Chapter 2 in Noel Malcolm's Kosovo, a short history (Macmilan, London, 1998, p. 22-40)
All origins become mysterious if we search far enough into the past. And almost all peoples, when we look at their earliest origins, turn out to have come from somewhere else. Before embarking on these origin quests, it is good to keep a few qualifying principles firmly in mind. First, it can never be said too often that questions of chronological priority in ancient history - who got there first - are simply irrelevant to deciding the rights and wrongs of any present-day political situation.
Secondly, accounts of the earlier movements of peoples or tribes give a very misleading impression when they treat them as if they were unitary items, with unchanging identities, being transferred from place to place in a game of ethno-historical pass-the-parcel. In many cases (such as the migrations of the Franks in early western Europe) it is the movement of a people into a new territory or society that gives it an identity it did not previously have. Identities continue to develop over time: 'Serb' was a tribal label in the sixth century but not in the sixteenth, so that to treat 'the Serbs' as an unchanging category is as foolish as trying to identify Jutes and Angles among the subjects of Queen Elizabeth I.
And thirdly, we should never forget that all individual ancestries are mixed -especially in this part of Europe. When a Serb today reads about the arrival of the early Serbs, he may not be wrong to suppose that he is reading about his ancestors; but he cannot be right to imagine that all his ancestors were in that population. The equivalent is true for the Albanians, and indeed for every other ethnic group in the Balkans.
While most details about the movement of the early Slavs into the Balkans are unclear, the basic facts are known. A large tribal population of Slavs - among whom the Serbs and the Croats were two particular tribes, or tribal groupings - occupied parts of central Europe, north of the Danube, in the fifth and sixth centuries ad. The Serbs had their power-base in the area of the Czech lands and Saxony, and the Croats in Bavaria, Slovakia and southern Poland. This central European location was not the earliest known home of the Serbs; most of the evidence points to an earlier migration from the north and north-eastern side of the Black Sea. At that earlier period the Serbs and Croats seem to have lived together with more warlike Iranian tribes, and their tribal names may derive from Iranian ruling elites: Ptolemy, writing in the second century ad, located the 'Serboi' among the Sarmatians (an Iranian grouping) on the northern side of the Caucasus. Little is known about the Slavs' way of life in these earlier periods. The first descriptions we have of them are by Byzantine writers, who portray them as a wild people, more pastoral than agricultural, with many chiefs but no supreme leader. 
For a tribal population with a fairly low level of material culture, reaching the line of the Danube and looking south was the equivalent of a hungry man pressing his face against the window of a grocery. The Balkans, fully restored to Byzantine control under the energetic Emperor Justinian (527-65), contained many flourishing towns and cities, supported by productive agriculture and active trading routes. The Slavs were not the first to cross the Danube in search of better things. Germanic Goths had done so (with Byzantine permission, at first) in the fourth century, and had gone raiding as far as Greece and the Albanian coast thereafter; Huns, under Attila, had attacked in the 440s, and Bulgars (a Turkic tribe) had started raiding at the end of that century.  But none of these earlier invaders left any imprint on the Balkans comparable to that of the Slavs. Indeed, by the time that the Turkic-speaking Bulgars came to settle permanently in the Balkans in the seventh century, the Slav element was already so well established there that the conquering Bulgars were eventually to lose their own language and be absorbed by their Slav-speaking subjects. 
The first major Slav raids took place in the middle of Justinian's reign. In 547 and 548 they invaded the territory of modern Kosovo, and then (probably via Macedonia and the Via Egnatia across central Albania) got as far as Durres on the northern Albanian coast.  More substantial invasions took place in the 580s, bringing Slavs deep into Greece. Historians used to think that it was only these later invasions that involved any permanent settlement; but there is evidence of Slav place-names in the Balkans - particularly along the river Morava - by the 550s, which suggests a more continuous process of infiltration.  One factor which may have turned the southward movement of Slavs from a trickle to a flood was the arrival, in the north-western part of the Balkans, of an especially warlike Turkic tribe, the Avars, who subjugated or coopted some Slavic tribes but drove many others away. By the early seventh century the Avar armies were raiding as far as the walls of Constantinople, and threatening the very existence of the Byzantine Empire.
It was at this point, in the 610s or 620s, that the Emperor of the day (according to a detailed but somewhat confused account by a later Emperor-cum-historian, Constantine Porphyrogenitus) invited the Croats to come down from central Europe and deal with the Avar threat.  This they did, bringing with them their neighbours, the Serbs. Both populations then settled in the territories abandoned by the Avars: the Croats in modern Croatia and western Bosnia, and the Serbs in the Rascia area on the north-western side of Kosovo, and in the region of modern Montenegro. In some of these areas they supervened on an already existing Slav population, which, as a result, must gradually have taken on a 'Croat' or 'Serb' identity. The Serbs did not have anything like a state at this stage, but they developed several small tribal territories, each called a zupa and ruled by a tribal chief known as the zupan. 
By the mid-seventh century, Serbs (or Serb-led Slavs) were penetrating from the coastal lands of Montenegro into northern Albania. Major ports and towns such as Durres and Shkodra held out against them, but much of the countryside was Slavicized, and some Slav settlers moved up the valleys into the Malesi. By the ninth century, Slav-speaking people were an important element of the population in much of northern Albania, excluding the towns and the higher mountainous areas (especially the mountains in the eastern part of the Malesi, towards Kosovo).  Slav-speaking people lived in the lowlands of this area, gradually becoming a major component of the urban population too, until the end of the Middle Ages. 
What had happened to the local populations of the western and central Balkans during and after the Slav invasions? Something is known about the urban inhabitants, but much less about the people in the countryside. Despite the apocalyptic tone of early Byzantine writers, who give the impression that all civilization came to an end here in about 600, there is good evidence that the main cities survived (or were revived), just as they had done after earlier sackings. Refugees from central Balkan towns such as Nis and Sofia fled to the safety of Salonica at first, but many must have gone back home later.  The main towns on the Dalmatian and northern Albanian coastline, too, retained their Latin-speaking populations and stayed under Byzantine rule. (For naval and commercial reasons, Durres was the most important Byzantine possession on the entire Adriatic coast of the Balkans.)  But outside the major cities there are signs of decline and contraction; typical of the seventh to ninth centuries are the remains of small townships based on hill-forts, such as the one at Koman in the mountains of north-central Albania, where a Christian and probably Romanized (Latin-speaking) population must have led a rather limited existence. 
As for the rural population, which was also mainly Latin-speaking in most of the territory of Yugoslavia and north-western Bulgaria, it is assumed that large numbers of people were driven southwards by the Avars, Croats and Serbs. Some evidence from place-names suggests a flow of such refugees down the Dalmatian coast into northern Albania; and a folk tradition set down by a later Byzantine writer referred to a large movement of native people southwards and eastwards away from the area of the Danube and the Sava - that is, from northern Bulgaria, northern Serbia and Croatia.  No doubt Latin-speaking peasants and farmers continued to live in many of these areas, especially where they were in contact with a large town or city. But sooner or later the majority of them were Slavicized, and the towns in the interior of the Balkans filled up with Slav-speakers too.
Only the remnants of a Latin-speaking population survived in parts of the central and west-central Balkans; when it re-emerges into the historical record in the tenth and eleventh centuries, we find its members leading a semi-nomadic life as shepherds, horse-breeders and travelling muleteers. These were the Vlachs, who can still be seen tending their flocks in the mountains of northern Greece, Macedonia and Albania today.  The name 'Vlach' was a word used by the Slavs for those they encountered who spoke a strange, usually Latinate, language; the Vlachs' own name for themselves is 'Aromanians' (Aromani). As this name suggests, the Vlachs are closely linked to the Romanians: their two languages (which, with a little practice, are mutually intelligible) diverged only in the ninth or tenth century.  While Romanian historians have tried to argue that the Romanian-speakers have always lived in the territory of Romania (originating, it is claimed, from Romanized Dacian tribes and/or Roman legionaries), there is compelling evidence to show that the Romanian-speakers were originally part of the same population as the Vlachs, whose language and way of life were developed somewhere to the south of the Danube. Only in the twelfth century did the early Romanian-speakers move northwards into Romanian territory. 
Finally, before turning to the most mysterious problem of all - the origin of the Albanians - it is worth looking once more at the pattern of settlement in the Kosovo area during the early Slav centuries. Kosovo did not fall within the Serb territory of Rascia, which was further to the north-west: the Serbian expansion into Kosovo began in earnest only in the late twelfth century. About the other early Slav settlers in this part of the Balkans we have much less information. Byzantine sources just referred generally to 'Sklaviniai', Slav territories, in the Macedonian region; in the few cases when they made more localized references they often used names derived from rivers, so that it is not clear whether these were the names of Slav tribes or just geographical labels. The 'Moravoi' or 'Moravlians', for example, who are first mentioned in the ninth century, lived somewhere near the river Morava, but that is all we know about them. Historical map-makers, who do not like leaving too many blank spaces, place these Moravlians over much of south-eastern Serbia from as early as the sixth century, with arrows showing them passing into Kosovo; real evidence for this is lacking. 
Obviously some Slavs did spread through all these areas sooner or later. But there is one intriguing line of argument to suggest that the Slav presence in Kosovo and the southernmost part of the Morava valley may have been quite weak in the first one or two centuries of Slav settlement. If Slavs had been evenly spread across this part of the Balkans, it would be hard to explain why such a clear linguistic division emerged between the Serbo-Croat language and the Bulgarian-Macedonian one. The scholar who first developed this argument also noted that, in the area dividing the early Serbs from the Bulgarians, many Latin place-names survived long enough to be adapted eventually into Slav ones, from Naissus (Nish), down through the Kosovo town of Lypenion (Lipljan) to Scupi (Skopje): this contrasts strongly with most of northern Serbia, Bosnia and the Dalmatian hinterland, where the old town names were completely swept aside. His conclusion was that the Latin-speaking population, far from withering away immediately, may actually have been strengthened here (and in a western strip of modern Bulgaria), its numbers swelled, no doubt, by refugees from further north. These Latin-speakers would have thus formed 'a wide border-zone between the Bulgarians and the Serbs'. 
Kosovo's protective ring of mountains would have been useful to them; and the Roman mountain-road from Kosovo to the Albanian coast - along which several Latin place-names also survive, such as Puka, from 'via publica' - might also have connected them with other parts of the Latin-speaking world. (The hill-top town of Koman, mentioned earlier, is only a few miles from Puka, and may well have had a Latin-speaking population too.) If this argument is correct, we might expect many of the ancestors of the Vlachs to have been present in the Kosovo region and the mountains of western Bulgaria; it may have been in these uplands that they developed their pastoral skills.
Only in the ninth century do we see the expansion of a strong Slav (or quasi-Slav) power into this region. Under a series of ambitious rulers, the Bulgarians - a Slav population which absorbed, linguistically and culturally, its ruling elite of Turkic Bulgars - pushed westwards across modern Macedonia and eastern Serbia, until by the 850s they had taken over Kosovo and were pressing on the borders of Rascia. Soon afterwards they took the western Macedonian town of Ohrid; having recently converted to Christianity, the Bulgar rulers helped to set up a bishopric in Ohrid, which thus became an important centre of Slav culture for the whole region. And at the same time the Bulgarians were pushing on into southern and central Albania, which became thoroughly settled by Bulgarian Slavs during the course of the following century. 
Kosovo was to remain under Bulgarian or Macedonian rulers until 1014-18, when the army of the Macedonian-based Tsar Samuel died, his empire broke up, and Byzantine power was fully re-established by a strong and decisive Emperor, Basil 'the Bulgar-killer'. For nearly two centuries after that, Kosovo would stay under Byzantine rule. 
One key element is missing from the picture presented so far. While the origins of the Vlachs are obscure enough, the origins of the Albanians have been the subject of a much more bewildering mass of conflicting claims and theories. The two main rival theories that have emerged identify the early Albanians as either Illyrians or Thracians: in pre-Roman and Roman times, Illyrians lived in the western half of the Balkans and Thracians in the east. Albanian historians, who like the idea that Albanians have always lived in Albania, prefer the Illyrian theory. Romanian scholars, who have to deal with the awkward fact that there are strong early links between the Albanians and the Vlachs, prefer to put them on the Thracian side of the divide (the ancient Dacians, who lived in Romania, were part of the Thracian group), and in this they are sometimes supported by Bulgarian experts. But there is really no point in going into this labyrinth of historical debate unless one is prepared to discard all national prejudices at the entrance.
The Albanians first emerge in the historical record in 1043, when Albanian troops appear fighting alongside Greeks in the army of a rebel Byzantine general. They are mentioned at Durres in 1078, and again in 1081, when they joined the Byzantine forces resisting an invasion there by the Norman adventurer Robert Guiscard.  (Bizarrely, a garbled list of Albanian place-names, picked up by the Normans on this expedition, was soon afterwards incorporated into the Song of Roland: one manuscript of that poem includes a reference to 'Albanie', implying that it was a place or area just north-east of Durres.) 
Over the next two centuries the references to Albanians gradually increase, until by 1281 we have a mention in an Italian document of a 'duca Ginius Tanuschus Albanensis', who ruled an area between Durres and Shkodra: 'Ginius' must be the Albanian 'Gjin' (John), and this 'duca Gjin' is presumed to be the founder of the famous 'Dukagjin' family.  By the early fourteenth century there are also signs of a long-established Albanian presence in the mountains of Montenegro, and as far north as the Ragusan hinterland. 
The name used in all these references is, allowing for linguistic variations, the same: 'Albanenses' or 'Arbanenses' in Latin, 'Albanoi' or 'Arbanitai' in Byzantine Greek. (The last of these, with an internal switching of consonants, gave rise to the Turkish form 'Arnavud', from which 'Arnaut' was later derived.) Nor is there any mystery about the origin of this name. In the second century Ptolemy referred to a tribe called the 'Albanoi', and located their town, 'Albanopolis', somewhere to the east of Durres. Some such place-name must have survived there, continuously if somewhat hazily, ever since; there was an area called 'Arbanon' in north-central Albania in the eleventh century, and in the early twentieth century 'Arben' was the local name for a region near Kruja (which lies just north of Tirana).  Linguists believe that the 'Alb-' element comes from the Indo-European word for a type of mountainous terrain, from which the word 'Alps' is also derived. (So too, coincident-ally, is the Gaelic word for Scotland, 'Albainn', which classicizing eighteenth-century Scots sometimes turned into 'Albania'.) 
The continuity of this name is a striking fact; but it does not amount to proof that the Albanians have lived continuously in Albania. Place-names can endure while populations literally come and go. In any case, the Albanians do not use this word to describe themselves: in their language, Albania is Shqiperia, an Albanian is a shqiptar, and the language itself is shqip. (The only Albanians to use the 'Alb-' root are the ones who emigrated to Italy in the fifteenth century, who call themselves 'Arberesh'.)  The origins of shqiptar, which first crops up as a personal name in late-fourteenth-century documents, are completely obscure: some think it means 'he who understands', from a verb shqipoj, while others connect it with the word for an eagle, shqipojne, which may have been the totem of an early tribe. 
Is there any way to bridge the gap between the 'Albanoi' of the second century and the medieval Albanians? The historical record is utterly silent: there is one apparent reference in a medieval document to 'Duchagini d'Arbania' warring against a king of Bosnia in the seventh century, but it must be discounted, as the document's chronology is completely unreliable.  For some scholars, the argument from silence carries a certain force of its own; it is suggested that any large-scale migration of the early Albanians into Albania would surely have been remarked on by Byzantine authors.  But the truth is that those authors were interested in alien tribes only when their actions impinged, militarily or politically, on the Empire. A small pastoral population, moving away from them into some remote mountain region, might never have attracted their notice.
Some Albanian archeologists have tried hard to show that the Koman hill-town culture of the seventh and eighth centuries is the essential proof of Illyrian-Albanian continuity; but material remains do not tell us what language people spoke (unless they include inscriptions, which these do not), and the main cultural affinities here seem to have been with the Latin-speaking Romano-Byzantine towns of the previous centuries.  And one other line of argument, which tries to find striking similarities between Albanian social practices and what classical authors tell us about the Illyrians, must also be described as inconclusive. Certainly the tribes of the ancient Illyrians, political groupings covering large areas and heavily stratified with a powerful ruling caste, were quite different from the modern Albanian clans. 
If there is any chance at all of solving this mystery, it lies in the study of the Albanian language. Historical linguistics is a complex science and not, in some of its activities, a very exact one. But by sifting through the evidence of vocabulary and place-names, and sorting out different layers of borrowings from other languages and cultures, linguists can often construct quite a detailed chronology, just like an archeologist examining different layers of wood-ash and broken pots. They can point out, for example, that the Albanian names for the fauna and flora of the high mountain regions are purely Albanian, while the low-altitude vocabulary borrows heavily from Slav; the words for ploughing are mainly Slav, and so are many words for weaving, masonry and milling. Much of the vocabulary of medieval government and society is also Slav-based.  This strongly suggests that the early Albanians led a mainly pastoral life in mountainous regions, before settling in lowland areas after the Slavs had extended their culture and rule. And the evidence of place-names shows that Albanian-Slav contacts in the northern Albanian region must have happened before 900 at the latest: a vowel-shift in the Slav language took place by the end of the ninth century, and some Albanian borrowings from Slav preserve the pre-shift form of the vowel. 
We have now got back to the ninth century, but that still leaves seven centuries unaccounted for. The most direct way of bridging the gap with the Roman world would be for the historical linguists to demonstrate a link between Albanian and one of the 'barbarian' Balkan languages of the region - either Illyrian or Thracian. It is clear that Albanian is indeed the only surviving representative (apart from Greek) of an ancient Balkan language: it belongs to the Indo-European family of languages, but exists in a sub-section of its own, with no immediate relatives. If either Illyrian or Thracian could be identified as its parent, this would at least set some fairly clear geographical limits to the early home of the Albanians: Illyrians lived in Albania and most of Yugoslavia, Thracians in Bulgaria and part of Macedonia, and the boundary between them ran approximately along the Morava valley and down the eastern side of Kosovo.  (Kosovo itself was part of the tribal land of the Dardanians, who almost certainly belonged to the Illyrian grouping.) 
Unfortunately, working out the relation between Albanian and Illyrian or Thracian is like trying to solve an equation with too many unknowns. We do not possess a single text in Illyrian. We have two short texts in what is presumed to be Thracian, but no one knows what they mean. The longer one, consisting of sixty-one Greek letters without any word-divisions, has been subjected to eighteen speculative and somewhat comically divergent translations: one version says 'I, Rolisteneas, son of Nereneas, eat the sacrificial meal; Tyiezypta, originally from Arazea, attached the golden objects to me', while another comes up with 'O Rolisten, I, Nerenea Tiltea, die peacefully next to you, my quietly deceased one, I who raised the children.'  The linguists who have offered these translations from the Thracian have at least fared better than the one who interpreted an 'Illyrian' inscription as 'Consecrated to the goddess Oethe': it was later pointed out that this inscription, if read from bottom to top, produced a perfectly normal Greek phrase, 'Lord help Anna'. 
Apart from inscriptions, there are a few 'glosses' (comments explaining the meanings of words) in classical authors: here the evidence is too slight to be conclusive. One Illyrian word, rhinon, glossed as 'mist', does resemble an old Albanian word for cloud, ren. A Thracian word for a blackberry, mantia, resembles the Albanian for a mulberry, man, and the Thracian for 'camomile' could perhaps be linked to the Albanian for 'sweet-tasting'; but those are the only clear resemblances, and the names of edible plants are in any case famously mobile across linguistic frontiers. 
Otherwise, the only evidence available consists of proper names: place-names, personal names and tribal names, preserved in Latin or Greek inscriptions and the works of ancient historians. There are several thousand such names altogether; but the difficulties of interpretation are immense. Trying to extract a language from such evidence is rather like some linguists of the distant future trying to work out the true nature of the English language on the basis of 'Edinburgh', 'Lancaster', 'Whitby', 'Grosvenor', 'Gladstone', 'Victoria' and 'Disraeli'. Place-names are often the remnants of an earlier language; personal names may reflect cultural influences (it has been observed that if future linguists knew only the names 'Carlo' and 'Lodovico', they would assume that the Italian language was a type of German); and in any case we have no reason to suppose that the ancient Balkans were any less of a linguistic hotchpotch than they have been for most of the rest of their history.  On balance, there are more examples of plausible links between Illyrian names and Albanian words than there are in the case of Thracian (though there are some of both, and some names were common to the two ancient languages). Most of these relate to place-names in the area of central and northern Albania, such as the river Mat (Alb.: mat, river-bank) or the town of Ulqin or Ulcinium (Alb.: ujk or ulk, wolf), or indeed the early name for the Kosovo area, 'Dardania' (Alb.: dardhe, pear). 
The strongest evidence, however, comes not from the meaning of the proper names (which is always open to doubt) but from their structure. Most Illyrian names are composed of a single unit; many Thracian ones are made of two units joined together. Several Thracian place-names end in -para, for example, which is thought to mean 'ford', or -diza, which is thought to mean 'fortress'. Thus in the territory of the Bessi, a well-known Thracian tribe, we have the town of Bessapara, 'ford of the Bessi'. The structure here is the same as in many European languages: thus the 'town of Peter' can be called Peterborough, Petrograd, Petersburg, Pierreville, and so on. But the crucial fact is that this structure is impossible in Albanian, which can only say 'Qytet i Pjetrit', not 'Pjeterqytet'. If para were the Albanian for 'ford', then the place-name would have to be 'Para e Besseve'; this might be reduced in time to something like 'Parabessa', but it could never become 'Bessapara'. And what is at stake here is not some superficial feature of the language, which might easily change over time, but a profound structural principle. This is one of the strongest available arguments to show that Albanian cannot have developed out of Thracian. 
Other linguistic arguments which have been deployed in this Illyrian versus Thracian debate are more technical. Much ink has been spilt, for example, on the question of whether Illyrian was a satem language or a centum language. This is a traditional classification of all Indo-European languages according to their underlying patterns of consonant development. (The labels are taken from the Old Iranian and Latin for 'a hundred'.) Albanian is a satem language, and Thracian is thought to have been one too. Most scholars believed that Illyrian was a satem language, until linguists analysed the surviving inscriptions in Venetic, a language of north-eastern Italy which was assumed (on the authority of ancient authors) to be related to Illyrian. This turned out to be definitely centum, and persuaded some experts that the whole Illyrian group must therefore have been centum too - in which case Albanian could not have come from Illyrian.  However, more recent research has shown that Venetic had nothing to do with Illyrian.  (Similar problems caused by another language thought to be related to Illyrian, the Messapian language of southern Italy, have also been resolved in the same way.)  Illyrian was probably satem after all.
And in any case, it is increasingly apparent that the whole satem/centum classification system does not correspond to the fundamental distinguishing features of the Indo-European languages: it may be the linguists' equivalent of one of those classifications of mammals by eighteenth-century biologists, which modern scientists have had to discard.  Another technical (and much more speculative) argument for identifying early Albanian with Thracian was put forward by the Bulgarian linguist Georgiev, who divided Thracian into two languages, one north-western, the other south-eastern, and argued on the basis of consonantal changes that Albanian must have come from the north-western one. But his arguments (at least in relation to the supposed Albanian connection) have been thoroughly dismantled by other scholars. 
Other linguistic arguments are more closely linked to geography. The place-names of the northern Albanian region offer a valuable linguistic testing-ground. We know what many of them were called in Roman times; it should therefore be possible to tell whether their modern Albanian form derives from a continuous Albanian tradition going back to contact with the Romans, or whether it is derived from the Slav form of the name. If the latter, then this might suggest that the Albanians entered this area only after the Slav immigration of the seventh century. The fact that Slavs developed their own forms of the urban names directly from the Latin (Skadar from Latin Scodra, for example, where the Albanian form developed as Shkoder/Shkodra) is not in itself significant; their contact in the urban areas would have been mainly with Latin-speakers anyway. But if, on the other hand, the Slav names for rivers or mountains show that they were borrowed from Albanian forms of those names, this would indicate that there were Albanian-speakers in the countryside when the Slavs first arrived.
The evidence is in fact very mixed; some of the Albanian forms (of both urban and rural names) suggest transmission via Slav, but others -including the towns of Shkodra, Drisht, Lezha, Shkup (Skopje) and perhaps Shtip (Stip, south-east of Skopje) - follow the pattern of continuous Albanian development from the Latin.  (One common objection to this argument, claiming that 'sc-' in Latin should have turned into 'h-', not 'shk-' in Albanian, rests on a chronological error, and can be disregarded.)  There are also some fairly convincing derivations of Slav names for rivers in northern Albania - particularly the Bojana (Alb.: Buena) and the Drim (Alb.: Drin) - which suggest that the Slavs must have acquired their names from the Albanian forms. 
Finally, one more common-sensical linguistic and geographical argument should also be mentioned: the claim, by the pioneering German Balkanologist Gustav Weigand, that the early Albanians must have lived a long way to the east of the Adriatic coast, because most of the Albanian words for fish, boats and coastal features are borrowed from other languages.  Sterling efforts have been made by Albanian scholars to find authentic Albanian fish-words, but the tally, though not insignificant, is still rather poor.  However, Weigand's argument could not be very powerful even if its basic observation were correct (as it may in fact be). A pastoral population might have lived only 50 miles inland in the Albanian mountains without having any contact with fishing or sailing; it is not necessary to push its location eastwards all the way to Thrace.  Of course Illyrians did once live on the coast, and would presumably have had their own maritime vocabulary.
But if Illyrian survived as Albanian, it did so only by means of physical contraction, withdrawal and isolation, which naturally would have taken place in mountain terrain. This is why the purest element of Albanian vocabulary refers to mountains, high-altitude plants and shepherding: the point is not that the proto-Albanians had never lived any other sort of life, but that the only ones who survived as Albanian-speakers did so precisely because that was the sort of isolated and independent life they led, probably for several centuries. The Illyrians who lived on the coastal plains were Romanized, like the ones on the Dalmatian coast and indeed in most areas of Yugoslavia. By the time the Slavs began arriving in the sixth century, there were only scattered pockets of speakers of the old 'barbarian' languages left anywhere in the Balkans, and all of them were in mountainous regions. 
Of these, the only population considered important enough to be mentioned by name in early written sources was the Thracian tribe of the Bessi, who lived in the western and southern mountains of Bulgaria. We know that their version of the Thracian language was still being spoken in the second half of the sixth century, and we also know that they had been converted to Christianity: the most striking piece of evidence refers to monks speaking 'Bessan', as well as Latin and other languages, in a monastery on Mount Sinai in the 560s.  Until very recently, this was treated by most scholars as just an intriguing oddity, a last lingering survival which must have been extinguished before long. However, a dazzling new piece of research and speculative reconstruction by the German scholar Gottfried Schramm has proposed that these Thracian Bessi were none other than the real ancestors of the Albanians.
According to Schramm, the Bessi must have moved out of their western Bulgarian homeland and into the northern Albanian region in the early ninth century, probably to escape the persecution of Christians by the still pagan Bulgar khans.  The early conversion of the Bessi to Christianity is indeed, in Schramm's view, the key to the entire question of how and why Albanian survived as a language. We know that the Bessi were converted by an enterprising bishop, Nicetas, in the late fourth century, and from the writings of a friend of Nicetas who celebrated this event we also know that he learned their language and taught them to practise their Christianity in it - in other words, that Bessan was used as a liturgical language. (The evidence of the Bessan-speaking monks supports this point.) Nicetas, whose own mother-tongue was Latin, may also have translated parts of the Bible; the obvious model - or competition - that he must have had in mind was the work of a heretical bishop, Ulfilas, who was using the Germanic Gothic language for liturgy and Bible-translation among the nearby population of Goths in northern Bulgaria. And, as comparison with other linguistic survivals (such as Armenian or Coptic) shows, nothing helps a language to survive quite so much as its use from a very early stage in a kind of national church. 
One thing is quite certain: the Albanians did acquire their Christianity from a Latin-speaking teacher or teachers. The Albanian language contains much Latin-derived vocabulary anyway, having obviously absorbed words from nearby Romans or Romanized barbarians from the second century bc onwards; but the Latin element is especially rich in the area of Christian belief and Christian practice. Thus we have meshe (mass), from missa; ipeshk (bishop), from episcopus; ungjill (gospel), from evangelium; mrekull (miracle), from miraculum; and a great number of other words, extending far into the vocabulary of psychology, morality and even the natural world (such as qiell, meaning heaven or sky, from caelum). 
Many of the words that would need to be put on such a list, in fact, are not special ecclesiastical terms, for which a non-Christian population would have no equivalent of its own; they are simple words such as 'spirit', 'sin', 'pray*, 'holy', and so on, for which most languages, even in pre-Christian times, have their own vocabulary. When other early evangelizers translated the Bible or the liturgy into Armenian, or Gothic, or Anglo-Saxon, they used local words for these things - that, indeed, is what is implied by the whole idea of translation. Why should Nicetas, translating into proto-Albanian, have simply transferred huge quantities of Latin words? Schramm notes the oddity of this in passing, and suggests unconvincingly that there must have been some special cultural reasons.  But the oddity is more overwhelming than he admits. For example, even the word for a flock, as used in Christian discourse, was taken from the Latin (grigje, from grex) - of all the things in the world, the one for which a shepherding population must surely have had its own word already. 
The solution to this puzzle is blindingly simple. These elements of Latin vocabulary have undergone exactly the same sorts of sound-changes, compressions and erosions as all the other Latin words which entered the Albanian language over several centuries; and the reason why those words entered the language was that the Albanians were in contact, over a long period, with people who spoke Latin. The existence of large quantities of such Christianity-related Latin vocabulary does not show that someone 'translated' Christian discourse into early Albanian. It shows the precise opposite - namely, that Albanians were for a long time exposed to the conduct of their religion not in translation but in the original Latin.
This can even be demonstrated grammatically. The term for 'Holy Trinity', Shendertat, bears a final 't' and an accent on the last syllable: this shows that it developed from the accusative, sanctam trinitatem, not the nominative, sancta trinitas. That is in fact the normal pattern of development in Romance languages, which gives us, for example, Spanish ciudad from dvitatem (not from civitas), or French mont from montem (not from mons). (There are many other Albanian examples too, such as grigje, mentioned above, which is really from gregem, not grex.) What this phenomenon reflects is a pattern of usage in spoken Latin: these words were heard much more often as the objects in sentences than as the subjects. If Nicetas had been coining new Albanian words out of Latin for the purposes of his translation, he would surely have taken them from the nominative form. These words entered Albanian because Albanians heard them, over and over again, in spoken liturgical Latin.
Schramm's theory fails, therefore; and in so doing it performs a signal service. Thanks to Schramm, the Thracians can now be eliminated from these enquiries. His research into Nicetas's activities does indeed show that the Bessi received their Christianity, so to speak, in translation; this must force us to conclude that the Albanians, who received theirs in the original Latin, cannot be identified with the Bessi. The language of the Bessi must eventually have perished. Since the Bessi were the only Thracian tribe known to have kept their language as late as the sixth century (and Byzantine sources are naturally more detailed on the Thracian areas, which for them were closer to home, than on the Illyrian ones), it is impossible to find any other Thracian candidates. The origins of the Albanians must be sought, therefore, on the Illyrian side of the divide - particularly in the mountains round Kosovo, in the Malesi, and in the tangle of mountains stretching north from there through Montenegro.
The Latin elements in Albanian help to confirm this location. From the fact that so much general vocabulary was absorbed into Albanian from Latin, and so little from Greek, it is clear that the proto-Albanians lived some way to the north of the Latin-Greek linguistic divide. This language frontier ran from the Adriatic coast near Lezha across the middle of Albania, then up to the line of the Sar mountains, curving southwards to take in Latin-speaking Skopje, and then running northwards roughly along the Serbian-Bulgarian border.  At the same time, the fact that the proto-Albanians never actually lost their language indicates that they were somewhat isolated from the main areas of Roman settlement - which included the lowlands and the major roads. One influential theory therefore places the early Albanians in the part of northern Albania which (according to archeological evidence and place-names) was the most untouched by Roman influence: the 'Mat' district north-east of Tirana and west of Debar. From there, according to this theory, the early Albanians were able to expand to fill the region bounded by the river Shkumbin, the Black Drin, the united Drin and the coast. 
What this theory fails to account for, however, is another key aspect of the Albanian language's connection with Latin: its intimate involvement in the development of the Vlach-Romanian language. Linguists have long been aware that Albanian and Romanian have many features in common, in matters of structure, vocabulary and idiom, and that these must have arisen in two ways. First, the 'substratum' of Romanian (that is, the language spoken by the proto-Romanians before they switched to Latin) must have been similar to Albanian; and secondly, there must have been close contact between Albanians and early Romanian-speakers over a long period, involving a shared pastoral life. (Some key elements of the pastoral vocabulary in Romanian are borrowed from Albanian.)  The substratum elements include both structural matters, such as the positioning of the definite article as a suffix on the end of the noun, and various elements of primitive Balkan pre-Latin vocabulary, such as copil ('child' in Romanian) or kopil ('bastard child' in Albanian).  If the links between the two languages were only at substratum level, this might not imply any geographical proximity - it would merely show that proto-Albanian was similar to other varieties of Illyrian spoken elsewhere. But the pastoral connections do indicate that Albanians and early Romanians lived for a long time in the same (or at least overlapping) areas.
This has some geographical implications. Late Latin developed in two different forms in the Balkans: a coastal variety, which survived as a distinct language (known as Dalmatian) until the end of the nineteenth century, and the form spoken in the interior, which turned into Romanian and Vlach.  From place-names it is clear that the coastal form, spoken also in Shkodra and Durres, penetrated some way into the northern Albanian mountains.  There are some traces of this variety of Latin in Albanian, but the Albanian language's links with the inland variety of Balkan Latin are much stronger. This suggests that the centre of gravity ofAlbanian-Vlach symbiosis lay a little further to the east. 
When and how did that symbiosis take place? Presumably the Latin-speaking proto-Romanians came to pastoralism later than the early Albanians. If they had been doing it for as long as the Albanians, and in similar areas, they would - just like the Albanians - have escaped Latinization altogether. Some historians have decided that the proto-Romanians must have been Latin-speaking city-dwellers, who somehow extricated themselves from their towns in the early Slav centuries and became long-distance travellers or shepherds instead; but this seems inherently implausible.  (Had they come from the towns, their Latin would surely have been closer to standard Latin in its structure, too.) There is in fact enough Latin agricultural vocabulary in Romanian -words for sowing, ploughing, harrowing, and so on - to show that they were farming in Roman times.  The shift towards pastoralism was probably quite gradual. One particular factor that may have helped to promote it was the practice of horse-breeding, which was, or at least became, a Vlach speciality: the medieval records are full of Vlach muleteers and Vlachs leading caravans of pack-horses.  Such an occupation requires contact with towns (where the trade is), and may be combined with some farming in the towns' vicinity; but it also involves a form of stock-breeding, which could have given the early Vlachs an entree into the higher-altitude world of Albanian flocks and herds.
The main area of the Balkan interior where a Latin-speaking population may have continued, in both towns and country, after the Slav invasion, has already been mentioned: it included the upper Morava valley, northern Macedonia, and the whole of Kosovo. It is, therefore, in the uplands of the Kosovo area (particularly, but not only, on the western side, including parts of Montenegro) that this Albanian-Vlach symbiosis probably developed.  All the evidence comes together at this point. What it suggests is that the Kosovo region, together with at least part of northern Albania, was the crucial focus of two distinct but interlinked ethnic histories: the survival of the Albanians, and the emergence of the Romanians and Vlachs. One large group of Vlachs seems to have broken away and moved southwards by the ninth or tenth century; the proto-Romanians stayed in contact with Albanians significantly longer, before drifting north-eastwards, and crossing the Danube in the twelfth century. 
Having reached these conclusions, it may be possible, finally, to draw some further implications from them that point back to a much earlier period of Kosovo's history. The point is a very simple one. If Albanian-speakers were able to live in this area without losing their language during the period from the sixth century to the twelfth, is there any reason to think that they could not have been there in the previous six centuries or more? The Roman province of Dardania contained some Roman towns and several large estates, but it was far from being utterly and homogeneously Romanized: frequent Roman references to Dardanian bandits and robbers, and the presence of many forts and watch-towers, suggest that it was never completely under control.  References to Dardanian cheese, a famous and widely exported product, also testify to a large shepherding population.  And if the shepherds in the hills were speaking proto-Albanian, then perhaps that is what the ordinary Dardanians had spoken in the valleys too, before the Romans came. This is more a speculation than a conclusion; and it is not meant to exclude other areas in the Albanian (or Montenegrin) mountains further to the west, given that 'Dardania' was, essentially, a tribal division, not a linguistic one. Once again it must be emphasized that such ancient history can have no implications for modern politics. Nevertheless, the idea that the Illyrian Dardanians were ancestors of the Albanians may be of some sentimental interest to Kosovo Albanians today.
Chapter 3. Medieval Kosovo before Prince Lazar: 850s-1380s
The previous chapter brought the political history (if such it may be called) of Kosovo up to the final period of Bulgarian-Macedonian rule, before the territory of Tsar Samuel was reconquered by the Byzantine Emperor Basil the Bulgar-slayer. Medieval Kosovo is often referred to in general terms as 'the cradle of the Serbs', as if it had been a Serb heartland from the outset; but the reality was rather different. Just over 800 years separate the arrival of the Serbs in the Balkans in the seventh century from the final Ottoman conquest in the 1450s: out of those eight centuries, kosovo was Serb-ruled for only the last two-and-a-half - less that on-ethird of the entire period. Bulgarian khans or tsars held Kosovo from the 850s until the early eleventh century, and Byzantine Emperors until the final decades of the twelfth.
Unfortunately there is very little direct evidence about conditions in Kosovo during those earlier centuries of Bulgarian and Byzantine rule. We can assume that the Slav population that had settled in Kosovo was brought within the cultural realm of the Bulgarian empire, which means that it would have been included in the Bulgarian dioceses of the Orthodox church. Thanks to the work of Saints Cyril and Methodius (and their followers) in the ninth century, the Slavs had a liturgy and other texts in their own language, written in either of two newly invented alphabets: Cyrillic and Glagolitic. The western macedonian town of Ohrid developed strongly as a cultural and religious centre in the ninth and tenth centuries, and by the end of Tsar Samuel's reign the archbishopric of Ohrid included bishoprics in Skopje, Lipljan (Alb.: Lipjan; a town just south of Pristina) and Prizren.  Although the formal division of the Christian Church into Roman Catholic and Eastern Orthodox did not occur until 1054, it would not be anachronistic to describe this Bulgarian Christianity as Eastern in the ninth and tenth centuries; the roots of the conflict between East and West went back a long way. (The Slav liturgy was at first violently rejected by the Roman Church, on the grounds that God spoke only three languages: Hebrew, Greek and Latin).
1. Fine, Early Medieval Balkans, pp. 26-7; Obolensky, Byzantine Commonwealth, pp. 85-6; Goldstein, Hrvatski rani srednji vijek, pp. 88-90.
2. Velkov, Cities in Thrace, pp. 31-46, gives a good summary; V. Popovic, 'L'Albanie', p. 259, discusses the Goth attacks on Albania in 459 and 479.
3. The Bulgar language disappeared by the mid-ninth century: Runciman, Bulgarian Empire, p. 93. From then on, the term 'Bulgarian' can be used generally to refer to the political unit which the Bulgars created, to its mainly Slav population, and to their Slav language.
4. Velkov, Cities in Thrace, p. 49; Skok, Dolazak Slovena, p. 105; Hammond, Migrations, p. 66 (suggesting that they entered Macedonia through the Kacanik pass).
5. Runciman, Bulgarian Empire, pp. 22-4 (traditional view); Fine, Early Medieval Balkans, p. 28 (new evidence).
6. Fine, Early Medieval Balkans, pp. 49-59. Goldstein, Hrvatski rani srednji vijek, pp. 87-92, gives a more hesitant treatment of Constantine's story, and Whittow, Orthodox Byzantium, p. 263, rejects it.
7. Fine, Early Medieval Balkans, p. 57; on the social organization of the early Serbs see Gimbutas, Slavs, pp. 140-1.
8. Makushev, O Slavianakh, p. 2; Stadtmuller, Forschungen, p. 128.
9. Nopcsa. 'Beitrage'. p. 238; Ducellier, Facade, pp. 70, 196; Selishchev. Slavianskoe naselenie, pp. 73-85.
10. Lemerle, ed., Les Plus Anciens Recueils, vol. 1, p. 186; Velkov, Cities in Thrace, pp. 52-3; Howard-Johnston, 'Urban Continuity'; Whittow, Orthodox Byzantium, pp. 267-8.
11. Jirecek, 'Die Romanen'; Sufflay, 'Stadte und Burgen', p. 36; Stadtmuller, Forschungen, p. 129.
12. Whittow, Orthodox Byzantium, p. 268; V. Popovic, 'Albanie', pp. 269-72.
13. Nopcsa, 'Beitrage', p. 238 (Dalmatians in Albania); Mirdita, 'Balkanski Vlasi', pp. 75-6 (citing and analysing the statement by the eleventh-century writer Kekaumenos).
14. For good general studies of the Vlachs, see Wace and Thompson, Nomads;Weigand, Aromunen; and Winnifrith, Vlachs.
15. Weigand, 'Albanische Einwanderung', p. 225.
16. For a powerful presentation of the evidence see Schramm, 'Fruhe Schicksale'.
17. E.g. Niederle, Slovanske starozitnosti, vol. 2, map opposite p. 296; Angelov, Obrazuvane, map on p. 155.
18. See van Wijk, 'Taalkunde gegevens'; quotation from p. 71. The modern dialect of Serbo-Croat which borders Macedonian and Bulgarian territory, the 'Timok-Prizren' dialect, does have some transitional features; but research has shown that it picked them up only after the medieval expansion of the Serbian state into Kosovo and the Morava valley, which brought its speakers into closer contact with Bulgarian (ibid., pp. 62, 71).
19. Runciman, Bulgarian Empire, pp. 88-93, 127-38; Selishchev, Slavianskoe naselenie, p. 61; Stadtmiiller, Forschungen, pp. 129-32.
20. Whittow, Orthodox Byzantium, pp. 387-8. Hilferding, Geschichte, p. 35 n., notes that after Samuel's death in 1014, the local Slav tribes in Eastern Kosovo formally submitted to Byzantine rule.
21. Stadtmuller, Forschungen, pp. 162-4.
22. Gregoire and de Keyser, 'La Chanson'; Gregoire, 'La Chanson'; Hammond, Migrations, p. 56.
23. Sufflay, 'Stadte und Burgen', p. 203; duca is the Italian title.
24. Jirecek, 'Albanien', p. 69; Sufflay, 'Povijest', p. 227.
25. For slightly different eleventh-century locations see Stadtmiiller, Forschungen, pp. 167-73, and Ducellier, Facade, p. 80; the twentieth-century name was noted by Father Gjecov, and reported in Dema, 'Shqypnija katolike', p. 532 n.
26. The same root gave rise to 'Albion', an early name for Britain, and another name derived from it, 'Albany'. There is also a territory known to classical geographers as Albania in the Caucasus; some earlier writers, such as the French diplomat de Pouqueville, supposed that the Balkan Albanians came from there, but there is no connection between the two areas.
27. This probably derives from another version of the place-name, Arberia, a term for the highlands between Vlora, Gjirokastra and the sea: see Stadtmiiller, Forschungen, p. 177.
28. Sufflay, 'Povijest', p. 200 (documents); Stadtmiiller, Forschungen, p. 70 (eagle). The 'he who understands' argument may possibly be the wrong way round; in Hungarian, for example, magyardzni means 'to explain', but only because its original meaning was 'to put into Magyar', Cabej notes that the presumed derivation of shqipoj from Latin excipere is very doubtful ('Zur Charakteristik', p. 194). Another derivation of shqip, suggested by Skok, from the place-name Scupi (Skopje; Alb.: Shkup), requires some unusual sound-changes (see Schramm, Eroberer, p. 361).
29. Curiously, it is treated as authentic by Hammond, Migrations, pp. 56-7. But the document, a Ragusan chronicle written probably in the fourteenth century and surviving only in an eighteenth-century copy, evidently described much more recent events: see Makushev, 'Issledovaniia', pp. 204, 303-32.
30. E.g. Cabej, 'Problem of place', p. 79.
31. Skendet Anamali has argued the Illyrian-Albanian case in a series of articles: see 'Problemi i kultures' and 'De la civilisation'. But as Vladimir Popovic points out, the finds at Koman and the Kruja necropolis are simitar to those at other semi-isolated Romano-Byzantine towns of this period in Corfu and Dalmatia: 'L'Albanie', pp. 269-72.
32. Durham, 'Antiquity'; for a strong argument against tribal continuity see Kaser, Hirten, Kampfer, pp. 39-43.
33. Stadtmuller, Forschungen, pp. 138-9, 145-7; Selishchev, Slavianskoe naselenie, pp. 176-81; Jokl, 'Slaven und Albaner', pp. 291-7, 315.
34. Schramm, Anfange, p. 151. The best evidence is from the area of Debar and the Black Drin.
35. Stipcevic, Iliri, pp. 27-30; Wilkes, Illyrians, p. 68; Wiesner, Thraker, p. 27.
36. Stipcevic, Iliri, p. 30 and n.; Mirdita, Studime dardane, pp. 7-46; Papazoglu, Central Balkan Tribes, pp. 210-69. As Papazoglu notes, most ancient sources classify Dardanians as Illyrians. Her reasons for rejecting this identification in a later essay, 'Les Royaumes', are obscure. There were Thracian names in the eastern strip of Dardania, but Illyrian names dominated the rest; Katicic has shown that these belong with two other Illyrian 'onomastic provinces' (see his summary in Ancient Languages, pp. 179-81, and the evidence in Papazoglu, 'Dardanska onomastika').
37. Decev, Sprachreste, pp. 575, 579 (by Decev and Georgiev respectively).
38. Katicic, Ancient Languages, pp. 169-70.
39. Ibid., p. 171 (cloud); Decev, Sprachreste, pp. 543, 544 (blackberry), (Cabej notes a Swiss (Rhaetoromance) word for a raspberry, mani, and suggests that the term was originally an Illyrian word, which spread both west into Alpine Latin and east into Thracian ('L'lllyrien', p. 52). Cimochowski also argues that 'mantia' could be Illyrian: 'Prejardhja', p. 38. The oft-cited link claimed by Weigand between the Thracian word for 'thyme' and the Albanian for 'peas' is now rejected: see Weigand, 'Albanische Einwanderung', p. 209; Philippide, Originea Rominilor, vol. 1, p. 375; Decev, Sprachreste, p.554.
40. Mladenov, 'Albanisch', p. 183 (Carlo, Lodovico).
41. Cabej, 'Problem of place'; Schramm, Eroberer, p. 293; Huld, Basic Etymologies, pp.48, 121-2.
42. For this important argument see Gjinari, 'De la continuation'. On Thracian compound names see Georgiev, 'Thrace et illyrien', p. 73; Katicic, Ancient Languages, pp. 139-41.
43. The best discussions of this issue are Pisani, 'Les Origines'; Cimochowski, 'Prejardhja', pp. 41-5. See also Mayer, Sprache der Illyrier, vol. 1, p. 12; Katicic, Ancient Languages, pp. 174, 184. One more recent attempt to prove that Illyrian was centum is by Schramm, Anfange, pp. 26-7. But his argument rests only on one speculative etymology for a river-name, connecting it with an Indo-European root for 'knee': this does not match the known derivation from that root in Albanian (see Huld, Basic Etymologies, p. 70).
44. Katicic, Ancient Languages, p. 163; Rosetti, Thrace, daco-mesien', p. 81.
45. Polome, 'Position'; Hamp, 'Position', p. 111. Based on the assumed Messapian link was another argument, about the accentuation of the first syllable in place-names (Brindisi, for example, preserves the Messapian accent): some Albanian names do this and others do not. Dropping the Messapian-Illyrian connection removes this problem from the agenda.
46. See Huld, Basic Etymologies, pp. 159-61. Huld finds the classification particularly unhelpful for Albanian, which diners in some ways from satem languages without being identifiable as centum.
47. Georgiev. 'Albanisch, dakisch-mysisch'. See Hamp, 'Position'; Rosetti, Thrace, daco-mesien'; and, for the fullest demolition, di Giovine, Tracio, dacio ed albanese'.
48. Cabej, 'Problem of Autochthony', p. 43; Katicic, Ancient Languages, p. 186; Mihaescu, 'Les filements', p. 325.
49. This claim is put forward as a prime argument against the 'Illyrian' origins of the Albanians by Schramm: Eroberer, pp. 33-4; Anfange, p. 23. It had already been answered by Cabej, who pointed out that the shift to 'h' belonged to a much earlier (pre-Roman) period of Albanian: 'Problem of Autochthony', p. 44. Schramm's case can be disproved by a series of Albanian borrowings from Latin, such as shkorse (rug) from scortea, shkendije (spark) from scantilla, shkemb (rock-formation) from scamnum, and shkop (staff) from scopae: see Capidan, 'Raporturile'. pp. 546-8; Philippide, Originea Rominilor, vol. 2, pp. 653-4; Cabej, 'Zur Charakteristik', p. 177; and the entries in Meyer, Etymologisches Worterbuch.
50. Jokl, 'Slaven und Albaner', pp. 287. 618, 627.
51. Weigand, 'Sind die Albaner?', p. 233.
52. See Cabej, 'L'lllyrien', p. 46, and the comments in Hamp, 'Position', p. 98.
53. It is sometimes imagined that the shepherds of the northern Albanian mountains must always have grazed their flocks on the coastal plains in the winter, but this is not correct. Many move only from summer pastures in the mountains to winter pastures in nearby valleys: see Kaser, Hirten, Kampfer, pp. 57-67.
54. Skok, Dolazak Slovena, p. 22 (Plovdiv area); Schramm, Fruehe Schicksale', (ii), p. 104 (Ohrid area); Schramm, Eroberer, pp. 115-30 (mountain areas). St Jerome referred to Illyrian-speakers in Dalmatia or Pannonia in the fifth century, but their location is uncertain: Mirdita, 'Ceshtja e etnogenezes', pp.638-9.
55. Schramm, Anfange, p. 232. In later Byzantine usage, 'Bessoi' became a general name for Vlachs (see Cankova-Petkova, 'La Survivance'). Perhaps because of this, Tomaschek argued ('Die alten Thraker', (1), p. 77) that these monks were speaking a Balkan Latin, and that bessam was just added in the manuscript as a gloss on latinam; but this is refuted by the evidence of the two earliest MSS (see Milani. ed., Itinerarium, p. 204). There is other evidence, of Christian 'Bessf in Constantinople and Jerusalem:' Schramm, Anfange, pp. 112-20. Irfan Shahid's attempt to identify Bessan here with Arabic is unconvincing (Byzantium and Arabs, Fourth Century, pp. 320-1), but his location of Lakhmids in the region (ibid., pp. 31-60, and Byzantium and Arabs, Sixth Century, p. 979) must overturn a recent claim that the 'Lachmienses' at Sinai were Vlachs (Nandris, 'Jebaliyeh').
56. Schramm, Anfange, pp. 149-56.
57. Ibid., pp. 48-77. Even the Goths held out for a long time: a small Gothic-speaking population existed in the Crimea as late as the sixteenth century.
58. For useful listings see Haarmann, Der lateinische Lehnwortschatz, pp. 105-8; Philippide, Originea Rominilor, vol. 1, pp. 665-76.
59. Schramm, Anfange, pp. 94-5.
60. There are two common Albanian words for flock, kope and tufe.
61. Jirecek, 'Die Romanen', (i), p. 13; Philippide, Originea Rominilor, vol. 1, pp. 70-2; Papazoglu, 'Les Royaumes', pp. 193-5. Albanian does preserve a very small quantity of borrowings from ancient Greek: see Thumb, 'Altgriechische Elemente'; Jokl, 'Altmakedonisch'; Cabej, 'Zur Charakteristik', p. 182. This low level of borrowing from Greek is a further argument against the identification of Albanians with Bessi, part of whose tribal territory was Hellenized: see Philippide, Originea Rominilor, vol. 1, pp. 11, 283; Velkov, 'La Thrace', p. 188.
62. Stadtmiiller, Forschungen, pp. 118-22; Zeitler, 'Das lateinische Erbe'.
63. Schramm gives a valuable survey of the literature and the evidence: 'Fruhe Schicksale', esp. pp. 112-15. See also Pipa's comments on the symbiosis in Albanian Literature, pp. 62-75.
64. On substratum vocabulary see Capidan, 'Raporturile', pp. 457-83. For an etymology of copil see Reichenkron, 'Vorromische Elemente', pp. 242-3. One key word, vatra (hearth), suggests that the original substratum may have been a widespread 'Albanoid' group, of which Albanian is the only survivor: Hamp, 'Distribution'. But below that there may have been a sub-substratum of pre-Indo-European words: for examples (connected with Basque) see Polak, 'Die Beziehungen', pp. 213-15.
65. On Dalmatian, which was recorded just in time from its last speaker in the 1890s, see Mihaescu, La Romanite, pp. 91-130.
66. Weigand, 'Sind die Albaner?', pp. 231-2.
67. Earlier studies linked Albanian exclusively with Romanian; more recent ones have tried to prise them apart, especially if written by Albanians trying to keep Albanian origins in Albania, or Romanians trying to keep Romanian origins in Romania: see Cabej, 'Zur Charakteristik'; Mihaescu, 'Les Elements' Mihaescu uses Latin Christian vocabulary in Albanian to emphasize its divergence from Romanian, but this is highly misleading: Romanian has a different vocabulary here simply because Romanians were later brought under the Orthodox Church.
68. Jirecek, Geschichte der Bulgaren, pp. 112-13 (admitting some agriculture too); Howard-Johnston, 'Urban Continuity', p. 251 (towns only).
69. Capidan, 'Romanii nomazi', pp. 205-8.
70. Dinic, 'Dubrovacka trgovina'.
71. The Montenegrin highlands are rather neglected in most studies of these issues; but they clearly had a well-established Vlach population by the early fourteenth century, when Vlach place-names are recorded there: see Sufflay, Srbi i Arbanasi, p. 75; Radusinovic, Stanovnistvo, p. 31 (and for Albanian names in Montenegro, see above, n. 24).
72. Capidan, 'Raporturile'; Schramm, 'Fruhe Schicksale'. Weigand, 'Albanische Einwanderung', shows that some Albanians went with the Romanians into Transylvania.
73. Cerskov, Rimljani, p. 54; Mirdita, 'Rreth problemit'.
74. Cerskov, Rimljani, p. 55.
1. Gelzer, Patriarchat, p. 4; Gjini, Ipeshkivia, pp. 79-80.