59
Luka Repanšek               UDK 811.163.6'373.21(497.473)
University of Ljubljana*   DOI: 10.4312/linguistica.55.1.59-72
TOWARDS A CLARIFICATION OF THE HISTORY 
OF THE SLOVENE RIVER NAME SOČA
1.  COMMON SLOVENE *SÓČA
The immediate source of the Common Slovene river name *Sa (= dial. (Natisone) 
Sóːča, Resian (Bila) Sóča)1 may be unproblematically reconstructed as Slavic *Sǫa, 
regularly continuing an older *Sunā. The original accentual pattern of this feminine 
noun cannot be recovered but such ambiguity is typical of the western and southern 
Slovene feminine a-stems with a long root vowel. It is likely that the name originally 
belonged to the immobile accentual paradigm I/1, so *Súnā > *Sa (> *Sa > 
*Sa), as is the case with, e.g., *Stъ/ьla (Sotla) < *Súntu/ilā and in clear opposition 
to the coherent group of river names with simple disyllabic structure *(C)C1VC2ā-, 
which feature regular circumflexion and accompanying (in my view undoubtedly 
internally Slavic) lengthening of the originally accented vowel, e.g. *Ziĺ < *Ʒĺā,2 
*Sa < *Sā, *Dra < *Drā, *Rab < *Rbā, and perhaps Γáːna < *Glána 
(< *Glan?) < *Glānā (= *Glnā?).3 The substitution of Slavic short *u for Romance 
* Oddelek za primerjalno in splošno jezikoslovje/Department of Comparative and General Lin-
guistics, Filozofska fakulteta, Aškerčeva 2, 1000 Ljubljana, Slovenia; luka.repansek@ff.uni-lj.si
1 See Šekli (2008: 166), Steenwijk (1992: 310).
2 The reconstruction is based on the recently acquired dial. attestation Zlíːca (Žabnice/Campo-
rosso) – cf. the Italianised form Slizza – with regular vowel reduction in the paroxytonon, which 
clearly points to the diminutive *Ziĺca (AP F1) ← *Ziĺ (AP b).
3 Note here that the specifically Slavic accent shift known as the Dybo’s Law must be chronologi-
cally posterior to the integration of the oldest core of pre-Slavic toponymic material. Consider in 
this respect a few truly unambiguous cases which show accent progression but were for one reason 
or another regularly exempt from the subsequent rule of accent retraction by Ivšić-Stang’s Law, so 
that subsequent integration into the productive patterns as they existed after the operation of both 
uniquely Slavic accent shifts is unthinkable. Such are Sln. Cèlje < *Celḗ < *Celь < *Celi < 
*Kelẽa ← Rom. *Keléa with contraction-induced inhibition of subsequent accent retraction, and 
the category of AP E accentual units such as Čak. Osȍr (Rom. *Ápsaru → Slav. *Ãsaru > *Osrъ), 
Štok. Mòsor (Rom. *Más(s)aru → Slav. *Mãsaru > *Mosrъ), Sln. Celôvec (< *Cь/iĺьcь < 
*Kĩ/ĺ(a)u ← Rom. *K/ĺ(a)-?), which by definition were incapable of subsequent accent re-
traction. I would also include here *ýdьnъ, *ýdьna > Common Slovene *Vídən, *Vídna with 
preserved quantity of the accented /íː/ throughout the paradigm. Note that Ter (Subid) Vid’an, Gsg 
Vid’na (quoted by Šekli 2009: 153) < *Vídən/*Vídna < *ydnъ/*ýdьna does not in fact point 
to AP B but rather points to *ydnъ/*ydna < *1dinu/*1dinā < *’ū1dinu/*ū1dinā ← 
Rom. *’Ut/dinu. Note that the latter (mis)interpretation is in fact impossible even under an assumed 
generalisation of the /íː/ < neo-acute *ý/recent Slovene acute *í from the nominative singular.
Linguistica_2015_FINAL.indd   59 14.3.2016   8:39:29
60
open *o (= /ɔ/?) < VLat. *ŏ (and subsequent development of the back nasal vowel *ǫ 
for the accented tautosyllabic sequence *un) is the theoretically expected outcome of 
phonetic adaptation before the rise of the new Slavic *o from Proto-Slavic *a. The 
regular substitution of the Romance affricate *c ~ * (< * < *t, *k) in the earliest 
layer of Romance loans into Slavic, which predate the rise of the genuinely Slavic 
*c by second and third palatalisation processes, is the corresponding Slavic affricate 
* (< *kE).4 Note that this is the only theoretically expected outcome, neatly corrobo-
rated by a set of early loans into the type of Slavic idioms that do not coalesce the older 
product of the first palatalisation * with the palatal stop *, e.g. Čak. Porȅč < *Porь 
← *Par’encu, *Brõč < *Bra ← *Br’acu etc., including the above-mentioned dialec-
tal Slovene forms.
The Slavic adaptation of the Romance prototype points to something like *’Sɔncọ or 
rather and given the likely point of contact, *’Sɔnọ. The morphological adaptation of 
the originally masculine name (as undeniably proved by the older sources) and its inte-
gration into the class of Slavic feminine a-stems is a fully systematised process, based 
on gender attraction of the premodifier to a feminine head-word in the underlying noun 
phrase (i.e. the onomastic model, for which see Šivic Dular 1998). A more serious point 
of disagreement between the Slavic adoption and the organic reflex of the hydronym as 
inferable from the contemporary Friulan outcomes Lusìnç (standard form), dial. L’Isùns 
(Aquileia), il Lusìns (Flumisel/Fiumicello),5 however, lies in the fact that the Slavic 
form presupposes a source with early aphaeresis of the initial vowel. But Friulan – the 
contemporary form of the donor language – seems to have preserved no trace of such 
a variant. It is possible that the internal Friulan otucomes such as Lusìnç & c. may go 
back to a later non re-analysable conglomeration of the Old Friulan masculine definite 
article lu (< *illum) and a form such as *Sɔnə < *Sontọ, but it is equally likely that 
they simply reflect a sporadic case of vowel metathesis (in the case of standard Friulan 
variant in a form theoretically borrowed from the sonziaco Friulan with u < *ɔ), or 
may even go back to a remodelled *Lu Z()ɪn(ə) for *Liz()in(ə) = *L’Iz()ɪn(ə)6 
< *Iz’ɔnọ. Note that il Lusìns attested in Flumisel/Fiumicello may speak in favour 
of the possibility that at least some of these forms may have undergone metathesis, 
granting that the expected outcome *L’Isùns (with expected u < *ɔ/__NT) would have 
undergone subsequent agglutination. Be that as it may, a pre-form with an anlauting 
(unaccented!) vowel, which seems to underlie the Friulan reflexes of the river name, 
cannot have been the source of the integration into Slavic as one would then expect 
Slavic **Zǫa – the process of sonorisation of intervocalic consonants in the precursor 
of Friulan, as is inferable from a number of transparent cases, predates the contact be-
tween the Romance and the Slavic idiom. It goes without saying that the attested forms 
4 Pace Ramovš HG II §152, id. (1936: 27, 51), who proposes to start from **Sǫa (similarly Šturm 
1927: 59–60, repeated in Furlan 2002: 33 and ESSZI s.v. Soča; sceptically Šekli 2008: 166).
5 For older attestations see Di Prampero (1882 s.v.) and Pirona, Vocabulario friulano (1871), who 
records Lusìnz and (L)isùnz.
6 The bilabial element of the original diphthong is only preserved after labials, cf. pɪnt < *pɔntə 
vs. frɪnt < *frɔndə etc.
Linguistica_2015_FINAL.indd   60 14.3.2016   8:39:29
61
also exclude the possibility of a specifically Slavic case of aphaeresis (cf. *Nynъ for 
Aenona, on which see Schramm 1981: 58, 262, 307).
If the manuscript tradition is to be trusted, the first Latinised attestations of the 
trisyllabic variant are Cassiodorus’ Sonti fluenta (Var. epist. I,18,10) and the famous 
prepositional phrase super Sontium (ib. I,29), reported in the early 6th century. All 
other attestations of Sontius which directly refer to the river, including the 8th-century 
account of Paulus Diaconus (Hist. rom. XV, 20, p. 214: iuxta Sontium flumen, qui non 
longe ab Aquileia labitur),7 may simply be reproductions of an established manuscript 
variant. Note that the entire corpus of the earliest literary attestations of the hydronym 
refers to or briefly recounts one and the same historical event with AD 489 as its in-
dubitable terminus post quem. The question of the source of the variant Sontius thus 
cannot receive an easy answer, especially in the light of the fact that Cassiodorus, who 
presumably would have been aware of the organic, autochthonous name of the river, 
seems inclined to indiscriminative use of the original, four-syllabic Isontius beside 
the truncated variant (see Cass., Chron. p. 159, 1319-1321: ad Isontium, and cf. gur-
gites Isontii in Paul. Aquilei. (8th c.), carm. II,1,v.5). Since aphaeresis is not strictly 
speaking a sound change, it most often goes back to deglutination, normally resulting 
from the reinterpretation of morpheme boundaries. A putative Sontius could easily 
have been produced by deglutination in cases such as *de Esontio > *de Sontio or 
through phonetic hyperinterpretation of Isontio as i(n) Sontio with VLat. is- < *ins etc. 
None of this can be verified, however. Another, rather obvious and hence already pro-
posed8 possible source of the alternative form Sontius is the contracted and presum-
ably reinterpreted form of the place name Ponte (= Lat. abl.-loc. sg.) Sonti(i) (= Lat. 
gen. sg.), first attested in Tabula Peutingeriana at segm. III,5 as the designation of 
a Roman mansio (see Vedaldi Iasbez 1994: s.v. Ponte Sonti). From an underlying 
*Ponte Esonti(i) > *Pontesonti the way to Ponte Sonti and subsequent extrapolation 
of Sontius, corresponding to the new genitival form of the river name, is easily envis-
aged. This scenario may be oversimplistic, however, since there can be no knowing 
whether the reported form of the place name is in fact organic or perhaps artificially 
(re)produced in transmission. It is noteworthy that the compiler of the Tabula seems 
to have had no notion of the actual river name, simplistically superimposing the fl. 
Frigido (III,5) on the lower course of the actual Isontius and ending it in a rather large 
lake (no designation) near Aquileia. This means that in the case of Sonti reported on 
the Tabula there would have been no source of possible (hyper?)correction. Note that 
Yordanes is the only author who treats the two components of the place name as in-
dependent units (Getica 57,292–293, p. 133: ad pontem Sontii), but this must surely 
be secondary. Both genitives pontis Sontis (Fasti vindob. prior. II.50, p. 317, 1. 490) 
and ponte Sontis (Auct. hauniense II.50, p. 317, 1. 490)9 clearly point to Pons Sonti-/
7 For the attestations see Vedaldi Iasbez (1994: 109–110).
8 See Brusin (1924: 225–226), Niedermann (1931: 3ff.).
9 For the sources see Vedaldi Iasbez (1994, s.v. (Ae)sontius/Isontius).
Linguistica_2015_FINAL.indd   61 14.3.2016   8:39:29
62
Ponte=Sonti-,10 which was by that time quite evidently acquiring the status (limited to 
manuscript tradition!) of a dithematic toponymical unit.
It is perhaps worthwhile to stress the fact that every single testimony of the river’s 
name, including the autochthonous Friulan outcomes, refers to the lower course of the 
Soča between Goriza and the Gulf of Trieste. The distorted image of the river system 
in this part of the Friulan plain typical of all classical geographers goes hand in hand 
with the fact that the upper course of the river was simply too insufficiently known to 
afford the right connection. While the Frigidus river must have been familiar even to 
an outsider due to its strategic position, the complicated net of tributaries dominating 
the (south-)eastern Friulan plain, starting at Sovodnje (Savogne dal Lusinç/Savogna 
d’Isonzo) will have remained rather perplexing. There is all the reason to believe, how-
ever, that locally the correspondence between the river’s name as handed down from 
prehistory and the entire stretch of its course was unquestioned. Since the contact of 
the incoming Slavs with the Roman-speaking population was not in fact limited to the 
south-western perimeters of the penetration wave (the contact area stretches from south 
to north in a more or less compact belt), the geographical parameters of this particular 
case of name-integration are nevertheless not determinable.11
2. ROMANCE *VEz’onọ
The Friulan continuations of the pre-Romance source, viz. Lusìnç/Lisùns, are silent 
as to their ultimate origin. Due to vowel harmony12 of unaccented e ~ i they are both 
10 Pace Vedaldi Iasbez (1994: 113) there is no need to assume a syncopated Venetic Sontis. Rather 
the genitive singular Sonti (for Sontii) was interpreted as an i-stem in the process of disintegration 
of the original syntactic relationship between both members, complete in ponte Sontis = virtually 
gen. sg. to *Ponte-sonti-. 
11 The case of Sotla < *Stъ/ьla (a tributary of the Sava in Eastern Slovenia) is most likely ir-
relevant for the question of the date and the source of the aphaeresis (similarly sceptical SVI 
II: 210–211 and ESSZI: 390 but cf. Furlan 2002: 33). Nothing is known of the history of this 
particular river name and the existing attestations are much too late to provide a better starting 
point than what can be inferred from the historical phonology of the contemporary forms. Slavic 
*Sá/úntu/ilā points to *S’a/’o/’untVlV – a form that is not immediately transparent. If the vowel 
in the second syllable is the result of later (but pre-Slavic!) anaptyxis (cf. in this respect the likely 
case of Οὐιστούλα (Ptol., pass.), see also Krahe 1964: 102–103), one could, e.g., think of an 
agent noun *st-ló- to PIE *sent- ‘move, go’. On the other hand, the *t could also be an integral 
part of the suffix such as *-tlo- ~ *-tleH2- (note that *t as the result of consonantal epenthesis 
in a *-n-l- cluster is not at all likely in a voiced environment). Alternatively, we could be deal-
ing with a complex suffix *-tú- + *-lo- to a verbal root such as *sneH2-, notably popular in the 
derivation of river names, but there is no knowing whether a putative tautosyllabic *H would 
result in something like *an/*on/*un in the particular linguistic system that generated our source. 
Note that both *Sa-a?s? and *Dra-a?s? are likely to be more specifically Pannonian rather than 
just broadly Old European and that it may be important in this respect to note that the Pannonian 
reflex of (at least) the plain syllabic nasal is in fact *uN.
12 I am grateful to Prof. Dr. Franco Finco for reminding me of this important fact of Friulan histori-
cal phonology.
Linguistica_2015_FINAL.indd   62 14.3.2016   8:39:29
63
derivable either from *Iz'onọ < *Īs'onto- or *Ez'onọ, the latter from *As'onto-, 
*Os'onto- or *Ĭs'onto- (assuming *i [– accented] > (*ẹ ~) *e rather than *Iz'onọ < 
*Ĭs'onto-, but neither of these developments is in fact demonstrable solely on the basis 
of Friulan historical phonology). The remaining alternative *Es'onto- is problematic 
because the reflex of *e in a name integrated into Latin after the monophthongisation 
of the inherited *e is simply irrecoverable. If such a sequence were borrowed before 
the 2nd half of the second century BC, which in the case of Soča is in fact not impossi-
ble, and if it was identified with the long mid-high monophthongisation product of Old 
Latin *e, then it would merge with the latter and eventually be raised to *ī. In unac-
cented position *ī would retain its quality throughout the development towards Vulgar 
Latin. If, however, the integration post-dated the final raising, a non-autochthonous 
unaccented *e would probably either be preserved or join the VLat. unaccented *e, de-
pending on how late the borrowing was made. Due to the lack of instructive examples, 
these developments ultimately have to remain speculative. Note that if the 5th-century 
attestation Isontius (Cass., Chron., v.s.) can be taken at face value, however, this would 
provide a more solid argument for a direct development of unaccented *e (~ *ẹ) (< *a, 
*o and possibly *ĭ) to *i (which in turn would then be lowered to *e if /__$V[+ low]).
The aphaeresis signalled by the resegmented Ponte Sonti in Tabula Peutingeriana 
may in fact point to a stage with a low front vowel in the anlaut. Such a putative 
Esontius = *Es'ontọ vel sim. (the exact developmental stage of the auslaut is irrelevant 
here), which is more than likely to represent a stage older than the one reported by 
Cassiodorus, may also be reflected in the oldest attestations of the hydronym in the two 
epigrapical sources Inscr. Aqu. 1.96 = AE 1926, 108 and Aquileia nostra 1996, 109 
= AE 1996, 695. As is known, both these sources refer to the originally indisputably 
dehydronymic river deity Aesontius.13 At least one of the inscriptions dates back to the 
end of the first century AD. If the sequence <AE> is the archaising graphic represen-
tation of the unaccented VLat. *e (perhaps still an open front vowel at this stage), it 
would easily point to a source with an anlauting *a, *o and possibly unaccented short 
*i (if subsequent lowering to *e is assumed). If, on the other hand, the attested spelling 
reflects the only other possible phonetic reality, the source will have had an analuting 
*a, with subsequent development towards *i through *e, possibly implicitly preserved 
in *Ponte Esonti(i). As the only truly unlikely source one should mention the long un-
accented *ī as this would have been reflected as *i throughout.14
13 The masculine gender of the deity itself is of course dependent on the grammatical gender of the 
Latinised river name, which is the direct source of the transonymisation.
14 The famous ethnonym Ambisontes (Plin., Nat. hist. III.137, cf. Ptol. II.13.2 Αμβισóντιοι) is ut-
terly ambiguous since the exact location of this Alpine tribe is still very much disputed (see Šašel 
1972: 140–144, Šašel Kos 1997: 23–24 and pass., contra, e.g., Alföldy 1974: 68, Scherrer 2002: 
32). Given the evidence of, e.g., Ambarri < *Amb-arar-o- ← dehydronymic to *Arar- (on the lat-
ter see K. Ihm, RE I,2 = 1894, pp. 1795–1796), it is clear that the i reported in the second syllable, 
should the name reflect a compound of *ambi- and *Isont- rather than *Sont-, reflects the vocalic 
anlaut of the governed member, the auslaut of the preposition being regularly elided in such 
cases. However, it should be remembered that the name reflects a Latinate form of the ethnonym 
(not at all necessarily Gaulish since *ambi- ‘around’ could just as easily have been produced in 
Linguistica_2015_FINAL.indd   63 14.3.2016   8:39:30
64
3. PRE-ROMANCE *(a/o/e)is-o/a(-)nt-
The correct etymological connection of the underlying hydronym with the PIE root (in 
modernised notation) *H1esH2- ‘kräftigen; antreiben’ (LIV
2: 234) has already been 
proposed by Krahe (see Krahe 1953 s.v. and cf. Krahe 1964: 56, IEW: 299–301, Bezlaj 
SVI II: 204, ESSZI: 386–387 etc.). Both the name’s phonetic and morphological struc-
ture are transparently Indo-European and unspecifically so, meaning that neither of 
its outward features are diagnostic in terms of a specific linguistic affiliation.15 If one 
also accepts the likely connection of the morphological structure of the suffix with the 
(nearly suffixal) element -antia attested in numerous European river names, simple 
forward reconstruction would demand a straightforward feminine present participle 
formation *H1isH2-t-iH2- (corresponding to the masculine *H1ésH2-ont-/*H1isH2-t-´ 
or *H1isH2-ént-/*H1isH2-t-´). There is no general consensus on the original ablaut pat-
tern of the present active participle (amphikinetic?/hysterokinetic?), but it seems safe to 
assume, at least on grounds of the external comparative evidence, a zero grade stem for 
the feminine counterpart, although the generalisation of the full-grade root cannot be 
excluded a priori (cf. Gr. *ϝεk-at-a < *éḱt-iH2- or ἔασσα = Myc. °e-a-sa
16 < *H1és-
t-iH2- vs. IIr. *uć-a-t, *sa-t etc.). Certain amount of caution should be exercised, 
however. The clearly transitive meaning of the root *H1esH2- as set up on the basis of 
the attested members of the averbo is as a present participle in its agentive meaning ‘the 
impelling, stimulating, invigorating one’ (the latter not in the sense of the otherwise 
acceptable and attested semantic motivation ‘nourishing’!) rather unsuitable for a rive 
name. The usually assumed motivation such as ‘the quickly moving one’ et sim. is of 
course completely ad hoc and seems to be based more on our own projection of the 
naturally assumed possible qualities of a river than on any positive evidence.
Sifting through the available comparative data, however, it immediately becomes appar-
ent that the underlying transitive semantics of the root may be misleading, since they seem 
any Venetoid linguistic system and nothing really tangible is known of the productive patterns 
in the formation of ethnic names in these languages), which is apparent at least from the fact that 
it has been integrated into the third declension. If the ethnic name reflects *Amb-isont- and if it 
should turn out that the underlying hydronym indeed refers to the Soča river, the implicit under-
lying form *Isont- is in fact the oldest available testimony of the pre-Romance pronunciation of 
the indigenous name. Most interestingly it would in fact point to an original *i.
15 Contrary to some recent attempts to offer a dissenting view (see, e.g., J. B. Trumper, Ce fastu 82/2 
(2006), pp. 151–169, id. in F. Finco (ed.), Atti del secondo convegno di toponomastica Friulana, 
Udine, 2007, pp. 279–326), there will of course be no need to invoke a specifically Celtic etymol-
ogy. Not least for the reason that there can be no talk of a Celtic toponymical layer in this part 
of the Southeastern Alps and that the assumption of the process theonym → hydronym (even if 
external derivation is assumed after all) does not in fact represent the typologically unmarked pat-
tern. Note that the putative pre-Romance etymon *ais- (sometimes quoted as zero-grade *is-) with 
the underspecified and rather random meaning ‘water’ or similar (cf. Frau 1978: 71 and repeated 
as a possibility in ESSZI: 387, see also DTFT: 438), proposed by a number of older authors (see 
SVI II: 203–204 for a useful overview) is of course nothing else but the inadequately interpreted 
*H1ésH2- appearing across a wide range of European river names.
16 See Morpurgo Davies (1978) with bibliography.
Linguistica_2015_FINAL.indd   64 14.3.2016   8:39:30
65
to have been (rather inevitably) based on the conspicuously causative/factitive meaning of 
the attested formations. As nasal-infix presents Ved. iṣ-ṇ-ti ‘bring sb. in quick movement, 
spur’ (co-occuring with the rare active *´-e/o-present íṣ-ya- ‘send’ = Av. iš-iia- ‘spur, set 
going’) and Gr. ἰνάο ‘expulse, empty out’ (both regularly < *H1is-né-H2-/*H1is--H2-), 
together with the old extended forms in *-e/o- (*H1is--H2-é/ó- > iṣ-an-yá- ‘stimulate, 
impel’, ἰαίνω ‘warm up, heat up’; cf. Ved. gbh-ā-yá- beside *ghbh-né-H- in gbh-n-ti 
etc.), carry a clearly derivationally motivated causative meaning (cf. in this respect the 
clearly causative Ved. p-ṇ-ti ‘fills (up)’ as opposed to the originally fientive Aktionsart 
of the verbal root *pleH1- etc.). The same secondary causative function may also be un-
problematically assumed for the reduplicated present  *H1i-H1ésH2-/*H1i-H1sH2´- in Gr. 
*īha- → ἰάομαι ‘heal’ < *‘vivify, envigorate’ (on the latter see García Ramón 1986), and 
of course the formal causative attested in Indo-Iranian.17 Quite expectedly, the only non-
causative formations are the internally IIr. intransitive zero-grade present in *-ée/o- (Ved. 
iṣ-áya- = Av. iš-aiia-) and the deverbative o-grade abstract noun *H1ósH2-o- > Ved. éṣa- 
‘running (after)’ (cf. RVS V.66.3: éṣe rathnām as clearly opposed to other instances of 
éṣa- ‘searching (for)’ to iṣ- ‘seek’). Ved. eṣá-, however, which is multiple times attested 
as the epithet of Víṣṇu, meaning ‘swift, quick’ (cf. RVS II.34.11, VII.40.5, VIII.20.3), 
may represent the participial agent noun *H1osH2-ó-, but it may just as well belong with 
Avestan aēša- ‘strong, potent’, which continues a possessive o-grade *H1osH2-ó- (in 
all probability a Gmc. *χraa-type adjective), completely parallel to the equally posses-
sive derivative in -ró- as continued by Ved. iṣi-rá- ‘lively, active, strong’ = Gr. ἱερός18 
‘holy ...’ < *H1isH2-ró- and neatly preserved in a number of European hydronyms such as 
Ἰσάρας (Strab. IV,6,9), Ἴσαρ(oς) (Ptol. II,10,6), Isara (Liv. 21,31,4), Germ. Isar, Thrac. 
Ἴστρος (see Янакиева 2009 s.v.), on which see Bichlmeier (2012b: 29–37) and, recent-
ly, Repanšek (2015: 786). Lastly, the primitive feminine root-noun *H1ósH2-/*H1isH2-´ 
continued by IIr. *iš- *‘strength, vigour, invigoration, stimulation’ (cf. the abstract femi-
nine noun *H1osH2-éH2-, probably preserved in the ON denominative eisa ‘to rush’ < 
*H1osH2=eH2-é/ó-) and the o-grade abstract in -mo- (*H1ósH2-mo- > Av. aēšma- ‘anger, 
rage’, Gr. οἶμα ‘rush, raging attack’) are silent as to the verb’s original voice, as are the 
possessive Caland adjectives. However, at least on the strength of the deverbative abstract 
éṣa- (and perhaps eṣá- if it represents its corresponding derivational pair) it is possible to 
suspect that the original meaning of *H1esH2- was indeed ‘to be/become (?) set in motion’ 
vel sim. and thus essentially intransitive. This would successfully account for the derived 
semantics (secondary semantic shifts included) and, most importantly, provide a suitable 
base for the active present participle *H1isH2-t-iH2-, which very much parallel to the pos-
sessive feminine adjective *H1isH2-réH2- could theoretically be pressed into service of an 
Eigenschaftsname, matched by the telling description μέγιστος ποταμός and ὁ χειμάρρους 
by Herodianus (Hist. rom. VIII.4.1/2).19
17 For the attestations see EWAia I: 271–272, García Ramón (1986), Gotō (1993: 128–133), Kel-
lens (1995: 13), Werba (1997: 450–451).
18 On the internal history of ἱερός < *-ero- see Peters (1980: 325).
19 For the identification see Vedaldi Iasbez (1994: 112).
Linguistica_2015_FINAL.indd   65 14.3.2016   8:39:30
66
As anticipated above, the alternative starting point *Asonto-/*Osonto- is only ex-
plicable, however, if one starts from a possessive denominal derivative  *H1osH2-éH2- 
(i.e. the Greek type *φονή ‘manslaughter’). As already correctly interpreted by Krahe 
(1953: 119; cf. Greule 2007: 117 et passim), the North-Germanic parallel *Eisandi 
is only indirectly comparable, seeing that the latter is in fact a deverbative participial 
formation to ON eisa < PGmc. *asōnan ‘rush forward’ (in turn built to the exact 
same deverbal abstract *H1osH2éH2-). There is no positive proof, however, that a hy-
pothetical possessive *-nt-adjective *H1osH2-o-nt- ← *H1osH2-éH2- would form its 
feminine in *-iH2- (or, for that matter, *-()eH2-). The case of clearly denominal river 
names such as Albantia, Aquantia etc. is ultimately indecisive in this respect: either 
the ubiquitous sequence -antia is interpreted as the reflex of the feminine form of the 
*-nt-possessive, or – in avoidance of circularity – as a ready-made toponymical suffix, 
mirroring a virtual *-tiH2- (cf. Repanšek 2015: 787).
20
The question is of course inextricably bound with the controversial problem of Old 
European hydronymy. It was silently assumed above that the a-vocalism of the theoreti-
cal starting point *H1osH2éH2- necessitates the same vowel quality in the suffix and that 
this somehow dictates the realisation of the suffix -iH2- as *- <ia>, so that one would 
be inclined to start from *Asant. The regular change of the short PIE *o to short *a21 
and the realisation of the syllabic nasal * as *an (if  *H1osH2-o-nt- is assumed as the 
strating point, the change of *-o-nt- to *-a-nt- would of course be equally regular), paired 
with the likely (morphological) substitution of the biphonemic sequence *-ā (or less 
likely *-ă) for PIE *-iH2-, are all characteristic of the Old European system if defined as 
an actual linguistic stratum with the pertaining set of system-specific phonological and 
morphological transformations.22 Note that we are only forced to presuppose an underly-
ing *Asant in case the epigraphic attestation of river name, viz. Aesontius*, is taken 
at face value. The same goes for *H1(e)isH2-t-iH2- > *(E)is-ant-, but only if one does 
not want to assume the tenuous possibility of the generalisation of the o-grade suffix 
(*H1(e)isH2-ont-iH2-). Any reconstruction that would start from an *-ant-, however, 
must assume that the name must rather early on have passed through the phonetic filter 
of an idiom that typically preserved the PIE *o intact. Given the areal distribution of the 
autochthonous linguistic systems in the wider region of South-Eastern Alps prior to the 
arrival of Gaulish, the only serious candidate is the Venetic complex. The integrated unit 
*Asant or *(E)isant would potentially be liable to renovation to *As/(E)is-ont-, 
but only under the assumption of possible interference from the autochthonous Venetic 
reflex of the present participle suffix *-o-nt-. As no feminine forms of the (thematic) pre-
sent active participle are attested in Venetic, however, the possibility of an assumed cor-
respondence between *-ant- and *-ont-ā is impractical to test in any more or less direct 
20 As is the case, for example, in North Germanic (see Greule 2007: 118, 121).
21 On the possible evidence of Νóαρος (Strab. VII,5,2; 12), if for *(s)nō-o- < *(s)noH2-o- to PIE 
*(s)neH2- ‘baden, schwimmen’ (LIV
2: 572–573), I would restrict this sound change to the system 
of short vowels (see Repanšek 2015: 785, ft. 22). 
22 Such an approach is now tentatively advocated by Bichlmeier (e.g. 2012a, 2012b et pass.).
Linguistica_2015_FINAL.indd   66 14.3.2016   8:39:30
67
way. Feminine personal names such as vho.u.go.n.ta (LVen., Es 85) are uneluciadating. 
Although diachronically such names are based on a true deverbal active participle, they 
in fact represent a productive pattern of feminine motion and function synchronically 
solely as feminine versions of masculine names.23 The form Voltaronti, attested at Ig 
(CIL III.3877), on the other hand, indirectly probably points to the preservation of the 
old shape of the feminine participial suffix *-ont-ī at least for this peripheral dialect (see 
Stifter 2012: 258 and Repanšek forth.). Since the Venetoid linguistic system attested in 
the ager is in many respects more archaic than the more central Venetic, however, the 
latter might have innovated in the direction of *-ontā. Be that as it may, an equatable 
sequence *-ontā- would certainly have been provided by the system through the oblique 
cases of the feminine paradigm.
It must in any case be assumed, however, that should the hydronym belong to a layer 
of toponymy older than the one generated by Venetic, the auslauting sequence  *-ant- 
will eventually have been remade into a masculine *-ont-o- – a form that is indeed re-
flected by all the historical attestations of the river name, including its modern Friulan 
reflexes. Whether this process is intrinsically bound with the ultimate Latinisation of 
the name (*-ont- > -ontia → -ont-ius) or should be recognised as the result of Venetic 
recharacterisation of *-ant is not unambiguously determinable either. It is true that 
a Venetic *-ont-os would result in the syncopated *-ont-is > *-ont-s, but any Venetic 
stem in -(i)s would still be interpretable as corresponding to the Latin -ius, not least for 
the fact that the oblique cases still regularly preserved the unsyncopated shape of the 
sequence *-o-. At least systematic masculinisation of feminine river names is well at-
tested for Latin, while the possibility that the same could be claimed for Venetic is of 
course completely open to conjecture.
If contrary to the above the Old European stratum is rather perceived as a static 
network of the oldest, undifferentiated Indo-European toponymical heritage, Venetic 
historical phonology (and morphology, for that matter, but here the involved mor-
phological patterns may have already become unproductive by the time of the first at-
testations, so that any potentially archaic trait is self-explanatory and useless for our 
purposes) should account for the pre-literary form of the name. In this respect only 
*H1osH2-o-nt-
? is the possible starting point, while the participial *H1(e)isH2-t-iH2- 
is only conceivable as an alternative under the assumption of the generalisation of 
an o-grade suffix based on the masculine/thematic stems, for which no supporting 
evidence can be adduced, however. Note that a putative Venetic reflex *(E)is-ant-? 
< *H1(e)isH2-t-iH2-
24 is far less likely as a starting point, as one would then have to 
assume some secondary process through which the resulting vocalism in the suffix 
would have been analogically adapted to the shape of the masculine participial stem. 
Although, admittedly, this possibility is not at all unimaginable, it must remain for 
the time being impossible to prove. Needless to add, both in terms of the shape of the 
23 Contrary to Lejeune (1974: 83), the possibility of a shared form of the present active participle 
for the masculine and the feminine in Venetic (as is the case in Latin) is virtually nil. 
24 For Venetic *aN < PIE * cf. iiuva.n.t.s. (Es 25) < *H2u-H3-t-.
Linguistica_2015_FINAL.indd   67 14.3.2016   8:39:30
68
final morpheme as well as the distribution of the inherited *-ont- vs. *-t-iH2- nothing 
decisive can be inferred from the available linguistic material.25
Abbreviations
AP = accentual paradigm; Av. = Avestan; Čak. = Čakavian; Gr. = Ancient Greek; Lat. = 
Latin; Myc. = Mycenian; ON = Old Norse; PGmc. = Proto-Germanic; PIE = Proto-Indo-
European; Rom. = Romance; RVS = gvedasahitā; Slav. = Slavic; Sln. = Slovene; Štok. 
= Štokavian; Thrac. = Thracian; Ved. = Vedic; Ven. = Venetic; VLat. = Vulgar Latin
Bibliography
AE = L’Année épigraphique. Paris, 1888–
ALFÖLDY, Geza (1974) Noricum. London/Boston: Routledge and Kegan Paul.
BICHLMEIER, Harald (2012a) „Anmerkungen zum terminologische Problem der 
 alteuropäischen Hydronymie samt indogermanistischen Ergänzungen zum Namen 
der Elbe.‟ Beiträge zur Namenforschung 47/4, 365–395.
BICHLMEIER, Harald (2012b) „Einige ausgewählte Probleme der alteuropäischen 
Hydronymie aus Sicht der modernen Indogermanistik – Ein Plädoyer für eine neue 
Sicht auf die Dinge.‟ Acta linguistica Lithuanica 66, 11–47.
BRUSIN, Giovanni Battista (1924) „Il nome dell’Isonzo.‟ Rivista della Società 
Filologica Friulana 5/3, 223–226.
CAMPANILE, Enrico (ed.) (1990) Rapporti linguistici e culturali tra i popoli dell’Italia 
antica, Pisa, 6–7 ottobre 1989. Pisa: Giardini editori e stampatori in Pisa.
CIL = Corpus inscriptionum Latinarum.
DI PRAMPERO, Antonio (1882) „Saggio di un glossario geografico friulano dal VI al 
XIII secolo.‟ Atti del Regio Istituto Veneto di Scienze Lettere e Arti 5/7–8, 1–236.
DTFT = CINAUSERO HOFER, Barbara/Ermano DENTESANO con la collaborazi-
one di Enos CONSTANTINI e Maurizio PUNTIN (2011) Dizionario toponomas-
tico. Etimologia, corografia, citazioni storiche, bibliografia dei nomi di luogo Friuli 
storico e della Provincia di Trieste. Udine: Edizioni Ribis.
25 An original *-ont-(i)o-, i.e. a deadjectival *o-derivative (?) based on a *-nt-participle, how-
ever, does not seem to be a promising alternative either. The evidence for the existence of the 
category of departicipial adjectives is in fact vanishingly small. The often quoted Ved. sáhan-
tiya- to *sáh-a-nt- < *seǵ
h-o-nt- ‘überwältigend’ is completely isolated and may just as well be 
analogical as has been suggested already by Hoffmann, see “Ved. santya- und ahd. samfti, ags. 
sēfte”, Münchener Studien zur Sprachwissenschaft 23 (1968), 29ff. Note that the Celtic congener 
*Segontio- may be due to the popularity of secondary and purely structural *o-derivatives to 
nouns/adjectives with overtly agentive semantics.
 On a side note and just to be exhaustive in the account of all the theoretical possibilities, one 
might also want to consider a departicipial feminine abstract noun *H1(e)isH2-(o)nt-(i)eH2- (cf. 
the popular Lat. (!) category of abstracts in -ent-ia < *-t-). The meaning would of course be 
close to the one proposed for the deverbal abstract *H1osH2eH2-, for which see above.
Linguistica_2015_FINAL.indd   68 14.3.2016   8:39:30
69
ESSZI = SNOJ, Marko (2009) Etimološki slovar slovenskih zemljepisnih imen. 
Ljubljana: Založba Modrijan in Založba ZRC, ZRC SAZU.
EWAia = MAYRHOFER, Manfred (1992–2001) Etymologisches Wörterbuch des 
Altindoarischen I–III. Heidelberg: Carl Winter Universitätsverlag.
FRAU, Giovanni (1978) Dizionario toponomastico Friuli-Venezia Giulia. Udine: 
Istituto per l’enciclopedia del Friuli-Venezia Giulia.
FURLAN, Metka (2002) „Predslovanska substratna imena v slovenščini.‟ Jezikoslovni 
zapiski 8/2, 29–35.
GARCÍA RAMÓN, José Luis (1986) „Griego ῑ̓άoμαι.‟ In: A. Etter (ed.), o-o-pe-ro-
si. Festschrift für Ernst Risch zum 75. Geburtstag. Berlin/New York: Walter de 
Gruyter, 497–514.
GOTŌ, Toshifumi (1993) „Materialien zur eine Liste altindischer Verbalformen.‟ 
Bulletin of the National Museum of Archaeology (Osaka) 18, 191–141.
GREULE, Albrecht (2007) Etymologische Studien zu geographische Namen in Europa. 
Ausgewählte Beiträge 1998–2006. Ed. by W. Janka/M. Prinz. Regensburg: edition 
vulpes.
IEW = POKORNY, Julius (1959–1969) Indogermanisches etymologisches Wörterbuch. 
Bern/München: Francke.
ЯНАКИЕВА, Светлана (2009) Тракийската хидронимия. София: Академично 
издателство „Проф. Марин Дринов‟.
KELLENS, Jean (1995) Liste du verbe avestiques. Wiesbaden: Dr. Ludwig Reichert 
Verlag.
KRAHE, Hans (1964) Unsere ältesten Flussnamen. Wiesbaden: Otto Harrassowitz.
KRAHE, Hans (1953) „Alteuropäische FluSnamen.‟ Beiträge zur Namenforschung 4, 
37–53, 105–122, 234–243.
LEJEUNE, Michel (1974) Manuel de la langue vénète. Heidelberg: Carl Winter 
Universitätsverlag.
LIV2 = RIX, Helmut/Martin KÜMMEL/Thomas ZEHNDER/Reiner LIPP/Brigitte 
SCHIRMER (2001) Lexikon der indogermanischen Verben. Die Wurzeln und ihre 
Primärstammbildungen. Zweite, erweiterte und verbesserte Auflage bearbeitet von 
Martin Kümmel und Helmut Rix. Wiesbaden: Dr. Ludwik Reichert Verlag.
LVen. = PELLEGRINI, Giovan Battista/Aldo Luigi PROSCODIMI (1967) La lingua 
venetica I–II. Padova/Firenze: Istituto di glottologia dell’Università di Padova/
Circolo linguistico fiorentino.
MORPURGO DAVIES, Anna (1978) „Thessalian εἴντεσσι and the participle of the 
verb ‘to be’.‟ In: Étrennes de septantaine. Travaux de linguistique et de grammaire 
comparée offerts à Michel Lejeune par un groupe de ses Élèves. Paris: Éditions 
Klincksieck, 157–166.
NIEDERMANN, Max (1931) „Zur altillyrischen Toponomastik.‟ Zeitschrift für 
Ortsnamenforschung 7, 3–8.
PETERS, Martin (1980) Untersuchungen zur Vertretung der indogermanischen 
Laryngale im Griechischen. Wien: Verlag der Österreichischer Akademie der 
Wissenschaften.
Linguistica_2015_FINAL.indd   69 14.3.2016   8:39:30
70
Ramovš, HG II = RAMOVŠ, Fran (1924) Historična gramatika slovenskega jezika II: 
Konzonantizem. Ljubljana: Učiteljska tiskarna.
RAMOVŠ, Fran (1936) Kratka zgodovina slovenskega jezika. Ljubljana: Akademska 
založba, 1936.
REPANŠEK, Luka (2015) „Καλαμαντία (Ptol. II,11,15).‟ In: N. N. Kazanskij (ed.), 
Индоевропейское языкознание и классическая филология/Indo-European 
Linguistics and Classical Philology – 19. Материалы чтений, посвященных памяти 
Иосифа Моисеевича Тронского/Proceedings of the 19th Conference in Memory of 
Professor Joseph M. Tronsky, 22-24 июня 2015 г./June 22-24, 2015, 780–790.
SCHERRER, Peter (2002) „Von Regnum Noricum zur römischen Provinz: Grundlagen 
und Mechanismen der Urbanisierung.‟ In: M. Šašel Kos/P. Scherrer et al. (eds.), The 
autonomous towns of Noricum and Pannonia – Die autonomen Städte in Noricum 
und Pannonien. Noricum. Ljubljana: Narodni muzej Slovenije, 11–70.
SCHRAMM, Gottfried (1981) Eroberer und Eingesessene. Geographische Lehnnamen 
als Zeugen der Geschichte Südosteuropas im ersten Jahrtausend n.Chr. Stuttgart: 
Anton Hiersmann.
SIMBECK, Katrin/Wolfgang JANKA (eds.) Namen in Altbayern. Gesammelte 
Beiträge zu Ehren von Josef Egginger und Günter Schneeberger. Regensburg: 
Edition vulpes.
STEENWIJK, Han (1992) The Slovene dialect of Resia: San Giorgio. Amsterdam: 
Rodopi.
STIFTER, David (2010) „On the linguistic situation of Roman-period Ig.‟ In: T. 
MeiSner (ed.), Personal names in the Western Roman world. Proceedings of 
a workshop convened by Torsten Meißner, José Luis García Ramón and Paolo 
Poccetti, held at Pembroke College, Cambridge, 16–18 September 2011. Berlin: 
Curach bhán, 247–265.
SVI = BEZLAJ France (1956–1961) Slovenska vodna imena I–II. Ljubljana: SAZU.
ŠAŠEL, Jaroslav (1972) „Zur Erklärung der Inschrift am Tropaeum Alpium (Plin. n. h. 
3, 136–137. CIL V 7817).‟ Živa antika 22, 135–144.
ŠAŠEL KOS, Marjeta (1997) „The end of the Norican kingdom and the formation 
of the provinces of Noricum and Pannonia.‟ In: B. Djurić/I. Lazar (eds.), Akten 
des IV. Internationalen Kolloquiums über Probleme des provinzialrömischen 
Kunstschaffens, Celje, 8.–12. Mai 1995 – Akti IV. mednarodnega kolokvija o prob-
lemih rimske provincialne umetnosti, Celje, maj 1995. Ljubljana: Narodni muzej 
Slovenije, 21–42.
ŠEKLI, Matej (2008) Zemljepisna in osebna lastna imena v kraju Livek in njegovi 
okolici. Ljubljana: Založba ZRC, ZRC SAZU.
ŠEKLI, Matej (2009) „On Romance-Alpo-Slavic substitutional accentology: the case 
of pre-Slavic masculine substrate place names in Slovene.‟ In: Th. Olander/J. H. 
Larsson, Stressing the past, Papers on Baltic and Slavic Accentology. Amsterdam/
New York: Rodopi, 145–160.
ŠIVIC-DULAR, Alenka (1998) „Lingvistička definicija vlastite imenice.‟ Folia ono-
mastica Croatica 7, 281–293.
Linguistica_2015_FINAL.indd   70 14.3.2016   8:39:30
71
ŠTURM, Fran (1927) „Refleksi romanskih palataliziranih konzonantov v slovenskih 
izposojenkah.‟ Časopis za slovenski jezik, književnost in zgodovino 6, 45–85.
TAVANO, Sergio (ed.) (1972) Aquileia e l’Alto Adriatico 2: Aquileia e l’Istria. Udine: 
Arti grafiche Friulane.
VEDALDI IASBEZ, Vanna (1994) Venetia orientale e l’Histria. Le fonti letterarie 
greche e latine fino alla caduta dell’Impero Romano dʼOccidente. Roma: Edizioni 
Quasar.
WERBA, Chlodwig H. (1997) Verba Indoarica. Die primären und sekundären Wurzeln 
der Sanskrit Sprache. Pars I: Radices primariae. Wien: Verlag der Österreichischer 
Akademie der Wissenschaften.
Abstract
TOWARDS A CLARIFICATION OF THE HISTORY 
OF THE SLOVENE RIVER NAME SOČA
The etymology of the Slovene hydronym Soča (Standard Friulan Lusìnç, Italian Isonzo) 
was satisfactorily elucidated more than half a century ago. The overall idea that the 
river name goes back to a derivative of a Proto-Indo-European verbal root *H1esH2- is 
undoubtedly correct (the exact meaning of the root in question has, however, not been 
sufficiently corroborated) but the details remain hopelessly obscure. Given the con-
temporary reflexes of the underlying name (viz. Friulan Lusìnç ~ dial. Lisùns etc.) and 
the sometimes conflicting or not at all easily reconcilable historical attestations (span-
ning nearly a millenium), no consensus can in fact be reached about the exact formal 
means involved in the derivation of the original deverbative formation – a fact which 
results from the overall inconvenience of sheer wealth of combinatorial possibilities 
that go hand in hand with the general ambiguity of the phonology more or less directly 
reflected by the later reflexes of the name. The present contribution is an attempt at 
unifying all the reliable data at our disposal into a coherent picture that would bring a 
satisfactory answer as to the exact phonological and morphological make-up of the pre-
historic source (alongside with the reconstructed meaning) and its subsequent history 
as it passed through the various linguistic strata.
Keywords: Soča, Indo-European hydronymy, Slavic, Romance, pre-Slavic substratum 
Povzetek
PRISPEVEK K ZGODOVINI SLOVENSKEGA REČNEGA IMENA SOČA
Etimologija imena Soča (standardno furlansko Lusìnç, italijansko Isonzo) je bila sicer 
zadovoljivo razjasnjena že pred več kot pol stoletja. Ime je brez dvoma treba izpeljevati 
iz praindoevropskega korena *H1esH2- s sicer težje ugotovljivo prvotno semantiko, 
vendar se v poizkusu natančnega spremljanja zgodovinskega razvoja imena od njegove 
prvotne podobe do sodobnih kontinuantov (predvsem furl. Lusìnç ~ narečno Lisùns 
Linguistica_2015_FINAL.indd   71 14.3.2016   8:39:30
72
itd.) izkaže, da kopica bistvenih dejstev ostaja nerazjasnjenih. Splošna večznačnost 
fonetične podobe organskih kontinuantov, vključno s skopimi in težje pojasnljivimi 
podatki, ki jih ponuja skoraj tisočletje zelo sporadičnih historičnih zabeležb imena, je 
v neposredni vzročni povezanosti z množico verjetnih (in mnogokrat povsem teoretič-
nih) kombinatoričnih možnosti rekonstrukcije formalnih sredstev, vpletenih pri tvorbi 
izhodnega deverbativa. V prispevku se pod vprašaj ponovno postavlja vse relevantno 
in zanesljivo gradivo, ki bi, povezano v smiselno in kronološko prepričljivo izgraje-
no celoto, utegnilo odgovoriti na vprašanje o natančni fonetični in morfološki podobi 
izhodne oblike (in posledično njenem prvotnem pomenu) ter pojasniti vse razvidne 
spremembe substitucijske narave, do katerih je zanesljivo prihajalo pri prehajanju hi-
dronima skozi različne jezikovne sisteme.
Ključne besede: Soča, indoevropska hidronimija, slovanščina, romanščina, predslo-
vanski substrat
Linguistica_2015_FINAL.indd   72 14.3.2016   8:39:30