PhiN 43/2008: 11

Keith Andrew Massey (Morris/New Jersey)

Further Evidence for an "Italic" Substratum in Romanian

This article presents further evidence for the theory that one of the substrata within Romanian is an Italic language other than Latin itself. Lexical items of uncertain etymological origin are explained as more directly derived from this source through a previously unrecognized sound rule. The article suggests that the Romanian substratum that bears these Italic traits could be one of the indigenous languages in the Balkan Peninsula. Finally, the article demonstrates that a presumably Thracian inscription produces a sensible reading if read as bearing more affinities with Italic than has previously been supposed for that language.

1 Introduction

Of all the Romance languages, the development of Romanian is the least understood (see Rosetti 1969; Du Nay 1996; Sala 2005). There is an interval of several hundred years in which no historical record mentions its existence, let alone providing scholars with samples to track its evolution (Rosetti 1969: 481; Sala 2005: 25). Considerable debate in Romanian studies continues over the various lexical items deemed to constitute substrata from the Thracian and Dacian languages attested in the area prior to the Roman occupation (Du Nay 1996: 46–72; Poghirc in Rosetti 1969: 319–320; Hubschmid 1983). Owing to the scanty inscriptional evidence for the pre-Roman indigenous languages in the Balkan peninsula, there is still little known with certainty regarding them (Russu 1967; Dechev 1957; Georgiev 1983; Duridanov 1985).

In this article I will argue that a number of words of uncertain etymology and presenting inconsistent phonology in Romanian will be better understood as borrowings from a language similar but not identical to the Vulgar Latin that constituted Proto-Romanian itself.

2 /kw/ > /p/

Romanian shares with the Sabellic branch of the Italic languages the shift of original /kw/ to /p/. This is documented in Oscan, for instance, with pis 'who' (cf. Latin quis [Buck 1904: 7]). Romanian examples are apă 'water' (cf. Latin aqua) and patru 'four' (cf. Latin quattor [Cioranescu 2003: 41]). Among the other Romance languages, Sardinian saw a similar development, e.g., ebba, 'water'.

Puşcariu (1973: 11, 29–30) has argued that this development in Romanian and Sardinian is related to influence from a substratum. On the other side of the debate, Sala (2005: 161) and Politzer (1953: 487–489) do not believe it is necessary to invoke a substratum and see the developments as parallel but unconnected.

It remains unclear whether Romanian limbă, 'tongue' (cf. Latin lingua), should be included as an example of the phenomenon. On the one hand, it would seem to present a voiced version of the rule, with subsequent shift to the bilabial nasal. But the word limbă, 'tongue', may have been current within Latin on a colloquial level, possibly by folk etymology from lambere, 'to lick'. (cf., Sardinian limba, 'tongue').

PhiN 43/2008: 12

In the Sabellic Italic languages, this shift is exhibited in every lexical item with original /kw/. Romanian, however, is less consistent. First off, Romanian /kw/ > /p/ occurs only where /kw/ preceded the vowel /a/:

apă 'water' (cf. Latin aqua)

patru 'four' (cf. Latin quattor)

iapă, 'horse' (cf. Latin equa)

păresimi, 'Lent' (cf. Latin quadragesima)

There are, however, at least as many exceptions to this rule as there are examples of it:

când, 'when' (cf. Latin quando)

cât, 'how much' (cf. Latin cuantus)

ca, 'like' (cf. Latin quasi)

cam, 'kind of' (cf. Latin quam [magis])

care, 'which' (cf. Latin qualis, 'what sort')

and încă, 'still' (cf. Latin unquam, 'ever')

This inconsistency has been observed, but as yet no definitive explanation has been forwarded as to what secondary rules could have conditioned it (Rosetti 1969: 123-124; Sala 2005: 161; Du Nay 1996: 43; Hasdeu 1972: 121), which makes influence by a substratum a compelling possibility.

3 /kt/ > /pt/

Romanian also displays a shift of the cluster /kt/ > /pt/. Examples are opt, 'eight' (cf. Latin octem) and fapt, 'fact' (cf. Latin factum). Unlike /kwa/ > /pa/, /kt/ > /pt/ occurs without exception in Romanian. (Items such as efect, 'effect', are either more recently borrowed from other Romance languages or are neologisms based on Latin.) Items such as coace, 'to cook', which keep the /k [če]/ in most forms, shift to /p/ in the closed environment produced in the past participle copt, 'cooked'. The alternation between /k/ and /p/, then, is phonologically conditioned and is not a change to the root of the word as such. Because it does not internally present any phonological inconsistency, I view this alternation as a separate phenomenon from the /kwa/ > /pa/ items discussed above. A similar phenomenon in Romanian exists with the shift of /sc/ > /şt/ before high vowels (e.g., sciō > ştiu, 'I know'). Here also, words alternate depending on environment, e.g., ceaşcă, 'cup', ceşti, 'cups'.

Oscan and Umbrian shifted the cluster /kt/ to /ht/. E.g., Oscan Úhtavis (cf. Latin Octavius) and Umbrian rehte (cf. Latin recte, 'rightly' [Buck 1903: 11]).

Even within Latin itself, there are isolated cases of a /k/ and /p/ interchange, e.g., columbus and palumbes, both meaning 'pigeon, dove'. These may be explained as borrowings from another Italic language such as Oscan, or borrowings from an otherwise undocumented dialect of Latin that exhibited such a shift.

PhiN 43/2008: 13

4 An "Italic" Substratum

Puşcariu has asserted that the inconsistent realization of /k/ > /p/ in Romanian is part of a substratal influence (1973: 129–130). I propose as a paradigm through which to explain the inconsistency, that Romanian merely borrowed some of these items. If this is the case, it would seem most logical to posit that words displaying /kw/ > /p/ are the imports. Every other word in Romanian could then be explained by the consistent shift /kw/ > /k/.

If words such as patru and apă are borrowings into Romanian, they must come from a language that is closely related to Vulgar Latin. As a hypothetical substratum, I will describe it as "Italic" in the broadest possible understanding of that term. This substratum could be from a previously undocumented dialect of Latin, another member of the Italic language family proper, or another language bearing strong affinities with Latin, the best understood and documented of the Italic languages.

5 /kw/ > /f/

All the items described above have been properly explained through their Latin counterparts. My principal contribution to identifying a wider scope to this substratum is the identification of lexical items within Romanian which point to a third and previously unrecognized realization of original /kw/.

Romanian fel, 'kind, sort', has been tentatively and unconvincingly explained as a derivation from Magyar -fele, 'similar' (Cihac 1879: 498).

I propose that Romanian fel is more likely akin to Latin qualis, 'what sort of, what kind of'. If fel is also a borrowing into Romanian from a language that had shifted initial /kw/ to a bilabial, it would represent a further phonological development. The following high vowel or liquid could have conditioned a fricatization of the /p/. (Oscan displays a similar shift in the cluster /pt/ > /ft/, e.g., scriftas [cf. Latin scriptae] Buck 1903: 8.)

Also suggestive of this development is Romanian sfert 'fourth, quarter', which has been explained from Slavic četvrutu (Cihac 1879: 387). This derivation, however, requires positing a phonological realization different from all other Slavic borrowing, because no other Romanian word with initial sf- is derived from Slavic initial č (e.g., sfat, 'advice' [cf. Slavic sŭvĕtŭ]; sfînt, 'holy' [cf. Slavic svętŭ]). Again, the form suggests a more straightforward relationship with Latin quartus 'fourth, quarter. It corroborates /kw/ > /f/ before the high vowel and/or the liquid.

Another possible case of the phenomenon is firetic, 'argumentative' (cf. Latin querī, 'to complain'). (Though not derived from an original /kw/ in Latin, note as well the similarity between afin, 'berry bush' and Latin acinus, 'berry'.)

In summary, by this understanding of the items explained above, Romanian itself underwent the following phonetic development regardless of environment:

/kw/ > /k/

Another language, from which Romanian borrowed a number of words, developed differently and preserved original /kw/ with either /p/ or /f/, depending on the following environment:

/kwa/ > /pa/

/kwe, kwi/ > /fe, fi/

PhiN 43/2008: 14

6 Possible Sources of the "Italic" Substratum

As stated above, these words could be preservations of a previously unknown dialectical variation in the Latin brought to the Balkans by the Roman expansion. The history of Roman settlement in Dacia involved numerous stages and waves of peoples. If one of them involved a populace with this particular dialect feature, it could produce the uneven attestation of the sound rule we observe in Romanian. Perhaps as well the populations involved in the Roman expansion included speakers of a still extant Sabellic Italic language.

A more intriguing possibility is that the items came from an indigenous language in the area of settlement. Wherever the Italic Urheimat was, those speakers who eventually would migrate into the Italian peninsula at least transited the Balkan area. A population of Italic speakers could have continued there until contact between the peoples was reestablished via the Roman conquest of the Baltic peninsula. Similarly the Venetic language, which exhibits affinities with the Italic languages, was still extant in northern Italy and Illyricum until the Roman expansion (Pellegrini 1967).

7 A Window into Thraco-Dacian?

Still more controversial is the possibility that the items in question come from the still poorly understood Dacian and Thracian languages. If that is the case, at least one of these languages was much more like Italic than current reconstructions based solely on onomastics and the isolated words cited by ancient sources would suggest.

The assertion that Traco-Dacian was similar to Latin is not new. The controversial Romanian writer Nicholae Densusianu argued that Dacian and Latin were merely dialects of the same language (Densusianu 1913). As evidence for this, he noted Ovid's claim to have learned the indigenous languages quite quickly during his exile to Dacia (see Ovid's Tristia 5.57ff.)

Also suggestive of this similarity is an inscription found at Sarmizegetus in Romania, which describes the Dacian patriot Decebal as "Decebalus per Scorlio" (interpreted to mean "Decebal, son of Scorlio" [Mackendrick 1975: 64–65]). If this is correct, the Dacian word for 'son', per, is close to Latin's puer (Georgiev 1983: 1181).

The picture gained thus far into Dacian and Thracian, mainly through onomastic items, does not preclude the possibility that one of these languages was more akin to the Italic language family than currently suspected. It is implausible, for instance, that a meaningful classification and description of the Italic language family could have been achieved solely from a list of Italian place and personal names, let alone details of the variations between Latin and Umbrian. We should not, then, assume we yet know enough about these languages to exclude them as a hypothetical source of words that are similar but not identical to Latin.

8 Conclusion

I have presented here further evidence that Romanian preserves traces of a language similar to Latin, but exhibiting a distinctive phonology that allows us to distinguish it from the Proto-Romance ancestor of Romanian itself. If this theory bears up under further research, it may point to the way to explaining further inconsistencies within Romanian phonology. If the source of these items does prove to be one of the languages in the Balkan peninsula before the Roman conquest, this theory would important implications for further research into the field of Thracian and Dacian studies.

PhiN 43/2008: 15

9 Excursus: An "Italic" Reading of the Ezerovo Ring Inscription

If a language bearing strong affinities to Italic had subsisted in the Balkan peninsula, whether that were a previously unknown survival of an Italic language there or known entities such as Traco-Dacian, the scanty inscriptions understood as Thracian or Dacian should be reassessed to see if an Italic language makes more sense of them. One of the longest of these is the Ezerovo Ring Inscription, found in 1912 in burial mound in Bulgaria. Numerous and wildly divergent decipherments of it have been proposed, indicative of the poor understanding at present of the Thraco-Dacian languages (Dechev 1957: 566–582). I propose, however, that an Italic reading of the piece yields compelling results.

A Romanized version of the inscription, which is in Greek letters, is as follows:


One recurrent series of letters in the text, suggestive of morphological significance, is the vowels ea, occurring five times total. Two of those are followed by repetitions that produce a sensible text, if understood as Italic:



In the Italic (and other Indo-European) languages, a 3rd person singular verb is marked with /t/, while the plural is /nt/. Note that these forms are then immediately followed by letters which additionally suggest corresponding singular and plural forms, il and ile. The common Latin demonstrative ille (singular) and illī (plural) 'that' could be a cognate to the forms in the inscription. I propose as a reading for the above forms, "May that one go" (eat il) and "May they go" (eant ile).

A forced rendering of the full text, which would resort to mere speculation of much of the inscription, will not be proposed here. (Indeed, not even 50% of the oldest Latin inscriptions, such as the Doenos Inscription [ca. 500 B.C.] can be confidently translated, though the later stage of the language is perfectly known.) Even so, an Italic understanding of the inscription would suggest other possible recoveries. The first incidence of the /ea/ series could be read as eas, 'may you go' or even teneas 'may you hold'. The series dom suggests a word akin to Latin domus 'house' (noted by Kretschmer 1915: 90–91).

Bibliography of Works Cited

Buck, Carl Darling (1904): A Grammar of Oscan and Umbrian: With a Collection of Inscriptions and a Glossary, Boston: Ginn and Company.

PhiN 43/2008: 16

Cihac, Alexandru (1879): Dictionnaire d'étymologie daco-romane, éléments slaves, magyars, turcs, grecs-moderne et albanais, Frankfurt: Ludolphe St-Goar.

Cioranescu, Alexandru (ed.) (2003): Dictionarul Etimologic Al Limbii Romane, Bucharest: Editura Saeculum I. O. 2003.

Dechev, Dimitur (1957): Die thrakischen Sprachreste, Wien: In Kommission bei R.M. Rohrer.

Densusianu, Nicholae (1913): Preistorica Dacia, Buchurest.

Du Nay, Andre (1996): The origins of the Rumanians: the early history of the Rumanian language, Buffalo: Matthias Corvinus Pub.

Duridanov, Ivan (1985): Die Sprache der Thraker, Neuried: Hieronymus.

Georgiev, Vladimir (1983): "Thrakish und Dakisch", in: Temporini, Hildegard (ed.), Aufstieg und Niedergang der romischen Welt. Geschichte und Kultur Roms im Spiegel der neueren Forschung, 1148–1194, Berlin / New York: Walter De Gruyter.

Hasdeu, Bogdan Petriceicu (1972): Etymologicum Magnum Romaniae. Editie ingrijita si studiu introductiv de Grigore Brancus, Bucharest: Minerva.

Hubschmid, Johannes (1983): "Zum Substrat und zur Vorgeschichte des Rumänischen (Probleme der Balkanlinguistik)", in: Zeitschrift für romanische Philologie 99, 497–511.

Kretschmer, Paul (1915): "Glotta", in: Zeitschrift für griechische und lateinische Sprache 7.

Mackendrick, Paul Lachlan (1975): The Dacian Stones Speak, Chapel Hill, University of North Carolina Press.

Pellegrini, Giovanni Battista and Aldo L. Prosdocimi (1967): La lingua venetica, Padova: Istituto di glottologia dell'Universita di Padova.

Politzer, Robert L. (1953): "On the Rumanian and Sardinian treatment of Latin qua and qua." Modern Language Notes 68–7: 487–489.

Puşcariu, Sextil (1973 [reprint of 1937 edition]): Études de linguistique roumaine, New York: G. Olms

Rosetti, Alexandru et al. (eds.) (1965): Istoria limbii române, Vol. II, 1969, Bucharest: Editura Academiei Republicii Populare Romine.

Russu, Ion, I. (1967): Limba traco-dacilor, Bucharest: Editura Academiei Republicii Populare Romine.

Sala, Marius (2005): From Latin to Romanian: the historical development of Romanian in a comparative romance context, Oxford/MS: University of Mississippi (Romance monographs 63).