Ethnic Continuity in the Carpatho-Danubian Area
1. THE ROMANIAN LANGUAGE
The Testimony of Language
Language Community: A General Consideration
The Apennino-Balkan Group of Romance Languages. The Descendants of East Latin
The Characteristics of East Latin Phonology
The Changes of Late Latin Continued in Romanian Phonology
Late Latin Characteristics Not Found in Romanian
The Relationship of the Territory of Former Dacia Traiana to the Roman Provinces South of the Danube During the Late Latin Period
2. THE PROBLEM OF THE ETHNO-LINGUISTIC SUBSTRATUM
Ancient Indo-European Elements in Romanian
The pre-Roman Languages of Southeastern Europe
The "Thraco-Dacian" Linguistic Data
Thracian and Dacian
The Problem of the Dacian Elements in Romanian
A Comparison of the pre-Roman Lexical Elements in Romanian with Thracian and Dacian Words
The Origin of Albanian
The Relationship of the Substratum of Romanian to Albanian (Phonetics and Phonology, Morphology and Syntax)
The pre-Latin Lexical Elements in the Romanian Language (The Semantic Aspect)
— TABLE I. The semantic distribution of Romanian lexical elements probably or possibly originating from the substratum
— TABLE II. The semantic distribution of a total of 198 Romanian words probably or possibly originating from the substratum
A Comparison of Two Groups of Words: Those with and Those Without an Albanian Counterpart
Theories Based on the Distribution of the Assumed Substratum Words Among the Romanian Dialects
— TABLE III. Substratum lexical elements assumed by Russu found only in Northern Romanian
Albanian-Romanian Contacts in the Late Latin Period
Common Romanian. The Early Slavic Influence
The Development of Dialects
3. THE ROMANIAN LINGUISTIC LITERATURE
Correspondences Between the Latin Elements of Albanian and Romanian
The "Carpatho-Balkan Territory" as the Assumed Area of the Origin of the Romanian Language
About the Relationship of Latin to Gothic and Its Assumed Relevance to the Romanian Language
The Alleged Gothic influence on the Romanian Language
The Inherited Latin Words in Romanian
4. THE THEORY OF THE "CORE REGIONS" (KERNGEBIETE) OF THE ROMANIAN LANGUAGE (Sextil Puşcariu)
— TABLES IV-V. Romanian dialectal words cited by Sextil Puşcariu to support his Kerngebiet theory
Other Proponents of the Kerngebiet Theory
The Kerngebiet Theory and the Romanian Dialects
Dardania and the Origin of Romanian
The Kerngebiet Theory in Romanian Historiography
1. THE ROMANIAN LANGUAGE
The Testimony of Language
The theory of Roman continuity north of the Danube emerged as a result of the discovery that Romanian, spoken in a territory that was once a Roman province (Dacia Traiana), is a Romance language. The early proponents of this theory (the 18th century Transylvanian School) were chiefly concerned with a political struggle, hut at that time the lack of sufficient data and scientific methods of investigation made it impossible adequately to analyze the origins of Romanian. With the accumulation of such data and better methods, it was again the facts of language that carried the problem further: the discovery of the ancient characteristics of Romanian and their relationship with several Balkan languages challenged the theory of continuity in the former Roman province of Dacia.
As mentioned previously, the material culture showing Roman provincial influence—-Roman patterns in pottery and other products, the circulation of Roman coins, and similar phenomena—does not prove the existence of a Latin-speaking population. Such phenomena can be observed in wide areas of Europe and to a comparable degree as in the former Dacia, since all the non-Roman peoples were strongly influenced by Roman culture and civilization during the period in question.
In contrast to the data provided by history and archaeology, the facts of language are specific. The Romanian language, with its ancient elements inherited from a pre-Roman population, as well as from those who spoke Late Latin, is the link between the speakers of present-day Romanian and their forebears. Consequently, knowing the antiquity of a certain linguistic phenomenon that exists both in Romanian and in one of these idioms (for example, a Late Latin lexical element), it can be postulated that the ancestors of the Romanians
must have been living under circumstances that permitted the incorporation of that element into their language. One or a few elements are not sufficient, of course; and in many cases difficulties arise about details; but this principle, if used judiciously and consistently, could give the most reliable results in resolving the difficult problem of the origins of Romanian. More concretely, the problem can be formulated as follows: is the theory that the ancestors of the Romanians lived outside the Roman Empire, beginning at the end of the third century A.D., compatible with the facts of the Romanian language?
There are several types of linguistic phenomena to be investigated in this connection. In this chapter, two of the most important will be presented: 1. the Latin elements (the vestiges in modern Romanian of a territorial variant of vulgar Latin, namely East Latin), as well as the vestiges of the Late Latin period, and 2. the pre-Latin elements (in other words, the vestiges from the substratum). This analysis will show the close relationship of the Romanian language with several idioms spoken in the central parts of the Balkan Peninsula. Considering them in the context of the historical situation in Southeastern Europe between the end of the third century A.D. and around 600 A.D. (or somewhat later), conclusions can be drawn about the territory in which the ancestors of the Romanians were then living.
Romanian scholars have produced several theories about the Romanian language that will be treated in the second part of this chapter; but since no really critical analysis of the problem has been permitted in Romania for several decades, these theories have not been exposed to normal critical scrutiny. The resulting gap cannot be compensated for by foreign studies, which are never as comprehensive or systematic as one would wish. The present analysis will point to the great need for international contributions to resolve these problems and will also suggest that political pressure on the Romanian scientific community has created a conflict between the officially desired conclusions and those based on facts.
It is well known that opinions differ among Romanian linguists and archaeologists on the subject of Romanian ethnogenesis. Linguistic research has brought more accurate and conclusive results, given the nature of that discipline. While in the past, linguists such as Alexandru Philippide and Ovid Densusianu could express their views freely, any deviation from the official line at this time is severely criticized. Such is the case with Ion I. Russus Etnogeneza românilor [The Ethnogenesis of the Romanians], which has been criticized by the archaeologist Nicolae Gudea.  Russu may justifiably be criticized, among other things, for the fact that he considers too many words of unknown etymology to be of substratum origin and in a significant number of
cases such an origin is dubious. Many of Russu's proposed Indo-European etymologies for the substratum words are uncertain and not accepted by most linguists.  Russu presents, however, a vast spectrum of material (partly incorporating his own earlier investigations into this monograph) and presents an interesting discussion of the relevant problems.
Russu's point of departure is that the only really specific element of documentary value about the early history of the Romanians is the Romanian language. Archaeological finds including those reported recently, are not able to throw too much light on the origin of the Romanians:  "We are still far from a reliable and evidenced, or at least probable, conclusion generally accepted by scholars. From the sparse records of the ancient chroniclers and from the often inconclusive archaeological remains, it never will be possible to know exactly what the extreme political-military act of official evacuation consisted of and what remained Roman in the old Carpathian province". 
In Gudea's opinion, Russu assigned too much significance to the autochthonous words—the presence of the substratum elements— which show a very close relationship with Albanian, that is, with a language spoken in a territory different from Dacia Traiana. In criticizing Russu, Gudea stated that "the way in which Russu conceives the contribution of the autochthon stratum of words [in reality, of the autochthonous population] in the process of ethnogenesis, an isolated and even unhistorical way, is another exaggeration and, because of the lack of connection with history, leaves the hypothesis open that the Romanian language and people may have been formed anywhere in the Balkans."  The archaeologist Gudea, in contrast to the linguist Russu, attributes great significance to archaeology; but he does not use the occasion to discuss the fact that linguistical and archaeological research give such different results.
In his monograph, Russu defends the hypothesis, as do most contemporary Romanian linguists, that the Romanian language and people were formed in a large territory both north and south of the Danube. This may be considered a compromise between the demands of the official concept and the conclusion suggested by the analysis of the Romanian language (in this case, the pre-Roman elements). This compromise is, however, quite fragile.
Another contemporary Romanian linguist considers Russu's Etnogeneza românilor to be a synthetic introduction into Romanian historiography, accepting Russu's historical concept that only the Romanian language may give concrete information about the ethnogenesis. 
Another illustration of the conceptual differences about Romanian ethnogenesis between archaeologists and historians, on the one hand,
and linguists, on the other, is given in a monograph by Iancu Fischer.  The author of this study, dealing with Romanization and its manifestations, considers, not without reservations, the official doctrine on the origins of the Romanians. Even when he states that the adoption of the Latin language must have been paralleled by an adoption of the Roman way of life, he continues that such an affirmation is by itself vague and refers to notions that are not only difficult to define but also impossible to investigate with scientific exactness. Fischer does not fully accept as immutable the conclusions reached by archaeologists adhering to the current official views on continuity. In his opinion, the elements of material culture may cross the borders of states and could, in ancient times, be adopted by non-Romanized populations.  A material object cannot give any information about the language of its user (or producer); and it is also a fact that the material (and spiritual) aspects of a culture are more easily spread than a language and consequently, the adoption of elements of Roman civilization would not presuppose a change as profound as linguistic Romanization. Archaeological vestiges may be used only as partial evidence in elucidating certain aspects of the process of Romanization. 
Language Community: A General Consideration
The characteristics of Balkan Latin that have survived are to be found mainly in Romanian, but also in the extinct Dalmatian language, the Latin elements of Albanian, New Greek, and to a certain extent in the Southern Slavic languages. A significant differentiation in languages occurred in the Balkan Peninsula. Through their specific characteristics, the Balkan languages constitute a "language community."  The term "language community" was first used by N.S. Trubetzky.  In the linguistic sense, G. Weigand considers Albanian, Romanian, and Bulgarian characteristic of Balkan languages, while Serbian, Greek, and Turkish are regarded as Balkan languages only in the geographic sense, because they are used on the Balkan Peninsula. The Bulgarian linguist Vladimir Georgiev places the Balkan languages in the following sequence: Romanian, Bulgarian, Albanian, and Greek. According to the German linguist Georg Renatus Solta, the true Balkan languages are Romanian (with its dialects), and Bulgarian; according to other researchers, also New Greek belongs to the Balkan languages.  The Romanian linguist Alexandru Rosetti agrees about the Balkan ties of the Romanian language. In Rosetti's view, the so-called Balkanisms may be traced to common substratum influences. This interpretation is also supported by other linguists including Ion I. Russu who, however, is otherwise opposed to the notion of Balkan linguistics.
According to Vladimir Georgiev, the substratum and adstratum influences on Balkan languages are not necessarily decisive but rather represent a "multiple convergence."  To support these views he points out, for instance, the post-position of the article—characteristic of Bulgarian, Macedonian, Romanian, and Albanian. The Romanian linguist Emil Petrovici believes that the Balkan characteristics of the Romanian sound system are due to Slavic influences.
One of the common phenomena of the Balkan language is the change from the unpronounced a to ă (Romanian and Bulgarian) and ë (Albanian). Rhotacism, that is, the transformation of the inter-vowel "1" into "r", is the result of the original, neighboring locations of Albanians and Romanians. 
The Apennino-Balkan Group of Romance Languages
The Descendants of East Latin
The first division of the Romance languages was made by F. Diez (1882). He distinguished an eastern group (Romanian and Italian), a southwestern group (Spanish and Portuguese), and a northwestern one (French and Provencal). His criteria, besides linguistic aspects, were literary, geographical, and political. The Italian linguist M. Bartoli, who described Dalmatian, which was not yet known by Diez, developed this division further, adding the criterion of the substratum. Bartoli places Middle and Southern Italian, Dalmatian, and Romanian, as well as the Latin elements found in Albanian, New Greek, and the Slavic languages of the Balkan Peninsula, in the eastern Apennino-Balkan group; and the rest of the Romance languages in another group called Pireneo-Alpino.  Later, the frontier between the western and eastern Romance language areas was established with a line drawn from La Spezia to Rimini. The territory of Italy south of this line is considered to belong to the eastern group.  This designation is used also by Rosetti, who believes that Romanian belongs to the Apennino-Balkan group together with Dalmatian, Albanian, and the central and southern Italian dialects (Abruzzian and Puglian). 
From these languages, as well from the Latin elements of the South Slavic languages and of New Greek, a variant of Latin spoken in the eastern provinces may be reconstructed; this can be called East Latin. The frontier is, of course, only approximate, the division not absolute, since the real situation is complicated both from the chronological and from the territorial points of view. Sardinian, for instance, often shows correspondences with the East Latin idioms, and there also are many cognates between northern Italian dialects and Romanian. On the other hand, Dalmatian also showed similarities to the western group.
In 1960 a monograph was published about the Latin language spoken "in the Danubian provinces of the Roman Empire."  It contains a study of the inscriptions and of the literary texts from the first six centuries A.D. found in Noricum, Pannonia Superior and Inferior, Dalmatia, Moesia Superior and Inferior, and Dacia Traiana (in the last-mentioned province only between the years 106 and about 260 A.D.). The population of this area of about 600,000 square kilometers is estimated to have been about three million during the Roman period. The inscriptions provide only limited information about the Latin idiom spoken in these provinces, since usage was in many cases inconsistent, final development often occurring very late.
As mentioned previously, Rome conquered part of the southern Balkan Peninsula in 168 B.C. and Illyria in 167 B.C. Dalmatia followed in 156 and 155 B.C.; Pannonia, from 12 to 9 A.D., and northwest of the Balkans, Noricum and Raetia, in 15 A.D. Moesia was created at approximately the same time; in 86 A.D., it was divided into Moesia Superior and Inferior. At the official division of the Roman Empire in 395 A.D. into eastern and western parts, Dalmatia remained with the western part and Preavalitana (Montenegro and northern Albania) went to the eastern part. The frontier between the two territories went from the gulf of Cattaro to the area west of Belgrade. Since most of Dalmatia was in the sphere of influence of Rome for almost another one-and-a-half centuries (in 535 A.D. it was conquered by Byzantium), it was able to take part in several linguistic developments that did not affect the rest of East Latin.  The Balkan provinces belonged to the empire until the end of the sixth century.
Roman life in this territory, south of the Danube, is well documented; and the Roman population was numerous, as indicated by a large number of roman towns: 24 in Dalmatia, 28 in Moesia Inferior, 13 in Moesia Superior, and 31 in Pannonia. Beginning with the fourth century, the organization of the Christian Church was intensive: the ruins of 45 buildings of religious character from the third to the sixth centuries inclusive have been found in the Balkan provinces, excluding Dalmatia. From the same territory, at least 57 bishops, all with Latin names, are known, beginning in the third century.  The official language of the Byzantine Empire was, until the early seventh century, Latin and was then replaced by Greek. The records of the fifth ecumenical council in Constantinople (553 A.D.) were written in Latin. The epigraphic material found in the Southeastern European provinces (a total of about 21,000 inscriptions) shows a Latin similar to that used in documents in the western parts of the Roman Empire, and the same is true about the texts of writers from the Balkan provinces.  The inscriptions were made, however, predominantly by a small
number of educated people near the top of social hierarchy; the texts were written by leaders of the Church, mostly for the masses, "who probably understood them without much difficulty." 
Until the sixth century the eastern part of the Roman Empire (Eastromania) was characterized by a definite degree of uniformity. After that, ties with Romanized peoples were broken. There is no doubt that the process of Romanization was brought to an end by the arrival of the Slavs around the year 600 A.D.; but as late as in the eighth century, Romania was the name of the Eastern Roman Empire and used by the Slavs and Latins in Thracia, Macedonia, and the Thessalian coastal regions.
In the former province of Dacia Traiana, which belonged to the empire for only about 165 years, no remains of religious buildings have been found, no written texts are known, and the 11 or 12 towns that existed during the Roman period had disappeared entirely by the second half of the fourth century at the latest. Of the more than 21,000 inscriptions found in the Southeastern European provinces, almost 3,000 came from Dacia. In spite of this apparently high number, this material does not provide information about the Latin spoken in the north-Danubian province. According to Haralambie Mihăescu, these inscriptions reflect a lingua franca, a Latin used by the Roman administration and the army. This is because those who ordered the inscriptions in Dacia Traiana belonged prevalently to these organizations. They rarely reflect the sincerity and speech of the simple citizen or the slave. Therefore, we lack any evidence that the Latin language of Dacia was different from the common Latin spoken in the other provinces of the Roman Empire. On the contrary, those almost 3,000 inscriptions from Dacia illustrate, in general, characteristics of language that appear in similar documents from the entire territory of the Roman Empire.  It may be added that the differentiation of Latin in the time that these inscriptions were made (106 to 260 A.D.) was probably relatively slight; more significant regional differences developed in the centuries that followed.
The review of the sources of East Latin shows that all data about its main differentiating characteristics derive from the territory south of the Danube, with the exception of the Romanian language, which today is spoken both north and south of the river.
The designation "Danubian Latin" is used by Romanian scholars  to describe "succinctly the Latin idiom on which the Romanian language is based. The description will focus on the main differentiating
characteristics of this territorial variant of the Latin language limited to both shores (Dacia and Moesia) of the Danube." 
The term "Danubian Latin" is not exempt from criticism. According to Iancu Fischer," we are dealing here with an hypothetical Danubian Latin, which may be nothing more than the Latinized reflection of the Romance data we possess. In other words, we consider that there was a Danubian Latin whose only descendant is Romanian, because we describe it on the basis of the characteristics of Romanian."  The risk would be avoided, argues Fischer, if we are able to define the chronological limits of our investigation (that is, of the Latin idiom from which Romanian developed), and to give the stages of evolution of each linguistic fact. Such a method would result in the establishment of really Latin phenomena (thus, from the period before the 8th–9th centuries) and of "a dialectal variant that will not be the creation of our fantasy or of faulty reasoning."
In this way, however, only the chronological problem can be solved. The question of the area in which this Latin idiom once was spoken is not answered; and in assuming that "Danubian Latin" was once also spoken in Dacia Traiana, because it was the ancestor of the Romanian language, spoken today mainly north of the Danube, one risks the circular reasoning that Romanian originates from the territory of former Dacia Traiana, because Danubian Latin was spoken there.
The term Danubian Latin is also inadequate because it could be applied to the provinces of Moesia Superior and Inferior and to Pannonia but not to the areas where most of the descendants of this Latin idiom were and are spoken: Dalmatian, the central and southern Italian dialects, and the Latin elements of Albanian and the South Slavic languages. (It is not known whether the speech of the Romans who were living in Dacia from 106 to 275 A.D. showed any regional peculiarities. There is no evidence to assume a special Latin idiom for the provinces of Moesia).
The Characteristics of East Latin Phonology
Vowels: The reduction of Latin ŭ to o, characteristic of the Western idioms, did not take place in the East: Lat. crŭcem > Northern Romanian cruce, Albanian kryqe (as opposed to Italian croce, French croix); Lat. pŭlverem, Dalmatian pulvro, Romanian pulbere, Albanian plluhur (cf., Italian polvere).  Another conservative trait in the East was the preservation of ŏ without diphthongation to uo, as occurred in Western Romance languages. Mihăescu remarks that uo in the place of o is also absent from the inscriptions found in Pannonia, Moesia Superior, and Dacia. 
The diphthong au was, in stressed syllables, preserved in Dalmatian, Friulian, Sicilian, Provencal, Catalan, and Romanian, for example, Lat. aur > Vegliotic (one of the Dalmatian dialects) yaur, Old Provençal aur, Romanian aur, but French or; Lat. taurus, Friulian taur, Old Catalan taur, Romanian taur, but Spanish toro. If the following syllable contained a u, the diphthong au was reduced to a: Lat. auscultate > Italian ascoltare, Old French ascouter, Rom. asculta; Lat. augustus > Italian agosto, Spanish and Portuguese agosto, Rom. dialectal agust. This phenomenon occurred in most of the Romance idioms, but in Romanian, it is also found in certain cases if the syllable is not followed by a u: Lat. repausare > Rom. (through *răpăsa) răposa; Lat. *plausare > Rom. plăsa. In Albanian, this is the rule: Lat. aurum > Alb. ar, Lat. aut > Alb. a, Lat. paucum > Alb. pak. 
Consonants: Here also are many instances of conservation as opposed to innovations in the West. The intervocalic voiceless occlusives (p, t, k) of Latin (with the exception of k before e, i) show, in the Western Romance idioms, a tendency toward sonorization. In the southern Italian dialects, in Dalmatian, partially in Sardinian, and also in Romanian, they were preserved: Lat. ripa > Dalmatian raipa, Sicilian ripa, N. Romanian rîpă; Lat. capistrum, rota, pecorarius > Italian capestro, rota, pecoraio, N. Rom. căpăstru, roată, păcurar, as opposed to Spanish cabestro, rueda, Portuguese pegureio, respectively.  The situation is similar with the occlusives before r, l (except cl): Lat. capra > Ital. capra, N. Rom. capră, but Spanish cabra, Lat. petra > Ital. pietra, N. Rom. piatră, but Spanish piedra, Lat. lacrima > Ital. lacrima, N. Rom. lacrimă, but Spanish lágrima. 
Latin -ct- changed in East Latin to -pt- or -ft-, and Lat. -cs- to -ps- or -fs-. The labial treatment of Latin -ct- is characteristic of the Balkan Peninsula.  (It has been explained by a possible Thracian influence, but it also appears in certain Greek dialects as well as in Macedonian.) With regard to Latin -cs-, it changed in all Romance languages to s or, if followed by i, to š: Rom. măsea from Lat. maxilla and ieşi, from Lat. exire. In Romanian, however, there are several words in which cs changed to -ps-: Lat. coxa > N. Rom. coapsă, Lat. fraxinus > Arumanian frapsin. Lat. toxicum > N. Rom. toapsec. This phenomenon has its counterpart in Albanian: Lat. coxa > Alb. kofshë (also koshë), Lat. metaxa > Alb. mëndafshë (also mëndafsh, mëndash). 
The final s was preserved in the West, while it disappeared in Italian and Romanian. Latin nos, minus, cantas > Sardinian nos, minus, cantas, French nos, moins, chantes, Spanish and Portuguese nos, menos, cantas, but Italian noi, meno, canti, N. Romanian noi, cînţi.  The final -i was not simply added, since it also appears in words in which it is not the inflectional ending of the plural (Lat. post - pos > N Rom.
poi), but must be explained by the substitution of one sound for another.  This change was very late, beyond doubt after the sixth century,  although it may be an independent reflection in Italian and Romanian of some older trends.  (In view of the numerous correspondences between Italian and Romanian, it seems at least as likely that also this phenomenon is an example of the close relationships between these two idioms.)
Late Latin had probably, at least in monosyllabic words, an extra vowel, (usually an -e) after -r, -l. This was preserved in Sardinian, Italian, and Romanian: Lat. cor, fel, mei, sal > Sardinian koro, fele, mele, sale, Italian cuore, fiele, miele, sale, N. Rom. fiere, miere, sare.  Latin -gn- corresponds in many of the Romance languages to a guttural or palatalized nasal occlusive: Lat. agnellus > French agneau, Italian agnello, Friulian añel, etc. In Romanian, Lat. -gn- changed to -mn-: Lat. lignum > N. Rom. lemn, Lat. cognatus > N. Rom. cumnat, Lat. signum > N. Rom. semn. This labial treatment is also found in southern Italian dialects: Lat. agnum > aunu. The area of this pronunciation in Italy was formerly much larger than it is today.  Latin cl > ch: Lat. clavis > Italian chiave, Rom. cheie (as opposed to French clé); Lat. Sclavus > Arum. scl'eau, N. Rom. şchiau.
There are also a few particular features of morphology and syntax, such as the plural of the nouns in the third declension in -i: Italian monti, Rom. munţi. Some changes characteristic of East Latin also appeared in the conjugation; in the first conjugation, the type izo / amus was spread in Moesia, in certain Rhaeto-Romanic dialects, partly in southern Italy, and is present also in Romanian. The construction of the future by the auxiliary verb corresponding to Latin vol o "I will" is found in Romanian, partly in Dalmatian, and in the Tosc dialect of Albanian, as well as in other Balkan languages. 
Lexical elements: More than 100 Latin words exist exclusively in Romanian, for example Lat. adjutorium > N. Rom. ajutor, blanditia > blîndeţe, lingula > lingură.  There are Latin words that appear exclusively in Romanian and Dalmatian, and another, larger group, found only in Romanian and Albanian, such as Lat. densus > N. Rom. des, Dalmatian dais; Lat. cerebrum > N. Rom. creier "brain," Albanian krie "head"; Lat. imperator > N. Rom. împărat, Alb. mbret "king." All these words must have existed in East Latin.
The Latin words that did not exist in East Latin are more difficult to determine. Beginning with Romanian, one may state that the number of Latin words found in most of the Romance idioms but not in Romanian is high. Of all Romance languages, it is, in fact, Romanian that has the fewest Latin lexical elements.  There are 214 such words, of which 129 are nouns, 26 adjectives, 5 adverbs, 13 numbers,
38 verbs, and 3 particles.  These words are largely grouped in certain semantic areas, and it is characteristic that the Latin terms regarding urban life are among them. There are, for example, 11 marine terms, 12 connected with farming, 8 terms of commerce, and 8 of civilization, as well as many important everyday words, such as amo "I love," corpus "body," dexter "right," hortus "garden," laboro "I work," nego "I deny," pauper "poor," sapio "I know," semper "always," uia "way." The lack of these words in present-day Romanian does not, of course, prove their lack in East Latin as well. It is not plausible that the Latin terms of commerce and urban life in general were lacking in the speech of the Romans living in the numerous Roman towns of the Balkan Peninsula. An attempt to determine which of these words really were absent in East Latin was made in Istoria limbii române (ILR).  One of the criteria was whether the word in question was, in Romanian, replaced by another Latin word or by a foreign borrowing; and the other, the situation in the Romanian language of the family of the words that disappeared.
The frequent use of certain suffixes is characteristic of East Latin: ex (de ex), extra, cf., Ital. scapeta, N. Rom. scăpata "to set down, to decline," Ital. stravecchio "old," N. Rom. străvechi "ancient"; in: Sicilian dialectal intiniriri "to rejuvenate," N. Rom. întineri.
The lack of a large number of urban terms suggests that the population's way of life was a rural one. This is also indicated by the changes of meaning of a number of Latin words, many of which are shared by Romanian and Albanian. There are, for example, N. Rom. pădure, Alb. püli "forest," from Lat. palus (genitive paludis), Vulgar Lat. padule "marsh"; N. Rom. şes, Alb. shesh "lowland," from Lat. sessum (sedere "to sit"). Other examples of changes of meaning shared by Romanian and Albanian are words of everyday use, such as N. Rom. cuvînt "word," cuvînta "to say, to utter, to speak," Alb. kuvendoj "I discuss," from Lat. conventus "district court, session, agreement"; N. Rom. cui "nail," Alb. kuj, with the same sense, from Lat. cuneus "wedge." The largest semantic group among these words concerns the pastoral way of life, that is, they are Latin words whose sense was adapted to the conditions of this life: Lat. meridies "middle day" > N. Rom. meriză "place where the cattle rests at midday"; animalia "animals" > N. Rom. nămaie "small cattle"; coccineus "scarlet red" > N. Rom. dialectal (Oltenia): coasin "sheep with reddish spots on its head."
There are several groups of words (and other elements of language) shared by Romanian, Italian, Sardinian, and Corsican; by Romanian and the dialect of Calabria; by Romanian and the southern Italian dialects; by Romanian and Sardinian.  An interesting group among
these lexical elements is that of thirty pastoral and peasant words shared by Romanian and the southern Italian dialects. Rosetti considers that these correspondences cannot be explained by simple coincidence but attributes them to the colonization of Dacia Traiana during the roman period by colonists from Dalmatia, who introduced the idioms spoken in Apulia to Dacia.  This is a possibility that cannot, in certain cases, be excluded; but it must be remembered that correspondences between Romanian and the idioms from the area in question exist also from the period after 275 A.D., when Dacia Traiana no longer belonged to the empire and such colonization was no longer possible.
The problem of the unity or diversity of the Latin language is a much debated question. Since the texts were in general written in a language that the writer or author considered correct, the speech of the common people was not directly preserved in them. One might say that most of the East Latin characteristics belong to post-classical Latin (from about 14 A.D. to around 200 A.D.) or Late Latin (after 200 A.D.).  Many elements of the popular or vulgar speech (sermo urbanus, rusticus, plebeius, vulgaris) were preserved in Latin grammars, whose authors tried to combat the "erroneous" forms, giving examples of the "mistakes" along with forms considered correct at the time. One of the most useful of these is the Appendix Probi, from the late sixth or the seventh century,  with 277 "incorrect" forms. Latin glossaries also give lists of Latin words, explaining them by terms generally known at the time. Such is the Glosses of Reichenau, written in northern France in the eighth century. Other sources of popular speech are inscriptions (often with "incorrect" forms), technical texts, and laws; the Christian authors, who used a language that was understood by the people, contributed much in this respect. From the sixth century on, historical treatises give valuable information about contemporary Latin. 
Late Latin forms can also be deduced from the Romance languages. In general, if a characteristic is found in all or most of these or in a geographically ascertainable group of them, it is considered to have existed in Late Vulgar Latin; there are, however, many exceptions to this rule.
Changes in language occur usually successively, over considerable lengths of time, and not necessarily simultaneously through the entire territory where the idiom is spoken. In Latin most of the innovations began in Rome and spread later throughout the whole territory of
the empire. It is often difficult to determine the earliest date of a particular feature. An innovation that became general in a certain century may have had much earlier precursors. Then there is also the possibility that different Romance languages changed independently in the same direction on the basis of a similar structure. It is, however, improbable that usages that were considered wrong as late as in the fifth to seventh centuries A.D. would have already existed before 275 A.D. The great majority of changes that lead from Latin to the Romance languages probably occurred after that year, that is, after the Roman retreat from Dacia Traiana.
Assuming that the speakers of the Latin idiom from which Romanian developed lived outside the Roman Empire after 275 A.D., one would expect considerable differences between Romanian and the rest of the Romance languages with regard to the Late Latin developments. A comparison from this point of view of Romanian with the other idioms of the Apennino-Balkan group and also with the rest of the Romance languages is therefore indispensable.
The Changes of Late Latin Continued in Romanian Phonology
Vowels: Stressed short Latin e was diphthongized, according to the grammars, in the fifth century A.D. The language of the ancestors of the Romanians took part in this change, as shown by Arumanian a(i)eri, from classical Latin heri, "yesterday," N. Romanian fier, from Lat. ferrum. This change in late Latin should not be confused with the pronunciation with an i̯ of initial e / i̯el, for el, which became the rule in N. Rom. (but not in Arumanian) after the eleventh century as a result of the symbiosis with South Slavs.
Latin malum "apple" had had a phonetic variant melum. This form became general in Late Latin and resulted in Italian melo, Rom. măr (both of which presuppose Latin melum).
Latin unstressed i was confused with e: Lat. vicinus, Spanish vecino, Provencal vezin, Rom. vecin; Lat. civitatem, Rom. cetate. Latin silvaticus (from silva "forest") had a popular variant: salvaticus. This variant is the basis of Italian salvatico, Friulian salvadi, French sauvage, and Rom. sălbatec. 
Consonants: c + e, i > č. - In the texts written in the western parts of the empire, some examples of this change are found as early as in the fifth century A.D.; but on the Balkan Peninsula, it is not found even in the sixth century.  Other indications of its late date are a number of German place names transferred from Latin during the fifth to sixth centuries: Lat. Celio Monte (in 470 A.D.) > German
Kellmünz, Lat. Celeusum (6th century) > German Kelsbach. Also German Keller (Old Germanic kelari) and Old Germanic kista, from Latin cellarium and cista, respectively, presuppose a k in the Latin words pronounced in the fifth century, when they were transferred to Old Germanic. 
ín Arumanian, Lat. c + e, i corresponds to ţ, in N. Rom., to č.  Dalmatian had č before i but preserved k before e.
Lat. t + e, i and d + e, i were assibilated after the sixth century: Lat. terra > N. Rom. ţară "land"; Lat. teneo > N. Rom. ţin "I hold"; Lat. decem > Arum. dzate, N. Rum. zece "ten"; Lat. deus > Arum. dzău, N. Rom. zeu "god". The assibilation of Lat. d + e, i, followed by a vowel started as early as the second century  (Lat. medius > N. Rom. miez "the midst").
Intervocalic v disappeared in certain situations: Appendix Probi writes: "rivus non rius," and v changed in certain cases to b: Appendix Probi: "alveus non albeus." Both innovations appear in several Romance languages; in Romanian: Late Latin rius > N. Rom. rîu, and Late Latin albeus > N. Rom. albie. The case of intervocalic b is similar: Lat. caballus > N. Rom. cal, Lat. vivus > N. Rom. viu; and in some instances -br- disappeared too: Lat. fabrum > N. Rom. faur.
Intervocalic b, pronounced as present-day English w, followed by u changed to g: Lat. nebula > N. Rom. negură, Lat. rubus > N. Rom. rug, Italian rogo. The w > g change after a, o and u is attested to in Late Latin texts.
Before a, Lat. kw changed to k in the West; in East Latin and in Sardinian, only in the following words:
This is explained by the hypothesis that these words were drawn into the sphere of the Latin pronouns quid > N. Rom. ce, quem > N. Rom. cine.  In all other instances, Latin kw developed in Romanian to p. in Sardinian to b: Lat. aqua > Sardinian abba, Rom. apă. In this respect, the Latin elements in Albanian are different from Romanian: Lat. quattuor > Alb. katrë, Lat. quadragesima > Alb. kreshmë (as opposed to N. Rom. patru and păresimi, respectively.) 
Morphology: The flexion of the noun shows the syncretisms of Late Latin, in the same way as in all the other Romance languages: the genitive merged with the dative and the accusative with the
ablative. The three genders are preserved in a way that may be deduced from the late texts and from the other Romance languages: "The tendency of grouping the inanimate objects as neutrals, the plural neutral in -a and -ora, the accord of the neutral nouns in the plural with the feminine form of the adjective." 
Since all the idioms that came from Common Romanian have the definite article, this must have existed in Common Romanian. The form illorum, genitive until the seventh century and after that period also dative, gives a more exact date: "... the definite article appears in Common Romanian after the seventh century, when positive signs of the reorganization of the flexion of the nouns are observed." 
Also the verbal flexion corresponds to the late patterns.  The Romanian compound perfect (habere + past participe) "came from constructions in which habere indicated possession and received the value of the perfect not earlier than after the fourth century."  For example: promissum enim habemus . . . nihil sine eius consilio agere "we have promised . . . not to undertake anything without his knowledge," in the text of Gregorius Turonensis, Gallia, from the years 538 to 594 A.D.; Matheum quem ante te ibi missum habui "Matheus whom I have sent there before you," in Acta Andreae et Matthiae apud anthropophagos, Italia, sixth to eighth centuries A.D.  Cf., the Romanian compound perfect: Arum. mi-amu dúsi-m-pizári "I went to the town" ("I have been in the town"), N. Rom. m-am dus la oraş "id." The auxiliary used here is, in Arum., amu, ai, am; avému, avéţî, au,  and in N. Rom., grammaticized, am, ai, a; am, aţi, au, from Lat. habeo "I have."
Lexicology: A series of new expressions appeared during the Late Latin period. A large group consists of terms used and partly created by the Christian Church.  This process was going on beginning in the first centuries A.D., although it became most intense during the fourth century. Basilica with the sense of "church," for example, is attested to beginning in the fourth century.  A term of Byzantine origin, it is found in Dalmatian (basalka), Albanian (bjeske), and Romanian (biserică): in the Western Romance languages, ecclesia, with the original sense of "community of the Christians" was continued: basilica is preserved probably in some place names and in the Engadin dialect (baselgia). Most of the basic Romanian terms of the Christian faith are of Latin origin: Lat. angelus > Rom. înger "angel," Lat. baptizo > Rom. boteza "to baptize, to name," Lat. paganus Rom. păgîn "heathen." It is sometimes asserted that missionaries from the empire were active in Dacia during the fourth century and that they could have taught these terms to the "Daco-Romans" living there. 
This is not a priori impossible, although there are only vague reports about early Christian missionaries north of the lower Danube.
It is not necessary, however, to assume missionaries north of the Danube to explain the presence of this group of words in the Romanian language. The general rule is that the Late Latin innovations in lexicology (as in the other branches of language) appear also in Romanian. The Glosses of Reichenau, written in Gallia in the eighth century, contains several words in use during the Late Latin period and different from the classical forms. These are found also in Romanian. The following examples are given by Iancu Fischer: 
This text also contains characteristics specific to Gallia, among them some transferred from Old Germanic. For example, instead of classical Latin dem "that I give," in the eighth century Gallia used donem, from which French donner derived, while Romanian has da; similarly, classical Latin opilio, eighth century Gallia (....) berbicarius, from which French berger derived, while Romanian has păcurar, from Latin pecorarius.
Appendix Probi lists many "erroneous" forms, giving what in the sixth to seventh centuries was considered correct by the grammaticians. The parallelism between Romanian and the rest of the Romance languages is evident also here: 
In Itinerarium Egeriae, written in about the year 400 A.D., probably in northwestern Spain, the verb "to eat" is expressed by manduco and the words edo and comedo do not appear. Spanish preserved, however, comedo, while manduco was preserved in Italian and in Romanian (mangiare and mânca, respectively). The verb plicate was, in the classical texts, used in the sense of "to bend, to curve"; in this text from 400 A.D. it means "to go toward." This sense was preserved in Portuguese, Spanish, and also Romanian (chegar, llegar and pleca, respectively).
In the Late Latin period, short words were increasingly replaced by longer ones, which were felt to be more distinct: for example, instead of classical Latin aes "metal, copper," one said aeramen, from which Italian rame, Northern Romanian aramă "copper." The original Romanian word for the Slavs (Northern Romanian şchiau, Arumanian scl'eau) derives from the Late Latin name given to the Sloven branch, of the Slavic peoples: Sclavus (or Sclavinus), attested to beginning with the sixth to seventh centuries.  This word is found also in Albanian: shqua "Bulgarian" (plur. shque). The correspondence with Albanian is in this case revealing also because the Romanians' ancestors, if they had lived north of the Danube in the sixth to seventh centuries, would have met other groups of Slavs than the Slovenes and would hardly have borrowed a designation used on the Balkan Peninsula to name them.
During the Late Latin period, several Latin words changed their meaningr'Reliable data about the century in which the change occurred are extant, for example, about hostis "enemy." This sense changed, beginning with the sixth century to "army." Romanian oaste, Spanish hueste and Portuguese hoste continue this sense.  Latin necare "to kill" received, also during the sixth century, a narrower sense: "to choke, suffocate, stifle," gradually developing the sense of "to drown," continued by French noyer and Romanian îneca.
Late Latin Characteristics Not Found in Romanian
It is evident that Romanian is based on Late (Vulgar) Latin like the other Romance languages and to the same degree. This applies to the entire structure of language as well as to particularities of minor significance. This was concisely stated by ILR, II, 1969 (p. 15-16): "The partial systems that constitute its [of the Romanian language]
grammatical structure continue without any interruption the original Latin systems, almost all of the linguistic changes that occurred between the fifth and the eighth century being found in one form or another in Late Danubian Latin."  (Here "Late Danubian Latin" stands for the idiom spoken by the ancestors of the Romanians).
The reason why a discussion of the exceptions to this rule is necessary is that Romanian linguists have repeatedly mentioned the isolation of Dacia in the third century A.D., which would explain the lack of some Late Latin innovations in the Romanian language:*From the third century onwards, Rome in fact no longer possessed the strength to impose its lexical innovations upon the remote and more autonomous provinces, such as Dacia. A series of words, known through the occidental Romance languages, do not penetrate into Dacia: aviaticus, carruca, (ex)iutare, sugia, and others".  A number of terms could not have penetrated into Dacia because they spread at a late period of time. 
The examples given are, however, neither significant nor numerous. All that Rosetti mentions in this context are those four lexical elements above. Mihăescu presents somewhat more; all of his examples will be presented and analyzed in the following:
1. The expression in se "together, in the same time" (cf., Italian insieme) appears in Gallia, Italia, Dalmatia, Pannonia Inferior, and Moesia Inferior beginning with the third century: "very probably, it was a late innovation that did not circulate in Dacia."
2. Two words of Old Germanic origin appeared in the documents after the third century and do not exist in Romanian: brutis "bride" and sculca "military guard."
3. Some lexical elements appear only in Dalmatia and in Noricum: anna "mother, nurse," socerio "brother-in-law," sponsa "spouse" (instead of marita or mulier).
4. The word mansio "building, post station" remained unfamiliar north of the Danube, "because Dacia was situated outside of the large trade and military routes that connected Constantinople and Asia Minor with Italia and Gallia."
5. The propagation of e in words such as espiritus, esponsa (cf., French esprit, épouse) did not go farther than Dalmatia.
6. The suffix -ment(e), originating from western Romania in the fifth century, "remained on the surface and could not penetrate into the masses in the Danubian provinces."
Some of even these few features represent differences between the Balkan Romance idioms (including Romanian) and the Western Romance
languages: anna, socerio, and sponsa, as well as e in such words as espiritus appeared in the Balkans only in Dalmatia; and the suffix -ment(e) is absent from all Balkan Romance idioms. The word brutis, of Old Germanic origin, exists only in Rhaeto-Romanic (brut) and French (bru).  The absence of numerous lexical elements does not have much importance here; every Romance language shows the absence of several words that exist in the other Romance idioms. Moreover, the Latin words that are lacking in Romanian are grouped in specific semantic groups, which suggests that they disappeared for social reasons rather than isolation from the rest of the Romance idioms.
The Relationship of the Territory of Former Dacia Traiana to the Roman Provinces South of the Danube During the Late Latin Period
The relationship between the former province of Dacia Traiana and the Roman provinces south of the Danube in the Late Latin period is an important, if not decisive, factor in the question of the territorial evolution of the Romanian language. First of all, it is necessary to clarify whether the relationship between the former province of Dacia Traiana and the Roman Empire, specifically to the Romanized regions south of the Danube, was retained even after the abandonment of the province and, if so, to what extent. Furthermore, it is needful to prove which specific linguistic differentiations occurred in the Late Latin in general, that is, in the period from the third to the sixth century.
The historical circumstances argue in favor of the presumption that no large-scale communication among peoples was possible between the area of former Dacia Traiana and the Roman Empire across the Danube, as had been the case earlier when the province belonged to the empire. Changes in the political situation much less radical than detachment from the Roman Empire resulted in the development of regional differences in the Latin language. The division of the empire in 395 A.D. into two halves is considered to have caused a divergent development of the Balkan Romance idioms from the language spoken in the West ; both halves were Roman states and had relations with each other quite different from those that prevailed between the empire and Barbaricum in the third century. The territory of the former Dacian province was, after 275 A.D., separated from the empire by the Danubian frontier, which was guarded by the Roman army. Communication with the empire was not entirely disrupted;
Christianity, for example, was spread during the fourth century among the Goths living north of the Danube. Also, as in the case of most other European areas, trade contacts have been attested to by archaeological finds. Roman merchants probably traveled through the territory to distant places, but such activity had no significant effect upon the language of the masses. Local trade involved more people and also those living in Barbaricum, who were allowed to visit, certain market places. This occurred under strict military supervision, not only for the security of the empire but also in order that the different customs regulations be enforced.  Such markets were permitted at only a few places; in 369 a.d., for example, there were only two along the entire Gothic frontier.  Even if more market places were permitted in the period when the frontier became weaker and trade increased, it is obvious that the contacts were restricted both to a very small proportion of the population and to the Roman market places along the (Danubian) border. Under these circumstances, no significant movement of peoples across the Danubian limes was possible, at least during the first centuries after the abandonment of Dacia by the empire. The occupation by the eastern empire of certain areas in southern Muntenia and OItenia and possibly also of the southern Banat for some decades in the fourth century cannot be considered as having promoted communication on a large-scale between the Latin-speaking inhabitants of the south and the populations living north of the Danube.
Consequently, if a Roman population remained in Dacia Traiana after the retreat of the Roman army and administration in 275 a.d., its Latin language could no longer have developed in the same way that it did earlier, in close contact with speakers of Latin in the Balkan Peninsula and in Italy. If there was such a language, it must have eventually disappeared, its speakers having been assimilated to the surrounding Goths, free (non-Romanized) Dacians, Gepidae, and the other populations known to have been living there after the Roman era; nothing will probably ever be known about it. But the ancestors of the Romanians, whose language contains all the Late Latin characteristics, must have been in close contact with the speakers of Late Latin, and thus can not have been living north of the Danube in the epoch of Late Latin (the 4th - 8th centuries A.D.).
Romanian is a Romance language that contains, to the same degree as the other Romance languages, all the characteristics of Late Latin, that is, the changes that appeared in Latin during the third to seventh
centuries. It also shows the peculiar characteristics of East Latin, that is, correspondences with the southern Italian dialects, Dalmatian, and the Latin elements of Albanian and Serbo-Croatian. This implies that the ancestors of the Romanians lived in close contact with the speakers of East Latin not only at the beginning of the Late Latin period (from around 200 A.D. to 275 A.D.) but also during the following centuries, probably up to the eighth century. An analysis of the historical records and the archaeological findings from the period shows that such contacts did not exist between the empire and the inhabitants of the territory of former Dacia Traiana, which was separated from the Roman world by the Danubian frontier. The assertions of Romanian scholars that after the Roman withdrawal Dacia could not have remained entirely isolated from the Romanized areas south of the Danube until the end of the sixth century and that there were religious contacts through the intermediacy of Byzantium  is sheer hypothesis.  Other circumstances, such as trade contacts and the occupation by the empire of some areas north of the Danube for short periods of time, do not suggest contacts of any linguistic significance. Consequently, the ancestors of the Romanians could not have been living, in the Late Latin period, in the territories north of the Danube; but they did belong to the Roman population living in the Balkan Peninsula.
2. THE PROBLEM OF THE ETHNO-LINGUISTIC SUBSTRATUM
Ancient Indo-European Elements in Romanian
Pre-Roman remnants in Romanian have been analyzed by several linguists, and many bold hypotheses have been advanced at the expense of scientific evidence.  A number of Romanian words may derive from Indo-European roots. Romanian rezema "to lean, to rest," for example, could derive from Indo-European reg' "right; to straighten, to raise helping,"  which developed into re(d)zem, ra(d)zem, in a satem language. The same Indo-European root is at the basis of Latin regem, regimen, and of Celtic rig-; but Romanian rezema cannot have derived from these languages because of the (d)z. Ion I. Russu gives 166 such Indo-European roots which, with more or less probability, could, on the basis of Romanian words, be of pre-Latin origin:  Indo-European bhas-k "band, bundle, handle," romanian bască "wool" (cf., Albanian bashkë, id.); Indo-European bhol "steam, smell," Romanian boare "breath of wind"; Indo-European dereu(o) "branch, tree," Romanian dialect druete "wood" (cf., Albanian dru, "id."); ster "sterile," Romanian sterp "sterile, lifeless."
These lexical elements probably originate from a pre-Latin language, the substratum of Romanian. This language contains, however, substratum elements in other areas as well: phonetics, morphology, and syntax. The criteria for deciding the substratum origin of any element of an unknown etymology are given by Cicerone Poghirc  and summarized in English by André Du Nay.  ideally, elements of Ancient Romanian (or Common Romanian), before the development of the dialects, should be used. One of the main criteria for establishing the substratum origin of any element of language is its existence in Albanian. Poghirc maintains, however, that other Indo-European languages,. including the old Balkan languages, must also be taken into consideration and, above all, the remains, even if questionable and uncertain, of Daco-Moesian, which have been almost totally neglected until now. Reference to Indo-European roots is not sufficient; words actually existing in one Indo-European language or another must be sought and the entire Romanian word explained.
The elements most probably from the substratum are: in phonetics, the phonemes /ă/ and /h/ and the kt > pt (Albanian ft) change; and in morphology, the definite article and the particle -ne of the accusative of the personal pronoun (the neuter gender in Romanian also probably has some connection with the substratum). About 10 to 13 suffixes probably originate from the substratum, of which -esc is very common. It is found in Thracian (-isko) with the same value as in Romanian and in Albanian (Alb. -ish); in Latin, this suffix appeared in words borrowed from Greek or Thracian. This problem will be discussed in some more detail below, in connection with the elements that Romanian shares with Albanian, including the important group of lexical elements from the substratum.
The pre-Roman Languages of Southeastern Europe
The few remains that have been preserved from the ancient languages spoken in Southeastern Europe before (and, for a considerable period of time, also during) the Roman conquest are not sufficient to determine even their basic characteristics. Thus, it is nearly impossible to set up a linguistic schema for this area. No sentences remained, and we therefore know almost nothing about their grammar. What we do know is a number of lexical elements, most of them without any known meaning, and something about the sound patterns, based on these lexical elements. In the period in question the main languages, after Greek, were Illyrian (in the west) and Thracian (in the east). In smaller, but not insignificant, areas Paeionian, Dardanian, and Scythian (in the northeast) were spoken. Getae and Dacians are
mentioned mainly north of the Danube, but their languages and their relations to the Thracians and the Illyrians are not clear. Some scholars also assume that there was a "Daco-Moesian" language. Among the remains of all these languages of the Indo-European family, there are certainly also pre-Indo-European elements. This complex problem can only be summarized briefly here.  For Romanian, the essential question is the possible dissemination area in which its substratum-language was spoken. An attempt will therefore be made to analyze the linguistic material from this viewpoint. Since the substratum of Romanian shows very close connections to present-day Albanian, the question of the ancient territories of this population will be investigated too.
Illyrian is generally considered to have been a separate language, spoken in the northwestern parts of the Balkan Peninsula (mainly the south Dalmatian coast and surroundings). About 300 place names, 600 personal names, 80 ethnic names (names of tribes), and 20 names of gods, all found in Greek and Roman texts, are attributed to Illyrian. This, in itself, would represent considerable material; but etymologies have been proposed for only about 10% of these lexical elements, of which 50 to 60 at most are acceptable or certain.  Most of the Indo-European sounds seem to have been preserved in Illyrian, as shown by these lexical elements. In general, Illyrian has been considered a satem language;  but the division centum/satem is, according to several more recent investigations, not applicable to this language, in which both developments appear. Illyrian and also Thracian are considered to have been spoken in the territory between the centum and the satem areas.  According to this view, Albanian shows a similar situation with regard to the phenomena in question. The frontier between Illyrian and Thracian went across Dardania and reached the Danube approximately at present-day Belgrade, but several areas were mixed Illyrian and Thracian. The relationship between these two ancient idioms is quite obscure and has been given various interpretations by different scholars.
Paeionian was spoken south of Dardania (north of present-day Thessaloniki). According to some scholars (Tomaschek, Jokl, Kretschmer), this language belonged to Illyrian; others (Detschew) believe it was similar to Thracian or was a completely separate language (like the Bulgarian linguist I. Duridanov).
Dardanian, spoken mainly in the valley of the Vardar River in the ancient province of Dardania (the region of present-day Niš and Skoplje), is also of unknown origin. Among the ancient authors, Strabon of Amascia  considered its speakers to be Illyrians while Polybios  said they were not. According to the Bulgarian scholar
Dimităr Dečev, this population was related to the Thracians but had been subdued at some time by the Illyrians. They also can represent an even older Balkan population. Another Bulgarian scholar, Ivan Duridanov, defined them as "Daco-Moesians." The name Dardania could have derived from Albanian dardhë "pear tree," and several place names there also seem to originate from Albanian (or from an ancient language but mediated to the Slavs by Albanians).
Thracian. The areas of dissemination of the Thracian language extended into the southeastern part of the Balkan Peninsula, reaching as far as north of the Danube and encompassing contemporary Bulgaria and parts of Greece and Turkey. The disappearance of this language occurred approximately in the middle of the sixth century A.D.; south of the Haemus (Balkan) Mountains it was Helenized and north of there Romanized. It is generally felt that Thracian belonged to the Indo-European languages and consequently must be investigated in this context on the basis of historical linguistic comparisons.
According to the Romanian linguist Ion 1. Russu, for instance, the remains of the Geto-Dacian language are fewer in number than the remains of the language of the Southern Thracians. The names of medicinal plants as well as personal and place names present special regional nuances; but the words (or names) with a clear or very probable etymology show that it is possible that only a dialectal differentiation was present between the Geto-Dacians and the Thracians south of the Danube and the Haemus Mountains.  In accordance with this view Ion I. Russu discusses the characteristics of "Thraco-Dacian," not differentiating between Thracian and Dacian. It must be stated, however, that most of the material preserved is clearly Thracian (originating from the Balkan Peninsula south of the Danube); the Dacian material is restricted to a few words. Russu's presentation can be summarized as follows.
The "Thraco-Dacian" Linguistic Data
Distinguished scholars, such as Wilhelm Tomaschek, Dimităr Dečev, Vladimir Beševliev, and, more recently, Vladimir Georgiev, Ion I. Russu, and I. Duridanov, have devoted their attention to the Thracian language. It is well known that the Thracians descended from various tribes and only few remnants of their language, in the form of inscriptions, glosses, given names, and place and ethnic names, have survived; even these remnants are sharply contested. From the standpoint of Balkan linguistics, it is first necessary to determine to what extent the Dacians and Getae may be subsumed under the Thracians.
Since no texts but only fragmentary lexical elements have been preserved in this language, virtually nothing is known about Thracian
grammar.  In Greek texts, 70 to 80 glosses have been found that could be attributed to this language. Only a few of these have a known meaning, and therefore they are not very helpful in the study of language. The main Thracian linguistic data consists of names: There are about 1,190 personal names (890 simple and 300 compound) and 910 place names (700 simple and 210 compound). 
Simple names are, for example, Bendis, Bithus, or Abro-zes, Dria-zis, the last two mentioned with suffixes. Compound names are Aulu-centus, Dece-balus, Epta-tralis. In contrast to Illyrian, in which compound names were rare, about one-fourth of the Thracian proper names are compound. The first element is often an adjective, for example, Germi-sara "warm water (source)", Diu-zenus "born from a god."
The endings of the place names in Thracia proper are mainly -berga, -bria, -burd, -cella, -diza, -pani, -para, -zura. In the territories north of the Danube, as well as on parts of the Balkan Peninsula, for example Dardania, Dacia Mediterranea, Little Scythia, Lower Moesia, and Thrace proper, the following endings predominate: -dava, -dina, -sara, -stur.
On the basis of these glosses, an attempt has been made, to reconstruct the phonetic features of Thracian, which Russu summarized as follows:
Vowels: a, e, o are largely preserved; as well as u, (in a few cases given as y). The diphthongs ai, ei (?), oi, au, eu were preserved.
Consonants: u̯ > v (or, in some cases, b). The voiceless occlusives k, p. t and the voiced ones, g, b, d, are preserved; "a consonant shift (Lautverschiebung) in Thracian does not appear probable." [92a] Indo-European bh, dh, gh lose their aspiration: b, d, g. The palatal occlusives k', g'(h) are represented by the post-dental spirants s and z (a satem phenomenon); a similar phenomenon is the delabialization of the labiovelars ku̯ and gu̯(h): k, g; the liquids are preserved, as well as m and n; s is preserved in all positions. The group sr is interposed with a t: Strymon, Istros.
Russu's conclusion is that most of the sounds of Indo-European were preserved in Thracian, according to the rules of a satem idiom of the East-European group. Other scholars believe that the centum / satem distinction cannot be applied to Thracian, this language, together with Illyrian, being in an intermediary zone. Thracian seems to have been related to Illyrian and also to Balto-Slavic and to Indo-Iranian. Even less is known about Dacian than about Thracian: The elements preserved from the Dacian language are few and often uncertain with regard to their form and meaning. They are words, proper names (place names, personal names, names of ethnic groups, names
of gods), and inscriptions (with uncertain interpretations).  A few semantically similar words, however, cannot be designated as Dacian.
With respect to the supposed Dacian names of plants, the linguist C. Váczy stated that the 35 names of plants contained in various versions of Dioscorides Pedanios and those 10 names of plants given by Pseudo-Apuleius could be considered to be of "probable" Dacian origin; but an even partial reconstruction of the language of the Dacians presents a major difficulty and would be extremely dubious.  The name of the Dacians, Δακοι, Δᾶοι, (Dakoi, Dakai) is not explained. The Greek daoi (wolf) or the Phrygian word daos (wolf) may be connected with it (dhăukos, Georgiev), but other etymologies have been proposed too. 
Similarly little is known about the origin of the ethnic name "Getae." (guet "to speak, to talk" ? , *ghend-, *ghed - "to grasp"?)  P. Kretschmer believed that the Getae were a Thracian stratum of common people with a Scythian ruling class. They were referred to in Greek texts by names of an Iranian type, such as Massagetai and Taurogetai.  Dacian personal names are extant, but their sense is unknown, and they are not helpful in the study of language.  The Dacian morphology, most important for the investigation of this language, is nearly unknown. 
With regard to the language spoken by the Getae, there is a single record by the Greek geographer Strabon of Amascia (60 B.C.-20 A.D.), who stated that "the Getae spoke the same language as the Dacians."  This assertion is, however, no longer generally accepted: "The affirmation made by Strabon (VII,3,10) that the Dacians spoke the same language as the Getae, who had the same language as the Thracians, (VII,3,13), should not be given more credence than the assertions of Italian travelers in the Romanian countries in the sixteenth century, who believed that Romanian was a dialect of Italian. "  The Bulgarian scholar D. Simenov also argued that Dacian and Getian were two different languages.  With the lack of sufficient data, this question cannot be clarified.
The existence of a "Daco-Moesian" language has been assumed by the Bulgarian epigraphist and linguist Vladimir Georgiev,  who believed that the speakers of this idiom migrated from the territories north of the Danube to the areas of the Roman provinces of Dardania, Dacia Mediterranea, and Dacia Ripensis as early as during the second millenium B.C. This population was, according to Georgiev, the ancestors of the Albanians and their language the substratum of Romanian. This migration is, of course, difficult to substantiate; and the linguistic arguments proposed by Georgiev in favor of a "Daco-Moesian" language, compared with Thracian, are vague and unconvincing. 
This does not, however apply to the idea that the language of the ancestors of the Albanians was the substratum of Romanian; this may be investigated by comparing the two languages.
Thracian and Dacian
Russu's opinion that Thracian and Dacian ("Geto-Dacian") were dialects of the same language is based on very weak evidence. Russu seems to be aware of this when he says that it is possible that this was so. The frontier between Thracian in the south and "Geto-Dacian" is also given only approximately (the lower Danube? The Haemus Mountains?)  The identity (or very close relationship) of these idioms is not generally accepted. On the basis of a comparison of what is known about the phonetical characteristics of the ancient Southeastern European languages (after V. Georgiev), Istoria limbii române (1969) considers that Illyrian, Thracian, and Dacian were separate languages: 
It must be emphasized that this table comprises only a few phonetical elements, that is, it is not exhaustive even with regard to the sounds and does not contain anything about other areas of language. Nevertheless, it seems reasonable to assume that the above-mentioned languages were distinct from one another. In the conditions of Southeastern Europe at that time, with many tribes living without any large and stable political organization in a vast and geographically heterogenous territory, it is not probable that a uniform language would have been spoken throughout any large area. Today six major languages (and other idioms with smaller communities of speakers) are spoken in that territory. The ethnic diversity of the Balkan Peninsula was certainly not less in pre-historic times than it is in the modern times. 
The Problem of the Dacian Elements in Romanian
One of the basic problems of the history of the Romanian language is that of the special idiom spoken by the ancestors of the Romanians before their Romanization. One approach to analyzing this problem is that used by the German linguist Günter Reichenkron. Departing from the belief that the pre-Latin elements of Romanian were necessarily derived from Dacian, Reichenkron constructed various theories that connected more than one hundred Romanian words of unknown origin with assumed Dacian lexical elements. Reichenkron also assumed a consonant shift from Indo-European to Dacian (bh > b, dh > d, g > k, b > p. and d > t). A typical example of these etymologies is the following scheme: 
Dacian barðo is, however, not attested to in any source and is also unnecessary; Romanian barză could as well derive from Indo-European bhərəg via prehistoric Albanian barðo. Deleting one of the assumptions in Reichenkrons scheme, it should be written as follows:
The vague etymologies of the German linguist Günter Reichenkron have been refuted by most scholars (Georgiev, Hubschmid, Russu, and Rosetti). Although Reichenkron's ability to investigate the substratum may be recognized, it is difficult to go along with him in the details; and, with regard to the hypotheses on morphology, it is not convincing. The way in which Reichenkron uses the declared principles in his etymologies, as well as the great majority of his examples, is unlikely and unacceptable, because they ignore Romanian etymological laws, in the first place, as well as the phonetic laws of Romanian and of Thraco-Dacian. [108a] Of those 130 Romanian words for which Reichenkron proposed a Dacian etymology, only four (leagăn, melc, viscol, and băiat) are accepted by Russu as "Thraco-Dacian" (not Dacian, thus not giving any indication of an origin north of the Danube). 
A Comparison of the pre-Roman Lexical Elements in Romanian with Thracian and Dacian Words
A basic question is whether those more than 100 pre-Latin words in Romanian (or any of them) appear in an ancient Southeastern European language and, if so, in which language? The possibilities of finding an answer are strongly restricted by the limited number words with a known meaning in Illyrian, Thracian, and Dacian. Moreover, the ancient lexical elements are preserved in texts written not by native speakers but mainly by Greek and Roman authors and in many cases are probably not quite correct. An analysis of the Thracian lexical data  gives seven words for which there exists a Romanian word from the substratum:
Only two of these seven Romanian substratum words show a similarity with the Thracian word mare, "big," and gard, "fence, enclosure, pilework." Romanian mare has been connected with one part of such compound personal names as Βηριμαρος Καρσιμαρος. It could derive from Indo-European *mēro-, mōro-, "big, stately"; cf., for example, Old High German *mār in names such as Volkmār, Hlodomār.  Romanian gard has been connected with names of towns: Gordion, Manegordum. The sense of these names is not known, however; and it therefore cannot be determined whether "gord" really has something to do with Romanian gard. Romanian gard could derive from Indo-European *gerdh, "to knit, to enclose"; it also exists in Albanian gardh, "fence, enclosure."
Istoria limbii române lists 35 words "for which a substratum etymology is probable or at least possible" and that have been "connected with words known in Daco-Moesian or other ancient Balkan languages."  Of these, 11 refer to Thracian words and 2 to Dacian ones; in one case, similar words are found in both.
Romanian argea: "room (made in earth)," Thracian αργιλος, "mouse"; Albanian ragal, "hut," Old Macedonian άργελλα, "id."
baltă: "swamp, marsh," Thracian toponym Di-baltum, "two marshes" (near Burgas, Bulgaria). Albanian baltë, "mud, swamp."
buză: "lip," Thracian Byzas, Byzos, Dacian Beusas; also Illyrian Buzos, Buzetius; Byzantion "town on the shore." Albanian buzë, "lip."
drum: "road," Thracian Ἀνά ζραιμος translated as "nine roads"; Greek δρόμος Albanian dhrom, "id."
mal: "shore, bank, edge." Thracian toponym Malua, Dacia Maluensis, translated to Dacia Ripensis, also Illyrian Malontum, Dimallum, cf., Alb. Dimale, "two mountains." The ancient Albanian form is reconstructed as *mol-no; modern Alb. mal, "mountain," Romanian mal, "bank, mountain."
(Northern măldac, maldac Rom. dialectal): "bunch of hay," Thracian μανζάκης "id." Modern Greek μανζακης "bundle of osier willow" is considered to be a loan from Thracian.
mînz: "foal," Thracian Mezenai, (the surname of the Thracian knight) and a reconstructed Thracian stem *mel(d)z- "to milk." Albanian (Tosc dialect) mës, mëzi, (Gheg dialect), mâz, maz, fem. mëzë.
murg: "dark," Thracian toponym Μηργίσκη and name of a tribe Μόργητες. Indo-European mər(ə)g, "to become dark," Albanian murk, murgu, "dark."
rămf rîmf, rimf, remf: (only in Northern Romanian, Transylvania) "Aristolochia clematitis," German "Osterluzei." Thracian ρομφαία, "lance." Transylvanian Saxon Rimf(ərt) "Tanacetum vulgare," German "Rainfarn."
scai: "thistle," Cirsium lanceolatum, Thracian σκάλμη, μάχαιρα Θρακία, Greek σκόλυμος;, "artichoke, Cynara scolymus." Albanian halë, "chip, sliver," hele, "lance."
şiroadă: "tub, vat," şirimpîu, "canal." Thracian σῖρος, σεῖρος "cavity in which cereals were laid down," probably of Thracian origin. Armenian širim, "cavity, tomb."
In most of these cases, the connection of the Romanian word with the Thracian one is questionable and is only one of several etymologies that have been proposed. Frequently, the meaning of the Thracian word is unknown (place names); but even if it is known and fits or may in some way be connected with the sense of the Romanian word, it is no proof of connection. Northern Romanian rămf, rîmf, remf, rimpf "Aristolochia clematitis," German Osterluzei, for example, has been connected with Thracian ρομφαία, "lance." The semantic explanation seemed obvious: the leaves of this plant are lanceshaped (Hasdeu). This Romanian word appears only in Northern Romanian, however; and even there, it is dialectal and found in Transylvania. Therefore, the etymology from Transylvanian Saxon Rimf(ərt) is by far more probable, as is recognized also by Poghirc. 
Among elements that are more likely to have derived from Thracian, one should mention the suffix -isko, which has the same value as Romanian -esc (and Albanian -ish), with the adverbial variant -eşte (Alb. -isht). In Greek the suffix does not have the same value as in Romanian; in Latin, it appears in words derived from Greek or Thracian.
With regard to Dacian, the Romanian linguist Cicerone Poghirc lists three words:
buză: "lip," Dacian, Beusas;
druete: "wood," Dacian toponym, Drobeta;
mic: "small, little," Dacian toponym, Micia (Veţel on the Mureş River), ethnic name Micenses, personal name Miccos, Miccas, Greek dialectal μίκκός.
The connection in these cases is purely conjectural, since the Dacian lexical elements are all place names whose meanings are not known. Besides these words, four appellatives (names of plants) found in the treatise of Dioscorides were compared with Romanian words:
- riborasta, "Filzkette, Maskenblume," with the variants ribobasta, peripobasta, peripomasta. This word has been connected with Northern Romanian brústur(e) (pl. brusturi), (Arumanian brostu, broştu, brustir, bruştură, bruştirusescu, broştur, brustur, brustură), "common burdock, Arctium Lappa." Albanian brushtullë "id.";
- guoleta "Lithospermum arvense," with Romanian gorun (variants gorón, gorîn, gurún), "species of oak";
- mozula, mizela, "thyme, Thymus vulgaris, savory, bean tressel Satureia hortensis," was compared with Northern Romanian mazăre (mazere, mazîne) Arumanian mádzăre, madzîre "pea, Pisum sativum." Albanian modhullë "id";
- pro-diarna "Nieswurz, Veratrum nigrum" with Romanian zîrnă "black nightshade, Solanum nigrum."
It must be emphasized again that any connection between these Romanian words and the Dacian names is hypothetical and only one of several proposed etymologies. Dicţionarul explicativ al limbii române,  considers gorun and zîrnă to be of Slavic origin (from Bulgarian and Serbo-Croatian gorun and from Slavic zrüno "grain, seed");  with regard to mazăre, it refers to Albanian modhullë and considers brusture to be of unknown etymology. In Russus opinion, "the connection of brusture with the Dacian medicinal plant riborasta is difficult to accept phonetically and is not very probable."  The identification of the Indo-European root *guel is problematic in the Dacian word guoleta; one would in Dacian expect *gul-, * gol-.  In the case of mazăre: "Thraco-Dacian mozula (mizela Dioscorides) meaning wild thyme (Thymus serpyllum), is semantically inadequate"; 
and Russu considers the etymology of Romanian mazăre unknown. At present "only the identity" of Romanian mazăre and Albanian modhullë are certain; the fourth word in this list, zîrnă, is not discussed by Russu.
After the conclusion of this analysis, it must be noted that there is not a single Romanian word that can be reliably demonstrated as originating from Dacian  or from Thracian, although some suffixes could originate from Thracian. The Romanian linguist Ovid Densusianu pointed out this fact: "Romanian phonetics or lexicology do not show any characteristic that would be found among the elements that are preserved from the Dacian language." 
In view of the fact that Northern Romanian is generally called in the Romanian historiography "Daco-Romanian," a term that suggests a Dacian origin of the Romanian language, it is important to establish the absence of any certain Dacian element in Romanian. There is, however, no proven Thracian or Illyrian element in this idiom either; and the theory that the ancestors of the Romanians were Thracians or "Thraco-Dacians" is pure hypothesis. Russu expressed this as follows: ..."the Thracian language had not (as far as we know today) any exclusive phonetic feature, existing only in our autochthonous words and entirely unknown in other Indo-European languages; the Thraco-Dacian language of the satem type had the same phonetic system as had Illyrian and had very many common elements with other languages of the satem type (Iranian, Baltic languages, Slavic). If we had, therefore, data indicating that the ethnic-social basis of the territory of Romanization in Moesia, Dacia, etc., was Illyrian, Iranian, or Balto-Slavic, one could admit the possibility of such an origin for the autochthonous Romanian words..." 
Russu asserts that the archaeological findings can never solve the problem of the substratum: such material, "in spite of its increasing abundance, will always, on the whole, remain of dubious or contestable ethnological significance."  Instead, he proposes that an analysis of the Romanian language should be used to appreciate correctly the value of the pre-Roman linguistic and ethnic elements.  But in his opinion it will be necessary to use (even if only as an hypothesis and in order to simplify the terms) the notion "Thraco-Dacian" (or "Thraco-Illyrian?") to designate the pre-Latin (autochthonous) Romanian words. 
It is obvious that problems of language, in this case the origin of the pre-Latin lexical elements of Romanian, can ultimately be solved only by linguistic methods; data and conclusions furnished by this discipline should be decisive. Neither historical sources nor the archaeological
material give any conclusive proof. Supposing that the lexical elements in Romanian were "Thraco-Dacian" (or "Thraco-Geto-Dacian, Thraco-lllyrian") implies that the ancient language was spoken in a vast territory of the Balkan Peninsula north of Greece as well as in the region north of the lower Danube. The term "Dacian" points especially to Transylvania. - As was shown above, however, there is nothing among the linguistic data that would indicate a north-Danubian origin of the Romanian language. Consequently, the term "Thraco-Dacian" is misleading and should not be used, even "as an hypothesis and in orther to simplify the terms."
The Origin of Albanian
Two main suggestions have been proposed for the ancient language that was the forerunner of present-day Albanian: Illyrian and Thracian. The opinion that Albanian is a continuation of Thracian and that the Albanians lived during the Roman period (and immediately before it) mostly in Dardania, east of contemporary Albania, was mainly defended by Alexandru Rosetti, Dimităr Dečev, Henrik Barić, Ivan Popović, and Gustav Weigand.  The Bulgarian scholar Vladimir Georgiev argues that most data indicate that Albanian developed mainly on a Thracian ethnic basis, in a region east of Albania.  Albanian scholars, especially the Albanologist Eqrem Çabej, have more recently emphasized the circumstances that support the theory of autochthoneity in Albania.  According to this view, Albanian would be the continuation of Illyrian. In Russu's opinion, the historical phonetics of the Albanian language as a satem type could be in accordance with the Illyrian as well as the Thracian [which are] both the same type, satem. 
It is quite likely that the peoples described by Greek authors as Illyrians and Thracians were a result of ethnic mixings, superpositions of heterogenous elements.  The frontier between them went through the towns Scupi (modern Skoplje) and Ulpiana; Dardania was divided between these two major areas. It was, however, not as clear-cut as a line drawn on a map would suggest. Thracian place names, for example, appear in Illyrian (for example, Thermidava, in the vicinity of Scodra [Skodër], Quimedava, in Dardania).  Several tribes are mentioned as being Illyrians in one source and Thracians in another (for example, the Triballes, the Tralles, the Istres, and the Daorses).  The people living in Dardania in ancient times were also described differently by different scholars: Strabon and Appianos of Alexandria considered them to be an Illyrian tribe, while Polybios distinguished them from the Illyrians. 
It is therefore not surprising that at least the possibility of both Illyrian and Thracian elements have been demonstrated in the Albanian
language. The etymologies are not, however, completely reliable for either Thracian or Illyrian: Everything is based on linguistic or ethnologic material, which frequently is doubtful.  The Bulgarian scholar Vladimir Georgiev defends the theory that Albanian derives from "Daco-Moesian" once spoken in Dardania and in parts of Macedonia and Moesia Superior. He argues that the sound changes from Indo-European to modern Albanian (for example Indo-European e > Albanian je; Ind.-Eur. ā > Alb. o) can also be demonstrated in "Daco-Moesian."  The linguistic data on which this hypothesis is based is scanty, and the examples are taken from the heterogenous and vast literature of the ancient era. The spelling of the ancient non-Greek and non-Latin names is often dubious. These etymologies are thus uncertain.
The lack of any decisive connection between Albanian and the remaining fragments of Thracian and Illyrian could be partly explained by the scarcity of documentary material; such connections may or may not exist. Albanian could, however, have derived from an ancient language (for example, a dialect of Illyrian or Thracian, or an idiom quite different from both) that simply remained unnoticed by the Greek and Roman scholars. This is, in fact, plausible with regard to the social situation of the population: they were a peasant population, predominantly pastoralists practicing transhumance. Living in the region of the high mountains, they and their small villages were scarcely a center of interest for Greek and Roman authors.
The problem of the territory in which the ancestors of the Albanians were living before and during the Roman period on the Balkan Peninsula has also been intensely debated. Albanians on the territory of present-day Albania were first mentioned in the twelfth century. Therefore, only an analysis of place names and of the Albanian language itself can give some indication of the origins. Attempts have been made to find place names of Albanian origin or showing a sound pattern characteristic of Albanian in Albania and in other areas of the Balkan Peninsula. Frequently, the difficulties are great and the problem cannot be settled. The name of the Serbian town Niš (Nish), for example, could have developed from Latin Naissus, with a typically Albanian hypheresis (especially usual in Latin loans: cf. Albanian pyll "forest," from *pëyll, Popular Latin pa[d]ule) and the s > sh change, similarly regular in Albanian. The German scholar Gottfried Schramm argues on the other hand  that the i in this name is a strong indication against this explanation (and the š may be an ancient phenomenon, common in place names and river names in Southeastern Europe).
Most of the place names in Albania are of Slavic origin, and most of the names known from ancient times do not show an Albanian sound pattern. According to Ivan Popović, in clear cases the Slavic mediation is certain and is not impossible in any case, while the same cannot be said about Albanian sound patterns.  The name of the town Durrës, for example, must be a loan from Slavic, since Latin ki in Dyrrachium would have resulted in Albanian *q, while in Slavic, this sound regularly developed to č. When such words were transferred to Albanian, this č changed, to modern Albanian s. There are, however, many geographical names in Albanian for which an Albanian origin or mediation seems probable: Lesh, in northern Albania, from Latin Lissus shows an Albanian evolution (cf. Latin spissus "dense" > Alb. shpesh) or the mountain name Shar, along and east of the northeastern frontier of Albania (Săr Pianina), from Shkárdos, Shkardon, with the sound change sk > sh. The names of some rivers in Albania can be of Albanian origin, such as Bunë (earlier Buenë), which could be compared to Albanian buenë, bujenë, boenë "inundation." 
Most of the place names with an Albanian sound pattern, however, seem to be found east of present-day Albania, in the ancient province of Dardania and in adjacent territories. The name of this province has been connected with the Albanian appellative dardhë "pear tree," and dardh-an "peasant," originally "producer of pears."  There are the names of the town Štip, from Astibos, with the typically Albanian elision of a- and s > sh; Štiponje, from Stiponion, also s > sh; of the mountains of Šar Pianina in the northwest (partly in Albania); Oxrid from Αυχνίς, with n > r after a velar consonant.
The Albanian language also gives some indication about the ancient regions of the Albanians: the oldest stratum of Greek loan-words in Albanian, for example, shows an ancient Greek sound pattern: Albanian (the Gheg [northern] dialect): mokën(ë), (Toské); mokërë "millstone," from Greek μήχανη with k for Greek x (which later, in the Byzantine era, changed to h); or bretëkë "frog," from Greek βρώταχος, also k and not h for Greek x-. 
The Austrian scholar Norbert Jokl summarized the criteria that determine the areas of the Albanians in the pre-Roman and Roman era on the Balkans as follows: 1. an area of contacts between Illyrian and Thracian; 2. an area that came under Roman influence relatively early; 3. an area in the vicinity of the ancient areas of the Romanians; and 4. an area under some Greek influence.  Since the third area refers to the very problem discussed here, it should not be considered; but areas one, two, and four can apply to the ancient province of Dardania.
The conclusion about the ancient territories of the Albanians can be recapitulated as follows: the prevalently pastoral population, whose ancient Balkan language is continued by modern Albanian, lived immediately before and during the Roman domination on the Balkan Peninsula east of present-day Albania, in Dardania, in parts of Macedonia, and probably also in parts of present-day Albania, although this last mentioned area is contested by several scholars (Ivan Popović, Gottfried Schramm, not to mention earlier writers on the subject who considered that only the area east of contemporary Albania that comes into question, in modern geographical terms, is southern Serbia and northern Macedonia).  For the problem of Romanian, the question of the autochthoneity of the Albanians in their present-day country is not very important. It is essential, however, that the central areas of the ancient Albanians were in Dardania, that is, in the region of the Vardar River; and the limits of their areas in that time, while not known exactly, cannot have been very far from the frontiers of that province. Nothing suggests, for example, that the ancient Albanians were also living in northern Serbia and northern Bulgaria. Ancient Albanian was thus a language of the central areas of the Balkan Peninsula, about which designations such as an idiom of the "Carpatho-Danubian region" cannot be applied.
The Relationship of the Substratum of Romanian to Albanian
Albanian, next to Greek, is the oldest language spoken today in the Balkans. It belongs to the Indo-European group but is not genetically related to any of the modern languages spoken in Southeastern Europe. The majority of the ancient, pre-Latin elements of Romanian, however, have their counterparts in Albanian. Evidently, these two languages have had a common substratum. Romanian and Albanian are more closely related than either of them is with Dalmatian.
Romanian-Albanian concordances are found in all areas of language: phonology and syntax, as well as lexical elements and phraseology.
- Phonetics and Phonology
Romanian contains a vowel ă, and Albanian has this vowel too (written ë). In both languages, it developed, along with other sounds, from unstressed Latin a, from a in front of a nasal, and in certain conditions from any other Latin vowel. The opposition a : ë is found in the category of determination (Albanian vajzë "[a] girl," vajza "the girl," Romanian fată "[a] girl," fata "the girl"), and in gender. The phonemization of ă developed under almost identical conditions in Romanian and Albanian.  In a later development, the Albanian ë underwent more change than the corresponding Romanian vowel. This is one of the circumstances supporting the idea that modern Albanian is a continuation of the language that possessed the vowel ë while the ancestors of the Romanians abandoned this language (becoming totally Romanized), preserving only some elements of it, such as the vowel ă . In its new surroundings, this vowel could be considered a borrowed sound. 
Latin -ct- shows similar changes in Romanian and Albanian. In the Romance languages, this consonant group evolved toward an open syllable: Latin lacte-, Romanian lapte, Italian latte, French lait. Albanian and Romanian reached the first stage of this development: Latin lucta > Albanian luftë, Romanian luftă (Arumanian), luptă (Northern Romanian). Latin -lv-, -rv- developed into Albanian and Romanian -lb-, -rb- : Latin servire > Albanian shërbenj, Romanian şerb.
These phonetical characteristics of Albanian and Romanian originate most probably from an ancient pre-Latin language, which, through almost total Romanization, developed into Romanian and through partial Romanization resulted in Albanian. Compared with the common elements in the fields of syntax, vocabulary, and phraseology, these phonetical concordances would seem less impressive. There is, nevertheless, a basic phonetical similarity between these languages, which they share with the other typical Balkan languages: The basis of articulation is very similar or almost identical; a Bulgarian, for example, easily learns Romanian, Albanian, or Greek and can speak these languages almost without any foreign accent. 
- Morphology and Syntax
In both Albanian and Romanian, the definite article is enclitic. They also have a proclitic article. How the postpositional article developed and in what way it evolved has not been definitively established. One explanation is that the article of the noun evolved
from a proclitic article of the adjective. Another opinion is based mainly on the fact that the postpositional article is stable and always has the same function, in contrast to the proclitic article. According to E. Çabej, this suggests that the postpositional article existed first and the proclitic article developed from that.  The postpositional article most probably existed in Proto-Albanian. 
The close connections between Albanian and Romanian can clearly be seen from an analysis of the use of the article: these two languages coincide in the use of this element of speech to the smallest details of its syntactical position, which contradicts the assumption of a spontaneous evolution in each of the two languages.  Many circumstances indicate that the Romanian constructions, which are to such a great extent similar to the Albanian ones, cannot be explained without them. 
It is significant that the postpositional article also appears in Bulgarian (and the Macedonian dialect) but not in Serbo-Croatian: It is a Balkan phenomenon, which did not reach the Serbo-Croatian language.  The particle -ne of the accusative of the personal pronoun (Romanian mine, tine "me," "you") appears in both Albanian and Romanian and is considered to have derived from Proto-Albanian.  Of 13 Romanian suffixes that are probably of pre-Latin origin, 6 are also found in Albanian (for example, the relevant suffix -esc, -eşte).
The pre-Latin Lexical Elements in the Romanian Language
A difficult problem is to establish which of the several thousand Romanian words with an unknown etymology originate from the pre-Latin substratum of the language. Definite conclusions do not appear to be possible at present, as evidenced by the large number of proposed etymologies and the widely divergent opinions among scholars who have studied the problem. Some investigators consider that not a single word could be attributed with certainty to the substratum; others believe that all Romanian words of unknown or uncertain etymology derive from it (10 to 15 percent of the entire Romanian word stock). Between these extreme views, the proposed figures vary: about 85 [words], according to Bogdan Petriceicu Hasdeu and Alexandru Rosetti; about 160, according to Ion I. Russu (72 also existing in Albanian, 90 only in Romanian); 30, according to Vladimir Georgiev.  In reference to Ariton Vraciu's statement, "the etymologies proposed until now for the elements attributed to the substratum are mostly inadequate." 
One of the basic questions in this context is the proportion of words from the ethnolinguistic substratum that Romanian has in
common with Albanian. It must be stated from the beginning that these common words are not loans from one of the languages to the other but chiefly derive from a common source, an ancient language once spoken in Southeastern Europe. As regards those of the assumed substratum words that do not exist in Albanian, I.I. Russu stated: "these do not make out a distinct group, deriving from another source than those which Romanian shares with Albanian, with which they in fact constitute a single pre-Roman, autochthonous lexical block."  This is very dubious indeed, because there are significant differences between these two groups from the semantic viewpoint, as will be shown below (see Table II, p. 240 and pp. 241-245).
Given the long period of time of separate development, one should, of course, not expect that all substratum words of Rumanian are also found in Albanian.
Another problem is the distribution of substratum words among the Romanian dialects. About 50 of those assumed by Russu to have originated from the substratum are found only in Northern Romanian, which has been thought to indicate "the greater resistence of the autochthonous population to Romanization north of the Danube" (compared with the Balkan Peninsula).
First of all, one may examine the present-day (or, in several cases, Northern Romanian from the seventeenth century) form of the Romanian words. On the other end of the time scale, there are ancient Indo-European stems that have been established by reconstruction from modern Indo-European languages, from Sanskrit, Old Iranian, Balto-Slavic, Old Germanic, Celtic, Latin, and Greek, languages.  Connections between the Romanian word and the reconstructed Indo-European stems are then decided on the basis of formal and semantic criteria. This means that there are five to six millenia between the reconstructed ancient Indo-European stems and the present-day Romanian words, an extremely long period of time, which only contributes to the uncertainty and precarious nature of the procedure.  The existence of words preserved from Thracian or other ancient Balkan languages could shorten this long period significantly; but, as was shown above, not a single Romanian word can reliably be connected with any such lexical element. Russu also admits that this method is uncertain;  and any attempt to explain the origin of the Daco-Getae, Thracian, Illyrian, (and others) or to search for connection between them and Indo-European is more or less hypothetical. 
An example of this method will be given here (the proposed derivation of the Romanian verb răbda "to endure, to bear, to tolerate"): (This word) belongs to Indo-European *orbho-, robh- "lacking something,
worn out, tortured, miserable," from which derived the Latin orbus "without sight," Greek όρφανός "without parents," German Arbeit "work," Slavic robu (robie); Romanian răbda is part of this etymological group, without the possibility of being a Latin, Greek, Slavic, or German word, originating from a language in which bh changed to b, o possibly to a; its morphological structure (the derivation with -d-) is specific and ancient.  Russu gives a thorough analysis of the words assumed to originate from the substratum of Romanian, including all of the more significant etymologies proposed in the literature and a critical discussion of each problem.  The conclusion with regard to the pre-Latin origin of the word in question, may be correct in several cases but in others, not.
Scholars are in general skeptical about the etymologies proposed by Russu. Cicerone Poghirc, using the criteria for considering a word of substratum origin, does not agree with Russu, giving only 125 Romanian words for which the origin from the substratum is probable or at least possible, of which about 100 are also found in Albanian. These words are divided into three groups: 1. Romanian words that have been connected with lexical elements from Daco-Moesian or other ancient Balkan languages; 2. words that Romanian shares with Albanian; and 3. substratum words deduced from a comparison with other Indo-European languages.  It is, however, questionable whether the words in this third group really come from the substratum - it seems more probable that most of them have a different origin. Istoria limbii române, 1978, p. 72, mentions 19 words which according to Russu, originate from the substratum - (băga, băiat, brînză, burtă, butuc, cîrlan, creţ,, doină, genună, melc, mişca, morman, muşca, niţel, prunc, răbda, şoric, ţăruş, zer) - but of which Dicţionarul Explicativ al Limbii Române (DEX, 1975) states: "of unknown etymology".  Also Istoria limbii române, 1978, states that some of the words given by Russu in Elemente autohtone ... as originating from the substratum could be explained from Latin.  In Vraciu's opinion, the vocabulary statistics of Russu must be reconsidered, many words are attributed arbitrarily to the pre-Roman substratum of Romanian.  (See list, p. 231-238.)
- The Semantic Aspect
Discussing the semantic aspect of these words, (see Tables I and II) according to Russu the specifically Romanian nature of the autochthonous words also appears from an observation of the groups of terms existing in both Romanian and Albanian, of which many denote plants, animals, natural phenomena; half of them refer to the human body, one in five refers to the age of man (moş), two in five denote psychological states (bucur-, mărat).  Those groups that are not found in Albanian are words denoting psychological states, diseases and features of the body, and many tools. [[ Continuation on page 241 ]]
A List of Northern Romanian and Arumanian Words Probably or Possibly Originating from the Substratum (On the basis of ILR, 1969, vol. II and Russu, 1981)
The following verbs are variants of verbs mentioned above or derive from nouns mentioned above:
descăţa - cf. acăţa; descurca - cf. încurca; desghina - cf. înghina;
deşela - cf. şale; îndopa - cf. dop; înşira - cf. şir; rezema - cf. razem
The semantic distribution of Romanian lexical elements probably or possibly originating from the substratum
Source: Ion I. Russu, Etnogeneza românilor, 1981, pp. 244-245 and C. Poghirc, in ILR, vol. I, 1969, pp. 327-356 [the list of words considered autochthonous]. The semantic categories are those used by Russu, to which "abstract notions" and "popular mythology" were added.
*. One of the Alb. words (kopil') may be a loan from Romanian.
**. Two of the Albanian words (batsë and urdhë) may be loans from Romanian.
***. Three of these words (brînză, zară and zăr) may have counterparts in Albanian.
The semantic distribution of a total of 198 Romanian words probably or possibly originating from the substratum on the basis of Table I in which the absolute figures are given
The difference from the semantic viewpoint between words that also exist in Albanian and those that are not known in that language is shown. This is a word stock of the most elementary notions of human life, notions connected with the primitive dwelling place etc. The two groups compared contain largely similar numbers of this kind of words. To this comes, however, a group of words of special interest for a pastoral population: in the first place phenomena met in everyday life by shepherds in high mountains. The number of these words is much higher in the group of words that also exist in Albanian. In contrast, the group of words that do not exist in Albanian contains many more verbs without a specific significance. The differences are too large to be coincidental.
If the terms of general significance for shepherds are common (baciu? ţarc, strungă, bască, as well as the names of the domestic animals), the terms for milk products are exclusively Romanian: străghiată, mărcat, brînză, zăr, zară, urdă.  Russu also notes that only a small number of the more than 40 verbs assumed by him to originate from the substratum also exist in Albanian. Thus, also Russu noted the striking difference from the semantic point of view between those lexical elements that Romanian shares with Albanian and those which only exist in Romanian. He did not, however, draw any conclusion from this fact, but expressed the view that "it is perhaps accidental" (Etnogeneza românilor, 1981, p. 132). He continued this sentence as follows: ..."the semantic grouping and the meaning of these words do not plead at ail for the thesis of a more close linguistic psychological, and socio-ethnic community of the Romanians with the Skipetars, for the pretended 'Albanian-Romanian symbiosis' during the Middle Ages". 
A list of the semantic distribution of the substratum words in Romanian can only be approximate. Many etymologies are uncertain, and no one is able to say exactly which words really originate from the substratum. The lexical elements also found in Albanian are relatively easy to classify; in most cases, the sound pattern of the word suggests its origin from a common, ancient language. There are, however, a few cases in which a borrowing can be discussed; baciu, according to Russu, Albanian batsë "shepherd in charge of a sheepfold," could either be inherited or "a loan from Romanian";  copil "child" corresponding to Albanian kopil', kopil "young boy, servant" (the Tosc dialect of the Albanian), bastard (the Gheg dialect of the Albanian) could be similarly a loan from Romanian or inherited;  mistreţ "wild boar," Albanian mistrets; the connection (of Romanian mistreţ) with Albanian mistrets is not at all clear;  Albanian urdhë "soft cow cheese" may be borrowed from Romanian urdă, according to C. Poghirc.  Words not existing in Albanian could be assumed to derive from a certain, reconstructed Indo-European stem, but it may only be a question of probability. In the following list the words proposed by Russu as well as many of those given by Poghirc are presented on the basis of Istoria limbii române (1969). From this last source, several words were omitted, mostly those that are not presented in Russu's list, especially in cases in which the connection with the proposed Proto-Indo-European seems poorly documented. The total number of words arrived at this way is really high, but it should be noted that these words have been assumed to derive from the substratum of Romanian by competent scholars; the probability of the derivation varies.
Another problem is the semantic grouping. While baciu "shepherd in charge of a sheepfold" is a typical shepherd word, ţap "male goat"
belongs to the group of animals, although it denotes an animal of great importance to shepherds. The word zgardă "dog collar" could be used by a peasant farmer population as well as by a pastoral population; căciulă "fur cap" is an item of clothing but one typically used by shepherds. The names of animals that appear here denote mainly animals existing in a pastoral society or wild animals. In his listing of the semantic groups, Russu classifies all animal names in a special group, but in the text (p. 132) he considers them, together with such words as baciu "shepherd in charge of a sheepfold," strungă "sheepfold," ţarc "fold, pen" as of general significance to shepherds. The semantic groups used by Russu  will be used here with slight modifications.
A Comparison of Two Groups of Words: Those with and Those Without an Albanian Counterpart
Of a total of 209 (4-11?)  lexical elements possibly originating from the substratum, 113 have also an Albanian equivalent while 96 ( + 11?) exist only in Romanian (half of them only in the Northern Romanian dialect). There is a clear difference with regard to the semantic areas presented by these two groups: of those 113 words also existing in Albanian, 22 (19.5%) are specific shepherd words and another 36 denote things and notions of special importance to a pastoral population: names of animals, plants, and natural phenomena, with all of which shepherds usually have everyday contact. The total number of lexical elements of special importance to the pastoral way of life is 58 (51.3% of the entire group). There are only 9 verbs (8%) of importance to all human beings and societies, without being specific to any special type of society.
On the other hand, the group of 96 (+11) words not found in Albanian contains only 8 specifically pastoral words  and another 15 ( + 3?) of special importance to pastoral way of life, making up a total of 23 ( + 3?), corresponding to 24% (27%?). The number of verbs without special significance is much higher in this group: 37 ( + 4?), corresponding to 38.5% (38.3%?) and the largest semantic group among these words. The proportion of words of special importance to pastoral life among words also existing in Albanian is thus more than twice that in the other group which, on the other hand, contains many more verbs of general significance (38% compared with 8%).
- The Lexical Elements of importance for a Pastoral Population
Of the 80 ( + 4?) lexical elements assumed to originate from the substratum of Romanian that are of primary importance to a pastoral
population (specific shepherd words, names of animals, plants, and natural phenomena), the majority (at least 58 [72%]) also exist in Albanian. In contrast, verbs of no specific significance appear only in a small number in Albanian (10 out of 47 [ + 4?], that is, 20%).
As mentioned above, no exact figures can be given about the semantic distribution of the substratum words. The differences discussed here between the two groups (Albanian and non-Albanian), however, are also recognized by Russu (see above). - Russu is also right when he states that these words "do not suggest an Albanian-Romanian symbiosis during the Middle Ages," since most of them are not loan-words.
- A Comparison of Two Groups of Words: Those Found in Northern Romanian Only and Those Also Existing in at Least One of the Southern Dialects
According to a list, Russu assumes a total of 169 Romanian words to be of substratum origin, of which he considers 12 as questionable, listing them with a question mark.  In the following analysis, the numbers of words considered by Russu as certain will be given first and the questionable words thereafter in parentheses. Of a total of 169 words in this list, three are not found in Northern Romanian (one exists only in Meglenitic and two only in Arumanian). Since the problem discussed here is a possible difference between Northern Romanian and the southern Romanian dialects, those not found in Northern Romanian are not taken into account. There are thus 166 words (169 minus 3) or 154, plus 12 with a question mark.
- The Presence of the Words in Albanian
Of 107 ( + 1?) lexical elements found in Northern Romanian and in at least one southern dialect, 62, or 58%, also exist in Albanian. Of the 48 (+11?) words that exist only in the Northern Romanian, only 16, or 33.3% (27.1%?), are also found in Albanian.
- The Presence of Words Found in Albanian and Also in the Southern Dialects
Of the words that are also found in Albanian (a total of 78), one finds that the great majority (62 or 79.5%) also exist in at least one of the southern dialects, only 20.5% being restricted to Northern Romanian.
- The Semantic Spheres
There are 32 (+11?) Romanian words that are assumed to have originated from the substratum but exist exclusively in Northern
Romanian (also lacking in Albanian). The largest semantic group among these words is that of verbs of general significance to all human beings, which make up about a third of the group: 11 ( + 4?). There are only three specific shepherd words (one of which is zară, possibly connected with Albanian) and 10 ( + 2?) that are of special significance to a pastoral way of life. The rest of the lexical elements are spread throughout all semantic spheres.
The etymologies of these lexical elements remain obscure. Russu collected all the more significant suggestions from the literature and gives a critical discussion of each.  It appears clear from this presentation that most of the proposed etymologies are uncertain, even unlikely; and not a single one could be considered correct beyond doubt. Russu tries to connect these words with Indo-European stems: in some cases, the connection seems plausible; in others, there are difficulties with the sound pattern and/or with the meaning. Many of these words could have an Indo-European origin, but we have no criteria whatsoever for deciding which language or languages were the direct source. The possibility of non-Indo-European words,  as well as of borrowings from such languages as Gypsy, for example, should also be considered. The distribution of the etymologies of these 32 words given by Dicţionarul explicativ al limbii române, 1975, is as follows: from Latin, 9; from Albanian, 3; from Slavic, 2; from Hungarian, 1; unknown, 16; and 1 of the words is not mentioned.
There must, of course, also be some difference among Albanian, Northern Romanian, and the different southern dialects with regard to this group of words, as there are differences in every area of the word stock. It would be illogical to believe that all these idioms have preserved exactly the same lexical elements from the substratum that they once shared. For this reason and with the uncertain origin of these few lexical elements assumed by Russu to derive from the substratum of Romanian (an assumption not accepted by several Romanian scholars), no conclusion about the Romanian language can be drawn from this group of lexical elements.
As mentioned previously, there is a correlation between words that 1. denote things and notions of specific importance to pastoral population; 2. also exist in Albanian; and 3. are found both in Northern Romanian and in at least one of the southern dialects. The other correlation found for the assumed substratum words of Romanian is that between verbs 1. of a general significance to all human beings and societies, without being connected to any specific activity or way of life; 2. of which only a few exist also in Albanian, and 3. which are mainly found in the Northern Romanian dialect. (All words for which the substratum origin is considered uncertain even by Russu
belong to those not found in Albanian [a total of 12], and 11 of them are found only in Northern Romanian).
It is unlikely that the relationship discussed here would be accidental, since the differences, especially between the words of significance to shepherds (most of them existing also in Albanian) and the verbs of unspecific character (only a small number of them being found in Albanian), are too large to be coincidental. A likely explanation of the findings is that at least a majority of the words whose group shows the first correlation really originate from the substratum, while a significant number of those that show the second correlation are in reality not substratum words but derive from some other sources. (See Table II)
Theories Based on the Distribution of the Assumed Substratum Words Among the Romanian Dialects
The Romanian scholar Theodor Capidan  advanced a theory to explain why the number of words assumed to derive from the substratum is much higher in Northern Romanian than in the southern dialects (so far as these have been investigated). Capidan assumed that the Northern Romanians had preserved more words from the substratum, because they were living in closer contact with the Albanians than were the speakers of the other dialects. This explanation implies, however, that the words in question are loans from Albanian, which is not tenable. Another, still current, theory is that the pre-Latin language was more vigorous north of the Danube and resisted Romanization with more success. This view was recently expressed by Russu in Etnogeneza românilor, 1981.  This could be explained, in Russu's opinion, by the circumstances that Dacia Traiana was conquered a century later and abandoned much earlier than Moesia; the "Geto-Dacian element" north of the Danube was more conservative and resisted Romanization for a longer period; and the position of Carpathian Dacians was more peripheral than that of the Romans living in the Balkan Peninsula, having more intense and more prolonged contacts with the Thraco-Dacian groups outside the frontiers of Dacia.
It must be noted that Russu only mentions this as a possibility, not as a concrete theory. The difference in this respect between Northern Romanian and the southern dialects is not established, since no one knows with any certainty which words derive from the substratum and because the word stock of the southern dialects is not very well known. Since, however, this view sometimes appears in writings about Romanian ethnogenesis, it may be conducive to discuss it briefly.
Substratum lexical elements assumed by Russu found only in Northern Romanian
The real question is the presence of these words in Common Romanian, that is, the idiom spoken by the Romanians before the development of the dialects. With the lack of any written document, Common Romanian must be reconstructed from the present-day Romanian dialects. In general, any ancient element in these dialects could be considered to have existed in Common Romanian; and this is most probably also true of the lexical elements. According to Russu's view, it could assumed that most (if not all) of the autochthonous words existed in the Carpatho-Balkan idiom, having been lost during the Middle Ages from the Arumanian, Meglenitic, and Istro-Rumanian dialects.  Citing Istoria limbii române, it must be assumed that the words that are lacking in Arumanian or in the other dialects have disappeared, and what is more interesting, that some of them were replaced by Albanian (or by Greek, in the case of Arumanian) elements. 
Assuming that all these words once existed in Common Romanian, the whole problem is solved and the theory about a "more vigorous substratum in the north" becomes unnecessary. The number of these words, about 50, is, however, perhaps too high to be explained in this way, even adding the possibility that some of the words exist in reality in one or another of the southern dialects but are unknown, not having been recorded by the investigators. It is this reasoning that prompted the elaboration of new theories about this question.
Considering the facts given above, it is likely that many of the discussed words do not derive from the substratum of Romanian: the majority of these lexical elements belong to those that do not exist in Albanian and denote notions of general significance for all human beings and societies (in contrast to those words which mainly denote things of special importance for a pastoral population and also exist in Albanian).
It is true that Trajan's Dacia was under Roman rule for a much shorter time (about 169 years) than were the provinces south of the Danube, (about 600 years) and that Dacia was a peripheral province. The great difference in history must have had its consequences on Romance languages developing in the two areas (if such languages existed). During the period of Late Latin, when numerous innovations and popular constructions came into use, the area of former Roman Dacia was divided from the Latin-speaking population in the Roman Empire by the Roman frontier. Even without "more conservatism and more prolonged resistance against Romanization" shown by the autochthonous population, this situation would have led to a Romance language substantially different from one that could develop on the Balkan Peninsula during the period of Late Latin. This difference
would not be restricted to the substratum words only but would also affect substratum elements in other areas of language and, most significantly, also the Latin elements of Northern Romanian. Such an effect is, however, not discernible.
The Romanian words originating from the substratum that do not exist in Albanian are probably much fewer than are assumed by Russu (about 90). The same can also be said of the substratum words found only in Northern Romanian and lacking in Arumanian, Meglenitic, and Istro-Rumanian. Any theory assuming a stronger substratum north of the Danube is unacceptable of this fact, not to mention the other difficulties.
Most of those Romanian words that are most likely to have derived from the substratum exist also in Albanian and in several Romanian dialects. The major semantic areas in this group of ancient lexical elements are those of special importance to a pastoral population. The pre-Latin linguistic material, existing in all areas of language, is of basic significance to the history of the Romanian language. It is shared to a great extent by Albanian, which suggests that the pre-Roman substratum of Romanian was the language spoken by the ancestors of the Albanians or an idiom very closely related to their language. Albanian is thus the direct continuation of this ancient idiom. The large proportion of words of primary importance to a pastoral population among the lexical elements from the substratum indicates that the people in question practiced a pastoral or transhumant way of life. During the centuries of Roman domination in the Balkan Peninsula, the population speaking this idiom was exposed to a prolonged Latin influence. Part of them borrowed a number of Latin elements but largely preserved the original language. Another part was entirely Romanized, preserving only several ancient elements (the most significant examples of which were given above). This last mentioned part of the ancient pastoral population developed into the Romanians. 
It has not been determined whether the ancient, pre-Roman language was similar to Illyrian, Thracian, a mixture of these two, or some other idiom; it has also been called Daco-Moesian. This much debated problem is not, however, decisive for the question of the substratum of Romanian, since this must have been Proto-Albanian. The territories of the ancient Albanians are known: they were living in southern Serbia, northern Macedonia, and some adjacent areas. In those same areas the ancestors of the Romanians also once lived. Pastoral terminology
demonstrates that the Romanization of the Albanians antedates that of the Romanians.
Albanian-Romanian Contacts in the Late Latin Period
Besides the substratum elements, there are also relevant correspondences between Albanian and Romanian with regard to the Latin elements.  There are, of course, also differences, mostly from an archaic period. Albanian has preserved numerous archaic Latin elements, borrowed during the first one or two centuries B.C., most of which have no counterpart in Romanian. This fact could shed light on the ancient territories of the Albanians through an analysis of place names and the Albanian language. The ancestors of the Romanians may also have used these Latin elements but lost them later in the course of total Romanization; but it is just as possible that their situation was somewhat different from that of the Albanians at that time.
Most of the Latin influence on Albanian, however, was exerted later, during the first centuries A.D. These elements show conspicuous similarities to corresponding elements in Romanian: Latin words only found in these languages, parallel changes of meaning, and so forth. These will be discussed below ("The Correspondences Between the Latin Elements of Albanian and Romanian.")
Common Romanian. The Early Slavic Influence
The elements in common with Albanian (the substratum elements and the similarities in the Latin influence) are, with some variation, extant in all Romanian dialects. They can be explained by the existence of the ancient Romanians in southern Serbia and northern Macedonia; nothing suggests that this population also lived north of the Danube. Some other phenomena in the Romanian language also attest to this.
The characteristics of Common Romanian or Ancient Romanian (română comună, străromănă) were established by an analysis of the present-day dialects: Northern Romanian with Istro-Rumanian, and Arumanian with Meglenitic.  This common language existed through the tenth century A.D. This fact is not compatible with the idea that Romanian is the descendant of the entire Latin-speaking population that once lived on the Balkan Peninsula. That population was exposed to the Slavic invasion and conquest, which divided its formerly large areas into smaller territories during the seventh century at the latest. The language of the Vlachs was not affected by this but continued its own development for another three or four centuries without any significant dialectal differentiation. The only effect that the Slavic
occupation of the Balkan Peninsula had on this idiom was a relatively weak Slavic influence. Living mainly as a pastoral population in the mountains of Serbia and Macedonia, this population also avoided assimilation to the Slavs, in contrast to the rest of the Romanized populations with the exception of those on the coast of Dalmatia. The habitat of the pastoral population was relatively isolated, the farmer peasant Slavs having preferred, for a long time, the lower mountains, the valleys, and the plains. This situation also made it possible for the Albanians to preserve their own language and identity.
A characteristic feature of all Romanian dialects is the similarity of the oldest Slavic influences. This would hardly have been possible if the ancestors of the Romanians had lived not only in Serbia and Macedonia but also north of the Danube, in Carpathian Dacia. All Romanian dialects contain a group of the same 70-odd words of Slavic origin, of a sound pattern characteristic of Slavic in the period before the tenth century. These words obviously existed in Common Romanian.  These lexical elements are not only found in all dialects; they also show some quite specific features, that is, changes compared with Slavic: there are, for example, several words formed from the Latin prefix in- and a Slavic loan-word such as Latin in + Slavic vr̥těti "to turn, to twist, to wind" which resulted in a Common Romanian word from which there are today the Arumanian anvîrtésc "to whirl, to spin; to dance; to wield"; or Latin in + Slavic plesti "to plait, to braid": Arumanian mpletesc, Meglenit amplites and Norhern Romanian împletesc "knit, weave." Changes of meaning also occurred when Slavic words were transferred to Common Romanian: Slavic koža "skin" resulted in Arumanian and Northern Romanian coajă "shell, crust, rind," Slavic loviti "to hunt, to seize" was borrowed by Common Romanian in the sense "strike" (Northern Romanian a lovi "to strike, to beat; to hit; to attack; to hurt"). 
These loans must originate from a special contact in a certain period and in a specific territory, indicating that the speakers of Common Romanian belonged to a homogenous community of people. It is most unlikely that their territory comprised, in addition to the central parts of the Balkan Peninsula, the plains north of the lower Danube and the territory within and east of the Carpathian Mountains.
The Development of the Dialects
It may be suggested that the uniformity of this language until the 10th century can be explained by the pastoral way of life of its speakers.  This may partly be true, but since dialects did develop later, it is probable that a more important reason was that they were living in a quite small
area. After the tenth century, dialects developed in a territory not larger than Serbia (the distance between the Aromanians and the southern part of the area with place names of Northern Romanian origin still existing in the Slavic toponymy is less than 150 km.) What conclusions may be drawn from the unitary character of the Romanian language (even today, there are only dialectal differences between the different areas) seen against the history of the region in question?
If Northern Romanian would be the continuation of Latin spoken in Dacia Traiana, it would have developed largely independently from Aromanian and Meglenoromanian on the Balkan peninsula from the end of the third century A.D. onwards. Common Romanian could not have been unitary in these circumstances, and in more than 1700 years without everyday contacts between the speakers of Northern Romanian and Aromanian, the two idioms would have diverged toward mutual unintelligibility, as shown, for example, by the case of Italian and French.
The dialectal differentiation of Common Romanian began in the tenth to eleventh century, following the emigration of the Romanians to the north, and resulted in two main dialects, each with one subdialect: Arumanian with Meglenitic and Northern Romanian with Istro-Romanian. The fact that Arumanian has no Hungarian loan-words indicates that the separation must have occurred prior to the tenth century.  Obviously, the cause of the diverging development must have been that contact between different groups of speakers was lost or weakened. Under the circumstances of the tenth through twelfth centuries this must have been caused by migration of groups of speakers in different directions. In this way, a southern area of Arumanian was created in the region today divided by the Yugoslav-Greek frontier and a northern one of Northern Romanian in Serbia, Montenegro, and eastern Bulgaria, where place names with the Northern Romanian sound pattern still exist in the Slavic toponymy. The Vlachs were first recorded in these areas in a report in 976 A.D., which mentioned them as being between Lake Prispa and Kastoria in northern Greece. At the beginning of the eleventh century, Byzantine sources recorded Vlachs "throughout Bulgaria." The Byzantine chronicler Kekaumenos described the Vlachs as living in Greece in the second half of the eleventh century.
The Northern Romanian dialect shows a very strong South Slavic (Bulgarian) influence from the sound pattern of Bulgarian in the eleventh through thirteenth centuries. The Bulgarian lexical elements transferred to Northern Romanian in that period are often terms dealing with social and state organization as well as with religion and Church hierarchy. In Arumanian the Slavic influence from that period is much weaker. Consequently, after the disruption of contacts
between the speakers of Arumanian and Northern Romanian, the former continued what was essentially their earlier way of life, relatively isolated from the Slavs, while the Northern Romanians changed their social situation to a considerable degree, taking part in the social life of the Bulgarian population. With this stage, the reconstruction of the main historical events can be corroborated by more detailed historical records; these tell us that strong groups of Vlachs living in Bulgaria took part in the organization of the Second Bulgarian Empire in 1186 A.D.
3. THE ROMANIAN LINGUISTIC LITERATURE
Contemporary Romanian historiographers consider the theory of the Romanian ethnogenesis mainly north of the Danube as axiomatic. Many examples have been given above to illustrate the effect of this fixation on the treatment of problems connected with the origin of Romanian in historical and archaeological works. The study of the Romanian language is also, however, of major significance to the problem. Historically, the discipline of linguistics determined that the speakers of early Romanians must have lived within the Roman Empire and in close contact with Italy for many centuries after the abandonment of Dacia Traiana by Rome in 275 A.D. With respect to the problems connected with the ethnogenesis of the Romanians, the Romanian linguistic literature contain more reliable data than most historical and archaeological works. Istoria limbii române  or Istoria limbii române,  for example, gives a very good picture of the Romanian language; and many other works and articles published in linguistic periodicals in Romania offer valuable information. Unfortunately, however, not even these publications are immune to influence from official policy. Because of the scarcity of materials, the problem of the origin of Romanian is hard to determine; and with regard to many details, one must be content with more or less plausible hypotheses. Debates and discussions, with the presentation of widely different ideas, are necessary and could lead to the clarification of problems. The opposite is achieved, however, by the systematic use of statements lacking any material evidence, by "solving" linguistic questions mostly or exclusively by means of archaeological finds, and by reaching conclusions that are in contradiction to the facts or to other conclusions presented in the same work. The subjective and one-sided treatment of the problems and the failure to draw logical conclusions from established facts also exacerbate the situation.
Correspondences Between the Latin Elements of Albanian and Romanian
One of the most important tools available to explicate the early history of the Romanian language is the study of its connections with Albanian. Numerous treatises have been written with the aim of diminishing their importance, since they obviously point to the early Romanians' prolonged presence in the central parts of the Balkan Peninsula.
Haralambie Mihăescu claimed the oldest Latin influence on Albanian showed that the ancestors of the Albanians were exposed to the Latin language as early as the beginning of the second century B.C., in contrast to Romanian, which lacks most of these ancient elements. Mihăescu discussed a total of 545 words of Latin origin, giving them the following classifications: 1. words in wide circulation, preserved in Albanian (with their original meanings), Romanian, and in the Western Romance languages (a total of 270); 2. words that are lacking in the Romanian language but have left traces in the Western Romance languages (a total of 151); 3. words preserved only in Albanian and Romanian (a total of 39); and 4. words found only in Albanian (85).
This classification is misleading. A large share of the words in Mihăescu's first group do not belong there, as can be seen in a comparison of Mihăescu's classification with the data on the Pan-Romanic words in Istoria limbii române.  The following are the words beginning with the letter "a" in Mihăescu's first group that are not found in the Istoria limbii romăne's list: 
There are 21 words beginning with "a" in the first group; take away these 9, which are not Pan-Romanic words, and almost half of Mihăescu's examples are shown to be misclassified. Far from being in "wide circulation," most of the 270 words of Latin origin and shared by Albanian and Romanian are characteristic of these languages
and appear at most in one or two dialects in Italy, Sardinia, and France. (Among those beginning with "a," Mihăescu remarks only about one [armăsar] that it also appears in a central Sardinian dialect [armissarius]). It also appears from these examples that the sound pattern of these words is in many cases similar or identical and expressly specific to Albanian and Romanian:
Mihăescu also included the Latin cuneus in the group of words in "wide circulation," in spite of the fact that this word means "wedge" and only Albanian kuj and Romanian cui have the sense of "nail" The case is the same with Latin hora "hour." In Albanian and Romanian, this word is used in the sense of "time": Albanian hërë, Romanian oară: de două ori "twice," de multe ori "many times," The only example of this sense besides Albanian and Romanian is found in the Venetian dialect (doi ora "twice").  In Mihăescus first and largest group of words of Latin origin in Albanian and Romanian, many specific Balkan-Latin elements are hidden, or classified there with a particular purpose, to deprive them of their significance. 
Mihăescu admitted that only 39 words belonged exclusively to Albanian and Romanian; but he did not allow even these to indicate a close relationship between the two languages, saying that 19 of the 39 words "were used widely." They are found in ancient sources. They could have developed "independently in Albanian and Romanian." This group omits, for example: Albanian krushk, Romanian cuscru "father of a son-in-law," from Latin cónsocer, although all other Romance languages, including Dalmatian, derive this word from consócer; Albanian mbret, Romanian împărat "emperor," the only popular surviving form of Latin imperator; and Latin pacare "to reconcile," which changed its meaning in the Western Romance languages to "pay" (cf., Italian pagare, French payer), with only Albanian and Romanian having preserved the original Latin sense of this word: Albanian pagonj, Romanian împăca "to reconcile."
Another 12 of the 39 words did not prove any similarity between Albanian and Romanian, according to Mihăescu,. because "they show either important morphological or semantic differences or analogies with the Western Romance languages."  This group includes, for example, Albanian pyll, Romanian pădure "forest," from Latin palude(m),
padule(m) "marsh." The sense of "forest" is found exclusively in Albanian and Romanian and appeared in this sense in records from Italy in the sixth or seventh century. Latin draco "dragon" appears in Albanian and Romanian in the sense of "devil" (Albanian dreq, Romanian drac), as it does in a southern French dialect (drac), which shows that one can find single examples in different Romance idioms and dialects for many of the Albanian-Romanian isoglosses. This is natural, since they all are Latin words. The point is that Albanian and Romanian not only share isolated elements but whole series of words, in most cases of a peculiar sound pattern, distinct from those found in the rest of the Romance languages. Moreover, and most significant, common changes of meaning are frequent. 
A third group in these 39 words consists of words from the Byzantine culture. The existence of such words supports the characteristic of the East Latin idiom. In several cases, the exact derivation of a word cannot be established; but its appearance in Albanian and Romanian is, nevertheless, significant: Independently from the etymology of Romanian codru "forest, mountain" (and also "big piece of bread"), the word cannot be separated from Albanian kodrë, kodër (with the same meaning).  Mihăescu's third group contains 151 words found in Albanian and in the Western Romance languages but not in Romanian, which can be explained by the fact that Albanian contains many Latin words borrowed during the first two centuries B.C. and that Albanian and Romanian history was not the same in later periods. Furthermore, the lack of a Latin word in a Romance idiom does not prove that the word did not exist at an earlier time. Many of the words in this group are represented in Romanian by Slavic loan-words. Therefore, only when one Latin word in one of the languages is represented by another in the other language may it be regarded as a certain difference.
The fourth group is 85 words of Latin origin that survived only in Albanian. What was said about the third group also applies to the fourth. Indeed, all Romance languages and idioms influenced by Latin contain several words do not appear in the rest of the Romance languages. It is therefore evident that the Latin elements of Albanian and Romanian are also closely related. 
The "Carpatho-Balkan Territory" as the Assumed Area of the Origin of the Romanian Language
In a linguistic textbook published in Romania, it has been stated that the theory of an exclusively north-Danubian origin of Romanian is no longer tenable: it could not be proved from the scientific viewpoint,
since the existence of the dialects south of the Danube (Arumanian, Meglenorumanian, and Istro-Rumanian) do not support this hypothesis.  According to authoritative Romanian historians and linguists (e.g., Ştefan Pascu and Alexandru Rosetti), an extremely large territory, both north and south of the lower Danube, was inhabited in ancient times by the ancestors of the present day Romanians.  This large territory is often called the "Carpatho-Danubian area,"  a designation unknown in ancient historical records. It is approximately 460.000 km2 large, and extends from Moldavia through Transylvania and Waliachia to Serbia and Macedonia. It is heterogenous from both the geographical and the historical viewpoints.
In works on the history of the Romanian people and language, it would be suitable to treat the entire area, both south and north of the lower Danube. One should, in fact, consider the area south of the Danube more important, because it is not only much larger than the Roman province north of the Danube (about 200,000km2 compared with about 80,000 km2) but was under Roman rule for six centuries compared to only 169 years for Dacia Traiana. This cannot be without significance for the development of a Romance language. It is, therefore, surprising to find that modern Romanian historiographers are interested mainly in the area north of the Danube, limiting most of their investigations to the present area of Romania As was shown in the chapter on archaeology, several excavations are being carried out with the aim of finding early vestiges of Romanians there.
Romanian linguists have studied the Romanian dialects in the central parts of the Balkan Peninsula (including the Northern Romanian dialect) and have collected data of great significance,  which have been presented in recent works, such as that by Alexandru Rosetti and the two volumes of Istoria limbii române. The results of these works have not, however, been used sufficiently in historical research in current Romania, which, as was previously stated, persistently concentrates on the territory of present-day Romania. A recent review of the development of Romanian linguistics  demonstrated this error. According to this treatise,"the Latin culture began to vanish in Dacia at the beginning of the third century after the withdrawal of the Romans (271 A.D.), while the language spoken in that time remained in direct contact with Western Latinity for about three or four centuries. The Daco-Roman population consisting mainly of poor people continued to exist after 271 A.D. but in double isolation—geographical and cultural (the cultural relationships with Western Romance-speaking peoples resumed only two hundred years ago).  The presumption
that the language of the "Daco-Romans" remained in direct contact with Western Latinity for about three or four centuries (after the abandonment of the province of Dacia Traiana), while these same "Daco-Romans" were isolated from Western Latinity both geographically and culturally, is a contradiction of fundamental significance. The evidence of early Romanian's development in close contact with the Western Romance languages (northern Italian dialects) and in the central Balkan Peninsula with Albanian and South Slavic is often mentioned in contemporary Romanian linguistic literature; but the available conclusions are ignored or denied, with reference be made, instead, to archaeological data. 
Ion I. Russu admitted that the sound pattern of the pre-Latin elements of the Romanian language was not specific to Dacian, Thracian, or Illyrian. He was, however, supposed to demonstrate the validity of the theory of continuity and therefore elaborated a hypothesis based on historical and archaeological data, independent from any linguistic considerations. He departs from the present territories of the Romanians and from the territory in Southeastern Europe that was once under Roman rule.  In his view, it was admitted as early as in the mid-nineteenth century that only Dalmatia, northern Thracia, Moesia, and Carpathian Dacia could be considered as the area of formation of the Romanian language.  Russu uses compound terms such as "Thraco-Dacian" to designate the pre-Latin (autochthonous) elements of Romanian. He states clearly that he uses these terms .."only as an hypothesis and in order to simplify the terms" (Russu, Etnogeneza..., 1981, p. 114).
There is strong evidence to doubt Russu's method that is outside of linguistic research. Investigations carried out on this field indicate a very close relationship between autochthonous elements of Romanian and Albanian language. Nevertheless, Russu prefers not to use the term "Albanian" or "Proto-Albanian" because of an alleged "general consensus" about the formation of Romanian.
Iancu Fischer admitted that the Romanian language contained all the changes that appeared in Late Latin from the fourth through the sixth centuries.  It may be concluded, therefore, that Romanian developed from Late Latin, which was spoken within the Roman Empire in the fourth through sixth centuries when the former province of Dacia Traiana was no longer part of that empire.
This problem of central importance for the history of the Romanian language is generally avoided in surveys and works published in contemporary Romania. All conclusions contradicting the theory of continuity are omitted, and the facts that are not in agreement with the theory are presented only fragmentarily. For example, a number
of sound changes in Late Latin are mentioned but without any indication of the fact that a large number of them appeared after the third century A.D. It is, of course, difficult to determine exactly the period of change in pronunciation or usage of grammar, and such changes are usually gradual. Several alterations becoming general in Late Latin could have started earlier, during the second or third centuries. The construction of the perfect tense with the verb habeo, for example, already appeared in the third to fourth centuries; but many others, such as the palatalization of Latin k in front of e, i (for example cellar, vicia); the assibilation of Latin t + e, i : Latin terra > Romanian ţară; the development of the syncretism of the genetive + dative; the characteristic patterns of declension of the verb in the modern Romance languages; and many other characteristics began in the period when the territories north of the Danube no longer belonged to the Roman Empire. All these Latin changes affected the language of the ancestors of the Romanians, since, as is also shown above, they all are present in the Romanian language.
An important problem of the South Slavic influence on Northern Romanian is its late character, the majority of the South Slavic (Bulgarian) elements in this language showing the sound pattern of Bulgarian in the eleventh to thirteenth centuries.  In that era, the Slavs north of the Danube did not, in contrast to the Bulgarian Empire, constitute a state and had no ecclesiastical institutions. This is one of the reasons for rejecting the theory of continuity north of the Danube. By formulating four theoretical possibilities, Ion Coteanu assumed that the late date of the strongest Slavic influence on Northern Romanian did not intimate that the ancestors of the Romanians were living south of the Danube. He considers that the contacts between the Slavs and the assumed Roman population north of the Danube were different from those existing between the Slavs and the Byzantines in the Balkan Peninsula. The four possibilities are that
1. Dacia was entirely uninhabited when the Slavs began to populate it. This hypothesis cannot be accepted, Coteanu continues, since the ancient names of rivers were preserved, which presupposes a local population from whom the Slavs could learn those names. The most significant statement in this passage is that these names show a Slavic sound pattern ("pronunţate după manieră slavă"). 
2. Only non-Romanized Dacians lived in Dacia at the time of Slavic colonization. In that case, either the Dacians would have been assimilated to the Slavs or the other way around. "Since none of these theoretical possibilities occurred in reality, this hypothesis too is baseless." 
3. In Dacia, only a small number of a Romanic population remained. In that case, "since in a certain period of time, terra Blacorum is mentioned in
Dacia"  it must be concluded that this scarce population, showing a great vitality, had Romanized at least a part of the Slavs and all Dacians.
4. Finally, there is the theory of Daco-Roman continuity, which assumes that Romanians were living in several "lands" (ţări) in the tenth century, when the Hungarians began to populate Transylvania.
These four theoretical possibilities do not, at any rate, clarify the issue. Hypothesis 1 is absurd; no serious scholar supports it. Hypothesis 2 corresponds in principle best to reality, but this is obscured by faulty formulation. It is surprising that Coteanu does not mention at least the Gepidae, a people known to have been living in several areas of former Dacia in the sixth century and who must has been the largest population when the Slavs arrived; Dacians no longer existed in significant numbers, if at all.  The Slavs consequently needed neither a Roman population nor Dacians in order to learn the ancient names of the rivers. The result was a Slavic population, from whom the Romanians later borrowed a vast number of geographical names, including those of the great rivers. Coteanu's statement that "none of these results has been produced" is based on the situation in a much later period, when Romanians already lived north of the Danube. The reasoning in 3 is also erroneous. Coteanu evades the fact that the Vlachs are not mentioned in Transylvania before the thirteenth century by using the vague formulation "in a certain period of time." That period is irrelevant in this context, since this mention of Vlachs does not prove that the Slavs found a Romanic population in Dacia some 600 years earlier. Finally, hypothesis 4, as it appears also from this text, is not supported by any evidence.
Discussing the question of the substratum of Romanian and its connection with Albanian, C. Poghirc does not exclude the possibility that "the substratum of the Romanian language is the language from which also modern Albanian derives"  Since the ancient Albanians lived in parts of present-day Albania and in areas to the east, this also implies the possibility that Romanian originated from the central parts of the Balkan Peninsula. Poghirc could not question the validity of continuity in Dacia Traiana; and such implications were therefore not discussed.
In the comments written to Ovid Densusianu's Histoire de la langue roumaine (1975 edition), it is admitted generally by Slavists that the Slavic language that influenced early Romanian was spoken south of the Danube (Old Church Slavonic).  These temporary admissions as well as the ambiguity of many texts indicate the tension felt by
Romanian linguists between objective research and the requirements of imposed official ideology.
About the Relationship of Latin to Gothic and its Assumed Relevance to the Romanian Language
Several Romanian lexical elements are assumed to have originated from Old Germanic. None of these etymologies has, however, proved valid; and Romanian linguists now believe that there are no Old Germanic elements in Romanian with the exception of a small number of words borrowed from Old Germanic by the East Latin spoken on the Balkan Peninsula.  Nevertheless, certain authors still return to the subject with new hypotheses. According to one assumption, the linguistic exchanges between Latin and Gothic occurred to the greatest extent in Dacia Traiana, between the middle of the third century A.D. and the end of the fourth century. The symbiosis between the two populations there resulted in the borrowing of Latin elements by Gothic of any period as well as in a Gothic influence on the Latin spoken in the area, the vestiges of which can be found in modern Romanian. 
In a critical analysis of this theory the difficult problem of the Latin influence on the Old Germanic languages must be viewed in its entirety, that is, without focusing or restricting attention solely to Dacia and to the third and fourth centuries. There are two main problems: The Latin influence on Gothic and its alleged significance for the history of Romanian, and the alleged Gothic influence on the Latin from which Romanian developed. The main difficulty here is the scarcity of data. The most important source about the language spoken by the Goths is the Bible translated by the Gothic Bishop Wulfila (or Ulfila) in the years 340 to 350 A.D.; data from later periods are very scarce. Wulfila wrote, no doubt, in a Gothic that was intelligible to his fellow countrymen; but his text also contains learned elements such as loan-translations of such words as misericordia "dispensation" or conscientia "conscience" that were not known earlier by the pagan population of Goths and were most probably created by Wulfila. Another problem, in many cases, is the fact that it is often hard or even impossible to determine whether a certain word was taken from Latin or from Greek. The Goths had also close contacts with the Greek population; and many loan-words exist, in several cases with the same sound pattern, in Greek as well. On the other hand, the sound pattern of several words gives an indication about the period of borrowing. It is also well known that the Goths had extensive trade contacts with the Roman Empire beginning in the first century A.D.
and that in the fourth century some of them were settled as foederati  in the Roman Empire, in Moesia, Beginning in the third century the Goths made frequent raids and expeditions from their settlements north of the lower Danube into the Roman Empire, reaching Asia Minor and the Greek islands. They thus came into contact not only with Roman civilization but also with Greek and various Oriental civilizations.
In the first two centuries A.D., while the Goths were living along the Vistula, several Latin lexical elements were transferred to their language.  Western Europe was under Roman rule, and Old Germanic-speaking peoples were living beyond the frontiers. The non-Roman populations of Europe used and often imitated the products of the superior Roman civilization. Roman merchants traveled throughout Europe, and members of the Old Germanic communities lived in the Roman Empire for varying periods of time as soldiers or political representatives of their tribes. The Italian linguist Vittoria Corazza mentions 26 words that most probably were transferred from Latin to Gothic during the first two centuries A.D., when the Goths lived along the Vistula. Early borrowing is suggested in these cases by several criteria: Latin sounds that later changed were preserved in wein (< Latin uinum; with the preservation of the semivowel u), kapillon (< Latin capillus "hair"; i preserved, later borrowings show e); kaisar (< Latin Caesar; the diphtong preserved). The loss of the final vowel is considered an indication of early borrowing; Latin lucerna > Gothic lukam, pondo > pund, mensa > mes. Several words (iasilus, ana-kumbjan, marikreitus, for example) show signs of having been very well assimilated into the Gothic language. Groups of semantically related words are likely to have been borrowed in the same period; for example, wein "wine," akeit "vinegar," alew "oil"; or the words related to eating habits: ana-kumbjan "to sit down to dinner," kubitus "group of people dining together," mes "dish, table, wine-press." The fact that a word exists in several modern Germanic languages also suggests an early borrowing by Old Germanic (in any case, a borrowing not restricted to the Goths or to a specific territory): wein, compare, for example, English "wine"; pund, cf., English "pound"; katils, cf., German "Kessel"; asilus, cf., German "Esel"; kaupo, cf., German "kaufen." The following Latin words were most probably borrowed by Gothic during the first and second centuries A.D. when the Goths were living along the Vistula: 
From the viewpoint of semantics, most of these words are connected with everyday life, first of all with fashion (5 words), followed by food, eating habits, and domestic objects. Two words denote Southern European products; another two are related to commerce; and five are foreign names (for example, Ruma "Rome") adapted by Gothic. According to Corazza, only 17 lexical elements were borrowed during the third century. This number itself can hardly be decisive; many words of Latin origin may have existed at the time without having been used by Wulfila. It is also necessary to analyse the semantic groups to which these words belong, since this aspect may be of more significance.
The following Latin words were most probably borrowed by the Goths in the third century A.D.
Vittorio Corazza believed that the contacts between the Goths and the Romans were closer than in earlier periods: such words as assarjus, drakma, unkja, and arka indicate that the Goths continued to have trade with the Romans; such words presuppose deeper contacts than those that existed along the Vistula. The penetration of the words of a military character is explained by the fact that Gothic soldiers fought as mercenaries in the Roman legions. Beginning in the middle of the second century, the barbarians made up the most important part of the Roman army. 
It is, however, difficult to see any significant difference between this list and the lexical elements transferred to Gothic during the previous period. Evidence of deeper contacts is, in any case, limited, despite the fact that it would be expected because of the increased number of Gothic soldiers serving in the Roman army in the third century. There are two words connected to the Roman monetary system: drakma "drachma" (a Greek coin, also widely used by the Romans), and assarjus (a bronze coin of low value)—such loan-words are usually indicative of commerce; there are also unkja "a measure of land" and arka "bag." Such borrowings are few and not even unique to the third century (pund "pound," thus, a unit of measure, had already been transferred to Gothic in the first two centuries A.D.). Moreover, unkja was not an isolated word in Dacia but was borrowed by most Old Germanic populatios: cf., Old High German unze, Old Icelandic unzia, Danish unse, modern English ounce. With regard to the rest of these words, balsan was probably (according to Corazza) borrowed by Gothic merchants somewhere in the region north of the Black Sea and ulbandus "camel" in Asia Minor; saur "Syrian" could have been borrowed anywhere in Europe after Syria became a Roman province in 62 B.C. and Syrian merchants traveled throughout the
continent. Skaurpjo "scorpion" and sabati, a linen cloth made at Saban, a locality near Bagdad, are also more likely to have been borrowed elsewhere than in Dacia, as are the three military terms (militon, "to serve in the army," anno "soldiers pay," and spaikulatur "watchman"), which were probably introduced into Gothic by soldiers returning from service in the Roman army. It should also be remembered that none of the Latin words from which these 17 Gothic loans originate exists in the Romanian language, not even platea, of Greek origin, which was widespread in the Roman world and appears in the major Romance languages as Italian piazza, French place, and Spanish plaza.
The following borrowings from Latin were most probably transferred to Gothic in the fourth century:
To this must be added the fact that Wulfila's text contains several loan-translations of Latin words:
Of these haiþno may be a very old, Proto-Germanic loan-translation from Latin (although there are other etymologies too, none of which implies Dacia). With regard to skillings, in the third century, clypeus, ciipeolus, was the popular name of an imperial Roman coin in the region of the Rhine River. (Latin clypeus "shield," Gothic skildus "shield"). Also, hundfaþs may be an ancient loan-translation from Latin (but could also originate from Greek). The rest of these loan-translations are learned constructions, most of them having been made by Bishop Wulfila.  In the Gothic language the number of loan-translations from Greek is much higher than those from Latin.  The borrowings from Latin dating from the fourth century are too few to warrant any historical conclusions (they also belong to a
variety of semantic categories). Obviously, our knowledge of the Latin borrowings by Gothic, at least from the third century A.D. and later, is extremely limited. At any rate, the linguistic material does not indicate anything about the territories in which the Gothic population was living during those centuries, and assumptions about Dacia being the main territory of borrowings lacks any evidence. 
This is also the case with possible Latin influence on the grammar of the Gothic language, including the Latin prefix dis- and the suffix -arius. Such cases in themselves never indicate a certain geographical area; they are significant only in determining the extent of the Latin influence on Gothic. It is claimed that Latin influenced the following grammatical elements in Gothic: the analytical past continuous and past passive; the present perfect; the accusative with the infinitive; and the accusative with the participle. This would be of great significance, since "the grammatical system of a language yields only to very high pressure from another language."  Not all of these phenomena were, however, with certainty borrowed from Latin (an Indo-European origin comes into question as well); and, more important, given the close and lengthy contacts between the Romans and the Old Germanic peoples, it is not surprising that the language of the latter was quite deeply influenced by Latin.
The Alleged Gothic influence on the Romanian Language
The Romanian linguist Ştefănescu-Drăgăneşti claimed that Romanian was the first Romance language to be influenced by Old Germanic. Without giving any evidence, he wrote about many Romanian words of Gothic origin, giving, however, only five examples: a găti "to prepare"; iubit "beloved"; leac "remedy, medicine"; lăutar "singer"; and isteţ "shrewd, cunning." He added that most of these borrowings were considered to be of Slavic origin, because they occurred in Slavic, too, which borrowed them, in its turn, from Germanic.  Standard Romanian monographs on the history of the Romanian language,  however, consider the Slavic origin of these lexical elements to be firmly established. (With regard to a găti, this could also be a pre-Latin substratum word or a Slavic one borrowed by Albanian and from Albanian by Romanian).
The Romanian grammatical structures that Ştefănescu-Drăgăneşti believes were borrowed from Gothic are in reality Balkanisms: the analytical future with the auxiliary a voi "will" is found not only in all Germanic languages and in Romanian but also in Greek, Bulgarian, dialectally in Serbo-Croatian, and in the southern (Tosc) dialect of Albanian. The analytical future with a avea "to have" (not meaning "must"):
am să fiu "I shall be" appears not only in Romanian but also in Bulgarian, Byzantine Greek, and the Tosc dialect of Albanian. In Old Romanian texts a avea also appears with the infinitive: n'am a te lăsa "I shall not let you," a perfect counterpart of Albanian kam + me + infinitive. The definite article in postposition is also a typical Balkan characteristic of Romanian that occurs in Albanian and Bulgarian as well. As the Albanian scholar, Eqrem Çabej,  pointed out, Romanian shows a concordance with Albanian in the use of the definite article in the smallest details of its syntactical position.
In conclusion, the Latin influence on the Gothic language was already strong in the first two centuries A.D., when the Goths were living along the Vistula. In the third and fourth centuries, the Goths were spread over a large area, from the Don River to Moesia, also including Dacia Traiana. The Latin loan-words from this time can be explained mainly by continuing trade contacts with the Romans, Gothic soldiers' serving in the Roman army, and military expeditions to the Balkan Peninsula and Asia Minor. The Latin influence on Gothic was exerted over several centuries and in a great territory, of which Dacia was only a small part. Of the 17 words probably borrowed during the third century, none suggests any connection with circumstances in Dacia; and none of the Latin words from which they derive exists in Romanian.
The assumption of a Gothic influence on the Romanian language is based on false etymologies and imaginary relationship of grammatical phenomena. The lexical elements assumed to derive from Gothic are mostly of Slavic origin, and the Romanian grammatical elements believed to be of Gothic origin are in reality a part of the many Balkanisms in the Romanian language and also are found in Albanian, Bulgarian, and other Balkan languages. There is nothing to support the idea that the Goths living north of the Danube in the third and fourth centuries would have had any influence on the speech of the ancestors of the Romanians. The assumed Old Germanic elements of the Romanian language are propagated mainly with the aim of defending the theory of continuity. Therefore, most of the proposed explanations are chosen in order to support a preconceived theory.
Assumptions often came to be accepted as more or less established facts, even by foreign scholars. Vittoria Corazza, for example, remarks that the Gothic domination over Dacia left interesting vestiges in some Romanian geographical names, such as Goteşti, Munte Gotului (correctly: Muntele Gotului or Muntele Gotul, cf. Iordan, 1952, op. cit., p. 230), Pârăul Gotului, and the personal names Gotes, Manea (from Gothic manna 'man'), and Goma, Guma, from Gothic guma 'man', from which
the place names Gomeşti, Gumeşti were created. Corazza refers to Romania Germanica of Ernst Gamillscheg. 
A detailed study of each individual element made by several Romanian scholars has shown that none of the Old Germanic etymologies (including those referring to place names) can be accepted. A study of these etymologies may, however, throw some light on the method by which such hypotheses are made plausible. There is first the historical basis: "Goths were once living in Dacia." There are also less rational ideas: Why would these names of different Germanic tribes have been preserved in France but not in Dacia?  With regard to the historical basis, one must remember that the Goths disappeared from Dacia at the end of the fourth century (and the Gepidae, in the seventh century). More important, no identifiable traces of these Old Germanic cultures are to be found in the popular traditions of any people now living north of the lower Danube. There is no historical tradition preserved in the folklore of the Romanian people about the ancient period of the formation of Romanian.  Because of these considerations, the scholar is forced to seek other, alternative explanations for these assumed Old Germanic names. In doing so, the phonetic laws of the Romanian language must be considered.
Romanian geographical names containing Got- are based mainly on personal names. Romanian, Slavic (Bulgarian), and Hungarian personal names come into question: Gotea could derive from the Romanian personal name Grigore-Gore, which in the speech of children is pronounced Gote.  The Romanian linguist Iorgu Iordan discusses an exhaustive list of geographical names of this kind: Muntele Gotului (Alba County), Pârăul Gotului (Alba), Goţi and Pârăul Goţilor (Sibiu County), Goţa (Slatina, Argeş County), Goţul (Păşcani, Iaşi County), Gotca (Vaslui County), Gotea (Mihăileşti, near Bucharest), Goteşti (Murgeni, Bîrlad County), Goteşul (Cislău, Buzău County), and Gotgoaia (Iaşi County).  In explaining names, with a t (not those with a ţ), Iordan refers to the theory of the Hungarian scholar István Kniezsa, who derived these names from the personal name Got (Goth, Gót) existing in Slavic and in Hungarian and probably a diminutive of Gotthard, Gottlieb, and similar names. The geographical names of this type found in Muntenia and Moldavia (and perhaps also in Transylvania) may, also according to Iordan, be based on the Bulgarian personal name Goto or Gota, the diminutive of Georgi. This Bulgarian name was borrowed by the Romanians in the form Gotea, from which developed Goteş, Goteşti. Alexandru Philippide proposed an explanation related to gotcă, the Romanian name for a mountain hen. In Romanian, got- could have existed as the name of a bird, and this got- may be the basis of geographical names containing Got-. 
The names with ţ (Goţ-) derive from the Romanian word goţa, monster, with which children are frightened when they are crying: "Taci, că vine Go/a/ţa!" (Don't cry, or Go/a/ţa will come). Iordan also adds, that if Goţa is the feminine of Goţul and this the name of the Goths, then it would contain the diphtong oa (with a, or, in the form without the definite article, ă, in the following syllable). 
Scholars who have critically analyzed the etymologies of Romanian lexical elements assumed to be of Old Germanic origin have also examined the personal names advanced in this connection and refuted these etymologies, in most cases concluding that the name derives from Slavic.  An example of this is Manea which was borrowed from the Bulgarian Manjo.  The Gothic language therefore does not contain any element that would suggest a symbiosis of Goths with a Latin-speaking population in the area of former Dacia Traiana; nor do the Romanian language and toponymy show any influence from Gothic.
With regard to the question of whether the Gothic language indicates . anything about the neighbors of the Goths in former Dacia, there are only hypotheses. Gottfried Schramm has advanced the hypothesis of an a > o change in the barbarian languages of Southeastern Europe during the first two or three centuries A.D. The Gothic word aikklēsjō, which ultimately originates from Greek or Latin (cf., Latin ecclesia), could have been borrowed from such a barbarian language.  Because of chronological considerations, those barbarians were probably living in a mountainous territory, possibly in Dacia Traiana. If this hypothesis were to prove correct, this would add another non-Roman population to the Goths, among whom Christianity was spreading north of the Danube as early as in the third or fourth century. It must be emphasized, however, that is only an hypothesis.
The Inherited Latin Words in Romanian
The study of the Latin vocabulary inherited by the Romance languages gives interesting insights into cultural history. In the Western Romance idioms, many terms connected with urban life have been preserved, indicating that at least some of the speakers of these languages lived in towns and practised trades already known by the Romans throughout the centuries. Romanian, however, did not preserve the Latin terms for urban life. The Romanian oraş "town" is of Hungarian origin; drum "road" derives from Greek dromos; cale "way," although a Latin word, originally meant the way used by transhumant flocks of sheep.  Latin pavimentum "floor" changed its meaning to "earth" in Romanian (pămînt). This suggests that the speakers of the
Romance language from which Romanian developed were, at least over a considerable period of time, a rural rather than urban people. Furthermore, the terminology of Latin origin of the most important agricultural plants (wheat, barley, rye, millet, and flax) and farming tools (plough [aratru], sickle, pitchfork, axe, and others) and farming activities (to plough, to sow, to sift) proves that the Romanian people practised farming without interruption. 
The Romanian language originates from one area of Roman colonization, whose southern frontiers in the Balkan Peninsula were established by Jireček and Skok.  South of this line, mostly Greek was spoken. The Vlach shepherds in their wanderings, however, reached areas south of the Jireček line in an early period. From the eighth century, there were Vlachs living in Greece, south of the Jireček line; and from the end of the tenth century onwards, groups of Vlachs were reported in almost all areas of the Balkan Peninsula. As the two main dialects of Romanian, each with a sub-dialect (Arumanian, with Meglenitic, and Northern Romanian [spoken north of the Danube] with Istro-Rumanian) developed, the language was disseminated by the wanderings of its speakers. The inherited Latin elements were spread with the language. No one has suggested that Arumanian, for example, originated in the area where its speakers are now living, simply because it contains several ancient, inherited Latin elements.
It is therefore strange that theories similar to this have been advanced about the Vlachs living north of the Danube. There is, for example, a group of hypotheses that connect the presence of certain physical remains (huts, objects of everyday use, such as pottery and open hearths) found in present day Romania with the corresponding words in the Romanian language.  The terms of Latin origin connected with agriculture and pastoral life are considered to indicate that the ancestors of the Romanians were peasants and pastoralists (not town-dwellers) in Dacia.  The words sat "village" and cetate "fortress" are also often given to prove the presence of the Romanians' ancestors in the former Roman province of Dacia Traiana after the retreat of the Romans.  Some Latin words, which changed their original meaning in Romanian (e.g., Latin veteranus "soldier who served his term" > Romanian bătrîn "old"), were also thought to be connected with Dacia Traiana. The German linguist, Günter Reichenkron, for example, claimed this indicated that veterans of the Roman army settled in the province north of the Danube and that this is still testified to in our day by such Romanian words as bătrîn < veteranus, bărbat < barbatus, mire "bridegroom" < miles as well as the somewhat rough
expression fată < foetus and, according to Tiktin, perhaps copil "child." 
In the same way, the religious terms of Latin origin have been connected with objects of Christian use found in former Dacia Traiana: oil lamps (lucernae), stamps for the fabrication of crosses, Byzantine amphoras on which a cross and the letters Alfa and Omega were painted, all dating from the fourth through to sixth centuries.  Istoria României. Compendiu went even further, claiming that Illyrian Christianity played an important role in the preservation of the Latin language in the Carpathian Mountains, because, beginning in the fourth century, missionaries spread Christianity in the Latin language, not only south but also north of the Danube. In this way one can explain the basic terminology of Daco-Roman Christianity, which is of Latin origin: biserică (basilica); dumnezeu (combination of dominus and deus, "Lord and God"); lege "law" in the sense of faith (legem); cruce (crucem) "cross." 
Haralambie Mihăescu, describing the dissemination of the Christian faith in the Latin language on the Balkan Peninsula beginning in the third century, asserted that in the fourth century, Christianity was generally accepted and reached Dacia north of the Danube; the word quadragesima, from which the Romanian word păresemi developed, is known, with a Christian sense, from that period. In Mihăescu's opinion, the Christian religion spread from the ecclesiastical centers south of the Danube to the populations north of the river; and this fact explains the preservation in the Romanian language of Christian terminology of Latin origin. 
There are several other hypotheses based on false premises. It must be noted that inherited Latin words do not indicate anything about the territory in which the ancestors of the Romanians were living. These words only show that the population from the time of Romanization onward, used the objects and at least knew religious concepts denoted by them. Such knowledge is totally independent of a people's territory. Physical remains connected with the rural way of life as well as objects of a Christian nature, have been also unearthed in many other territories, such as southern Poland and Slovakia, which had a Vlach pastoral population for several centuries; but no one would argue that these finds proved that the Vlachs had always lived there.
Such words as zână, zănateci, and Sînziana do not indicate that the cult of Diana survived the Romanization of Dacia. The word sat and the change of pavimentum to mean "earth" (pămînt) are connected with the rural way of life of the speakers of Romanian but tell us nothing about the territory in which they had lived earlier in their
history. The basic Christian terms in the Romanian language only show that the ancestors of the Romanians were Christianized during the Roman period, which could have occurred on the Balkan Peninsula as well. Indeed, as even Reichenkron has remarked, these terms must have been adopted by the ancestors of the Romanians in an area in which the Latin language dominated.  The Christian Church was widely established beginning at the time of Constantine the Great (fourth century) when Dacia Traiana no longer belonged to the empire and would not have been affected.
The conditions for the development of basic religious terms in the Romanian language were really favorable in Dardania, the central part of the Balkan Peninsula. According to historical records as well as archaeological finds (ruins of several churches), this province was the center of religious life for a large part of the Balkan Peninsula for about three centuries. With regard to the cult of Diana (or Artemis), for instance, it is assumed that this cult survived the Romanization of Dacia and that Northern Romanian sînziana derives from the Latin Holy Diana "of Sarmizegethusa." It is also claimed that a religious and linguistic continuity was assured by the fact that the process of transformation took place in a rural milieu. This is probably true, but there is nothing to suggest that the process of transformation occurred in Dacia Traiana and that the cult of Diana was more widespread or more intense there than in other parts of the Roman Empire. The cult of the goddess Diana was widespread throughout the Roman Empire. It is clear that "les monographies soulignent à l'envi la diffusion du culte d'Artemis-Diane au Sud du Danube, en Pannonié, sur la côte Adriatique, en Norique."  Diana appears in antique place names in Africa and Syria, as well as on the Balkan Peninsula (cf., ad Dianam in Epirus, mentioned in the Peuteringian Tables, and Ζάνες given by Procopios). Today, in the Balkans there are Dzîna and Zona (probably from Arumanian dzînă "fairy"). North of the Danube, however, there are no such vestiges. In addition to Romanian, the name Diana is preserved in several Romance languages, as well as in Albanian: "Les langues roumaine et albanaise ont gardé le nom de la déesse avec la signification de 'fée': zână, zánë."  Romanian folk traditions connected with the wild flower Gallium verum (Romanian sînziana) were even taken over by the Slavs living in the Balkan Peninsula.  There is nothing to imply that this cult survived the Romanization of Dacia and that the religious and linguistic continuity of this cult refers to that territory. On the contrary, all evidence is found in the Balkan Peninsula and other parts of the former Roman Empire. The ancestors of the Romanians inherited this cult in the central part of the Balkan Peninsula, where they also
inherited their Romance language, and took it with them in their wanderings: the Northern Romanians north of the Danube and the Arumanians south of the Jireček-line.
The word sat, from Latin fossatum, reveals a borrowing south of the Danube: its sound pattern does not agree with an inherited Latin word, because Latin o did not disappear in Romanian in this position. The disappearance of o occurred most probably in Albanian, cf., Albanian fsat "village" (from Latin fossatum). As late as in the sixteenth century, fsat was recorded in the Northern Romanian territory. This word is thus probably an Albanian loan-word in Romanian. The meaning "village" is found only in Albanian and Romanian.
Another example is Romanian bătrîn "old" from Latin veteranus "soldier who has served his term" (classical Latin) or "old" (Vulgar Latin) in several texts, veteranus = antiquus vel vetustus (cf., Vegliotic vetrum, Friulian vedran).  Obviously, veterans settled all over the Roman Empire and the change of the sense occurred in Late Latin.
The derivation of the Romanian word mire "bridegroom" from Latin miles "soldier" is considered dubious.  Rosetti mentions two possibilities: 1. a loan-translation on the pattern of Serbo-Croatian vojtio, vojino "husband," from voinu "soldier" (the original sense of this Serbo-Croatian word would have been "courageous"); 2. an evolution of sense in the liturgical language (the young bridegrooms form an "army"). 
Romanian bărbat "man, male, husband, manly, manful, virile" is of Byzantine origin (βαρβᾶτος,) and meant "man" in contrast to eunuch.  None of these words mentioned above contains any element connected with Dacia Traiana; but they all show characteristics that point toward South, the vicinity of Albanian and Byzantine Greek, as well as the area in which Late Latin was spoken.
4. THE THEORY OF "CORE REGIONS" (KERNGEBIETE) OF THE ROMANIAN LANGUAGE
Some historians advance the hypothesis that certain peculiarities of speech of the Romanians living in the region of the Apuseni Mountains (Erdélyi-szigethegység) indicated that Romanian had been spoken there uninterruptedly since the Roman period (106-275 A.D.)  and even after the abandonment of Dacia Traiana.  Departing from the observation of Bartoli, according to the Romanian scholar Sextil Puşcariu, there was a correlation between the geographical position and the age of a linguistic phenomenon.  Puşcariu concentrated on the area of the Apuseni Mountains (northwestern Transylvania), disregarding the fact that the words in question are found in a much
Romanian dialectal words cited by Sextil Puşcariu to support his Kerngebiet theory: the words, their Latin origin, and their meaning
larger area than those mountains. In his view, the Roman settlements were most dense in this region and, consequently, Romanization most intense. "A comparison of Giurescu's map  with our map, [Puşcariu continued], the words aiu, pedestru, nea, păcurar, june, and cuminecătură made it clear that the Romans extended from western Transylvania partly toward the west (where we find the words citeră and cotătoare preserved), and toward the northwest, including all the territory north of the Mureş River and reaching towards the east in Moldavia and Bessarabia up to the Prut River and even beyond it, a process that may be followed historically and that is reflected in its final stage by linguistic maps such as those for moare, curechiu, and cute." 
Three of the words mentioned in Table IV and Table V—moare, curechiu, and cute—induced Puşcariu to construct another hypothesis. In Muntenia the corresponding words are zeamă de varză (zeamă is of Greek origin), varză (from Latin viridia-virdia) and gresie (from Albanian gërëse).
Romanian dialectal words cited by Sextil Puşcariu to support his Kerngebiet theory
The distribution of moare and zeamă in Romania (1936). North of the departing line the ancient word of Latin origin, moare (< Latin mŭria) "sauerkraut brine" is used; south of the line, this word was replaced by a Greek loan (zeamă) (zeamă de varză "sauerkraut brine"). (After S. Puşcariu: Les enseignements de l'Atlas Linguistique de la Roumanie, 1936, map No. 12 (ALR-I, 755).
On the basis of this difference between Muntenia and northwestern Romania, Puşcariu assumed a "barrier" of ancient Latin words between Muntenia and northwestern Transylvania: All these innovations emanating from the south, in their expansion toward the west and the north, came up against a relatively powerful wall behind which the Latin words curechiu, moare, cute, păcurar, Sânicoară, Sămedru, and cuminecătură resisted them; and Puşcariu found on the same side of this wall such words as nea, cotătoare, and aiu, instead of zăpadă, oglindă, and usturoi, borrowed from the language of the Slavs north of the Danube. 
There are several reasons why these hypotheses cannot be accepted.
1. The most numerous Roman settlements during the Roman era were probably in western Transylvania, while in other parts of Dacia Traiana, especially eastern Transylvania, the population was always mainly non-Roman. Compared with many large areas on the Balkan Peninsula, however, the Roman settlements in Dacia Traiana were not particularly dense and, moreover, existed for only 170 years
compared with 600 years or more in the Balkans. In spite of this fact, most Roman settlements in the Balkan Peninsula disappeared within a few centuries after 600 A.D. Consequently, there is no general correlation in Southeastern Europe between dense Roman settlements during the Roman period and a Romanic population today.
2. Puşcariu's argument is illogical even when viewed by itself: The sine qua non of his theory would be that a certain area in which more inherited Latin words are used than in other areas coincides reasonably well with a Romanized area during the era of the Roman Empire. Of course, no exact coincidence should be expected; but what we find is rather a lack of any regularity: a large part of Dacia Traiana (the territory west of the limes Alutanus, present day Oltenia) is not among the areas in which Puşcariu finds ancient Latin words preserved. Moreover, and more important, all of these words are found in a much larger territory than that referred to by Puşcariu in his conclusion (the area of the Apuseni Mountains). These words are also used in western Romania along the Hungarian frontier, as well as in Maramureş; and most of them also in Moldavia and even in Bessarabia, beyond the Prut River, i.e., in large territories that never belonged to the Roman Empire and consequently never had a Roman population. Puşcariu stated that this distribution was the result of an extension of the Roman population; but if these Latin words could spread over large areas that were not Romanized during the era of the Roman Empire, then their presence in the region of the Apuseni Mountains too may be explained by later expansion.
Puşcariu refers to 12 appellatives and two personal names (names of Saints), of which, however, only about half seem to be found exclusively in the northwestern and northern dialects of Northern Romanian. At least seven of the appellatives also exist in Arumanian:
Of the rest of these lexical elements, Sânicoară is probably not even an inherited element in Romanian: It has been assumed to have derived from the Latin Sanctus Nicolaus. When Greek Nicoláos was transferred to Latin, however, its accent moved to the long -a-: Nicoláos (as shown by the Old Italian form Nicoláo and modern Italian Niccoló, with the accent on -lao, -lo). The unstressed -o- of Latin changed in Romanian to -u-, and so the Latin Nicolaus would thus have become the Romanian *Nicuráu and not Nicoară.  Five appellatives remain, of which one has a special meaning in the northwestern dialect, different from the rest of the Northern Romanian territory (pedestru "poor man"). Two appellatives have, in other areas, corresponding appellatives also of Latin origin: ceteră corresponds to vioară (from Latin uiola)  and curechiu to varză (from Latin virdia). In these cases one can hardly speak of the preservation of Latin elements compared with foreign loans. Only two appellatives remain that do not exist in Arumanian and are replaced in Muntenia by foreign loans: cute, from Latin cos, cotem, is replaced by gresie, of Albanian origin; and instead of cotătoare, the Muntenian subdialect has oglindă, a back-formation from the verb oglindi, of Slavic origin.
3. If the ancestors of the present-day Romanians had been living in certain areas of modern Romania since the period of Dacia Traiana, their language would have developed largely independently from the rest of the Roman provinces. (Puşcariu mentions "the isolated Romanians of Dacia Traiana.")  The dividing line during this period (largely that of Late Latin) was the Danubian frontier of the Roman Empire. Innovations would thus have easily spread all over the Balkans but with much more difficulty or not at all across the Danube (assuming that a Romance language was spoken there); but if varză, gresie, and zeamă once passed the frontiers of the empire (they are found today in Muntenia), what kind of a barrier could prevent them from also spreading to the speakers of Romanian in Transylvania? An analysis of the Romanian language from the viewpoint of the period in which its Latin elements originated shows that most of the new constructions and new lexical elements that appeared in Late Latin (that is, in the speech of Italy from the fourth through the seventh centuries) are, indeed, found in it  (including its subdialect spoken in northwestern Transylvania). In other words, there is no difference whatsoever in this respect between the subdialects of Northern Romanian; the differences are of a much later date.
4. Five principles (four areal and one historical) have been established for linguistic geography: a) isolated areas (those with fewer contacts, mostly islands, such as Sardinia) and b) lateral areas (in the Roman Empire: Iberia, the Balkans; in Northern Romanian: the
Banat, Crişana, Maramureş, and certain areas in Moldavia and Bessarabia) preserve usually earlier phases of language, c) A phase spread over a larger territory is usually older than the corresponding phase in a smaller area. This principle can only be applied, however, to the Roman Empire in toto.  d) Areas conquered (and Romanized) later often preserve the earlier phases. (In the Empire, Italy was usually more innovative than the provinces), e) If a phenomenon has two phases of which one is disappearing or has disappeared, this phase is the older one. The Latin ignis, "fire," for example, is not found in the Romance languages, all of which preserved the more recent word, focus.
It is now generally recognized that these principles are to be understood only as general tendencies. "All of them are full of exceptions and contradictions,"  and the opposite (especially of principles [c] and [d]) often occur. More important in the context of the Kerngebiet theory is that these principles at best can only give some indication of the relative chronology of changes in language.  They are not helpful in determining the absolute age of linguistic phenomena. One linguistic phase may indeed be earlier than another, but this does not indicate that it has existed in the area in question ever since the time of the Roman Empire. It is also needful to take into account the continuous change of language. Most Romance languages, for example, preserve the Latin caballus instead of equus, "horse." One could conclude from this that equus or equa survived only in lateral areas: Spanish yegua, Romanian iapă (< Latin equa, both with the sense of "mare"). Through ancient texts, however, it is known that the word derived from Latin equa was once used throughout the Gallo-Romance territory; Old French ive "mare" (from Latin equa) was still widely used in the fourteenth century. The areal principles should be examined, therefore, by chronological data.
The most plausible division of the Northern Romanian dialect distinguishes five areas with subdialects: 1. Muntenia, 2. Moldavia, 3. the Banat, 4. Crişana, and 5. Maramureş. Transylvania lacks any particular Romanian subdialect; the Muntenian subdialect extends over southern Transylvania and the Moldavian over northeastern Transylvania. Beyond this division, the Northern Romanian territory can be divided into two, three, or four areas, on the basis of several lexical elements. For example, instead of ficat, varză, and năduşeală in the southern area, there are mai (from Hungarian máj "liver") curechiu, and sudoare in the north;  or there are three areas: in Muntenia, one says os (from Latin ossum "bone"), in Moldavia, ciolan "animal bone with or without flesh on it" (from Slavic članú), and Crişana and Maramureş, ciont, with the same meaning as ciolan, from
Hungarian csont "bone." For the word meaning "cemetery," for example, four areas can be distinguished: cimitir (Muntenia), ţintirim (Moldavia), mormînt (the Banat), and temeteu, from Hungarian temető "cemetery" (Crişana). In Daco-Romanian the same notion can be expressed by one, two, or more words, as a consequence of the meeting within the same linguistic area of some old words with new ones, of some words of primary form (Latin, Slavic) with words created on Romanian territory, of some more or less recent loans from other languages (Hungarian in the west, Ukrainian in the east, Bulgarian in the south). 
These facts leave no room for the theory of ancient Roman-Romanian Kerngebiete; and Matilda Caragiu Marioţeanu, describing the Romanian dialects and subdialects in detail, does not even mention this theory. The appearance in certain subdialects of the words nea, arină, and ai is mentioned by this author only as an example of the preservation of some archaic words in lateral areas. 
5. In Muntenia Latin words were preserved, words that in other areas were replaced by foreign loans. Some investigators drew the conclusion that the presence of such words indicated that Muntenia (or part of this province) was another ancient Romanian Kerngebiet. The arguments were similar to those used by Puşcariu; one as that advanced by Eugen Lozovan is based on Latin words in Muntenia for which foreign loans are used in Transylvania and on geographical names derived from the Slavic name of the Romanians: codrii Vlăsiei, and Vlaşca. In Muntenia, argued Lozovan, one finds the ancient terms of Latin origin: celar "cellar," and cuptor "oven, kiln," as well as several words from the substratum of Romanian: argea "room made in the earth" and vatra satului "the precincts of the village" (cf., Albanian vatër, vatra [Tosc dialect] and votër [Gheg dialect], "hearth, fireplace, dwelling")? These are modest words, of Latin or of Thracian origin (argea, vatra), referring to the settlement of a stable population"  Other differences compared with Transylvania are words of Latin origin that were replaced by loans in northwestern Transylvania: 
Moreover, in Muntenia, the iotacized verb-forms (rîz, văz, auz) are used, which in Transylvania and other areas were replaced by analogical forms (rîd, văd, aud). Pointing out that a part of Muntenia is called Vlaşca, indicating the presence of Vlachs there, Lozovan concluded: "To summarize, linguistic geography and toponymy not only do not contradict our theory about the existence of an ancient Danubian zone where a Roman population could survive but can even be used to delineate the frontiers of such a zone."  Consequently, the area considered ancient by Puşcariu is now shown to be one in which foreign words have replaced Latin expressions preserved in Muntenia.
6. As shown by the first written texts in Northern Romanian, the speakers of this language in Maramureş and in southeastern Transylvania in the sixteenth century spoke a language much closer to Latin than any present-day Northern Romanian dialect. The so-called rhotacizing texts were translated in the north of Transylvania (Maramureş) beginning with the sixteenth century. One stratum of their language is characterized by the subdialect spoken there and another by that spoken in southeastern Transylvania, from where the copyst came. In these texts there are certain phonetic, morphological, and lexicological characteristics that bring to light a stage of the language closer to the Latin prototype. 
Phonetic features: Latin d > dz (a stage between Latin and modern Northern Romanian); the e after r is preserved: întîniu, spuniu (with a palatalized n) instead of the modern Northern Romanian întîi, spui (except in the Banat).
Morphological features: the final u was preserved for the nouns of the second declension (Latin -us, -urn): domnu, împăratu (today domn, împărat in Northern Romanian, but Arumanian has still final syllabic u); the simple perfect of Latin is not entirely abolished: Latin feci— sixteenth century Maramureş, Bucovina, and northern Moldavia: feciu, Latin venemus > venremu (today făcui, venirăm). A form of the conditional inherited from the perfect of the Latin subjunctive, as well as other morphological pecularities, was still in use.
More interesting, however, is the vocabulary, which shows an unsuspected richness of words inherited from Latin, words that have disappeared from circulation in the language today or that perhaps survive as vestiges in some remote region of the country: agru < agrum, "tilled land"; ariră < arenam, "sand"; auă < uvam, "bunch of grapes"; deşidera < desiderare, "to desire, to wish"; fuşte < fustem, "staff, rod, baton"; gerure < gironem, "depth, abyss"; gintu < gentem, "people, family, relative"; ğune < juvenis, "young man"; măritu < maritum, "bridegroom"; a se număra < nominare, "to mention by
name, to give a name to"; păsa < passare, "to go" (today păsa < pensare, "to weigh on, to press" [now preserved only in such expressions as nu-mi pasă]; urăciure < orationem, "wish, congratulation"; vipiu < victus, "wheat, corn, victuals"; vărgură < virgulum < virgo, "virgin, maiden"; opu iaste < opus est, "it is necessary, it is needed."  To these may be added several words that have preserved the original Latin meaning rather than that of modern Romanian: for example, codru (today "forest") with the sense of "mountain"; fămeaie (today "woman") with the sense of "family"; gǔdeţ (today "county") with the sense of "judgment"; săruta (today "to kiss") with the sense of "to greet, to salute." 
One must note that only a few centuries ago, the subdialect of Northern Romanian (and Northern Romanian in general) spoken in Maramureş and in adjacent areas of Moldavia, that is, in territories in which Roman settlements never existed, was much closer to Latin than any Northern Romanian dialect is today.
7. Returning to present-day Romanian, the most conservative dialect is Arumanian, which shows many archaic characteristics in all areas of language: phonology, morphology, syntax, and lexical elements. "Its archaic features bring Arumanian near to Ancient Romanian (română comună)."  Many of the archaisms found in present-day Arumanian still existed in the sixteenth century in Northern Romanian. In the following, a short description of the main archaic characteristics of this dialect will be given, based on the presentation by Matilda Caragiu Marioţeanu in Compendiu de dialectologie română (1975).
Phonology: The vowel system of the Farserot subdialect of Arumanian is identical to that of Ancient Romanian (română comună):
Arumanian has no i̯ before an initial e-, as Northern Romanian has through Slavic influence: Arum. eşti, eşţi, N. Rom. ieşti, "you are." The palatal consonants l', ń reappear: Latin leporem > Arum. l'épur, N. Rom. iepure. The vowels are not affected by the character of the vowel in the following syllable: Lat. pilu(m) > Ancient Romanian (română comună) *peru Arum. peru, N. Rom. păr. The final -u of Latin is preserved: Lat. lupu(s) > Ancient Rom. *lupu > Arum. lupu, N. Rom. lup. A characteristic feature of Arumanian (as well as of Meglenitic) is the treatment of the Latin velars c, g + e, i. In N. Rom., these developed to ĉ, ĝ; Arum. has ţ, ḑ : Lat. cepa > Arum. ţeápî, N. Rom. ceapă; Lat. gelu > Arum. ḑeru, N. Rom. ger. – Latin
c, - t, followed by io, iu, changed to ĉ, and j + o, u, and d + ió, iú changed to ĝ.
(An innovation of Arumanian is the prothesis of an a before words that begin with a consonant, particularly r, l : Lat. rivus > Arum. ariu̯, (N. Rom. rîu). The palatalization of the labials is general (all labials are affected in all subdialects) and total (there are no intermediary stages).
Morphology: The Latin plural suffix used in the words tata and mamma (tatanis > tatane, mammanis > mammane) is preserved: mumî, "mother"; mumîńi, "mothers." The dative is expressed by analytic means, with the preposition a (Lat. ad), which was also extended to the genitive.
The accusative of the place names is expressed without a preposition, as in Latin: (direction): mi ducu Hrúpişti "I go to Hrupişte," in Northern Romanian: mă duc la Hrupişte; (existence in a place): éscu Sîrúnî di dáu̯î ḑili "I have been in Saloniki for two days"; in Northern Romanian: sînt la Salonic de două zile. 
Another archaic feature is the fact that the accusative is expressed without pe (used in N. Rom. in certain cases—an innovation after the sixteenth century). The indefinite article is ună (uní) (< Lat. una): unî casî, unî fe̯atî (N. Rom. Istro-Rumanian have o). Arumanian has preserved Latin viginti "twenty": yńǵiţî, yíyinţî.
The verb: (The infinitive does not exist with a verbal value, only the long form, used as a noun.) The simple perfect is, the usual tense of the past in Arumanian, in contrast to Northern Romanian, where it is no longer used in speech (except in subdialects). The conjugation of the verbs shows the following archaisms: the third person plural indicative is etymological: Lat. cantabant > Arum. cinta; we find this in the sixteenth century texts of N. Rom.; present-day N. Rom. has, however, -au. The first and second person plural continue the Latin endings: -mu, tu: cîntămu, cîntatu (present-day N. Rom. has cîntarăm, cîntarăţi; but in the sixteenth century texts forms similar to Arum. are found.). Like Latin, Arumanian still has the two forms of the simple perfect: "strong" forms, with the stress on the stem, and "weak" forms, stressed on the inflexion. (The strong form also existed in the sixteenth century in N. Rom.) The auxiliary of the compound perfect (perfectul compis) is not grammaticized as in N. Rom.: Arum. amu, ai̯, ari, avému, aveţî, au̯ (N. Rom.: am, ai, a, am, aţi au). Arumanian has created a synthetic conditional from the confusion of the Latin prefect conjunctive (-verim) and the anterior future (viitorul anterior): -vero.
Lexical elements: Many lexical elements of Latin origin that have disappeared from the other dialects still exist in Arumanian.
M. Caragiu Marioţeanu gives the following "few examples as an illustration": 
Several words of Latin origin have, in Arumanian, a sense closer to the original Latin than is the case in N. Romanian; others show a special evolution of meaning (often influenced by the other Balkan languages). Fumeál'i, for example, means "family, children," as in Latin (familia); (Northern Romanian femeie, "woman"). The dialectal sense of june (this word is among those referred to by Puşcariu) exists also in Arumanian: ǧoni, "young boy," although it also has the meaning of "brave" and "bridegroom" (probably developed under Albanian and Serbian influence).
Arumanian has also preserved several Latin adverbs, prepositions, and conjunctions that do not exist in the other Romanian dialects: i̯u "where" (< Latin ubi; this word appears in the sixteenth century Northern Romanian texts but exists today only in the region of Crişana (i̯uă); díndi, didíndi "there, before" (< Latin de inde ± de); lárgu "far away" (< Latin largus); and many others. 
If the existence of ancient Latin elements in a Romanian dialect indicates that the speakers of that idiom have been living in their present-day area since the Roman period, then the Arumanians (together with the Meglenitic) would be the most likely candidates for this group. In other words, one could use Puşcariu's own arguments to claim that the present Arumanian territory is an ancient Roman - Romanian Kerngebiet, in which the Romance idiom has been spoken uninterruptedly ever since the time of the Roman Empire. The
reasoning would in any case be similar to that of Puşcariu for the region of the Apuseni Mountains. As is known, however, the Arumanians migrated to their present territory. Their Romance language could not, therefore, have originated there, far south of the territory influenced by Latin on the Balkan Peninsula, in the region of the Greek language. When migrating there, the ancesstors of the Arumanians took with them, of course, their Romance idiom; that they preserved the Latin elements to a much higher degree than did the Northern Romanians can be explained by their relative isolation in a lateral area but has nothing to do with the territory from which they originated.
Other Proponents of the Kerngebiet Theory
In light of these facts, it is peculiar that the Kerngebiet theory, instead of being criticized and rejected at an early stage, was advanced further. The German linguist Ernst Gamillscheg  and his disciple, Günter Reichenkron, accepted Puşcariu's theory and cited other linguistic phenomena that they felt would strengthen the hypothesis of ancient Romanian core areas, not only in the region proposed by Puşcariu but also in some areas along the Danube in Muntenia and along the Olt River.
Gamillscheg observed that native speakers in the region of the Apuseni Mountains put a k between the consonants s and l (also in words of Slavic origin): instead of slab ("weak"), they say sklab; instead of sloată ("sleet"), skloată. This phenomenon also occurred in Vulgar Latin, for example, insula > isla > iskla.
A variant of the diminutive suffix -culus: -unculus, was used in Late Latin in certain territories: in northern France, in the Alpine region (Noricum and Raetia), and among "a part of the ancient Romanians." The suffix now common in Northern Romanian originates from Latin -culus: for example, Romanian rinichiu, from Latin reniculus, a diminutive form of ren "kidney." In the region of the Apuseni Mountains, however, rărunchiu (and mănunchiu) forms derived from the suffix -unculus, are used. Gamillscheg is of the opinion that from this single circumstance it follows that the Romanian people must have come from at least two Kerngebiete: one was situated southward and re-established the connections with the south-Danubian Romanians; and the other was situated within the renunculus area, or, if the entire population that used renunculus immigrated, was geographically separated from the first named region. 
Like Puşcariu, Gamillscheg believed that there were other circumstances to support this theory. He still tried, for example, to find
inherited Latin place and river names in the region in question: Abrud, from Latin Abruttus; Ampoi, from Latin Ampeium; Turda, from a hypothetical Dacian Turidava; and the name of the Criş River from ancient Grisia. Moreover, Gamillscheg argued, the majority of the names of mountains were of Romanian origin: of 17 such names mentioned by Gustav Weigand,  13 were Romanian and 4 Hungarian.
The weakness of the theories based on the survival of the diminutive suffix -unculus, the insertion of a k between s and l, and similar isolated phenomena is that certain elements of language are taken out arbitrarily, without regard to the distribution of the subdialects in Transylvania within Northern Romanian. The consonant group skl is not specific to Vulgar Latin but also existed in Greek; and if this phenomenon appeared after the third century A.D. (according to Gamillscheg, it is attested to from the fifth century), one might also ask how it could reach an area outside the frontiers of the Roman Empire? If this phenomenon is connected with that found in some areas of Late Latin, it could indicate that some of the ancestors of the Romanians, living on the Balkan Peninsula within the Roman Empire, had acquired it; being a part of their speech habits, they took it with them in their later migrations. This also applies to the diminutive suffix -unculus.
On a close examination of linguistic facts it appears that Gamillscheg's additional arguments are based on obsolete assumptions. Romanian Abrud cannot have derived directly from the Latin Abruttus, since the Latin -b- disappeared in Romanian: e.g., Latin februa(ris) > Romanian făurar. The Latin Ampeium could not result in Romanian Ampoi, because the Latin a in this position changed to ă > â (cf., Latin campus > Romanian cîmp). *Turidava is pure, unattested conjecture. That the majority of the names of the mountains in this region are of Romanian origin can be explained by the fact that the Apuseni Mountains have been populated since the thirteenth century mainly by a peasant and pastoral Romanian population. The names of the mountains do not date from before the fourteenth century. (The river names of this region, like those of Transylvania in general, are of either Hungarian or Slavic origin).
Largely in agreement with Gamillscheg, Reichenkron  distinguished two Kerngebiete of Romanian north of the Danube: the region of the Apuseni Mountains and a strip of territory along the lower Danube, ("the Kerngebiet of Gamillscheg up to Cernavodă") to which Reichenkron added the course of the Olt River. He called this idiom "Geto-Romanisch." A third Romanian Kerngebiet was, in Reichenkron's opinion, in Dardania.
Besides advancing the theory of a Romanian Kerngebiet in Muntenia and along the Olt River, Reichenkron tried to find similarities to the phenomena cited by Puşcariu in the rest of the Romance languages. On the basis of the word-pairs ai and nea, and usturoi and zăpadă, he believed in an east-west division. The use of rărunchiu, mănunchiu, as well as that of curechiu in the northwest compared with the south, where rinichiu and varză are found, defines, in his opinion, a north-south division as well. This continues, according to Reichenkron, in the Western Romance languages (words derived from renuncuius are found in Northern France, in Noricum, and in Raetia, while reniculus is the basis of words denoting the kidney in the southern area of the Western Romance languages). The east-west division proposed by Reichenkron cannot apply to the Western Romance languages, but Reichenkron believed that it continued in Bulgarian. He referred to the tendency of the end vowels to closeness in the Moldavian subdialect (e > i, ă > î). The frontier line between this tendency and other areas of Northern Romanian goes from Czernovitz in the north along the Eastern Carpathians and reaches the Danube at Orlea, continuing toward the south in Bulgaria, since a similar phonetical tendency is found in Bulgarian to the east of this line.
These theories, however, are also based on a few arbitrarily selected linguistic phenomena. As shown above, the Latin image of Transylvania is artificial and does not exist in reality; and the Romanian subdialects do not support the idea of an "east-west division." The phonetical tendency to closeness of the end vowels is a peculiarity of the Moldavian subdialect. One of the bases for the north-south division theory in Northern Romanian is the corresponding word-pair curechiu and varză; but in the Western Romance languages we find only words that correspond to curechiu (derived from Vulgar Latin coliclus (< cauliculus). The other word, used there to designate "cabbage," has nothing to do with varză but is a continuation of the classical Latin caulis, which is not preserved in Romanian. The other example, the use of the diminutive suffix -unculus or -culus, does show something like a north-south division in the Western Romance languages; but in Romanian the frontiers of such a division should be sought along the line of the lower Danube, the frontier of the Roman Empire for several centuries after the Roman retreat from Dacia Traiana, if indeed the ancestors of the Romanians had been living in present-day Romania during the time of Late Latin when the use of the diminutive suffixes became widespread. This also applies to the distribution of curechiu and varză, cited in support of a north-south division and to Northern Romanian in general. If the Romanian language had developed, from the period of Dacia Traiana onwards, both north and south of the
Danube, an important dialectal frontier should go along the Danube, as the consequence of the different historical developments of the territories north and south of the river.
The Kerngebiet Theory and the Romanian Dialects
The hypothesis of ancient Romanian Kerngebiete in Dardania, in parts of Muntenia along the Danube and Olt rivers, and in western Transylvania assumes that in those territories Latin and, later, Romanian, have been spoken since the time of the Roman Empire. This situation should have left some traces in the present-day dialects of the Romanian language. Romanian spoken in the central parts of the Balkan Peninsula (in the territory between Skoplje, Niš, and Sofia, that is, in a part of present-day Macedonia) should show some differences in comparison with that spoken in Muntenia, along the Danube; the dialect found in western Transylvania would differ from both. The differences would not necessarily be great; one might also argue that no significant dialectal differences are to be expected because of a decisive influence (certainly exaggerated) from the wanderings of the shepherds which acted as a leveling factor. The wanderings and migrations could have wiped out differences or prevented them from developing.
The real situation, however, is neither a dialectal differentiation according to the assumed Kerngebiete nor a more or less uniform language. There are dialects: there is Northern Romanian ("Daco-Romanian") in present-day Romania and south of the Danube in the Timok Valley. During the Middle Ages, the population speaking this dialect occupied a much larger territory of the central Balkan Peninsula. This dense population of Northern Romanians is attested to by written documents (hrišovs) of the Serbian kings and monasteries written in the twelfth through the fifteenth century, in which personal names of a Northern Romanian pattern appear. In the central parts of the Balkan Peninsula, the Northern Romanians left their traces in the shape of place names and names of mountains, which still exist in the Serbian and Bulgarian toponymy, especially in the region bordered by the towns of Prizren, Skoplje, Sofia, and Niš, but also north and northwest of this region. The clearly Northern Romanian (not Arumanian) sound pattern of these names is still discernible. Part of this population migrated toward the northwest from the twelfth to the fifteenth century and reached the Istrian Peninsula. The Istro-Rumanians living there today are their descendants; they speak a special dialect, related more to Northern Romanian than to the southern dialects.
The southern dialects of the Romanian language are Arumanian and Meglenitic. The speakers of these dialects are now living on both sides of the Greek-Yugoslav frontier and in parts of Albania and Bulgaria. Since these territories are situated to the south of the ancient frontier between Greek and the Latin languages (the Jirecek or the Skok line), their original areas must have been further to the north.
To this comes the fact that until the eleventh century, Romanian was essentially a uniform language, with no significant dialectal variation. The reconstruction of this stage of Romanian (străromână, română comună, Ancient Romanian) shows, for example, that it was much closer to Latin than the present-day dialects (of which Arumanian is closest to Ancient Romanian) and that the Slavic words borrowed before the eleventh century are the same in all four Romanian dialects.  The existence of such a stage fits in badly with assumed Kerngebiete a considerable distance from one another.
Dardania and the Origin of Romanian
Reichenkron, like Gamillscheg, recognized the importance of the province of Dardania in the history of Latin on the Balkan Peninsula and in the development of the Romanian language. Reichenkron emphasized that Dardania was intensely Romanized and the cultural center of a large area. In 535 A.D. Emperor Justinian made Iustiniana Prima (Scupi, modern Skoplje) the see of the archbishopric, to which belonged the dioceses of Dacia Ripensis and Mediterranea, Moesia Superior, Praevalis, Macedonia Secunda, and the eastern part of Pannonia Inferior.  Dardania wras also the territory in which the Albanians were living during Roman times on the Balkan Peninsula,  and Reichenkron emphasizes the significance of the close and multiple connections between Albanian and Romanian.  To all this comes the Latin vocabulary of Christianity in the Romanian language, which the ancestors of the Romanians could only have adopted in a territory in which the Latin language dominated. 
On the basis of these circumstances, Reichenkron stated, "it was probably from Dardania that the first expansion of the ancient Romanians ["Urrumanentum"], as they have been called since Puşcariu, occurred."  With regard to the common lexical elements found in Albanian and in Romanian, Reichenkron considered that only a smaller part of them might be loans: It is more probable that most of the Albanian-Romanian correspondences in the field of the lexical elements developed south of the Danube, in so-called Dardania in the southern part of the Roman province of Moesia Superior, thus, approximately in Yugoslav Macedonia. 
The Kerngebiet Theory in Romanian Historiography
Romanian linguists no longer defend the Kerngebiet theory. Historians, however, continue to support it. According to Constantin C. Giurescu, for example,"the maps of the Romanian Linguistic Atlas (Atlasul Lingvistic Român) show the presence of some terms of Latin origin, such as nea, pedestru, and june, (which are found) only in the western parts of Transylvania, and (are) lacking in the rest of the Carpatho-Danubian area—something that would not be possible if the ancient Romanians had come from the Balkan Peninsula''  In the opinion of Ştefan Pascu, "also from the linguistic point of view, the German scholar Gamillscheg has identified a Romance group between Giurgiu and Cernavodă (Muntenia). 
A Romanian scholar living in the West, Vasile Arvinte, refers to the Kerngebiet theory as an important proof of Roman continuity north of the Danube. He asserts that recent archaeological excavations "confirm the conclusions of the linguists." The assumed Kerngebiet along the lower Danube and in Dobrudja, for example, is said to be confirmed by the Dridu culture, which proves the ancient Romanian element.  Eugen Lozovan pointed out the weaknesses of Puşcariu's theory: Puşcariu and his followers have committed two major errors: first, to change from the actual dialectal synchrony (opposition: archaic areas/innovating areas) to diachrony or to an earlier synchrony; secondly, to raise diachrony into a system, because to accept the idea of conservative areas as a more or less exact picture of an earlier phase means to accept the concept of the stability of diachrony, which is absurd. 
The Romanian linguist Alexandru Rosetti refutes the theory of ancient Romanian areas based on Latin words that were replaced in other areas. In his opinion, this geographical distribution may have occurred as well in a later period: "a series of other Latin terms that appear in the Romanian Linguistic Atlas subvert the proposed Latin image of Transylvania by their presence in Wallachia and Moldavia and prove, by their appearance in regions that were not Romanized, that later linguistic extensions are involved." 
The Kerngebiet theory no longer seems to be mentioned in Romanian textbooks or linguistic studies. No reference is made to this theory, for example, in the 1969 edition of Istoria limbii române,  in a recent history of the Romanian language,  or in Studii de dialectologie română.  Matilda Caragiu Marioţeanu gives a detailed and comprehensive analysis of the Romanian dialects and subdialects.  It is
noteworthy that the author does not mention the Kerngebiet theory: the dialectal characteristics taken out from a vast material of linguistic phenomena to support this theory show themselves to be insignificant details, a small share of the examples of linguistic change, and conservatism mostly connected with the situation of lateral areas.
[Back to Index]
1. Nicolae Gudea, "Cîteva observaţii şi note critice cu specială privire la partea istorică a monografiei Etnogeneza românilor de I.I. Russu" [Some Remarks and Critical Notes With Special Reference to the Historical Part of Etnogeneza românilor by I.I. Russu], in Acta Musei Napocensis, XX, 1983, pp. 903-916.
2. A review about this book was published in a linguistic journal: Studii şi cercetări lingvistice, Bucharest, XXXIII, 1982, 3, pp. 278-282.
3. Ion I. Russu, Etnogeneza românilor. Fondul autohton traco-dacic şi componenţa latino-romanică [The Ethnogenesis of the Romanians. The Autochthon Thraco-Dacian Ethnical Basis and the Latin-Romance Element], (Bucharest: 1981) p. 155.
4. Ibid., p. 196.
5. Gudea, 1983, p. 904.
6. Liviu Franga, in Studii şi cercetări lingvistice, XXXIII (1982), 3, p. 279.
7. Iancu Fischer, Latina dunăreană [Danubian Latin], (Bucharest: 1985).
8. Ibid., p. 16.
9. Ibid., p. 20.
10. K. Sandfeld, Linguistique balkanique (Paris: 1930); Alexandru Rosetti, Istoria limbii române (Bucharest: 1978), 2nd ed., pp. 247-289.
11. Actes du premier congres international des linguistes, (Leiden: 1928), p. 18.
12. Georg Renatus Soita, Einführung in die Balkanlinguistik mit besonderer Berücksichtigung des Substrats und des Balkanlateinischen, (Darmstadt: 1980), p. 7.
13. Siegfried Riedl, "Der Artikel im Bulgarischen," in: 1300 Jahre Bulgarien. Studien zum I Internationalen Bulgaristikkongress Sofia 1981, p. 335, note 47, cf., Trost P., "Zur Kritik der Substrattheorie," in: Les études balkaniques tchéchoslovaques III, 1968, p. 48.
14. Alexandru Rosetti, Étude sur le rhotacisme en roumain, (Paris: 1924).
15. Matteo Giulio Bartoli, "Das Dalmatische," in: Schriften der Balkankommission, vols. IV-V, (Wien: 1906); sardo, dalmatico, albano-romanico. Atti del IV Congr. nazion. dei arti e tradizioni populari, Roma 1942.
16. A presentation of this problem is given, for example, by B.E. Vidos, Handbuch der romanischen Sprachwissenschaft (München: 1975), pp. 319-335.
17. Alexandru Rosetti, Istoria limbii române, 1986, p. 78.
18. Haralambie Mihăescu, Limba latină în provinciile dunărene ale Imperiului Roman [The Latin Language in the Danubian Provinces of the Roman Empire], (Bucharest: 1960).
19. At the official division of the empire in 395 A.D. an eastern and a western part, Dalmaţia remained with the western part and Preavalitana (Montenegro and northern Albania) went to the eastern part. The frontier between the two territories went from the gulf of Cattaro to the area west of Belgrade. Since most of Dalmaţia belonged for almost another one and a half centuries to the sphere of influence of Rome (in 535 A.D. it was conquered by Byzantium), it was able to take part in several linguistic developments that did not affect the rest of East Latin. Cf., B.E. Vidos, Handbuch der romanischen Sprachwissenschaft, first edition 1968, p. 300, referring to M. Valkhoff, Latijn, Romaans, Roemeens (Amersfoort: 1932), pp. 18-19.
20. Mihăescu, Limba latină, 1960, op. cit., p. 267.
21. Ibid., p. 36.
23. Ibid., p. 278.
24. Istoria limbii române, II, ed. Coteanu et al., (Bucharest: 1969), chapter A, pp. 21-186. The Istoria limbii române will henceforth be referred to as the ILR.
25. Iancu Fischer, Latina dunăreană, op. cit., p. 6.
26. Ibid., p. 8.
27. There are exceptions to this: Lat. autumn Northern Romanian toamnă, and others.
28. Mihăescu, Limba latină, 1960, op. cit., p. 67; Rosetti, ILR, 1968, p. 108.
29. Rosetti, Istoria limbii române 1986, pp. 108-109.
30. Ibid., p. 78.
31. V. Väänänen, Introduction au latin vulgaire, 3rd edition, 1981, p. 57.
32. Rosetti, Istoria limbii române 1986, pp. 119-120.
33. Ibid., pp. 120-121.
34. Väänänen, 1981, op. cit., p. 68.
35. Fischer, Latina dunăreană, op. cit., p. 199.
36. Väänänen, 1981, p. 68; Fischer, op. cit., p. 199.
37. Fischer, p. 199.
38. Väänänen, 1981, pp. 69-70.
39. Rosetti, Istoria limbii române, 1986, p. 121.
40. Fischer, 1985, op. cit., p. 202.
41. ILR, 1969, II, p. 118. The exact number of such words is given: 107.
42. Ibid., p. 123.
43. Ibid., p. 124,
44. Ibid., p. 124 et seq.
45. Rosetti, Istoria limbii române, 1986, p. 79.
46. Ibid., p. 82.
47. Or are conservative traits as compared to innovations that appeared in the West in those periods. Väänänen's division of the ages of Latin is used here, cf., Väänänen, 1981, p. 13.
48. Väänänen, 1981, op. at., p. 200. Väänänen quotes C.A. Robson, "L'Appendix Probi et la phiiologie latine" in: Le Moyen Âge, 69, (1963), pp. 37-54. Robson shows that, in contrast to what has been assumed earlier, this text was written "in the Christian centuries, under Lombards, thus, after 568!"
49. Väänänen, 1981, pp. 17-18.
50. Rosetti, Istoria limbii române, 1986, p. 104.
51. Mihăescu, 1960, op. cit., p. 96.
52. Rosetti, 1986, op. cit., p. 114.
53. Rosetti, 1986, p. 343, considers that Arumanian ts is more recent and developed, as stated by Skok, Zeitschrift für romanische Philologie, XLVIII, under Greek influence. (In certain areas of Arumanian, c appears).
54. Haralambie Mihăescu, La langue latine dans le sud-est de l'Europe, (Bucharest-Paris: 1978), p. 198.
55. Rosetti, 1986, op. cit., pp. 116-117.
57. Iancu Fischer, "Aspectul linguistic al romanizării Daciei" [The Linguistic Aspect of the Romanization of Dacia], in Limba română, 27, 2, 1978, pp. 190-191. See also I. Fischer, Latina dunăreană, 1985, op. cit., pp. 200-201; Rosetti, 1968, op. cit., 136-148.
58. ILR, 1969, II, p. 232.
59. Fischer, 1985, op. cit., pp. 201-204; Rosetti, 1968, pp. 148-165.
60. Mihăescu, Limba latină, 1960, op. cit., p. 145.
61. Ibid., p. 146.
62. Matilda Caragiu Marioţeanu, Compendiu de dialectologie, op. cit., p. 249.
63. E. Löfstedt, Late Latin. Instituttet for sammenlignende kulturforskning, serie A: Forelesninger, XXV (Oslo: 1959), p. 68: "The new system of thought called for and created not a quite new language but certainly new forms of expression."
64. Mihăescu, 1960, op. cit., p. 277.
65. Cf., for example, Istoria Romîniei, I (Bucharest: i960), p. 631.
66. Fischer, 1985, op. cit., p. 54.
67. Ibid., p. 55. The text of Appendix Probi is reproduced in Väänänen, 1981, op. cit., pp. 199-203.
68. G. Mihăilă, Studii de lexicologie şi istorie a lingvisticii româneşti (Studies of Lexicology and of History of Romanian Linguistics], (Bucharest: 1973), p. 16.
69. Löfstedt, Late Latin, 1959, op. cit., p. 17.
70. ILR, II, 1969, pp. 15-16.
71. Rosetti, 1986, op. cit., p. 81.
72. Mihăescu, 1960, op. cit., p. 278.
73. Väänänen, Introduction au latin vulgaire, 1981, op. cit., p. 83.
74. ILR II, 1969, p. 15.
75. H. J. Dölle (red.), Römer und Germanen in Mitteleuropa, (Berlin, East Germany: 1975), p. 88.
77. Mihăescu, 1960, op. cit., pp. 267-268; Rosetti, 1986, op. cit., p. 82; Russu, Etnogeneza românilor, op. cit., p. 199.
78. G. Straka, in Revue de linguistique romane, Paris, XXIV, 1960, Livres, comptes rendues sommaires, p. 405.
79. For more detail on this subject in: Werner Bahner, Das Sprach- und Geschichtsbewusstsein in der rumänischen Literatur von 1780-1880, (Berlin: 1967).
80. Ion I. Russu, Etnogeneza românilor [The Ethnogenesis of the Romanians], (Bucharest: 1981), p. 125. The Dicţionarul explicativ al limbii române, (Bucharest: 1975) considers that the origin of this word is not known.
81. Russu, Etnogeneza românilor, pp. 118-121.
82. ILR, pp. 319-320.
83. The Early History of the Rumanian Language (Lake Bluff: 1977) p. 47.
84. The details and further literature on the subject can be found in several recent studies: Ivan Popović, Geschichte der serbokroatischen Sprache (Wiesbaden: 1960); Ion I. Russu, Limba traco-dacilor [The Language of the Thraco-Dacians], (Bucharest: 1967); id.: Illirii. Istoria—limba şi onomastica—romanizarea [The Illyrians—History, Language and Onomastics, Romanization], (Bucharest: 1969); Id.: Etnogeneza românilor [The Ethnogenesis of the Romanians], (Bucharest: 1981); Georg Renatus Soita, Einführung in die Balkanlinguistik mit besonderer Berücksichtigung des Substrats und des Balkanlateinischen, (Darmstad: 1980); Vladimir Georgiev, The Thracians and Their Language [in Bulgarian], (Sofia: 1977); Gottfried Schramm, Eroberer und Eingesessene. Geographische Lehnnamen als Zeugen der Geschichte Südosteuropas im ersten Jahrtausend n. Chr. (Stuttgart: 1981).
85. Russu, 1981, op. cit., p. 75.
86. Russu, Illirii, 1969, op. cit., p. 107. In ancient Indo-European, *kmtom "hundred" was spelled with a palatal k. In an ancient Indo-European dialect, this palatal k changed to č which in certain areas later became s. In Avestan (an ancient Iranian idiom), "hundred" is satem. Those Indo-European languages, including the Indo-Iranian, Armenian, Albanian, and Balto-Slavic subfamilies, which originate from this dialect with č are called "satem languages," the rest, "centum languages."
87. Solta, 1981, op. cit., pp. 119-120. Cf., also Popović, 1960, op. cit., p. 65.
88. Geographia, first century B.C-first century A.D.
89. Historia, second century B.C.
90. Russu, 1981, op. cit., p. 79.
91. Detailed data are provided in: Dimiter Detschew (ed.), Die thrakischen Sprachreste, (Wien: 1957).
92. Russu, 1981, op. cit., p. 81; in Limba traco-dacilor, 1967, pp. 89-130, Thracian (Thraco-Dacian) proper names and lexical elements are given, with assumed etymologies.
92a. Russu, 1981, op. cit., p. 81. In Russu's opinion, the lexical elements from which a consonant shift was deduced are in reality Grecized forms of Thracian names. Other scholars do not agree with this; cf., for example, I. Popović, 1960, op. cit., p. 74, who cites such forms as Μήτοκος – Μήδοκος, Κραστονία – Γραστουία.
93. ILR, 1969, op. cit., p. 314.
94. C. Váczy, "Nomenclatura dacică a plantelor la Dioscorides şi Pseudo-Apuleius," in: Acta Musei Napocensis (Cluj), part IX, 1972, p. 107.
95. Solta, 1980, op. cit., p. 19.
96. Ibid., p. 20.
98. ILR, 1969, pp. 315-316. See further, Dimiter Detschev, Die thrakischen Sprachreste, op. cit., pp. 542-565.
99. Ariton Vraciu, Limba daco-geţilor [The Language of the Daco-Getae], (Timişoara: 1980), p. 85.
100. Geographia, VII, 3, 10, 3, 13.
101. ILR 1969, p. 318, note 2.
102. SCIVA, vol. 28, 1, 1977, p. 15.
103. Linguistique Balkanique, II, 1960, pp. 1-19; Vladimir Georgiev, The Thracians and Their Language [in Bulgarian], (Sofia: 1977); Solta, 1980, op. cit., p. 117.
104. Cf., the critical discussion by Solta, 1980, op. cit., pp. 117-118.
105. Russu, 1981, op. cit., p. 79.
106. ILR, 1969, p. 317.
107. Solta, 1980, op. cit., p. 11.
108. Günter Reichenkron, Das Dakische (rekonstruiert aus dem Rumänischen), (Heidelberg: 1966), p. 92.
108a. Russu, 1981, op. cit., p. 62.
109. Ibid., p. 63.
110. Russu, Limba traco-dacilor, 1967, pp. 138-143, has compiled the Thracian lexical data with a probable or assumed sense. The seven words here are from his list.
111. Ibid., p. 112.
112. ILR, vol. II, 1969, p. 327 et. seq.
113. Ibid., p. 333.
114. Explanatory Dictionary of the Romanian Language, (Bucharest: 1975).
115. The Slavic origin of zîrnă is not probable, the meaning of the Slavic word is far from that of zîrnă and Romanian zîrnă is in its first written forms written with initial dz (Dzărnă, in a document from Suceava, dated 1488). Cf., Rosetti: DR. zîrnă "plante vénéneuse," in: Mélanges linguistiques (Bucharest: 1977), p. 159; in Rosetti 1985, p. 322.
116. Russu, 1981, op. cit., p. 275.
117. Ibid., p. 323.
118. Ibid., p. 346.
119. Ibid.. p. 113.
120. Ovid Densusianu, Histoire de la langue roumaine, 1901, I, p. 20, quoted by Russu, 1981, p. 47: "La phonétique et le lexique roumains ne nous offrent aucune particularité qui se trouve en mérne temps dans les restes de la langue dace, qui nous ont été transmis. Il serait inutile de batir des hypothèses fantastiques et de chercher des éléments daciques en roumain. On ne surait toutefois contester l'existence de tels éléments, mais tout philologue doit y renoncer à les admettre lâ où ils ne peuvent pas être prouvés par la science."
121. Russu, 1981, op. cit., pp. 115-116.
122. Ibid., p. 166.
124. Ibid., p. 114.
125. Gustav Weigand, "Sind die Albaner die Nachkommen der Illyrer oder der Thraker?," in Balkan-Archiv, 3, 1927, pp. 227-251; Dimităr Decev, Charakteristik der thrakischen Sprache, (Sofia: 1952); Id.: Die thrakischen Sprachreste, (Wien: 1957); Henrik Baric, "Albanisch, Romanisch und Rumänisch," in: Godišnjak, ed. Balkanološki Institut (Sarajevo: 1956), pp. 1-16.
126. An up-to-date presentation of this problem is given by Solta, 1980, op. cit., pp. 108-123.
127. Eqrem Çabej, "L'illyrien et l'albanais," in Studia Albanica, VIII, 1, 1970, pp. 157-170.
128. Russu, Limba traco-dacilor, 1967, p. 183.
129. Eqrem Çabej, "De quelques problémes fondamentaux de l'histoire ancienne de l'Albanais," Conférence des études albanologiques, Université d'état de Tirana, Institut d'Histoire et de Linguistique (Tirana: 1962), p. 6.
130. Ibid., p. 14.
132. Solta, 1980, op. cit., p. 36.
133. Russu, Illirii, op. cit., p. 111. In Etnogeneza românilor, 1981, p. 94, a similar statement is found: the phonetics of Albanian agrees with both Illyrian
and Thracian, "but in their lexicology, no common element can be found; the Illyrian and Thraco-Getian grammars [declension and conjugation] are entirely unknown, and the Albanians do not seem to have inherited any cultural element [for example, a proper name or a place name] attested to in the ancient era, in Thracian or in Illyrian."
134. Vladimir Georgiev, The Thracians and Their Language (in Bulgarian). (Sofia: 1977), p. 286. Cf., also ILR, 1969, p. 317.
135. Gottfried Schramm, Eroberer und Eingesessene, 1981, op. cit., p. 34.
136. Ivan Popović, 1960, op. cit., p. 79.
137. Çabej, "Le problème du territoire de la formation de la langue albanaise," in Bull. Association International d'Études Sud-Est Européennes, 1972, 2, pp. 87-89.
138. Popović, 1960, op. cit., p. 84.
139. Cf., for example, Popović, 1960, op. cit., p. 80.
140. Cf., Soita, op. cit., p. 115.
141. Popović, 1960, op. cit., p. 84; Solta, 1980, op. cit., p. 115; Schramm, 1981, op. cit., p. 34.
142. G. Brâncuş, "Albano-romanica, III. Vocala ă în română şi albaneza," [The vowel ă in Romanian and Albanian], in Studii şi cercetări lingvistice, XXIV, 3, 1973, Bucharest, p. 294.
143. Ibid., p. 295. A recent presentation of this subject was given by Solta, 1980, pp. 180-183.
144. Solta, 1980, op. cit., p. 184, quoting V. Georgiev, Linguistique Balkanique XX, 1977, p. 8.
145. Çabej, "Unele probleme ale istoriei limbii albaneze," in Studii şi cercetări lingvistice X, 4, 1959, p. 534; Soita, 1980, op. cit., p. 191.
146. Ibid., p. 193.
147. Çabej, "Unele probleme," op. cit., 1959, p. 531.
148. Solta, 1980, p. 189.
149. Ibid., pp. 195-196.
150. ILR, 1969, vol. II, p. 326. The expression "autohton" is used.
151. Ibid., p. 327.
152. Ariton Vraciu, Limba daco-geţilor, op. cit., p. 165.
153. Russu, 1981, op. cit., p. 110.
154. Ibid., p. 111.
155. The method is far from having been generally accepted; cf., for example, C. Tagliavini, Originile limbilor neolatine, (Bucharest: 1977), p. 113, note 174; p. 120.
156. Russu, 1981, p. 111.
157. Ariton Vraciu, Limba daco-geţilor, 1980, op. cit., p. 88, pp. 57-58, 61, note 3. See also Vasile Arvinte, Die Rumänen. Ursprung, Volks- und Landesnamen [The Romanians. Origin, Ethnical- and Landnames], (Tübingen: 1980), pp. 14-15.
158. Russu, 1981, op. cit., p. 112.
159. Ibid., pp. 245-426.
160. Cicerone Poghirc, ILR, 1969, op. cit., pp. 319-320; and pp. 327-356; see also Istoria limbii române, Florica Dimitrescu et. al, (Bucharest: 1978), p. 72.
161. The Explicative Dictionary of the Romanian Language, (Bucharest: 1975).
162. ILR, 1978, p. 73, notes 1 and 2.
163. Vraciu, Limba daco-geţilor, op. cit., pp. 141-142.
164. Russu, 1981, op. cit., p. 132.
165. Ibid. The last statement is quite dubious: brînză could be connected with Albanian bréndésa-t; zár with Albanian drä (and zär and zară are connected); urdă has an Albanian counterpart: urdhé, and it is not clear whether it is a loanword or not in Albanian.
166. Russu, 1981, op. cit., p. 132.
167. Ibid., p. 253.
168. Ibid., p. 295.
169. Ibid., p. 355.
170. ILR, 1969, vol. II, p. 352,
171. Russu, 1981, op. cit., pp. 244-245.
172. In the following analysis, all numbers given in parentheses and with a question-mark refer to words whose origin from the substratum is considered questionable by Russu.
173. Of which, however, four are questionable. The real number of specific pastoral words not existing in Albanian may thus be even lower.
174. Ion I. Russu, Dacoromania, 1973, pp. 191-192; also referring to Etnogeneza românilor, 1981.
175. Russu, 1981, pp. 245-426.
176. For example, note 158,
177. Theodor Capidan, Dacoromania II, 1921-22, pp. 482-487, quoted by Russu, Dacoromania 1973, p. 191; Russu, 1981, p. 135.
178. Russu, Etnogeneza românilor, 1981, op. cit., p. 137.
180. ILR, 1978: Viorica Ilea-Pamfil, p. 72.
181. Eqrem Çabej, "Unele probleme ale istoriei limbii albaneze," op. cit., 1959, p. 551: ". . . one can presuppose a non-Latin substratum, in the Romanian language, but not in Albanian. What for Romanian can represent the substratum is, for the Albanian language, only an ancient, earlier stage," Vladimir Georgiev (1977, p. 287) also considered that Proto-Albanian was the substratum of Romanian and called this ancient language "Daco-Moesian": "The Daco-Moesian tribes probably infiltrated into the central parts of the Balkan Peninsula (Moesia Superior, Dardania, Dacia Mediterranea, Dacia Ripensis) during the second millennium B.C. (the Dardanians), but also later, in successive waves, during the first millennium B.C. (the Triballes), and even later. Some of these tribes were Romanized during the first centuries A.D. but others were able to preserve their language, which developed under the powerful influence of Latin spoken in the eastern part of the Balkans. This
coexistence of the Romanized and the Proto-Albanian populations continued in the period from fourth to the sixth centuries, when Latin spoken in the eastern part of the Balkan Peninsula began to develop into Romanian."
182. Cf., the last sentence in the above note (V. Georgiev, 1977, p. 287): "This coexistence of the Romanized and the Proto-Albanian populations continued in the period from the fourth to the sixth centuries, when Latin spoken in the eastern part of the Balkan Peninsula began to develop into Romanian."
183. The vowel ă and the diphthongs, the phonemes /č/, /ǧ/, /ţ/, /š/ , existed in this idiom; the definite article appeared after the seventh century. Certain changes that did occur in East Latin developed at this stage in the pronoun as well as in the future tense, with the auxiliary verb uolo and the subjunctive with se (modem Northern. Romanian să). A detailed description of Common Romanian, as reconstructed from the present-day Romanian dialects, is given by M. Sala in ILR, 1969, vol. II, pp. 189-309.
184. Cf., for example, ILR, 1978, p. 76, where the following examples are given: babă, "old woman, mother," boală, "disease," nevastă "married woman, wife," scump "expensive, dear, valuable," sută "hundred," rană "wound," lopată "shovel," plăti "to pay."
185. Cf., for example, Popović, 1960, op. cit., p. 200. Popović points out that these loans, with "some distortions and changes, prove that there was one single point of borrowing."
186. Cf., for example, Solta, 1980, op. cit., p. 70 and 113.
187. Sextil Puşcariu, Die rumänische Sprache, (Leipzig: 1943), p. 396; Cf., Horedt, Siebenbürgen im Frühmittelalter, op. cit., note 259.
188. Alexandru Rosetti, Boris Cazacu, and Ion Coteanu (eds.), (Bucharest: 1965 and 1969).
189. Alexandru Rosetti, (Bucharest: 1968).
190. A. Rosetti, B. Cazacu, and I. Coteanu (eds.), vol. II, pp. 110-116.
191. This list was constructed on the basis of the dictionary of Ernout-Meillet and checked by the authors of ILR.
192. Cf., Ovid Densusianu, Histoire de la langue roumaine, Opere II, Lingvistica, ed. by V. Rusu et al., (Bucharest: 1975), p. 218.
193. Solta, 1980, op. cit., p. 131.
194. Haralambie Mihăescu, "Les éléments latins de la langue albanaise," I, in: Revue des etudes sud-est européennes 4, Bucharest, 1966, pp. 5-34; II pp. 323-354; quoted p. 27.
195. Cf., for example, Soita, 1980, op. cit., p. 125.
196. Solta, Einführung in die Balkanlinguistik, 1980, op. cit., p. 128. Solta also points out that it is wrong to believe that only common innovations are conclusive for the establishment of close relations between different idioms; quoting C. Watkins (Ancient Indo-European Dialects, Berkeley and Los Angeles, 1966, p. 30): "At any given stage of language, retentions and innovations are part of the same synchronic structure."
197. Solta, 1980, op. cit., p. 129: "In spite of Mihăescu, the existence of the Albanian-Romanian isoglosses must be admitted."
198. Florica Dimitrescu et al (eds.) Istoria limbii române, (Bucharest: 1978). p. 32.
199. Ştefan Pascu, (ed.), Istoria României. Compendiu, 3rd ed., 1974, p. 68. Unlike the first edition (1969), this text also includes in the area of formation of Romanian those territories of present-day Romania that never belonged to the Roman Empire (parts of Transylvania and most of Muntenia and Moldavia). We are dealing here with the phenomenon of a contemporary tendency to stretch the linguistic territory over the largest geographic area possible. Among others, for example, Constantin C. Giurescu, Dinu .C Giurescu, Istoria românilor, 1975, p. 30.
200. Cf., for example, Florica Dimitrescu et al (eds.). Istoria limbii române, (Bucharest: 1978), p. 30.
201. Among others, Ovid Densusianu, Alexandru Philippide, Sextil Puşcariu. A recent monograph about the Romanian dialects: Matilda Caragiu-Marioţeanu, Compendiu de dialectologie română (nordşi sud-dunăreană), (Bucharest: 1975).
202. Alexandru Rosetti, Sandra Golopenţia Eretescu (eds.), "Current Trends in Romanian Linguistics," in Revue Roumaine de Linguistique, 23, Cahiers de Linguistique Théorique et Appliquée, 15; History of Romanian, by Florica Dimitrescu and Mihaela Mancaş, (Bucharest: 1978), pp. 7-8.
204. Rosetti, Istoria limbii române, 1986, op. cit., p. 75; see also Ion Nestor, "Les données arhéologiques et la probleme de la formation du peuple roumain," in Revue Roumaine d'Histoire, III, 1964, pp. 383-423 and 407-410.
205. Ion I. Russu, 1981, op. cit., p. 166.
206. Ibid. Russu returns to the question of terminology on p. 164: "If the terms "Daco-Roman," "Thraco-Dacian," "Carpathian," etc., were frequently used on the preceding pages, they should not be interpreted as a silent pleading for the autochthony and the ethnic-social and territorial "continuity" of the Romanians, are used for the simple reason that they have been in use before and are more convenient and clear, covering certain historical-linguistic notions generally admitted by scholars, notions for which other, more accurate and unequivocal terms have been sought but not found." (Russu, Etnogeneza românilor, 1981, p. 164).
207. Iancu Fischer, "Aspectul lingvistic al romanizării Daciei" [The Linguistic Aspect of the Romanization of Dacia], in Limba română 27, no. 2, 1978, pp. 189-191.
208. Du Nay, 1978, p. 88.
209. Ion Coteanu, Structura şi evoluţia limbii române (de la origini pînă la 1860) [Structure and Evolution of the Romanian Language (from Its Origins to 1860)], (Bucharest: 1981), p. 74.
212. Cf., for example, Alexandru Vulpe, in Dicţionar de istorie veche a României (Paleolitic-sec X) [Dictionary for the ancient History of Romania], ed. by D.M. Pippidi, (Bucharest: 1976), p. 218: "The last mention of Dacians
is from the fourth century A.D., their language disappeared probably in the sixth to seventh centuries."
213. Cicerone Poghirc, in ILR, vol. II, p. 314.
214. Ovid Densusianu, Opere [Works], ed. by B. Cazacu, V. Rusu and I. Şerb, II: Lingvistica. Histoire de la langue roumaine, (Bucharest: 1975), p. 866.
215. Cf., for example, Rosetti, Istoria limbii române, 1968, op. cit., pp. 239-243 with the presentation of earlier theories (mostly of C. Diculescu, G. Giuglea, S. Puşcariu and E. Gamillscheg).
216. Virgiliu Ştefănescu-Drăgăneşti, "A New Look at the Socio-Linguistic and Historical Implications of the Latin Borrowings in Wulfila's Gothic Bible (fourth century A.D.)," in Forum Linguisticum, (Lake Bluff, USA), vol. VI, 3, 1982, pp. 265-269.
217. Barbarian peoples of the Roman border-area.
218. Vittoria Corazza, "Le parole latine in gotico," in Atti della Accadémia Nazionale dei Lincei (Roma: 1969), VIII, vol. XIV, Fa sc. 1.
219. Ibid., p. 7.
220. Ibid., pp. 77-78.
221. Ibid., p. 80.
222. Ibid., p. 85.
223. This assumption is from Ştefănescu-Drăgăneşti, 1982, op. cit., p. 267.
225. Ibid., p. 268.
226. Cf., for example, Rosetti, Istoria limbii române, 1968; Istoria limbii române, ed. A. Rosetti, B. Cazacu and I. Coteanu, vol. II, 1969; for a discussion of the origin of a găti, see p. 341.
227. Eqrem Çabej, "Unele probleme ale istoriei limbii albaneze" [Some Problems of the History of the Albanian Language], in Studii şi cercetări lingvistice, 1959 X, p. 531.
228. Ernst Gamillscheg, Romania Germanica, vol. II, (Berlin-Leipzig: 1935), p. 245 et seq. Gamillscheg also proposed Old Germanic origins for 30 Romanian words.
229. Iorgu Iordan, Nume de locuri româneşti în Republica Populară Română [Romanian Place Names in the Romanian People's Republic], (Bucharest: 1952), p. 230, quoting Gustav Kisch who published several studies on Transylvanian place names in the period between the two World Wars.
230. Iordan, 1952, op. cit., p. 230.
231. Rosetti, 1968, op. cit., p. 241.
232. Iordan, 1952, p. 230.
234. Ibid., p. 231.
235. Rosetti, 1968, p. 240.
236. Ibid., p. 327.
237. Cf., G. Schramm, Eroberer und Eingesessene, 1981, op. cit., p. 101; and G. Schramm, "Frühe Schicksale der Rumänen," in Zeitschrift für Balkanologie, Band XXI/2, (1985), pp. 236-237. (In the case of a direct borrowing from Latin or Greek, one would expect *aikklēsja in Gothic.
238. Istoria limbii române, ed. by A. Rosetti et al, vol. II, 1969, p. 162.
239. Ştefan Pascu et al., (eds.), Istoria României. Compendiu, 1974, p. 94.
240. Konstantin Josef Jireček, Geschichte der Serben, vol. I to 1371, (Gotha: 1911), pp. 38-39; a new ed., (Amsterdam: 1967),
241. Pascu (ed.), Istoria României. Compendiu, 1974, pp. 75-76,
242. Ion Coteanu, Structura şi evoluţia limbii române, 1981, op. cit., p. 73.
243. Dumitru Berciu, Daco-Romania, 1976, p. 85.
244. Günter Reichenkron, "Das Ostromanische," in Völker und Kulturen Südosteuropas, (München: 1959), p. 167.
245. Berciu, Daco-Romama, 1976, p. 138.
246. Ştefan Pascu,. (ed,), Istoria României. Compendiu, 3rd edition, 1974, p. 77.
247. Mihăescu, Limba latină în provinciile dunărene, 1960, op. cit., pp. 35-36.
248. Reichenkron, "Das Ostromanische," op. cit., 1959, p. 169, (referring to S. Puşcariu, Die rumänische Sprache, p. 456).
249. Eugen Lozovan, "Aux ori gines du christianisme daco scythique," in Geschichte der Hunnen, ed. by Franz Altheim (Berlin: 1959-1962), p. 159; with several references to the literature of this cult in the Roman Empire.
250. Ibid., p. 160-161.
251. George Ivănescu, "Les plus anciennes influences de la românite balkanique sur les Slaves: luna "knie," lunatik "somnambule," in Romanoslavica 1 (1958), pp. 44-51; quoted by Lozovan. 1959-1962, p. 161, note 81.
252. Densusianu, Histoire de la langue roumaine, ed. by B. Cazacu et al, (Bucharest: 1975), p. 194.
253. Rosetti, 1986, op. cit., p. 179.
255. Solta, 1980, op cit., p. 168.
256. After the appearance of the Atlas Lingvistic Român, published first in 1936, Bucharest.
257. Sextil Puşcariu, "Les enseignements de l'Atlas Linguistique de la Roumanie," in Revue de Transylvanie, 3, 1, 1936; Puşcariu, "Le role de la Transylvanie dans la formation et revolution de la langue roumaine," in La Transylvanie, Bucharest 1938,
258. Ibid., p. 15.
259. Constantin C. Giurescu, Istoria Românilor, (Bucharest: 1935).
260. Puşcariu, "Les enseignements," 1936, pp. 19-20.
261. Ibid., p. 21.
262. Christian lonescu, Mică enciclopedie onomastică [Concise Onomastic Encyclopaedia], (Bucharest: 1975), p. 219,
263. According to W. Meyer-Lübke, Romanisches Etymologisches Wörterbuch, 3rd edition, (Heidelberg: 1935), this word is of learned origin, something which was questioned by A. Graur (Cf., Istoria limbii române, ed. by A. Rosetti, B. Cazacu, I. Coteanu, (Bucharest: 1969), vol, II, p. 148.
264. Puşcariu, "Le rôle de la Transylvanie," 1938, op. cit., p. 63.
265. Cf., for example, A. Du Nay, 1977, pp. 34-41.
266. B.E. Vidos, Handbuch der romanischen Sprachwissenschaft, (München: 1975), translated from Dutch, p. 102.
267. Ibid., p. 101.
268. Matilda Caragiu Marioţeanu, 1975, op. cit., p. 73.
269. Ibid., p. 178.
270. Ibid., p. 177.
271. Ibid., p. 72. Obviously, the existence of words of foreign origin, replacing ancient Latin words in Transylvania (which have been preserved, for example, in Muntenia) "subverts the proposed Latin image of Transylvania," (Rosetti, 1986).
272. Eugen Lozovan, "Byzance et la Romanité Scythique," in: F. Altheim, Geschichte der Hunnen, 1959-1962, vol. II, p. 222.
275. N. Cartojan, Istoria literaturii române vechi [The History of the Old Romanian Literature]., 2nd ed. (Bucharest: 1980), p. 88.
276. Ibid., p. 89.
278. Matilda Caragiu-Marioţeanu, Compendiu de dialectologie, 1975, op. cit., p. 222.
279. Ibid., p. 238.
280. Ibid., pp. 256-257.
281. Ibid., p. 254.
282. Ernst Gamillscheg, "Über die Herkunft der Rumänen," Sitzungsberichte, (Berlin: 1940), p. 130.
283. Ibid., p. 122.
284. Ibid., p. 131; referring to G. Weigand, Balkan-Archiv I, 31 and 34.
285. Günter Reichenkron, "Die Entstehung des Rumänentums nach den neuesten Forschungen," in Südost-Forschungen, XXII, 1963, pp. 61-77.
286. Cf., for example, Rosetti, 1986, op. cit., p. 283.
287. Reichenkron, "Das Ostromanische," 1959, op. cit., p. 169.
288. Ibid., p. 160.
289. Reichenkron, "Vorrömische Bestandteile des Rumänischen," in Südost-Forschungen. XIX, 1960, p. 349. An interesting example of the borrowing of a typical Albanian expression by Northern Romanian is given on this page, note 26: Albanian ha, "I am eating" + suffix -me: hame, "the eating" + the suffix used for the construction of nomina agentis: -ës: hámës, "the eater," from which Northern Romanian hămesi "to be awfully hungry" (and hămeseală "canine appetite"').
290. Reichenkron, "Das Ostromanische," 1959, op. cit., p. 169, referring to S. Puşcariu, Die rumänische Sprache, p. 456.
291. Ibid., p. 169: '"Von Dardania dürfte so einmal die erste Ausbreitung des 'Urrumänentums,' wie man es seit Puşcariu nennt, vollzogen haben."
292. Reichenkron, "Die Entstehung des Rumänentums," 1963, op. cit., p. 66: "Es dürfte vielmehr so sein, dass die meisten der albanisch-rumänischen Wortgleichungen, im Süden der Donau, in der sog. Dardania, als dem südlichen
Teil der römischen Provinz Moesia Superior, also etwa im jugoslawischen Mazedonien, entstanden sind."
292. Giurescu, 1975, p. 151.
293. Pascu, (ed.) Istoria României. Compendiu, 1974, p. 77.
294. Vasile Arvinte, Die Rumänen, 1980, op. cit., p. 27.
295. Eugen Lozovan, "Diachronie et géographie linguistique roumaines," in Acta Philologica II, 1959, p. 162.
296. Alexandru Rosetti, Istoria limbii române de la origini pînă în secolul al XVII-lea [The History of the Romanian Language From the Origins to the 17th Century], (Bucharest: 1968), p. 214.
297. Ed. by A. Rosetti, B. Cazacu, I. Coteanu (Bucharest: 1969), vol. II.
298. Istoria limbii române, ed. by Florica Dimitrescu et al, (Bucharest: 1978).
299. Boris Cazacu, (Bucharest: 1966).
300. Compendiu de dialectologie română (nord- şi sud-dunăre ană), (Bucharest: 1975).