A. The innovations of Latin in the 1stB3rd centuries AD

i > e : magister > maester (Northern Rumanian m|estru ´master´).

e > a , by assimilation: passere > passar (N. Rum. pas|re).

i > a , by assimilation: silvaticus > salvaticus (N. Rum. s|lbatic ´wild´).

o > e : rotundus > retundus (N. Rum. r|tund, Old Italian ritondo, Proven al redon, Spanish and Portuguese redondo).

B. THE INNOVATIONS OF LATE LATIN (4th B 7th CENTURIES)

The assibilation of Latin t + e, i, as well as that of Latin d + i followed by a vowel appeared, according to grammars, in the 5th century. (For more details about these changes see below: Common Rumanian, pp. 115B119).

The late date of this change may explain its absence in certain dialects spoken in remote areas, such as Logudorian in central Sardinia and the same is the case of the Latin elements of German and Basque. In the northern dialect of Dalmatian, k was assibilated in front of i but not in front of e. In Arumanian, ts stands on the place of Latin k (+ e, i), in Northern Rumanian, … . (For further discussion of this problem, see below: Common Rumanian, p.115).

The use of the genitive for the dative can be found occasionally already in classical Latin. This trendency spread continually and became Aexceedingly widespread during the period of transition from Latin to Romance.@ The documents of later periods (the Merovingian period and Medieval Latin) show Athe use of the genitive of the pronouns B sing. illius, ipsius, plur. illorum, eorum, etc., B to supply the function of the dative.@ The process has been of great significance for the Romance languages:

Latin had no definite article. The Greek article was, in translations made at the end of the 3rd and the early 4th century, rendered by ille, ipse, hic, iste, or idem. The Rumanian definite atricle represents ille. The change demonstrative pronoun > definite article was protracted; in the texts, a definite article in its present day form appeared IN THE 7TH CENTURY, in everyday speech probably somewhat earlier.The phases of this process were essentially the same in Rumanian as in the entire territory of the Neolatin languages. (Regarding the postposition of the article cf. below, p. 64).

The Latin name of the Slavs (originally, only of the Slovenes), Sclavus or Sclavinus is found in texts from the 6th century on. The Northern Rumanian form of this word is Õchiau, in the plural, Õchei; the Arumanian form is Õcl´eau. In late Vulgar Latin, periphrastic contructions appeared for hieme ´during the winter´, vere ´during the spring´, and estate ´during the summer´: tempore hiemis, hiberno tempore, verno tempore, primo vere, aestivo tempore. This development finally resulted in the Romance terms for the seasons of the year, e.g. Italian inverno, French hiver, Rumanian iarn|; Italian primavera, Rumanian prim|var|. (Lat. aestas survived in Italian estate, French été, etc.).

In Late Latin texts, very often the derivations of aeramen and aeramentum are found instead of classical Latin aes. As judged by the texts, for example, Peregrinatio Aethereae, both variants were equally popular; however, in the Romance dialects, only aeramen survived: Italian rame, Rumanian aram|, etc. A specific East Latin feature is the preservation of Latin imperator ´emperor´. It is explained Aby the continued existence, in Byzantium, of the Roman Empire; the Occidental forms are of learned origin@ (cf. N. Rum. împ|rat, Albanian mbret).

Classical Latin hostis ´enemy´ changed its meaning to ´army´ B Aan eloquent testimony, be it said in passing, to the general attitude towards armed forces at that time.@ Some of the earliest proofs of this change are given by Löfstedt; all are from the 6th century. From hostis with this meaning developed Portuguese hoste, Spanish hueste, and Rumanian oaste, ´army´.

Clasical Latin necare ´to kill´ became more specialized, i.e., it received the meaning ´to drown´. This was, of course, a gradual process, perhaps through the intermediate stages of Aersticken, erwürgen.@ In Greg. Tur. Hist. Franc. VI 35, necare is used in both the old general meaning and the new, more narrow one. ADie endgültige Einengung und Spezialisierung des Sinnes ist also erst spät und ganz allmählich vor sich gegangen@... The Latin word necare exists in Rumanian in the form îneca, with the meaning ´to drown´ (cf. French noyer ´to drown´).

During the first five centuries AD, the populations living in Italy and in the Balkan peninsula belonged to the same state, the highly organized Roman Empire. Close contact existed between the Romans living in Italy and those in adjacent areas of the Balkans. The vestiges of this long period of community are found in the languages which carry on the Latin from those centuries spoken in Italy and in the Balkan peninsula: in Friulian, Venetian, and other Italian dialects, in Rhetoromance, in Dalmatian (until the end of the 19th century), in the Latin elements of Albanian B and in Rumanian. These are phonetical peculiarities, special expressions, and unusual meanings of certain words (changes of meaning only found in these idioms).

La Rhétie parait avoir pris part au contraire à la diphtongaison initiale en ie, uo, qui de là, par le Frioul et l´Istrie, s´ ést propagée le long de la côte de Dalmatie, même dans le cas d´ entrave (istr. mierlo, kuorno, puorta, vegl. fiasta, puarta). Dans la péninsule proprement dite des Balkans, où „ seul entre en ligne de compte (pour  , voir ' 153), la diphtongaison se produit également devant l´entrave latine: on a non seulement roum. ier/ (h�ri), mais

aussi piept (p�ctus)

Of all Romance languages, only Rumanian has a counterpart to the use of ora in this sense (doi ora ´twice´): N. Rumanian de dou| ori ´twice´.

Greek kottizw was transferred to ancient Venetian (scotezar), to Istria (kutisa), to Albanian (kudzon), and to Rumanian (cuteza ´to dare´) during the Middle Ages. This word, among many others,

Latin levare ´to lighten, to alleviate´ has, in Friulian, Rumanian, and Albanian, among other senses, received also that of ´to buy´: Friul. jeva, N. Rumanian lua, Albanian bl´ën, (from. Latin *ablevare) ´to buy´. Friul. no puess jevalu, no ai vonde bez, N. Rumanian nu-l pot lua pentru c| n-am destui bani ´I cannot buy it because I don´t have enough money´.

Latin intelligere is preserved in the Rhetoromance dialect of Engadin: incler, and in Rumanian: înÛelege ´to understand´. In the other Romance languages, this word was replacd by capere, comprendere, or intendere.

An expression existing in the Tyrol dialect of Rhetoromance and in N. Rumanian is soredle da, respectively soarele d| ´the sun appears´.

Rumanian shows a number of correspondences also with different southern Italian dialects: there are, for example, nzurare (Napoli), nzurar (Abruzzes), and N. Rum. (a se) însura ´to get married´ (from Latin inuxorare); ammessarum (southern Italy, in the Codex Cavensis) and N. Rum. arm|sar ´stallion´ (from Lat. admissarius); ammisteka (Abruzzes) and N. Rum. amesteca ´to mix´; ceppe (Abruzzes) and N. Rum. cep ´bung, plug, spigot; tap´ (from Lat. cippus ´rectangular pillar; fortification made of pointed sticks´).

Some of these correspondences may represent preserved archaic elements, others are innovations, many of them of a highly specific nature. In their entirety, they indicate that the ancestors of the Rumanians were, from the 2ndB3rd centuries AD until the Slavic occupation of the Balkan peninsula, i.e., until about 600 AD, in close, everyday contact with the speakers of Latin dialects in Italy. As many circumstances pertaining to the problem of Rumanian ethnogenesis, this is not something new, B in fact, Gaston Paris arrived at the same conclusion more than one hundred years ago.

According to Rosetti, East Latin ("grupul oriental al limbii latine") was spoken in the Danubian provinces and along the shores of Dalmatia, as well as, until the 2nd half of the 3rd century AD, in Italy. Rumanian belongs to this, AppenninoBBalcanic linguistic group, together with Dalmatian, the Latin elements of Albanian, and the central and southern Italian dialects (Abruzzian, Sicilian, and Puglian). However, there are concordances between Rumanian and northern Italian dialects, and also Calabrian, Sardinian, and southern Apulian. Thirty words pertaining to shepherding are shared by Rumanian and several southern Italian dialects.

Theoretically, the division of the Romance languages may be made according to the substratum: Gallic, Iberic, Italian, etc.; the ancient political and geographical division: Gallia, Italy, Raetia; or only on the basis of geography: Balkan Romance or Balkan Latin. This Romance idiom has neither an ethnic basis (the substratum is Thracian and Illyrian), nor a political one. However, it reflects a geographic and cultural unity within a well-defined territory, distinct from the Western Romance languages. Rosetti considers that Balkan Latin was spoken in the Balkan peninsula from the 5th century AD onwards, after the Roman Empire was divided (395 AD). This idiom is on the basis of Rumanian, and of the Latin elements of Albanian and SerboBCroatian. Dalmatia belonged after the division of the Empire to the western empire, and received from that time on mostly influences from the west. But even Albanian shows some innovations from the west (for example Latin u > ü). Although Balkan Romance does not contain Italian, there are, as shown above, influences from Italian dialects in the Balkan idioms, thus also in Rumanian.

The frontiers of Balkan Latin may be drawn with considerable accuracy. Towards the west, Latin spoken in the Balkan peninsula was connected through Istria with the region of Friuli and other parts of northern Italy and by sea, with other regions of Italy. The intermediary area towards the west was thus Istria. The southern and southeastern frontiers were, on the basis of inscriptions, milestones, and coins made by the towns, determined by Jire…ek as follows:

Map 6. The Roman Empire in southeastern Europe in the period of East Latin (the 4thB6th centuries AD). The dividing line between the area of Greek and that in which mostly Latin was spoken is shown by the Jire…ek- and the Skok-line.

This is called the AJire…ek line@ (see map 6). Skok considered that the western part of this line should be drawn somewhat more to the south. Of course, the frontier between Latin and Greek was probably not as clear-cut; bilingual areas certainly existed both north and south of the Jire…ek line.

The eastern part of the northern frontiers of Balkan Latin: Roman colonization reached the lower Danube during the 1st century BC; until 106 AD, the Danube was the northeastern frontier. Between 106 and 271B275 AD, the territories of present day Oltenia, part of the Banat and part of Transylvania were a Roman province: Dacia Traiana. Thus, during 169 years, the frontiers of the Empire were pushed northward as far as to northern Transylvania. However, from 275 AD until the withdrawal of the Byzantine army at the beginning of the 7th century, i.e., for more than three centuries which roughly correspond to the period of Balkan Latin (and in entire Romania, to Late Latin), the Danubian limes was the northeastern frontier of the Empire, thus also of Balkan Latin.

(a) ß was preserved in Rumanian, Albanian, and Sardinian: Latin gula > Rumanian gur|, lupus > lup, numerus > num|r.

(b) The pronunciation o of the diphthong au is found in inscriptions: Olii instead of Aulii in Pompei, oricla in Gallia; it is criticized in Appenix Probi (written after the 3rdB4th centuries): Aa colore auri quod rustici orum dicebant.@ This pronunciation was introduced in Rome by Umbrians and other groups after the war in 89 BC. The diphthong au was, however, preserved in Rumanian, Dalmatian, Friulian, Sicilian, Provencale, and Catalan, in a stressed position: Latin aur > N. Rumanian aur, Vegliotic yaur, Old Proven al aur; Lat. laudare > N. Rum. l|uda, Friulian lauda, Prov. lauzur. In Albanian, Lat. au > a : aurum > ar, paucum > pak.

(c) Examples of the change of voiceless -p-, -t-, -k- to voiced consonants are found already in inscriptions from Pompei (pagatus, megum [instead of me-cum]). In the west, this became general after the end of the 5th century, with the exception of a number of dialects (Mozarabic, Upper Bearnese, Upper Aragonese). In Balkan Latin, the voiceless stops were conserved "avec une remarquable fidélité": Latin pacare > N. Rum. împ|ca, ripa > rîp|, Vegliotic raipa; and also Sicilian ripa.

gw changed in the same way in Sardinian and Rumanian: Lat. lingua > Sard. limba, Rum. limb|, (Lat. sanguine > Sard. sambiene, Rum. sânge is an exception.)

(f) The loss of final s and its replacement by i in Italian in nouns of the third declension: N.Rum. munÛi, Ital. monti, (but French monts).

Also regarding vocabulary, there are similarities between East Latin (including southern Italy) as opposed to the west. But even here, the situation is complicated. A detailed study of the East Latin vocabulary is found in Istoria limbii române (edit. by A. Rosetti, B. Cazacu & I. Coteanu), 1969, vol. II, pp. 110B173; written by I. Fischer. The criterion for deciding whether a word belonged to the East Latin vocabulary was its existence in at least one of the Rumanian dialects.

The PAN-ROMANIC STOCK comprises 488 words. Among these, only seven changed their meaning in Rumanian.

A total of 107 words were PRESERVED BY RUMANIAN ONLY. New formations, unknown or unusual in Latin, as well as semantic changes are numerous in this group.

A total of 214 Pan-Romanic words do NOT EXIST IN RUMANIAN. This is a very high number, "considerably higher than the number of those absent in any other Romance language, including the Iberian languages, in the extreme west of the Romance territory." Many of these words belong to certain well-defined semantic spheres, which indicates that they are not lacking from Rumanian by chance. A large part of them are technical terms, marine, military, or commercial, and also agricultural. Also many Latin words pertaining to general civilization, such as balneum ´bath´, lanterna ´lamp´, lectus ´bed´, littera ´letter´, regula ´rule´, are lacking in Rumanian. The process of simplification and impoverishment, characteristic of Popular (Vulgar) Latin, was here more pronounced than in any other Romance idiom. Many synonyms disappeared, leaving one expression where earlier several existed. Thus, fleo, lacrimo, lamento, and ploro were all replaced by plango: N. Rum. plînge, Istrorum. pl|nze, Arum. plîngu, Meglenorum. pl ng, ´to weep, to cry´.

Also the number of inherited Latin terms pertaining to art and science, administration and religion, as well as some complex activities, such as iron manufacture and wooden handicraft is very low. THE LATIN WORDS CONCERNING URBAN LIFE ARE ENTIRELY ABSENT IN THE RUMANIAN LANGUAGE.

Of course, the Latin spoken by the Roman inhabitants of the numerous towns in the Balkan peninsula must have contained these and many other lexical elements which do not exist in Rumanian. Their absence in the language of the Vlachs is in accordance with information from other sources indicating that they were not town-dwellers.

(a) ex (de ex-): Italian, N. Rum. scurta ´to shorten´, Ital. scapeta, N. Rum. sc|p|ta ´to set down; tro decline´.

(b) extra- : Ital. strabello ´very beautiful´, stravecchio ´old´, N. Rum. str|vechi ´ancient´, str|luci ´to shine´.

On the basis of this process a SPECIALIZED GROUP OF PEOPLE DOES EXIST. The nature of this group (in the above example, farmers) determines the direction of the change. Other factors are also at work; for instance, in the case of trahere ´to draw´ > traire ´to milk´, a homonymic clash between moudre ´to milk´ and moudre ´to grind´ (from Latin mulgere and molere, respectively), made the elimination of one of the homonyms neccessary.

Remarks: turm| is an example of a word with a special sense (military) being used in a different special sense (shepherd). Latin stimulus had a similar, but broader sense: pointed stake used in battle; and driving stake, with an iron point, used to drive oxen; as well as figuratively ´stimulus, irritation´.

Another Latin word, mixticius ´mixed, crossed, hybrid´ may be added, probably > N. Rum. mistreÛ ´wild boar, (Sus scrofa)´; in French, Proven al, Spanish, and Portuguese with the original Latin sense (ILR 1969, p. 150). Althoug this is not a specific shepherd term, it belongs to the life of shepherds.

E. The question of the substratum

It is often said that the substratum of Rumanian is AThraco-Dacian@, ADaco-Moesian@ or AIllyrian@. Very little of these languages is known, however, and, moreover, nothing of what is known (mostly placenames without a known sense) can with certainty be shown to correspond to elements in the Rumanian language.

(b) Comparison must at first be made Awith the rests, no matter how precarious and uncertain, of DacoBMoesian.@

(c) Because of the paucity of the material, any of the old Balkan languages may be used when no DacoBMoesian equivalents are known: AIf no sure phonetic feature which excludes a Daco-Moesian origin exists in a word, we must not conclude that it was borrowed from another old Balkan language, but assume that it once also belonged to DacoBMoesian, even if it at persent is only attested in Thracian, Macedonian, or Illyrian.@

(d) Correspondences with Albanian (or with loans from Albanian or from the substratum of other modern Balkan languages) must be considered to originate from the substratum. ABut the comparison must be made between the Common Rumanian (româna comun|) and Common Albanian (albaneza comun|) forms and it must be extended to other Indo-European languages, with the aim to find possible relations between the respective forms and sounds.@

| also found in Albanian and Bulgarian, may originate from the substratum.

h present in the substratum, although a reinforcment by numerous Slavic loanwords is admitted.

Rhotacism is considered to be the consequence of the weakening of Latin intervocalic n and it is stated that Athe only fact to bear in mind is that also in Albanian, the simple n has had the same evolution as in Rumanian, i.e., that the distinction between strong and weak n existed.@

kw, gw > k, g; p, b B the non-uniform treatment of these Latin consonant groups.

ks > ps in Rumanian and ks > fs in Albanian is explained by the fact that this consonant group was weak also in Latin (cf. coxim > cossim). AOnly in a few cases is this group treated as ps in Rumanian and as fs in Albanian (Latin coxa > Rum. coaps|, Alb. kofshë; Lat. fraxinus > Rum. dialectally frapsin; Lat. toxico > toapsec; Lat. laxa > Alb. lafshë).@

The definite article, which is enclitic in both Albanian and Rumanian, Ashows in Rumanian and Albanian similarities in details which prove a connection between these two systems@ (ILR 1969, p. 324).

The neuter gender: AThe many correspondences in details with Albanian (among others, the appearance of some neuters as masculines in the singular and feminines in the plural) shows that the influence of the substratum is possible also here, but neither the linguistic material nor the structural oppositions lead to a definitive conclusion in favor of the substratum.@

The personal pronoun: in Rumanian, the particle -ne in cine, mine, tine, sine is considered to derive from the substratum, cf. Alb. -në, Old Greek -nh and New Greek -na. AThe perfect parallelism between the Rumanian, Albanian, and New Greek forms points to an autochthonous origin of -ne in Rumanian@:

AThe Albanian pronoun presents also other interesting parallelisms with the Rumanian pronoun (the confusion between dative and accusative of certrain atonic forms, etc.) which would be worth to study@ (ILR 1969, p. 326).

The confusion between the dative in both Rumanian and Albanian (also present in Greek, Armenian, and Iranian) is not considered to originate from the substratum because a similiar tendency appeared also in Vulgar Latin.

There are several thousands of Rumanian words of unknown etymology, 10 - 15% of the Rumanian word stock.A number of these originate from the pre-Roman substratum of the language. The criteria to decide which words belong to this category have been and still are discussed. Since the publication of the first edition of the present monograph, I.I. Russu published the results of his study of the pre-Latin elements of Rumanian (Etnogeneza românilor, 1981) with an extensive review of what is known about the subject. Russu uses the method of reconstruction of Indo-European roots and words: AAncient Indo-European archetypes: roots and intermediary forms, primary formations, ´reconstructed´ and hypothetical, corresponding to forms unquestionably related in Sanscrit, Old Persian, Balto-Slavic, Germanic, Celtic, Latin, Greek, etc.@ Poghirc (in ILR 1969, p. 319) used essentially the same principles (to look after etymologies among the rests of ADaco-Moesian@, the Baltic languages, Armenian, as well as other Indoeuropean languages from which Rumanian could not borrow). He adds that only words may be taken into consideration, I.E. roots are not sufficient. Poghirc divided the Aprobable or possible@ Rumanian substratum-words into 3 groups:

(1) AWords for which an etymology has been sought among words known from Daco-Moesian or other ancient Balkan languages.@ Thirty five words are placed in this group. The etymologies proposed are from Armenian, Balto-Slavic, Celtic, Dacian (threee words), Greek, Macedonian, Thracian, Thraco-Dacian (one word), etc. Most of them are very uncertain. Half (17) of these 35 words also exist in Albanian.

(2) ARumanian words also existing in Albanian.@ This group is the largest, with 101 words.

(3) AWords of the substratum deduced from a comparison with other Indo-European languages.@ Five words with obscure etymologies are given here. They may originate from a variety of languages, including Germanic, Greek, Hungarian, Lithuanian, and Sanskrit. Two of them (brînz| and burt|) may be connected with Albanian.

The result of this broad approach to the problem of the substratum of Rumanian is far from convincing. The origin from the substratum of the geographical names cannot be accepted (cf. below, pp. 245B246). But also group 1 of words, with a proposed etymology from DacoBMoesian or other ancient Balkan languages, and group 3, with proposed Indo-European connections, contain very many dubious and even certainly false etymologies (cf. Russu, Etnogeneza, pp. 59B66). Also in the list of Russu, in which about 180 Rumanian words are given as probably deriving from the substratum, the etymologies are not more reliable. Illyés analysed this word stock from the semantic viewpoint. He found that those Rumanian words which have an Albanian counterpart belong mainly to specific semantic categories: parts of the human body, shepherd terms and names of animals and plants encountered by a shepherd population in a mountainous region.On the other hand, words not found in Albanian are predominantly verbs and substantives of more general semantic content. This circumstance is, besides the uncertainty of etymologies, another reason for scepticism regarding the substratum origin of words not found in Albanian. One difficulty is to decide whether a given Rumanian word may be a loan from Albanian or vice versa (cf., however, below, pp. 71B72). There are no unequivocal criteria; it is not entirely clear how to interpret the sound pattern of these words. Many of them have exactly the same form in both languages: Rum. buz|, Alb. buzë, Rum. c|lbeaz|, g|lbeaz|, Alb. këlbazë, gëlbazë; others show sound changes: Rum. abur, Alb. avull, Rum. mînz, Alb. mës. Albanian has sounds which do not exist in Rumanian, where they correspond to one or several Rumanian sounds: Alb. th B Rum. Õ, Û, or c(i), Alb. dh B Rum. z, or d , etc. Rosetti assumes that the Rumanian sounds represent different sounds in Indo-Euroepan, which in Albanian evolved to a single sound, according to the sound laws of this language. It is today generally accepted that most of the lexical elements existing in both Rumanian and Albanian are not loans but derive from a common language.

For all elements of language (phonetics, morphology, and vocabulary) which probably originate from the substratum of Rumanian, it is true that a large number of them also exist in Albanian: of seven phonetic features discussed above, at least three: the phonemes /|/ and /h/, and the kt > pt (Alb. ft) change. Out of five morphologic elements discussed by Poghirc, the definite article and the particle -ne of the accusative of the personal pronoun are explained by the substratum, and also the neuter gender is in some way connected with it. In all these three cases, equivalents to the Rumanian forms are found in Albanian, reaching often into the smallest details. Out of 13 suffixes considered to originate from the substratum, six also exist in Albanian. Regarding vocabulary, the existence in both languages may be considered the most certain criterion for the inclusion of a certain word into the group of those coming from the substratum. There are about one hundred such words.

Thus, the method of broadening the field of investigation and taking a large number of (extinct and living) languages into consideration, gives many more or less possible etymologies. They are, however, very uncertain and are distributed over many different languages. On the other hand, there is one single language in which a large part of the elements considered to originate from the substratum of Rumanian are found B and that is modern Albanian.

THE ORIGINS OF THE RUMANIANS