (never attested as +Mogetomarus) to very much the same factors that produced the likes of Orgetirix, which, incidentally, is indeed attested in its “expected” form Orgetorix. Local peculiarities in popular names are known otherwise (cf. Sims-Williams, op. cit.), cf. Magimarus (several attestations in Noricum and Pannonia, see Corpus (NOR 47, 50, PAN 4, 27, 83, 125)) vs. Magiomarus (CIL III/11579), which cannot represent a morphological doublet. On the other
hand, there are other similar cases of -o------i- alternations that are difficult
to account for phonetically, cf. Uictimaria (CIL XII/344) for Uicto- < *uik-to- (Stuber 2005: 97) etc.
the latter then being clearly separable as a independent structural element in all other cases too. Nevertheless, the semantic interpretation of the underlying athematic *brig-et-, *bom-et- (-o- can now be interpreted as the composition vowel, cf. Cinget-o-rix beside OI cing
< *king-et-)43, *gabr-et-, *kun-et-, *sal-et-, *tasg-et- is difficult as there seems to be no further internal evidence that would help to regulate the exact meaning of these derivatives. If, however, *Bageton- is interpreted as an agent noun to *baga ‘battle’ = OI bag ‘id.’44, a common noun such as *bago- ‘beech’ being less probable in a personal name, it is possible to immediately compare it with such secondary desubstantival agent-noun formations as evidenced by Gr. inno-x-n^ ‘eques’ < *inno-x-a^ (— *inno-x- *‘eques’)45. As these could in turn be easily interpreted as possessive derivatives, the way was open for a generally possessive suffix *-et- to establish itself beyond its original domain.
Although a basically possessive meaning of these onomastic derivatives in *-et-46 is beyond question, it cannot be entirely ruled out that they in fact represent collective nouns. A supposed collective function of *-et- would easily derive from its more elementary employment as a possessive suffix, cf. OI daire ‘grove’, Gaul. *dario- (Cf. Hamp 1988: 125—127) *‘quercetum’ < *‘belonging to *daru- , Gaul. lN BayaKov (Pt. II, 9, 6) ‘fagetum’ = *Bag-ako-*‘belonging to *bago-, cf. OI craebach ‘having branches’/ ‘branches’ (Russell 1990: 92—93) etc. However that may be, a suggestion that we may be (at least in some if not all cases) dealing with collectives would also allow for a straightforward interpretation of the ubiquitous sequence -i-on-47 as a formal substantivization/ concretization of possessive adjectival derivatives, so that ultimately *X-et-i-on- = ‘the (place) to which belongs a collectivity of X’.
The final question is whether it can be well maintained that the combination of suffixes *-io- + *-on- was the result of an actual two-
43 Contrast DCC (Principal Elements, p. 10), where it was decided, in my view erroneously, to list *borbeto- as a separate element (similarly, however, De Bernardo Stempel 2005: 79, 90).
44 As indeed correctly surmised by Meid (KPP: 219). For Celtic *baga see LEIA: B 4—5, DLG2 s.v. bagauda.
45 For the latter see Risch 1974: 31 ff.
46 Along the same lines already Anreiter (2001: 203 et passim) and Greule (2007: 200), who considered ‘durch erhohte (Lage) characterisiert’ and ‘Stelle, wo es Salz gibt’ (for Brigetio* and Saletio* respectively), albeit without any prerequisite formal backing for their predictive assumptions.
47 Were the underlying derivatives still adjectival in nature,*-io- would have to be interpreted as a purely structural element.
grade derivational chain, or whether we are in fact dealing with an autonomous conglomerate suffix, identical to the desubstantival agentival *-ion- of OHG burgio ‘citizen’ or Goth. fiskja ‘fisherman’. Its function in Celtic, however, would apparently still have equalled the sum of the functions of its constituent parts, which in turn makes it difficult, if not impossible, to decide in favour of either of the two options. The existence of an independent suffix *-ion- has been claimed by Stuber (1998: 102) in order to account for the derivation of Amirov (OI Albu*) (PNRB: 247), which, as she pertinently remarks, cannot be satisfactorily explained as a nasal derivative of
*Alb(i)io- ‘world’ (on *albiio- see Hamp 1989). There is an alternative solution, however. In view of the Old Celtic hydronym rXaviq (AcS I: 2024), which, going back to a deadjectival i-stem derivative (*glano—► *glani-), is immediately comparable to OI glain ‘glass, clearness’, it would not be inconceivable to see in *Albi-io- (to which a derivative in *-on- would represent a simple formal substantivization) a possessive adjective and trace it back to *albi- ‘whiteness’ uel sim. (to-*H2el-bho-), cf. RN Albis* (gen. A^Pio^, Ptol. II, 11 passim).
Abbreviations
CIb. = Celtiberian; DN = divine name; EN = ethnic name; Gaul. = Gaulish; Goth. = Gothic; Gr. = Ancient Greek; Lat. = Latin; LN = place-name; OI = Old Irish; OInd. = Old Indic; OW = Old Welsh; PN = personal name; RN = river-name; Ved. = Vedic; W = Welsh.
Bibliography
AcS — Holder A. Alt-celtischer Sprachschatz I—III. Leipzig: Verlag von B. G. Teubner, 1896—1907 (Graz: Akademische Druck- und Verlagsan-stalt, 1961—1962).
Corpus — Sims-Williams P., Raybould M. E. A Corpus of Latin Inscriptions of the Roman Empire containing Celtic Personal Names. Aberystwyth: CMCS, 2007.
DCC — Falileyev A. in collaboration with Gohil A. E. and Ward N. Dictionary of Continental Celtic Place-Names. Aberystwyth: CMCS, 2010. DLG2 — Delamarre X. Dictionnaire de la langue gauloise. 2nd edition. Paris: Editions Errance, 2003.
IdgGr II/1 — Vergleichende Laut-, Stammbildungs- und Flexionslehre nebst Lehre vom Gebrauch der Wortformen der indogermanischen Sprachen II: Lehre von den Wortformen und ihrem Gebrauch. 1. Teil: Allgemei-nes. Zusammensetzung (Komposita). Nominal stamme. Strassburg, 19062.
IEW — Pokorny J. Indogermanisches etymologisches Worterbuch. 2vols, Bern and Munchen: Francke, 1959—1969.
KGP — Schmidt K. H. Die Komposition in gallischen Personennamen // Zeitschrift fur celtische Philologie. 1957. Bd. 26. S. 33—301.
KPP — Meid W. Keltische Personennamen in Pannonien. Budimpesta: Archeolingua, 2005.
/
LEIA - Vendryes J., Bachellery E., Lambert P.-Y. Lexique ёtymologique de l’irlandais ancien. Dublin: Dublin Institute for Advanced Studies / Paris: Centre National de la Recherche Scientifique, 1959-1996.
LG - Lambert P.-Y. La langue gauloise. Paris: Editions Errance, 2003.
LIV2 - Rix H., Kummel M., Zehnder Th., Lipp R., Schirmer B. Lexikon der indogermanischen Verben. Die Wurzeln und ihre Primarstamm-bildungen. 2nd edition. Wiesbaden: Dr. Ludwik Reichert Verlag, 2001.
NWAI - De Bernardo Stempel P. Nominale Wortbildung des alteren Iri-schen. Stammbildung und Derivation. Tubingen: Max Niemeyer, 1999.
OPEL - Onomasticon Provinciarum Europae Latinarum / Barnabas Lorincz et alii (eds.). Vol. I-IV. Wien, Budapest, 1994-2002.
OWED -Falileyev A. Etymological Dictionary of Old Welsh. Tubingen: Max Niemeyer Verlag, 2000 (= Buchreihe der Zeitschrift fur celtische Philologie 18).
PNRB - Rivet A. L. F., Smith C. The Place-Names of Roman Britain. London, 1979.
Ptolemy 2000 = Patrick Sims-Williams, David N. Parsons (eds.). PTOLEMY: Towards a Linguistic Atlas of the Earliest Celtic Place-Names of Europe. Papers from a workshop, sponsored by the British Academy, in the Department of Welsh, University of Wales, Aberystwyth, 11-12 April 1999, Aberystwyth: CMCS, 2000.
Ptolemy 2005 - de Hoz J., Lujan E. R., Sims-Williams P. (eds.). New approaches to Celtic place-names in Ptolemy’s Geography. Madrid: Ediciones Clasicas, 2005.
RIB - Colingwood R. G. et al. The Roman inscriptions of Britain. Vol. I-II. Stroud: Sutton, 1990-1995.
RIG - Recueil des inscriptions gauloises. Vol. I: Lejeune M. Textes gallo-grecs. 1985. Vol. II, 1: Lejeune M. Textes gallo^trusques. Textes gallo-latins sur pierre. 1988. Vol. II, 2: Lambert P.-Y. Textes gallo-latins sur instrumentum. 2002. Paris: CNRS Editions, 1985-2002.
TLG - Billy P.-H. Thesaurus linguae Gallicae. Hildesheim; Zurich; New York: Olms-Weidmann, 1993.
Anreiter 2001 -Anreiter P. Die vorromischen Namen Pannoniens. Budapest: Archaeolingua, 2001.
Bichlmeier 2009/10 - Bichlmeier H. Bairisch-osterreichische Orts- und Gewassernamen aus indogermanistischer Sicht // Beitrage zur Orts-namenforschung. 2009/10. Bd. 46. S. 3-63.
De Bernardo Stempel 2000 - de Bernardo Stempel P. Ptolemy's Celtic Italy and Ireland: A linguistic analysis // Ptolemy. 2000. P. 83-112.
De Bernardo Stempel 2003 - de Bernardo Stempel P. Die sprachliche Analyse keltischer Theonyme // Zeitschrift fur celtische Philologie.
2003. S. 41-69.
De Bernardo Stempel 2005 - de Bernardo Stempel P. Ptolemy's evidence for Germania Superior // Ptolemy. 2005. P. 71-94.
Delamarre 2004 - Delamarre X. Gallo-Brittonica: Transports, richesse et gёnёrositё chez les anciens Celtes // Zeitschrift fur celtische Philologie.
2004. Bd. 54. S. 121-132.
Delamarre 2006 - Delamarre X. Gallo-Brittonica (suite: 11-21) // Zeitschrift fur celtische Philologie. 2006. Bd. 55. S. 29-41.
Delamarre 2007 - Delamarre X. Noms des personnes celtiques dans l^pigraphie classique. Pariz: Editions errance, 2007.
Delamarre 2012 - Delamarre X. Noms de lieux celtiques de l’Europe ancienne (-500/+500): Dictionnaire. Pariz: Editions errance, 2012.
Delamarre 2013 - Delamarre X. Longidienus, faber navalis, a Ravenne, le toponyme Lombard et le theme longo- ‘navire’ en vieux celtique // Zeitschrift fur celtische Philologie. 2013. Bd. 60S. 19-26.
Фалилеев 2006 - Фалилеев А. И. Галльск. s, и.-е. ♦p и некоторые аспекты интерпретации контынентальных кельтских данных // Индоевропейское языкознание и классическая филология 2006. T. 10. C.306-313.
Falileyev 2007 - Falileyev A. Celtic Dacia. Aberystwyth: CMCS, 2007.
Falileyev 2013 - Falileyev A. Celtic Balkans. Aberystwyth: CmCs, 2013.
Greule 2007 - Greule A. Etymologische Studien zu geographische Namen in Europa. Ausgewahlte Beitrage 1998-2006 / Janka W., Prinz M. (eds.). Regensburg: edition vulpes, 2007 (= Regensburger Studien zur Namenforschung 2).
Hamp 1988 - Hamp E. Varia XXXIII: Old Irish daire ‘grove’ // Etudes Celtiques. 1988. Vol. 25. P. 125-127.
Hamp 1988a - Hamp E. (Aber) Gefenni // Bulletin oft he Board of Celtic Studies. 1988. Vol. 35. P. 53-54.
Hamp 1989 - Hamp E. Welsh elfydd, elfydden, Scottish Gaelic Alba // Bulletin oft he Board of Celtic Studies. 1989. Vol. 36. P. 109-110.
Hamp 1990 - Hamp E. Brigetionem, Bspyrnrov // Nyelvtudomanyi Kбzlemёnyek. 1990. Vol. 91. P. 59-62.
Hamp 1991/92 - Hamp E. British Celtic BRIGE and morphology // Studia Celtica. 1991/92. Vol. 26/27. P. 9-11.
Irslinger 2002 - Irslinger B. S. Abstrakta mit Dentalsuffixen im Altirischen. Heidelberg: Universitatsverlag C. Winter, 2002.
Isaac 2004 - Isaac G. R. Place-Names in Ptolemy’s Geography. An electronic data base with etymological analysis of the Celtic name-elements (CD-ROM). Aberystwyth: CMCS, 2004.
Krahe/Meid 1967 - Krahe H., Meid W. Germanische Sprachwissenschaft III: Wortbildungslehre. Berlin: Walter de Gruyter & Co., 1967.
Lambert 2000 - Lambert P.-Y. Remarks on Gaulish place-names in Ptolemy // Ptolemy. 2000. P. 159-168.
Lambert 2005 - Lambert P.-Y. The place-names of Lugdunensis [Лоиубо'цу'^а] (Ptolemy II 8) // Ptolemy. 2005. P. 215-251.
Lambert 2006 - Lambert P.-Y. Grandes textes magiques: Chamalieres, Larzac // Die grosseren altkeltischen Sprachdenkmaler. Akten des Kolloquiums Innsbruck, 29. April-3. Mai 1993 / Meid W., Anreiter P. (eds.). Innsbruck: Verlag des Instituts fur Sprachwissenschaft der Universitat Innsbruck, 1996. P. 51-85.
McCone 1995 - McCone K. OIr. senchae, senchaid and preliminaries on agent noun formation in Celtic // Eriu. 1995. Vol. 46. P. 1-10.
McCone 1996 - McCone K. Der Teller von Lezoux // Die grosseren altkeltischen Sprachdenkmaler. Akten des Kolloquiums Innsbruck, 29. April-3. Mai 1993 / Meid W., Anreiter P. (eds.). Innsbruck: Verlag des Instituts fur Sprachwissenschaft der Univ. Innsbruck, 1996. P. 107-117.
Negre 1990 - Negre E. Toponimie ^^rale de la France I: Formations preceltiques, celtiques, romanes. Zeneva: Libraire Droz, 1990.
Pokorny 1950 - Pokorny J. The river-name Argita // Journal of Celtic Studies. 1950. Vol. 1. P. 130.
Rieken 1999 — Rieken E. Untersuchungen zur nominalen Stammbildung des Hethitischen. Wiesbaden: Harrasowitz, 1999 (Studien zu den Bogaz-koy-Texten 44).
Risch 1974 — Risch E. Wortbildung der homerischen Sprache. Berlin; New York: Walter de Gruyter, 1974.
Russell 1989 — Russell P. Agent suffixes in Welsh: Native and non-native // Bulletin of the Board of Celtic Studies. 1989. Vol. 36. P. 30—42.
Russell 1990 — Russell P. Celtic Word-Formation. The Velar Suffixes.
Dublin: Dublin Institute for Advanced Studies, 1990.
Sims-Williams 2000 — Sims-Williams P. Degrees of Celticity in Ptolemy's names: Examples from Wales // Ptolemy. 2000. P. 1—16.
Sims-Williams 2007 — Sims-Williams P. Common Celtic, Gallo-Brittonic and Insular Celtic // Lambert P.-Y., Pinault G.-J. (eds.). Gaulois et Celtique continental. Geneve: Droz, 2007. P. 309—54.
Sims-Williams 2013 — Sims-Williams P. The Celtic composition vowels -o-and -io-, forthcoming.
Stuber 1998 — Stuber K. The Historical Morphology of n-Stems in Celtic, Maynooth. Department of Old Irish, National University of Ireland, Maynooth, 1998 (Maynooth Studies in Celtic Linguistics III).
Stuber 2002 — Stuber K. Die primaren s-Stamme des Indogermanischen.
Wiesbaden: Dr. Ludwig Reichert Verlag, 2002.
Stuber 2005 — Stuber K. Schmied und Frau. Studien zur gallischen Epigraphik und Onomastik. Budapest: Archeolingua, 2005.
Uhlich 1997 — Uhlich J. Zur sprachlichen Einordnung des Lepontischen // Akten des zweiten deutschen Keltologen-Symposiums, Bonn, 2.—4. April 1997 / Zimmer S., Kodderitzsch R., Wigger A. (eds.). Tubingen, Niemeyer, P. 277—304.
Uhlich 2002 — Uhlich J. Verbal governing compounds (synthethics) in Early Irish and other Celtic languages // Transactions of the Philological Society. 2002. Vol. 100/3. P. 403—433.
Vine 1998 — Vine B. Aeolic opnexov and deverbative *-eto- in Greek and Indo-European. Innsbruck: Institut fur Sprachwissenschaft der
Universitat Innsbruck, 1998.
Zair 2012 — Zair N. The reflexes of the Proto-Indo-European laryngeals in Celtic. Leiden, Boston: Brill, 2012.
Ziegler 1994 — Ziegler S. Die Sprache der altirischen Ogam-Inschriften. Gottingen: Vandenhoeck&Ruprecht, 1994.
L. Repansek Two notes on Old Celtic Morphology
The article discusses two aspects of Old Celtic morphology, focusing on the derivation of place-names in *-ati- and *-et-. As both suffixes share a complicated history, which due to the sparse attestation of Old Celtic can only be glimpsed from an occasional set of (mostly onomastic) data, their morphological and semantic functions are more narrowly specified with the help of external data in combination with careful observation of internal linguistic facts.
Keywords: Continental Celtic, toponymy, word-formation, dental stems.
