Invisible Britons: the view from linguistics* Richard Coates University of Sussex This version was made available

at the conference Britons in Anglo-Saxon England, organized by Dr Nicholas Higham, Manchester, 14-16 April 2004. Introduction It is common scholarly knowledge, and an essential element in the English national foundation story, that the Angles, Saxons and other peoples did not enter an empty landscape when they hauled their ships up on British beaches in the fifth century with the intention, for the first time, of making a more lasting impact than a simple raid. The land was occupied by people speaking a Celtic language ancestral to Welsh and Cornish, called by scholars British or Brittonic, depending on the exact period. British is the language of Britain during the Roman period, and Brittonic is the term used for this language in its developed state after a large number of important changes had affected it around the years 450-600 C.E. (Jackson 1953; Schrijver 1995, 2002). The problem is to ascertain what relations between the Brittonic and English communities were like in this crucial period. Traditional accounts, as is well known, focus on the expulsion, extermination and enslavement of Britons. This is held to explain the marked lack of Brittonic place-names surviving in much of England, especially in the south and in the east (where an overlay of Scandinavian complicates the interpretation of the absence of Brittonic names), and for the lack of Brittonic lexical borrowings into English. On the face of it, this view is not unreasonable; such a scenario will account for the facts, and an uncritical assessment of the evidence transmitted from and about these times broadly encourages us to adopt this view. Recently, a revisionist account (Higham 1994, 2002a; Ward-Perkins 2000; Matthews 2001) has gained in popularity, according to which the Britons would not have been entirely silenced in these ways, but would have continued to form a majority of the population. Hard evidence for this view could be: the archaeological evidence of continuing Brittonic funerary customs, DNA evidence from burials, evidence for the continuation of farming practices and of other customs (e.g. religious ones), and linguistic evidence from inscriptions and from borrowings of place-names and vocabulary. Mere continuation of farm boundaries tells us nothing, of course. It appears that the strongest reason for thinking the Angles and Saxons did not take over in the conventional sense is the perception that they just couldn't have done it, i.e. either there couldn't have been enough immigrants, or they couldn't or wouldn't have displaced practically all the Britons even with sufficient manpower. These objections to the traditional view seem fundamentally to be logistic ones. Whilst it is perfectly possible in principle that archaeological evidence will eventually show much Brittonic survival, it is not available yet; Yorke (1995: 69) affirms Brittonic numerical dominance for Wessex, despite the fact that so far there has been little archaeological evidence to support the contention. The application of DNA study to skeletal material has been patchy, and we await a fuller representative survey of burial-sites across the country. The possible continuation of religious practices in Wessex is interestingly set out by Yorke (1995: 155-65, 177-81). My purpose in this paper is to argue that, whatever may come from archaeology, the linguistic evidence favours the traditional view, at least for the south and east.

The reasoning will be based not merely on the relatively small amount of place-name and vocabulary borrowing in this area, but on comparison with the linguistic consequences of other invasions and conquests by military aristocracies and the settlers who may or may not have followed them. I argue that there is no reason to believe large-scale survival of an indigenous population could so radically fail to leave linguistic traces. The linguistic literature on contact We must look at the literature of linguistics which deals with the effects to be observed when languages come into contact. Moravcsik identifies lexical borrowing as a prerequisite for any other type of borrowing (Moravcsik 1978: 110; Trask 1996: 314), and this testifies to contact of the lowest intensity, where all contact is essentially about concrete situations and physical or conceptual necessities. Thomason and Kaufman (1988: 74) also rate a situation in which only vocabulary (and specifically content words) is borrowed as representing the lowest, least intense, degree of contact. Borrowing will not take place at all without the prospect of projected gain for the borrower (Winter 1973: 201; McMahon 1994: 201), and equally borrowing will be avoided in situations where the unconscious use of borrowed material will result in stigma for the borrower. Borrowing must be socially and culturally risk-free in situations where something more than need-driven communication is at stake. For cultural and functional (rather than typological) reasons, non-basic vocabulary will be borrowed before basic vocabulary - or rather, vocabulary will not be borrowed where it would cover part of the denotational range of existing vocabulary, which has to do with the distribution of expertise and knowledge in speakers of the source and the borrowing language (Croft 2000: 205). McMahon (1994: 203-4) suggests that this basic level of contact typically results in the borrowing of place-names and terms denoting topography and landscape, together with other terms contributing local colour - essentially, those terms which have no equivalent in the borrowing language because its speakers have never encountered the topography or objects in question before. This results in a demonstrable gain in the expressive power of the borrower's language, and cannot produce the stigma of using borrowed vocabulary in place of native vocabulary. This is what we see in the cases of the material to be discussed below. Johanson (2002) examines contact-induced language-change within a framework representing what it is possible to copy from one code to another, and views copying at any linguistic level as a reducing of the difference between the codes in the repertoire of a particular user. It is clear that the material we are going to examine is the least theoretically difficult from Johanson's point of view. Borrowing the sorts of item in the list above simply absolves them from belonging necessarily and solely to one code and makes them available to both, though their use in the borrowing language may at first be felt to be quotative (spoken in inverted commas), marked and exotic (and therefore risk-free and not stigma-producing) until the concepts which these words denote become themselves an accepted part of the discourse of users of that language. In talking about borrowing, we accept that what is borrowed is not identical, in a linguistic sense, with its source, and that use of this term may lead to misleading metaphorics (Johansson 2002: 288). In the light of this literature, let us consider in turn what is to be learned from the borrowing by the Anglo-Saxons of Brittonic toponymic and lexical material, seeing that they did not borrow (or at any rate cannot be proved to have borrowed) any features of morphology or syntax.

Borrowing involving Brittonic and Old English: some general comments What, if anything, the Britons and the English said to each other during these earliest contacts has become controversial. The Britons certainly communicated some of their place-names to the newcomers. In recent work, I (like many predecessors) have acknowledged this Celtic contribution in the landscape to the emerging polities that became England, but have also argued that there are more such traces than has generally been recognized (see the papers in Coates and Breeze 2000 and subsequent papers by the same authors), but not a huge amount more. It is possible that the Britons also gave some of their personal names to the English. If so, that would be an indication that the Britons were a sufficiently positive outside influence in English society for the English to want to imitate their naming practices and therefore for Clarks First Law of Applied Anthroponymics to apply (Clark 1979/1995).1 Unfortunately we cannot tell whether a Brittonic name in an Anglo-Saxon source is truly the name of a Briton or a Saxon even where it is ostensibly that of a Saxon, as in the case of Cerdic or Cdmon (Coates 1990 vs. Parsons 1997; Clark 1992: 463), although Clark plausibly suggests that Brittonic names may have been chosen for the fruits of early dynastic marriages between Brittonic and English royals. It is safest to assume that Brittonic names in such texts as the Durham Liber vitae (Sweet 1885: 153-65) belong to Britons. Also, we cannot reliably identify Brittonic names in the form in which they were used by the English. Redin (1919) contains material that might serve as a starting-point for identifying such names in use in the Anglo-Saxon period. But he observes in relation to one of the most plausible candidates, Tuda (1919: 71) that it will be safest to assume that Tuda is a short form of ... Celtic compounds [in *Touto-, RC], perhaps coined by the Anglo-Saxons, and that leaves us unable to be sure what the available anthroponymic material tells us about ethnicity and therefore about what the English actually adopted from the Britons. The survey of Clark (1992) does not address the question of which of otherwise obscure English short-names might be of Brittonic ancestry. Given these difficulties, no conclusions will be drawn in this paper from personal-name evidence. We need to confront the apparent paradox that the Angles and Saxons seem content to have taken some place-names from the Britons - not an enormous number, but, overall, not negligible either and yet took practically no Brittonic vocabulary in the earliest centuries of settlement. There was practically no early lexical traffic in the other direction either (Parry-Williams 1923: ch. 2), and all we have for sure is the talismanic word cyulis (Saxon long)ships in Gildas (de excidio Britanniae 23), which is actually nothing more than a mention rather than a use - Gildas glosses it in the running Latin of his text - and therefore not certainly a borrowing. This all appears to suggest little contact in which meanings were exchanged. Vocabulary-borrowing presupposes meaningful human interaction between speakers of different languages and is therefore a secure sign that that has happened. When European powers set up colonies and imperial administrations, the areas administered typically gave English and the other languages a considerable amount of vocabulary, notwithstanding the degree of technological development of the area which they ruled. At one end of the scale is the legacy of imperial rule in India which is evident from the huge Hobson-Jobson vocabulary, with sources in many different languages of India (represented in the 1000 pages of Yule and Burnell, ed. Crook, 1903); at the other is the uptake of an estimated 200 words (from a wide range of indigenous languages and of wildly differing text-frequency) into Australian English from the various Aboriginal languages (Yokose 2001). Many of the expressions found in Indian English are also found in the general standard language of the high imperial period and not just in the discourse of expatriates or local adopters of English, and they belong to a wide 3

range of semantic fields (sahib, raj, raja, rani, nawab/nabob, brahmin, khidmutgar, ayah, (punkah-) wallah, nautch(-girl), dhobi, mahout, pundit@, sadhu, swami, yogi, pariah@, thug@, sari@, dhoti, pyjama(s)@, cashmere@, khaki@, sati/suttee, ghat, nirvana, karma@, yoga@, swastika@, mongoose@, cheetah@, gaur, jairou, nilgau, bandicoot, jungle@, bhang, char(@) `tea', chota peg `whisky and soda', betel, ghee, curry@, korma, chapati, dal, naan, rupee, pice, chit@, pukka(@), cushy@, dekko(@) look, sitar@, tabla, polo@, gymkhana@, durbar, bungalow@, Blighty). Words marked with @ are exceptionally well-embedded (by my subjective judgement) in the modern standard language even in Britain, and those marked (@) were formerly. Those due to native Australian languages tend to be more clearly concentrated in those semantic fields identified by McMahon (1994), namely (1) native flora and particularly fauna and (2) aspects and objects of Aboriginal culture (mulga, budgerigar, dingo, koala, wombat, wallaby, kangaroo, barramundi; corroboree, boomerang, didgeridoo). The same might be said of borrowings from Native American languages, including Inuktitut (Eskimo), directly or indirectly into (North American) English. Such borrowings may pass into the common standard language, at least to the extent that exotic locations become part of everyday discourse as informed by literature, film and the media generally. I have no expert knowledge or even direct experience of America, but I can rehearse over 30 such borrowings unprovoked: canoe, kayak, tepee, wigwam, igloo, totem, potlatch, toboggan, papoose, squaw, wampum, tomahawk, moccasin, anorak, parka, G-string, pow-wow, hominy, tapioca, squash `gourd', pecan, potato, tobacco, terrapin, moose, caribou, woodchuck, chipmunk, skunk, opossum, raccoon, sequoia, catalpa, hickory. The English took over Indian toponymy practically wholesale because they took on the pre-existing nucleated settlement pattern of India practically wholesale (few English names appear on maps of the Sub-Continent), and allowed Australian and Amerindian languages to have a proportionally smaller but still very significant impact on place-naming, as a glance at maps of these areas reveals even before one enters the specialist literature. In such a light, one might think that the amount of lexical borrowing and the uptake of toponymy are related in easily graspable ways to the degree and nature of the interaction between the invaders and the invaded. We must explore what this might mean in post-Roman Britain. No-one, to my knowledge, has demonstrated conclusively that Brittonic had an impact on English grammar. In the longer term, the Celtic languages have certainly had deeper effects, it has been argued, on the development of both dialect and the standard language - for England itself see especially the papers by White and Klemola in Filppula et al. (2002), and Tristram (1999) - and there may be significance in the similarity of the typological shifts undergone by both languages (Tristram 2002). Most provocatively, Klemola draws attention speculatively to the West Wiltshire heartland of DO-support in declarative sentences, and compares the toponymic evidence for the survival of a late variety of Brittonic in the same area adduced by Coates (2002). But since it is hard to be absolutely sure of causal connections between Brittonic and English grammatical phenomena because of the time-depth and the silence of the record at key periods, my discussion of the linguistic literature on borrowing will focus on lexical borrowing which has undoubtedly happened, be it never so meagre in amount. The borrowing of place-names by the English @@@ For basic etymological information about place-names, see Ekwall (1960), Watts (2004) and the volumes of the Survey of English Place-Names mentioned in the references to this paper.

First of all, I shall briefly review the evidence that Brittonic place-names were borrowed in England, indicating how scholarship about this matter has changed since the 1920s when it was first begun. I shall include river-names not as examples of Celtic names, since large numbers of river-names show evidence of being coined in a language spoken in Britain even before British, but as evidence that such material was transmitted by Brittonic-speakers to the English. I also specifically exclude Cornwall, where the presumption must be that a given place-name is of Brittonic origin unless demonstrably English, and where naming in Cornish was an active process whilst naming in English was taking place elsewhere, as demonstrated by the structurally late character of most of the surviving names. In early English toponymic scholarship, the default hypothesis for an etymology was that the name was English. The most Celtically-inclined of the first generation of place-name scholars, and the one with competence over the widest range of languages, Eilert Ekwall, advised that once students have analysed the names of rivers, hills and forests, they ... would do well to try as far as possible to explain place-names belonging to other categories [i.e. mainly names of inhabited places, RC] with the help of Germanic material. The fact that a name is difficult to explain or has an unusual appearance should not be taken to point to pre-English origin, unless there is some special circumstance to render it plausible (1924: 27). Leaving aside the primary topographical features, this means that unless Celticity leaps out and hits one between the eyes, the approved strategy is to argue for Germanic origin. The reasoning underlying Ekwall's position is presumably as follows: (1) mass English immigration was assumed, and (2) the absence of Brittonic habitation-names is only to be expected given that their villages were non-nucleated and that their dwellings were ephemeral. (Gerald of Wales, writing as late as the last quarter of the 12th century, noted that Welsh houses may consist of stacked branches.) The English would therefore have been able to destroy them, or ignore them, at will, even where they did not enslave the inhabitants. But the fact remains that there are Brittonic names widely spread through England whose original denotata were not rivers, hills or forests but human artefacts including dwellings and other buildings. The hall at Liss (Ha), the small hall at Beccles (Sf), the earth-house at Priddy (So), the big artificial mound at Penkridge (St), and those of indeterminate size at Crick (Nth) and Creech (So), the kiln at Onn (St), a church of a named person at Landican (Ch), a possible village of Britons at Carburton (Nt) and the fort at Caerlanrig (Rox), a little settlement at Wigan (La), James's house at Penteiacob (now Eddleston, Pbl) and names in tref `farm, village' in Lancashire and several counties of southern Scotland - these form a set of names which is not geographically restricted. This suggests that where an obscure and difficult name is to be analysed, there is no solid reason to assume a priori that it must be Germanic; a Celtic origin should be considered equally, and any greater probability of Germanic is a matter, at this stage, only of statistical expectation deriving from already established patterns, not a probability inherent in the very name to be analysed. In other words, if it seems Celtic, there is no reason to presume it isn't. Accordingly, we now find a greater open-mindedness about the possible survival of Brittonic names, and Coates and Breeze, in their book and in subsequent papers, offer possible additions to the canonical list of survivors. The distribution of Celtic place-names Although the absolute number of surviving Brittonic place-names may have been underestimated, it has been known for a long time that their distribution is uneven. The state of knowledge about this in the mid-twentieth century is enshrined in Jackson's well-known and much-reproduced map of Celtic 5

river-names in LHEB (1953: 220), which divides England into four zones of increasing Celticity as one travels westwards. The eastern boundary of his area III, which cuts off much of the north-west, the Welsh Marches, and the south-western peninsula, bears some relation (especially in the south) to the line on his other well-known map (1953: 208-9) which marks the limit of the Anglo-Saxon political and military advance by the year 600 or so. The general picture indicates a correlation between that advance and the density of Celtic (and also pre-Celtic) river-names, suggesting a problematic relation between English progress and Brittonic survival. It has generally been assumed that what is true of river-names is also true of other categories of place-names, though no nationwide mappings of other categories of early place-names exist. Partial information is given by Hogg (1964), who maps surviving RB place-names in England (amended in Gelling (1988); NB not Celtic ones unrecorded in RB sources), and by Gelling (1992: figs. 29-34), who gives maps showing Brittonic and other ancient names in the counties of the west midlands (exemplified by her fig. 30), and (1988: 91) a map showing names indicating the presence of Britons, some of which of course are English names. Otherwise, such information is available only in county maps accompanying the volumes of the Survey of English Place-Names (SEPN). Over the last twenty years, I have drawn attention to a number of place-names possibly of Brittonic origin. These include Leatherhead (Sr; Coates 1980)).2 The SEPN solution of the name (Gover et al. 1934: 78-9) had been, predictably, English. The cost of this solution was to adduce an element which had never been discovered in place-names before or since, OE lode people, and to invent another one, *ride place suitable for riding. Much more satisfying, and more prudent, is an explanation involving known elements compounded in a known fashion, obeying the phonological development of both Brittonic and English, being consistent with the record of medieval spellings, and being topographically appropriate for a place centred on a ford across the River Mole: Brittonic *ld-rd grey ford. (The proposed etymologies of this and the other names I mention are given, in the order of their presentation, in the appendix.) Somewhat later I examined an Anglo-Saxon period name for Bath (So), Acemannesceaster, which could plausibly be shown to have the RB name of the place, Aquae (Sulis) waters (of Sulis), as its first element, though the morphology of the entire name-form is not without problems. Other suggestions involve Clovelly (D), where I reject an SEPN English solution and give substance to a suspicion of Brittonic origin articulated by Jackson (1953: 226); Ower(moigne) (Do), where assumption of Brittonic origin permits the explanation of two separate medieval spelling-traditions of the name (pace Kristensson 2000; 2001); Merrow (Sr), where the SEPN suggestion of the OE word for pith is a poor etymon, and a Brittonic solution is good in the geological environment of this place; and Helvellyn, a mountain on the boundary of Westmorland and Cumberland, for which a simple Cumbric solution can be proposed despite the late sixteenthcentury first appearance of the name in the record. The name Tric (Coates 1988b; Owen and Coates forthcoming), applying to a location near Skegness (L), can be shown to be of Latin origin but mediated by Brittonic pronunciation. Those difficult names such as Clovelly and Owermoigne which appear in the record for the first time in the Anglo-Saxon, or even the post-Conquest, period offer the most problematic category. In the overwhelming majority of cases there is no evidence in Welsh sources, and the English, medieval Latin and French record has to bear the full weight of the interpretation. Nevertheless, a good case can be made for the Brittonic origin of a not inconsiderable number of names, of which a few are mentioned here. Coslany, in Norwich (appears to contain a form identical to the ancestor of Welsh cystlynnedd kindred, relationship, an apparently unique instance of an abstract term used as a place-name, though no doubt acting as a surrogate ethnonym (i.e. the people called The Kindred)). 6

Still in Norfolk, Trunch involves, with topographical appropriateness, PrW *trum ridge and *cd wood. Several Lancashire names are given topographical interpretations by Breeze, e.g. Cuerden rowans, and Wilpshire, an English name based on *wlb damp. Rollright in Oxfordshire is made up of the Brittonic phrase *rodland rx, where *rodland wheel enclosure refers to the circular megalithic monument the Rollright Stones and *rx groove to the unusually-shaped narrow gorge by which Great Rollright village stands. No grand claims for the number of Celtic survivals should be made. There are, I believe, quite a lot more than has been acknowledged hitherto, but the number is still not enormous and (with some local exceptions that I discuss immediately below) the newly-recognized names do not radically distort the accepted picture, which is consistent with de-celticization. Early contacts, for whatever reason, seem to have afforded relatively little opportunity for Britons to pass on place-names to the English; later and more westerly ones afforded more. Interesting Brittonic place-name distributions The analyses that Breeze and I have offered permit some interesting observations about the distribution of Celtic names, and I shall mention just two here to give some substance to the point that however barren of such names much of the south and east are, there are areas where their incidence points decisively to the survival of Brittonic-language enclaves. We shall look at instances in the region south of Sheffield and in north-west Wiltshire. There is a small knot of wholly or partly Brittonic names in the Derbyshire/Nottinghamshire boundary area: the river Poulter, the stream which gives its name to Clowne (Db) and Clumber (Nt), and Mansfield (Nt). Clowne is just five miles from the township of Wales (YWR), whose name means exactly what it appears to: it is an ethnonym meaning The Welsh used as a place-name. In the middle of this area is Carburton, which has no satisfactory explanation. In DEPN, Ekwall refers to a considerable hill in accounting for the first element; the SEPN editors for Nottinghamshire deny the hills existence and offer no solution of their own. The simplest course is to take it to be *Cair Brtton Britons village, and accept the consequences. Coincidentally, this name-form is actually on record, in the list of the 28 cities of Britain in Nenniuss Historia Britonum, though it cannot be taken as referring to Carburton there. The name must be late, firstly since it includes the word *cair in its late sense village (there is no hillfort here), and secondly since it is of a Brittonic rather than a British compound-type; Padel (1985: 50-4) believes that such names do not date from before the sixth century. This tells us something, though not novel, about the date of the Anglian advance into this region, well before the eventual colonization of Derbyshire at least. We can also infer that Britons were conspicuous, and arguably independent or simply not submerged numerically, here on the later county boundary at least as late as 600.3 An extremely interesting collection of Brittonic names may be found in Wiltshire, as mentioned above in connection with the possible survival of Brittonic grammatical features. It was remarked long ago by the SEPN editors for this county that the number of Brittonic names was quite large, but this insight can be deepened. There is a body of evidence that suggests the late persistence of Brittonic in the north-west, but it needs careful handling since the etymology of some of the key names is not secure. The evidence is phonological, lexical and structural, and the relevant names are presented on maps in Coates and Breeze (2000: 116 and 391). The phonological evidence is mainly

concerned with the reflex of British intervocalic */m/, which becomes *[v] by lenition, though it continues to be heard by the Anglo-Saxons as [m] until about 600, and thereafter treated as either as [m] or [v], depending in part on dialect and in part on chronology, with some indeterminacy of outcome (Jackson 1953: 491-3). It is generally reckoned that, politically at least, Wiltshire was in English hands by 600, the battle of Dyrham in 577 being decisive. It follows that any borrowings by the English in Wiltshire which show [v] imply contact with Britons after 600. Two names may indeed show [v] for */m/: Cheverell, if based on the ancestor of MW kyfa(i)r piece of land ploughed in common, and Keevil, if borrowed from the ancestor of Welsh cyfyl border, neighbourhood; the latter would show ME open-syllable lengthening which would not be expected in the trisyllabic Cheverell, hence the difference in the treatment of the first Brittonic vowel. The lexical evidence involves the names Chitterne, Minety and Idover. Chitterne, as Breeze has argued, may represent the name-type which has as its modern Welsh form Coetre(f) woodvillage/farm; this is common in (minor) place-names in Wales ([d]igwydd yn dra chyff[redin] mewn enwau lleoedd), according to Geiriadur Prifysgol Cymru (535), but otherwise unknown in England. Minety, whatever its first element (kid or edge), appears to have Brittonic *t house as its second element (though the non-lenition of the first consonant needs special pleading, unless, as is perfectly consistent with the medieval record, this is really a testimony to cross-cultural miscommunication, and means my house, *men t, i.e. Welsh 'y nhy, with regular OE raising of first vowel before a nasal group!). Names for inhabited sites as such, as opposed to names transferred from topographical features, are commonly held to be of late appearance, and if that is so both of these names suggest relatively late Brittonic-speaking in this area. As for Idover, I argue that the most plausible interpretation of the first syllable is that it represents the ancestor of the modern Welsh preposition or prefix uch(-) above or the comparative adjective uwch upper, which share an etymon and have acquired functionally divergent meanings for phonological variants. The name would thus mean `above the water' or upper water. Whether one accepts that etymology or not, the second element is clearly Brittonic *dr water. It is a remarkable fact that this name appears 11 times in north-west Wiltshire; nine of the instances are field-names, and six of these fields contain the source of a stream (a seventh being close to a source). The other three all lie on streams. The tenth name is that of an islet in a stream. The eleventh is the ancient name of an actual stream recorded in pre-Conquest charters. The structural evidence concerns element-order in compounds. There is highly interesting evidence of late-type compounds of the characteristically Brittonic form with the specifier after the head, found outside Wales only in Englands western counties in a very few major names like Pensax (Wo) Englishman's head and in minor names close to the border itself. Freakishly, we find preserved in a continental source an ancient name of Brokenborough, Kairdurberg. The -berg is presumably an English addition to a native name, and whatever the exact etymology of that, it must contain cair in the sense of village, like Carburton (Nt), with a following specifier, there being no hillfort here. One would be inclined to be mistrustful if that were all the evidence we had to go on, but we also have Chittoe. The SEPN editors identified this as Brittonic, but their etymology, a suffixed form of *cd, is indefensible because intervocalic voiceless /t/ would not persist, but be lenited to /d/. Chittoe appears rather to be *Cd tew thick wood, with the cluster /dt/ accounting for the voiceless /t/ implied by the record of English spellings; and the almost universal medieval final <-e> suggests a relic of a plural adjectival inflection -on such as is found in parallel modern Welsh names of the type Coed-duon (coed of course being grammatically plural). A further surprising detail, which cannot be said to confirm beyond dispute the proposed origin of Chittoe but

which illuminates the ecological conditions of the area in a remarkable way, is the existence of the hamlet and Domesday manor of Thickwood in the parish of Colerne, just eight miles to the northwest. This is the only ancient place of this name recorded in the entire OS gazetteer, and must be under suspicion therefore as being a translation of the Brittonic name formerly applying to a much larger area. Taken together, the names Kairdur(berg) and Chittoe are quite compelling evidence for Brittonic of a late type in this area. These place-names suggesting something special about the north-west Wiltshire region become even more interesting in the light of the archaeological evidence assembled by Dr Bruce Eagles for Brittonic material culture in contexts later than the presumed date of the Anglo-Saxon landtaking in this area (Eagles 2001). We may infer with some confidence that there was a small area of Brittonic culture persisting here into the seventh century, of a type denser than that represented by the general rather high background level of Brittonic place-names in Wiltshire. For the purposes of this paper, then, it is clear that the Britons gave alarger number of names to the English than has been generally recognized, but that that number is by no means large, especially in the south and east. Their distribution suggests more persistent survival of cohesivegroups of Brittonic-speakers in a limited number of areas. A fuller discussion of surviving names, their distributions, and the genesis of Celtic voices,English places is given by Coates (2002). Borrowed words and place-name elements We now need to consider what information can be gleaned from the borrowing by the English of functioning Brittonic place-name elements and general lexis, which will be treated as a single class for the purposes of analysis. I mentioned earlier that the tally of lexical borrowings into OE from early Brittonic is small - in Baugh and Cables words, almost negligible (2002: 76). The canonical list is brief; Frster (1921) recognized 14, of which only 3 are still generally accepted: binn manger brocc badger cumb valley, but found only in place-names until its descendant Welsh cwm was reborrowed late in the second millennium There is a native English word covering part of the denotational range of what we now call a valley, namely denu. Gelling (1984: 88-94, 97-9) and Gelling and Cole (2000: 103-9, 113-22) have shown convincingly that a cumb had a strikingly different profile from a denu. It seems sensible to believe that denu was a term which could be applied in the landscape of the ancestral homes of the English in what are now southern Denmark and northern Germany, and that there was a sort of landscape feature in Britain which they felt to be inadequately described by this term. On this reasoning, they borrowed the word cumb for a feature with a bowl-shaped end to contrast with a long, narrow, relatively steep-sided denu typically found for instance in Chalk downland. We get a hint here of the kind of PN-element borrowing found in other situations where unfamiliar topography, geology and ecology presented themselves to English-speakers, and which resulted in the borrowing (with different degrees of discourse frequency) of such terms as mesa, kopje, volcano, pingo, arroyo, bayou, wadi, corrie (coire), karst, tundra and taiga. 9

Frster's canon of borrowed words had not quite been published when Ekwall proposed the Brittonic origin of torr outcrop, peak (1920). This suggestion found general favour (including by Frster himself (1922)); the borrowing may also have been encouraged by the absence of comparablyshaped features on or near continental North Sea coasts. He also suggested funta, a Brittonic mediation of Latin fontana, and this is also universally accepted. First attested after the OE period, coble (ferry-)boat can probably be added, possibly because this inland-waterway form of transport was of a different construction from the Saxon boat, possibly wickerwork as opposed to planking. There has been a recent argument (Breeze 1993; Horovitz and Coates 2000) that OW/MW genou mouth was borrowed early, principally (but not only) as a place-name element. It has long been thought that crag rock is a further possibility, though aspects of the phonology of this word are obscure (cf. and ctrst. MW creig). Frster's original work was discussed further by Ekwall (1922b) and Pokorny (1923), and Frster added another supposed borrowing from Irish to his list (1936: stor) which has now been convincingly shown to be Brittonic by Breeze (1998). Ten items from Frster's original list have been questioned and/or rejected by later scholarship, as referenced here: assa ass, now believed to be more likely from Irish, for morphological reasons bannuc bit (questioned by Brunner 1960: 27; might also be claimed as Goidelic on the basis of the modern (re-)borrowing in the sense flat oatcake (for which see OED-2)) becca fork, Gabelaxt (questioned by Brunner 1960: 27) bratt cloak, now believed more likely to be from Irish (Breeze 1995b) carr rock (for discussion see Parsons and Styles 2000: 143-4) dunn dun (colour-term) (counterargument based on Weyhe 1905: 56-9) gafeluc spear (see now Breeze 1993b, and further comment below) hogg hog (claimed as OE by Coates 1982) mattuc mattock (questioned by Brunner 1960: 27) toroc bung (whose status as an English word has even been questioned, though I have recently argued that it is an English word meaning throat but not of Brittonic origin (Coates forthcoming b)) The state of scholarship on these items in the 1930s was usefully reviewed by Serjeantson (1935: 5560). Kastovskys conservative survey (1992: 318-20) acknowledges as true Brittonic borrowings binn, bannoc, gafeluc, dunn, broc, assen, and, from the glosses to the Lindisfarne gospels, bratt, carr and luh, as well as the toponymic elements torr and cumb, plus funta as a Brittonic-mediated Latin loan (of which others might have been mentioned: see Gelling 1997: ch. 3). He observes that the Celts have left remarkably little behind in English, a phenomenon that has not really been explained satisfactorily. True, the surviving Celts were a conquered race, but their culture must have been more developed than that of the German invaders due to the 400 years of Romanisation, and from that point of view more loans would not have been completely unlikely ... contrary to all expectations, [Celtic] has not really left its mark on the English language[.] (1992: 319-20). 10

But recent archaeological work suggests that at the end of the Romano-British period, on the whole, there was not much to pass on materially, if a band across central southern England from the eastern boundary of Oxfordshire to Essex was typical (Baker 2001). Over a wider landscape, Wacher (1998: 297-9) notes a slow and patchy decline in Romano-British culture after the late fourth century, and refers to its eventual almost complete eclipse (1995: 409). If there is no important material culture, there is nothing to offer by way of lexis to incomers except landscape terms, including place-names. It does not follow that the material culture of the invaders must have been technologically superior. However, Kastovsky's assessment that British culture must have been more developed appears unsound, and possibly dependent on the glories of Celtic or Celtic-derived Christian sculpture and book-creation which are new during the Anglo-Saxon era rather than present at the dawn of it. Work by Breeze during the 1990s has suggested that some words of OE, most of which have not survived into modern English, may expand the list somewhat. The possibilities he has identified include: trem pace and trum strong (1993a), wered sweet drink (1993c), stor incense or better medicinal wax (1998; cf. Frster 1936), dor brave (1997), and perhaps even baby (1993a); see the fuller list of claims and the wider discussion in Breeze (2002). He has also (1993b) returned to an earlier view that at least one of the controversial ones, gafeluc javelin, is after all ultimately Celtic but shows that it is actually Goidelic, and must have entered English via Old Norse. Borrowed early non-Brittonic Celtic material is equally sparse. One might adduce a couple of possibilities from the period of continental contact: rce realm, dn hill(fort), sern iron; and some early words authoritatively and traditionally reckoned to be of Irish origin, in some cases possibly mediated by Brittonic: ancor(a) hermit, assa ass, bannoc bit, bratt cloak; child (acc. Breeze), clucge bell, dr magician, str history and probably even Christ (on the basis of thelong vowel, a Celtic phenomenon: OW Crist, OIr Crst); and also a single-figure list of other words attributable to the influence of Irish missionary Christianity, of which the only two remaining in use in Modern English are curse (acc. Tolkien; Breeze 1993b) and cross (and that, like gafeluc, via Old Norse) - and some added to by Breeze in recent work: Beltancu Beltaine cow, heriot (1996) and deorc bloody(red) (1995a). But even taking all these into account it is clear that the total impact of Celtic on OE in the era of colonization, expansion and consolidation is extremely small, and that of Brittonic only a part of it. But to concentrate on Brittonic: two words were borrowed which became fully lexicalized, joined the English onomasticon and were used as place-name generics: cumb and torr, as already mentioned. Additionally, some words were borrowed for which there is no evidence outside placenames, and about which it must be suspected, unless further information comes to light, that the English took over monomorphemic Brittonic words for landscape features as if they were proper names, possibly many times over, in such a way that no new lexical item of the relevant form entered English. Into this category seem to fall genou and crag, as already mentioned, and the following: *cd wood, frequently found in wood-names in forms like cheet, chet, chat; the southwestern term *cors reeds, bog, e.g. in names in Somerset, Worcestershire and Gloucestershire; *crg barrow, frequently found countrywide in barrow-names in forms like crook, crick, creek, creech, these names collectively deserving further study because of their perplexing historical phonology; *egls church, regularly found as eccles; *lux lake, in some Northumbrian local coastal names [possibly lexicalized] and in Lutton (L, Holland); *penn head, top, end (confusable with certain OE words); *poll/*pull pool (phonological uncertainties, unclear connection with modern pool with which it may sometimes have become identified); and *ros moor, widespread though with a regional phonological complication concerning the length of the vowel in borrowed names. 11

Significantly, none of these words is ever found as the generic in an early two-element English place-name, though luh may have been so used in early Northumberland if the suggestive evidence of some modern coastal names is taken into account, whilst some are used as single-element names (Creech, Crick, Eccles, Penn, Ross, Roos). That reinforces the view gained from the minimal amount of lexical borrowing: that Brittonic was not much understood by the incomers, and most items that were borrowed were understood as being used in order to make reference, but not as denoting by virtue of their lexical content. They performed the task of naming, but had in all or most instances no lexical status. This is illustrated perfectly by the fate of Brittonic *aon river, which has given English no lexical item, but appears as the proper name of, not a word for, six rivers of England. If they had become true English words, there is no reason why they should not, like cumb and torr, have served as generics in the same way that their English translation-equivalents did. Parallels for near-zero lexical borrowing and sparse omomastic borrowing? The situation we are seeking to explain, then, is one in which the Britons transmitted to the English a non-negligible but rather small number of place-names, some southern and eastern counties, especially Bedfordshire, Cambridgeshire, Hertfordshire, Sussex, Middlesex, Leicestershire and Rutland, being practically devoid of Brittonic names (apart from those of rivers) altogether, and an extremely small number of lexical items. Are there any other situations in recorded history where a military aristocracy, whether or not backed in due course by settlers, spoke and maintained a language which owed so little to its substrate? The scenario for which we need to find parallels is as follows: A new language arrives borne by military conquest, arguably followed by mass immigration (though this assumption has been vigorously questioned recently, and that is of course a major reason for this conference) The indigenes have a material culture technologically no more advanced than that of the newcomers (Baker 2001: passim) The indigenous language gives practically no vocabulary to the newcomer The indigenous language gives the newcomers only a rather small number of place-names and personal names The indigenous population can be seen documentarily only in scattered (relict) statelets (as for example with small kingdoms recorded by Bede such as Cornwall, Deira and Elmet) and by their concentrated onomastic traces in such areas as north-west Wiltshire and the Pennine Wales (see above), of course Cornwall and Wales (including marcher counties) where their language persisted, and perhaps south Lancashire and parts of the Lake District DNA evidence is equivocal but may suggest survival of the indigenous population over at least some of the conquered/ resettled area (as has recently been claimed by Sykes and others) Some, perhaps many, indigenes were enslaved

12

Possible parallel case no. 1: Basque and Latin My colleague Larry Trask has observed that the romanization of the Basque Country offers some striking parallels to the case of English, though it should be remembered that the securest evidence for early Basque comes from Aquitania, i.e. north of the Pyrenees, and that it is not known for sure how most of the modern Basque provinces of Spain came to be populated by Basques. <> A new language arrives borne by military conquest, arguably followed by mass immigration Latin. <> The indigenes have a material culture technologically no more advanced than that of the newcomers There is no evidence for Basque cities (or even settlements of significance), coinage, writing, roads, political entities beyond tribes, armies, substantial seafaring, or a legal system. This echoes the archaeologically-verified collapse of Romano-British material culture, and the historically-verified collapse of its central administration, in the fourth and fifth centuries. <> The indigenous language gives practically no lexis to the newcomer There appears to be not a single Basque word in Latin. Indeed, the total Basque lexical contribution to Romance before about 1900 probably does not reach a dozen words, except in the provinces of Alava and the Rioja (excluding Basque words used in Romance only by native speakers of Basque, i.e. a clear bilingual phenomenon). Possibly ancient words of Castilian which have been claimed to be of Basque origin are discussed in depth by Trask (1997: 415-21). He concludes that only about 7 of the claims are even remotely plausible (chamorro close-cropped (semantics very dubious, since the proposed Basque etymon samur means tender, soft, delicate)), chaparro dwarf evergreen oak, izquierdo left [as in hand], laya spud [agricultural tool], mogote hummock, knoll (phonology very dubious; Basque moko peak or muga boundary), pizarra slate (if for *lapitz-arri, as suggested by Corominas and Pascual 1980), zarza blackberry, bramble) and that about three of these are reasonably strong possibilities (izquierdo, laya and zarza). None approaches certainty. Some may be common borrowings by Basque and Castilian from some other language; early Iberia was a patchwork of languages whose affiliations are unknown. This low number closely parallels the number of reasonably secure early English borrowings from Brittonic. Professor Trask comments in correspondence (10/03/2004) that if Basque had died out between 400 and 600 C.E., as its neighbours probably did, there would be little hard evidence for its former existence. (See also Trask 1997: 9-12.) <> The indigenous language gives the newcomers only a rather small number of place-names and personal names There is no evidence for Basque or Basque-mediated personal names in use among non-Basques before well into the Middle Ages, and even then there were not many. There are very few toponyms used by the Romans which can be shown to have existed before the Roman settlement, and a number of these are non-Basque. Of course, there is a documentation problem here, but there do not appear to have been many Basque settlements. However, the modern toponymy of the Basque Country is substantially, though by no means completely, Basque. Brittonic toponymy in England is much less conspicuous than Basque toponymy in the Basque country. 13

<> The indigenous population can be seen documentarily only in scattered (relict) statelets The Basques in the Roman period can be seen only as a collection of tribal names, and we cannot even tell which of those tribes spoke Basque, for lack of evidence of language. Most of the personal names recorded from the region are non-Basque. We only know the Basques were there at all because they were there later, and because some Aquitanian personal and divine names can be readily interpreted as Basque. By contrast, there are ample references in Old English sources to Britons, their activities, their ethnicity, and the social and legal status that derive from it. <> DNA evidence is equivocal but may suggest survival of the indigenous population over at least some of the conquered/ resettled area Genetic evidence is not even equivocal: the Basques are in certain respects definitely very distinct from their neighbours. See Cavalli-Sforza et al. (1988), and Cavalli-Sforza (2000: 21-2) on the high incidence of the negative allele of the RH gene among the Basques. By contrast the most that has been claimed for Britons in England is that their descendants represent one genetic strain, even if a very important one. <> Some, perhaps many, indigenes were enslaved Professor Trask knows of no evidence for the person of a single Basque slave, but the Romans practised slavery on a vast scale, and it seems very unlikely that the Basques escaped the practice. We know that the English enslaved Britons, and the word w(e)alh notoriously means both Briton and `slave' (see the extensive discussion of this topic by Cameron (1980)). South of the Pyrenees, with the exception of the city of Pamplona, of Calahorra (previously Celtiberian) and of Oiasona whose status as a Roman town is conjectural, there is actually not much evidence for Roman settlement or romanization at all in the Basque Country. It would therefore be hard to regard (most of) the area as Latin-speaking even in the loosest sense, and in that case the opportunity for Latin-speakers to borrow from the substrate vocabulary would hardly exist, whilst this point does not militate against Basque borrowing from Latin since one would expect borrowing from the materially higher culture even if the borrowings were words only for portable objects that would not necessarily need Latin-speakers to transmit them to Basque-speakers. In Trasks words (1997: 10-11) there was a lack of Roman interest, few Romans settled in the territory and the area was hardly worth the trouble of colonization. He cites Roger Collinss conjectural view of the existence of Basque-Roman trading relations, which is reasonable enough but which remains hypothetical. This scenario is still at variance with the English conquest of Britain, therefore, where active English overlordship is not in doubt and it penetrated deep into the local earth. That ought to have afforded the opportunity for substrate borrowing amounting to more than the three certain and the handful of possible words in the canon and the miserable number of place-names compared with the very (but by no means completely) Basque toponymy of the Iberian Basque Country. However, as noted above, practically all the evidence for the existence of Basque or its ancestor in Roman times comes from north of the Pyrenees, an area which was unquestionably romanized, though Basque speech persisted, and persists, in the northern foothills of the mountains. The Romans built a fort at Lapurdum, modern Bayonne. So romanization of a sizeable Basque-speaking area is

14

assured, as far north as the Garonne to judge by epigraphic anthroponymic evidence, and the opportunity for lexical borrowing from the ancestor of Basque must have existed here if nowhere else. Gorrochategui (1995: 49) argues that the available evidence favours shallow latinization in the Pyrenees and western Aquitania in the Roman period: the adoption of a Latin anthroponymy but none of the other cultural backup to this such as the ability to write Latin full funerary epigraphs. Roman administration in at least a loose sense over what is now the southern Basque Country is not in doubt. The Romans even raised military units there. The difficulty lies in establishing that the local speech was Basque in Roman times. Evidence that it was consists of three brief inscriptions and the observation that the Roman name for the inhabitants of the larger part of the territory was Vascones, the name later applied universally by Romance-speakers to the Basques. Possible parallel case no. 2: Norse and Gaelic The pattern involving Brittonic and English may also appear consistent with what happened in Norse-Gaelic contacts in Scotland. Regrettably, as with the Basque Country, the sequence of historical events there is not known for sure, but it seems likely that, in the Western Isles at least, Gaelic replaced Scandinavian (Nicolaisen 1976: 138, quoting an anonymous source, ?Magne Oftedal); whether this was a reintroduction or a new introduction, i.e. whether Gaelic was also substratal to Norse, is not clear. The number of Scandinavian borrowings in Gaelic is pretty small, reckoned not to exceed 50 (Oftedal 1961; Jackson 1962), with a possible few more in island dialects (see e.g. Coates 1988a), though this is a much larger tally than that of borrowings from Brittonic into English. But topographical terms, especially as applied in place-names, constitute the bulk of them, as with: b bay, cleit rock, sgeir skerry, acarsaid anchorage, gil narrow valley, gerraidh home pasture, and seabird-names such as sgarbh cormorant, lmhaidh guillemot, and mall gull on St Kilda. A few more enter place-names but not the lexicon: Scand. nes headland, fjall fell, and blstar farm. Discussion The pattern in both the Norse and the Brittonic cases appears consistent with withdrawal of speakers of the previously dominant language, rather than assimilation of the dominant classes by the incomers. Withdrawal can be achieved in a number of ways: murder (ethnic cleansing), enslavement (resulting in zero cultural impact), flight, exile, negotiated withdrawal. But what we see in eastern England, as in the Scandinavian kingdom of the Hebrides, is the lexical and onomastic evidence that the incomers moved into a landscape from which a major withdrawal had taken place. There is no reason to reject this older view in favour of the newer one which asserts the survival of a substantial local population having the option of cultural assimilation (Higham 1992: 209-36; I follow Yorke 1995: 48), especially given the undeniable historical evidence for flight across the Channel (in the existence of the Bretons), massacre of local Britons, and the use of the word wealh, primordially Briton, to mean slave.4 Ward-Perkins (2000: 526) is aware of the tension between his essentially Highamist viewpoint and the invisibility of the Britons when he says that "what needs to be explained is why ... the necessary cultural changes all occurred in one direction: in favour of the conquerors." Given his assumptions, we do not have an answer.5

15

Dr Higham argues that a political lite governing a large underclass may be sufficient to account for the facts as presented. But evidence from elsewhere suggests a political ascendancy absorbs local vocabulary at least, if not, in the fullness of time, the entire vernacular. Western Romance has a significant Celtic substrate (Lambert 1997: 185-204). Local Castilian borrows richly from Basque (Zrate 1976, and see the comments implicit in Trasks remarks drawn on above). Hiberno-English has a huge regional lexicon (Crystal 1995: 338) with many borrowings from Irish. Contact in differing degrees may have consequences ranging all the way from such vocabulary adoption to full language-shift. Language-shift by an lite is perfectly possible. French yielded before Middle English in England, and Cloviss descendants spoke Gallo-Roman not Frankish. I know of no case where a political ascendancy has imposed its own language without significant impact from the language of the conquered. Entwistle (1936: 41) noted that there is a fairly substantial amount of Celtic vocabulary in Iberian Romance even though it betray[s] a culture below that of the invaders; where there are practically no borrowings at all, the conclusion to be drawn is obvious: there was practically no culture to betray, that is there were practically no speakers in a position to betray it, whatever its material level. The situation we are contemplating bears comparison with the English advance into sparsely-populated Australia where murder and displacement of the Aboriginal population were normal. There, numbers of place-names are adopted, and a vocabulary for unfamiliar flora and fauna; the European flora and fauna of Britain would have been largely familiar to the English, and predictably therefore such borrowing from Brittonic is small. But one does not have to be an essentialist, i.e. to believe in blood- or gene-given ethnicity, to question the proposition that many Britons voluntarily became English (and were allowed the opportunity to do it), because, if they did, so little linguistic material went with them. Where they had the opportunity, as the pace of westward conquest relented, we do unsurprisingly find signs of greater Brittonic impact, especially in the retention of larger numbers of place-names, though only in Cornwall does lexical borrowing become really significant (Wakelin 1975: ch. 7). It is of great interest that some of the now defunct borrowings claimed for Middle English in Breezes studies (1991; 1993a; 2002: 177) appear to have a West Midland incidence; they appear for instance in Ancrene Wisse and its associated texts in Tolkiens AB language (possibly written in or near Herefordshire), in the lyrics in MS. Harley 2253 (copied at Ludlow), and in other texts which appear to have spent some period of their textual history in Mercia (see especially his comments 1991: 438). It is also possible that an English word may have taken on the sense of a term of Welsh hospitality in a Flintshire context (Breeze 1996b). But this apparent geographical effect may be an artefact of Dr Breezes as yet incomplete coverage of ME texts (as he observes in correspondence, 02/04/2004), and it is true that he finds good evidence for borrowings in texts that appear to originate elsewhere, such as Beowulf (1993a, 1993c). Moreover, evidence for significant numbers of lexical survivals in the modern dialects of the Marcher counties is lacking. Leeds (1972: 1), for example, claims that the number of words of Welsh derivation in general use is small and confined almost entirely to the western and north-western areas [of Herefordshire] (and cf. Charles 1963). Where discussion turns on the survival of Brittonic in England (e.g. Stevenson 1904: 248-50 for Dorset; Jackson 1953: 234-43 for the whole of England), it does not engage with the question of its traces in the local English dialect. It was Jackson's view (1953: 242) that the natives learned AngloSaxon thoroughly and accurately, so accurately that they had to mangle their own names to suit the new language rather than the new language to suit their own sound-system ... [i]t is impossible to point to any feature about Anglo-Saxon phonology which can be shown conclusively to be a modification due to the alien linguistic habits of the Britons ... they must have learned the new phonology very completely. He compares the few Gaulish words in Romance (1953: 243), but

16

Lambert (1997: 186-203) actually identifies 40-odd Gaulish borrowings into Latin and 116 into French or Provenal (counting only those not bearing a question-mark in Lambert's survey), and this is a far greater tally than that of supposed Brittonic survivals in English. Jacksons scenario depends on the integration of individual Britons into the English-speaking community and the disappearance of Brittonic communities that might sustain the ancestral language. This cannot easily be squared with the idea that genetic Britons formed a very large element of the population of England. I believe therefore that Wordsworth was essentially correct as far as the east was concerned when he wrote, in the poem Monastery of Old Bangor (1821/2): .... Mark! how all things swerve From their known course, or vanish like a dream; Another language spreads from coast to coast, Only perchance some melancholy Stream And some indignant Hills odd names preserve, When laws, and creeds, and people all are lost! Several papers, both at the Mekrijrvi conference in 2001 (Filppula et al. 2002) and recently published elsewhere, have argued for early Brittonic impact on English at levels other than the lexicon: phonology (Schrijver and Laker, both Mekrijrvi) and grammar (Klemola 2000; Tristram 1999: 19-30), even if some of the impact remains submerged till the Middle English period (White, Mekrijrvi; in the spirit of Dal 1952). It is of great interest that many of the phenomena discussed relate to areas other than my proposed vacated east (Klemola; Laker; White); White is quite explicit about the resistance of south-eastern English to innovations he believes to originate through contact with Celtic. Conclusion It is evident that the level of Celtic survival in those parts of England nearest the Continent, as shown by place-names, is somewhat greater than has been admitted until now. The evidence for Brittonic impact on English vocabulary, both general and toponymic, is still minimal. I submit that the most sociolinguistically persuasive account of these facts is that, after initial contact in which little more was achieved than the transmission of certain place-names and of words construed as names, the invisibility of the Britons was due to some or all of the following factors: the lack of a need for the English to borrow topographical or toponymic vocabulary because their own sufficed the lack of a need for the English to borrow vocabulary for flora and fauna because their own sufficed the non-perceptibility of Brittonic social and cultural institutions (even the OE word dr druid is an Irish borrowing) The last point can only mean that in the initial contact period Brittonic society and culture were either indistinguishable from that of the English (which is scarcely credible) or literally invisible

17

(because it was not there). If it had been there, even marginalized as that of the Native Americans or Australians was, one would have expected borrowed vocabulary to describe it, or at least to betray that it was there. If a Brittonic population had redefined itself as English through being absorbed by degrees into the population of the English settlements (Collingwood and Myers 1936: 318; cf. implicitly Yorke 1995: 69 and the work of Higham in this area in general (1994; 2002a; 2002b: 1314, 17), and assuming that the English would have tolerated or condoned this), experience in other parts of the world suggests that this could not have happened without substantial lexical copying from the substrate language(s) as the newcomers language was appropriated (as in India) and/or without structural and lexical transformation of the dominant language (as throughout the world, for example by pidginization and subsequent creolization, and by the relatively subtle local changes due to substrate effects seen in e.g. Irish and other extraterritorial Englishes, South African Dutch (Afrikaans), Latin American Spanish and Portuguese, Southern Bantu, the Indo-Iranian languages in India). As to the latter point, it is clear that Brittonic underwent far more massive structural transformation than English did during the initial contact period in areas including those where it vigorously survived to become modern Welsh and Cornish. My inclination is therefore to accept the traditional view that in certain parts of what became England there were few visible Britons, and that this state might in principle have been achieved by emigration, annihilation or enslavement, for each of which there is evidence in English sources, though much hinges, of course, on whether these sources are viewed as credible witnesses. The Chronicles accounts of flight by and extermination of the Britons are undermined somewhat, as Dr Higham reminds me in correspondence (07/04/04), by Bedes (HE I, 34) declaration that thelfrith (i.e., as late as 600) ravaged the Britons more extensively than any other English ruler ... for no ruler or king had subjected more land to the English race or settled it, having first exterminated or conquered the natives. He uses the word exterminare, and what he meant depends on our view of his latinity. This word does not mean to exterminate in the classical language, but to drive off, as the etymology, involving the word for boundary, would suggest. A conquered native might or might not be a slave. My reasons for holding this view inclination are essentially sociolinguistic. Of course one must allow for the possibility in these sources of rhetorical, politically-motivated exaggeration of the severity of what happened to the Britons, and one must also allow that some of the principal sources were written (in their current form) over 300 years after the events they purport to describe. Could this have been time for a borrowed vocabulary to have disappeared as the Britons institutions disappeared along with any English interest in maintaining a record of them? The English may not have had the anthropological impulses which writers of the later second millennium had when encountering peoples with different customs and institutions. Even Bede is interested principally, and not dispassionately, in the Britons Christianity, and has no particular interest in describing their other customs. Any conclusion is speculative, but if one wants to maintain that the coming of the English was essentially the displacement of one aristocracy by another, with minimal genetic impact on the general population, one must have an explanation for the paucity of lexical borrowing. The most basic level of lexical borrowing, which is all we need be concerned with here since it is presupposed by all other types of linguistic borrowing, involves the borrowing of culturally and technologically significant terminology and special words for distinctive flora and fauna. A low degree of technological advancement in the subjected people is not a necessary condition for this (witness the

18

Celtic borrowings in Latin). Marginalization of the subjected people and their institutions is also not a necessary condition (witness the contribution of other subjected peoples to general modern English vocabulary), though perhaps the passage of time before the appearance of the known texts may make marginalization sufficient for silence. The lack of an anthropological turn in early English writers might have had the same effect, in that they may have failed to record any interest in the customs of the Others. But our explanation must also take into account that the Britons, in accepting English, seem to have left no imprint of their own on it, which gives us a major problem. They were an ethnically and therefore legally distinct(ive) group deep into the recorded Anglo-Saxon period from the English point of view (as in the laws of Ine and even of Alfred: Attenborough 1922: 23; Yorke 1995: 72, 259-60, 285), which need not absolutely, but presumably does, imply they were still Brittonic-speaking. When did they give up Brittonic, and how, whilst remaining distinctive, did they fail to evolve a distinctive variety of English as they passed through the bilingual stage of language shift? How come, if they were a majority, our present language does not have the general character of a Brittonic variety of Germanic? The major pronunciation and grammar features which distinguish English from Continental Germanic cannot be explained as Celtic (though, as noted above, there are very interesting local effects which do suggest Brittonic influence (e.g. Klemola 2002; Laker 2002), and Vennemann (2000; 2002), out on a limb, argues that English does indeed have Celtic traits some of which can be traced back still further to Semitic (cf. also Preusler 1956)). Palatalization of velar consonants before front vowels is shared with Frisian, not Welsh. Breaking of vowels before certain consonant clusters is unparalleled in Welsh. Conversely, what we know of Welsh innovations from 450-600 reveals no similarity to those of English. The Welsh reduction of clusters of a nasal and a stop has no parallel in English, nor does intervocalic lenition of stops, and nor does the Welsh thorough redistribution of vowel length according to syllable structure. There is a similarity between the phonetic effects of OE i-umlaut and Brittonic i/j-affection, but they do differ, most strikingly in that the Brittonic change affects only short vowels, which is quite untrue of iumlaut. Brittonic changes typologically in radical ways at the time it is meeting OE, and OE shares none of them: for instance, Brittonic becomes firmly a verb-first language, loses its case-system and adopts quite rigid noun-adjective order in noun phrases. This could hardly differ more from early Old English (though Tristram (2002) notes some typological similarities in the developed forms of Welsh and English). How did the majoritarian Welsh adopt English in such a Germanic form without formal instruction, without a substantial immersion programme for all speakers, and without making it more like Welsh? These are the questions that need to be answered by those who propose a massive contribution of Britons to the English gene-pool. On the other hand, absence of Britons is a sufficient condition for the absence of Brittonic-coloured English! The common ground between the positions discussed in this paper lies in the possibility of enslavement, that is cultural annihilation, for in enslavement in its classical form the masters will make no effort to communicate with the slaves in the slaves languages; the linguistic onus rests on the slaves themselves. If slavery meant the breakup of communities, the probability of rapid skilled mastery of the conquerors language becomes greater on the grounds of sheer necessity for individuals to form human and economic relations of any sort; but to the extent that slave communities might be retained, as is implied by the frequent place-name types Walton and Walcot (Cameron 1980), that would act as a retardant to language-shift with the possibility of a slavecoloured variety of English emerging. It may be that the illiteracy of the crucial period (450-600) may be another important factor in the invisibility of the Britons. In recent cases of colonial conquest, especially by Western colonial powers, writers have been on hand from the start of the 19

process to record native customs and terminology which only became important in the light of later historically-oriented social and political thought (as for instance in Australia). If Whites had started writing about Australian Aboriginals only in 1970, much less would have been recoverable about their anthropology and languages perhaps practically nothing and certainly that would have been the case if Whites had been as aggressive towards Aboriginal communities everywhere as they were in Tasmania (and of course elsewhere; Dixon 1980: 78-9). If this imperfect analogy holds, the most likely reasons for the invisibility of the Britons appear to rest on: enslavement and dispersal; and the lack of contemporary chronicling. But the view from linguistics clearly suggests that fragmentation of communities dispersal needs to be reckoned with as an agent in rapid language-shift. There is still a tension to be resolved: we know that Britons were identifiable as such to lawmakers till at least 900, by which time there is no evidence at all for the continuation of even the most meagre linguistic impact. Communities versus dispersal: theres the crux. Notes * This version of the present paper has been prepared for circulation at the conference Britons and Saxons organized by Dr N.J. Higham at the University of Manchester, 14-16/04/2004. The paper contains some material previously published in Coates (2002), but that has been completely reviewed and revised as necessary. The present state of the paper is that it is not quite complete. For a full catalogue of literature before 1988 dealing with the Celtic impact on English place-names, see appendix 1 to Coates (2002), and for work thereafter see the relevant articles in Coates (1988) and the text of Coates (2002). I am extremely grateful for comments on a draft of this paper or its content by Nick Higham and Max W. Wheeler. 1 This law states: In any homogeneous community, naming-behaviour will remain constant, except when disturbed by outside influence. The Third Law adds a rider about the strength of such influences. 2 Note that for English county-names I put the abbreviation used by the Survey of English PlaceNames. 3 It might be thought that the non-lenition of */b/ after the feminine word cair destroys the proposed etymology. But (1) lenition sometimes fails after ker in Cornish (see Cardew versus Carthew in Padel (1985: 90)), and (2) [rvr] may have been rejected in OE just as [rr] was in the name Ower(moigne) (Coates 1995: 245). Note also that the account of the regions settlement given here is not inconsistent with the English having discovered the Brittonic name of the highest mountain of the Peak District, Kinder, in a phonological form *Cnderx look-out point which was available only during a brief window in the mid to late sixth century (as argued in Coates (2000b)). The English were obviously scouting in Brittonic territory before they settled. 4 The direct evidence for the withdrawal from parts of England includes: the alleged facts that the Britons forsook Kent (Anglo-Saxon chronicle MS. A 457); the Welsh fled from the English like/ as from fire (A 473); the English slew all the inhabitants of Pevensey (A 491); slaughter of countless Welsh at Chester including 200 priests (A 607), fulfilling a prophecy of Augustine. Even allowing for rhetorical overstatement, the general picture is pretty clear. The alternative is to believe that this is merely a conventionalized landtaking myth. In corroboration from the Brittonic side,

20

Gildas notes deaths, enslavements, flight to mountains, forests and coasts and overseas. (It should be recognized, as Nick Higham has emphasized to me, that the independence of the Welsh and English testimonies may be questioned, since the compilers of the Chronicle knew Gildas indirectly, through Bede.) There are Brittonic missions to Brittany after 500, suggesting attempts to consolidate the faith, and there are Brittonic hosts on the continent according to Jordanes, writing in 551 (both these facts being interpretable as self-perceived Roman citizens going to the aid of other Roman citizens). Jackson, on the basis of the evidence of continental writers, hypothesizes two peaks of emigration to Brittany, the first from 450-550, and the second after 577. There are subject populations referred to in the laws of Ine (688-725). General considerations pointing in the same direction are the fact that flight is a constant after battle, and that raiding has among its purposes to to damage agriculture, giving another incentive for flight, namely from famine; any famine must have been exacerbated by the world-wide climatic downturn of 535-545 (whatever the ultimate origin of that may have been: possibly volcanic activity in the Far East (Keys 1999)). 5 Whilst I have spoken of the vacated east as the main domain of early Germanization, I should note that the case of Dorchester-on-Thames, with sub-Roman and English archaeology overlapping, is a special one (Yorke 1995: 30; mentioned in Matthews 2001; archaeological references in Coates, in preparation). Appendix Concise etymologies of Brittonic names mentioned in the paper (all are discussed or mentioned in Coates and Breeze 2000 except the Scottish ones, for which references are given) Liss - (1086 Lis) PrW *lss hall Beccles - (1086 Becles) PrW *bax-lss `small-hall' Priddy - (1180 Pridi) PrW *pr-dy earth-house Penkridge - (958 Pencric, RB Pennocrucium) head-mound Crick, Creech etc. - (various dates) PrW *crc mound Landican - (1086 Landechene) PrW *lann (sacred) enclosure + personal name Carburton - (1086 Carbertone) PrW *cair brtton Britons village (with non-lenition, as sometimes in Cornwall Caerlanrig - (1610 Carlanerik) PrW *cair lanerx clearing village (Watson 1926: 368) Wigan - (1199 Wigan) PrW *wg + -an little settlement Penteiacob - (no date, spelt so) PwW *penn t Iacob head of Jamess house (Watson 1926: 354) Clovelly - (1086 Clovelie) PrW *cl earthworks, fort + personal name

21

Ower(moigne) - (1210 Ogres, 1219 Our) PrW *oir-rus wind-gap Merrow - (1185 Marewe) RB *Margio marl-place Helvellyn - (1577 Helvillon) PrCumb *hal meln moor yellow Tric - (1086) Lat *Triectus, British *Trjextos Brokenborough: Kairdur(berg) (date uncertain) PrW *cair village + unidentified element

References Attenborough, Frederick L. (ed. and trl.) 1922. The laws of the earliest English kings. Cambridge: Cambridge University Press. Baker, John T. 2001. The transition from Romano-British to Anglo-Saxon culture in the Chilterns and Essex region. Birmingham: University of Birmingham PhD dissertation (Medieval History). Baugh, Albert C., and Thomas Cable. 2002. A history of the English language (5th edn). London: Routledge. Breeze, Andrew. 1991. Exodus, Elene and The Rune Poem: milp army road, highway. Notes and Queries 236, 436-8. Breeze, Andrew. 1993a. (1) Beowulf and The battle of Maldon: trem pace and Welsh tremyn journey. (2) Welsh baban baby and Ancrene Wisse. (3) Welsh geneu mouth, jaws and the Middle English Seinte Margarete. (4) Old English trum strong, truma host: Welsh trwm heavy. Notes and Queries 238, 9-10, 12-13, 13-14, 16-19. Breeze, Andrew. 1993b. Celtic etymologies for Old English cursung curse, gafeluc javelin [etc.]. Notes and Queries 238, 287-97. Breeze, Andrew. 1993c. Wered sweet drink at Beowulf 496: Welsh gwirod liquor, drink. Notes and Queries 238, 433-4. Breeze, Andrew. 1995a. Deorc bloody in The Dream of the Rood : Old Irish derg red, bloody. igse 28 (1994/5), 165-8. Breeze, Andrew. 1995b. Irish brat cloak, cloth: English brat child. Zeitschrift fr celtische Philologie 47, 89-92. Breeze, Andrew. 1996a. Irish Beltaine May Day and Beltancu, cattle-rent in pre-Norman Lancashire. igse 29 (1995/6), 59-63. Breeze, Andrew. 1996b. Old English estas relishes in the Flintshire Domesday. Notes and Queries 241, 14-15. 22

Breeze, Andrew. 1997. A Celtic etymology for Old English dor brave. In Roberts, Jane and Janet Nelson (eds.), with Malcolm Godden, Alfred the Wise: Old English studies presented to Janet Bately on her sixty-fifth birthday. Cambridge: D.S. Brewer, 1-4. Breeze, Andrew. 1998. A Brittonic etymology for Old English stor incense. Anglia 116, 227-30. Breeze, Andrew. 2002. Seven types of Celtic loanword. In Filppula et al. (eds), 175-81. Brunner, Karl. 1960. Die englische Sprache: ihre geschichtliche Entwicklung. Tbingen: Max Niemeyer. Cameron, Kenneth. 1959. The place-names of Derbyshire (3 vols). Cambridge: Cambridge University Press (SEPN vols. 27-29). Cameron, Kenneth. 1980. The meaning and significance of Old English walh in English placenames. Journal of the English Place-Name Society 12 (1979/80), 1-53 [with appendices by Malcolm Todd and John Insley]. Cameron, Kenneth. 1991. The place-names of Lincolnshire, vol. II. Nottingham: English PlaceName Society (SEPN vol. 64-5). Cavalli-Sforza, Luigi Luca. 2000. Genes, peoples and languages. London: Allen Lane (The Penguin Press). Cavalli-Sforza, Luigi Luca, A. Piazza, P. Menozzi and J.L. Mountain. 1988. Reconstruction of human evolution: bringing together genetic, archaeological, and linguistic data. Proceedings of the National Academy of Sciences 85, 6002-6. Charles, B.G. 1963. The Welsh, their language and place-names in Archenfield and Oswestry. In Henry Lewis (ed.) Angles and Britons [six ODonnell Lectures]. Cardiff: University of Wales Press, 85-110. Clark, Cecily. 1979. Clark's first three laws of applied anthroponymics. Nomina 3, 13-19. Reprinted in Peter Jackson (ed.) Words, names and history. Selected writings of Cecily Clark. Cambridge: D.S. Brewer (1995), 77-83. Clark, Cecily. 1992. Onomastics. In Richard M. Hogg (ed.) The Cambridge history of the English language, vol. I: The beginnings to 1066. Cambridge: Cambridge University Press, 452-489. Coates, Richard. 1980. Methodological reflexions on Leatherhead. Journal of the English PlaceName Society 12 (1979/80), 70-4. Coates, Richard. 1982. Phonology and the lexicon: a case study of Early English forms in -gg-. Indogermanische Forschungen 87, 195-222, esp. 203. Coates, Richard. 1988a. Notes on the past of the Gaelic dialect of St Kilda. Brighton: University of Sussex (Cognitive Science Research Paper 081).

23

Coates, Richard. 1988b. Celtic etymologies of four place-names in England: Acemannesceastre, Helvellyn, Talke and Tric. In Richard Coates, Toponymic topics. Brighton: Younsmere Press, 24-38 and (for Tric) map on 124. Coates, Richard. 1990. On some controversy surrounding Gewiss/Gewissei, Cerdic and Ceawlin. Nomina 13 (1989/90), 1-11. Coates, Richard. 1998. Merrow and some related Brittonic matters in Surrey. Journal of the English Place-Name Society 30 (1997/8), 16-22. Coates, Richard. 1995. Owermoigne, Dorset. Indogermanische Forschungen 100, 244-51. Reprinted with minor changes in Coates and Breeze (2000), 100-5. Coates, Richard. 2000. Kinder, the highest mountain in the Peak District. In Coates and Breeze, 165-6. Coates, Richard. 2002. The significances of Celtic place-names in England. In Filppula et al. (eds.), 47-85. Coates, Richard. forthcoming a. The pre-English name of Dorchester-on-Thames. Coates, Richard. forthcoming b. Torksey. In Reflections on Lincolnshire place-names [provisional title]. Coates, Richard and Andrew Breeze, with a contribution by David Horovitz. 2000. Celtic voices, English places. Stamford: Shaun Tyas. Collingwood, R.G. and J.N.L. Myers. 1956. Roman Britain and the English settlements (2nd edn). Oxford: Clarendon Press. Corominas, Juan and Jos A. Pascual. 1980. Diccionario crtico etimolgico castellano e hispnico. Madrid: Gredos. Croft, William. 2000. Explaining language change: an evolutionary approach. London: Longmans. Crystal, David. 1995. Cambridge encyclopedia of the English language. Cambridge: Cambridge University Press. Dal, I. 1952. Zur Entstehung des englischen participium praesentis auf -ing. Norsk Tidskrift for Sprogvidenskab 16, 5-116. Dixon, R[obert] M.W. 1980. The languages of Australia. Cambridge: Cambridge University Press. Eagles, Bruce. 2001. Anglo-Saxon presence and culture in Wiltshire, A.D. c. 450- c. 675. In P. Ellis (ed.) Roman Wiltshire and after. Papers in honour of Ken Annable. Devizes, 199-233. Ekwall, Eilert. 1920. Zu zwei keltischen Lehnwrtern im Altenglischen: (1) Ae. funta Quelle; (2) Ae. torr , ne. tor a high rock, a pile of rocks; ... a hill. Englische Studien 54, 102-10. 24

Ekwall, Eilert. 1922a. The place-names of Lancashire. Manchester: Chetham Society and Manchester University Press. Ekwall, Eilert. 1922b. Review of Frster 1921. Anglia Beiblatt 33, 73-82. Ekwall, Eilert. 1924. The Celtic element. In Mawer, A.H. and F.M. Stenton (eds.) Introduction to the Survey of English Place-Names, part I. Cambridge: Cambridge University Press (SEPN vol. 1, part 1), 15-35. Ekwall, Eilert. 1960. The concise Oxford dictionary of English place-names (4th edn). Oxford: Clarendon Press. [1st edn 1936.] Entwistle, William. 1936. The Spanish language. London: Faber and Faber. Filppula, Markku, Juhani Klemola and Heli Pitknen (eds). 2002. The Celtic roots of English. Joensuu: University of Joensuu, Faculty of Humanities (Studies in Languages 37). Frster, Max. 1921. Keltisches Wortgut im Englischen: eine sprachliche Untersuchung. In Frster, M. and K. Wildhagen (eds.) Texte und Forschungen zur englischen Kulturgeschichte. Festgabe fr Felix Liebermann. Halle an der Saale: Niemeyer, 119-242. [Also published separately.] Frster, Max. 1922. [Title not ascertained.] Englische Studien 56, 204-39. Frster, Max. 1936. Altenglisch stor, ein altirisches Lehnwort. Englische Studien 70, 49-54. Geiriadur Prifysgol Cymru. See Thomas. Gelling, Margaret 1984. Place-names in the landscape. London: Dent. Gelling, Margaret. 1997. Signposts to the past (3rd edn). Chichester: Phillimore. Gelling, Margaret. 1992. The West Midlands in the early Middle Ages. Leicester: Leicester University Press. Gelling, Margaret, and Ann Cole. 2000. The landscape of place-names. Stamford: Shaun Tyas. Gildas. De excidio Britanniae. [Edition ad libitum.] Giraldus Cambrensis [Gerald of Wales]. Descriptio Walliae. [Edition ad libitum.] Gorrochategui, Joaqun. 1995. The Basque language and its neighbours in antiquity. In Jos Ignacio Hualde, Joseba A. Lakarra and R.L. Trask (eds) Towards a history of the Basque language. Amsterdam and Philadelphia: John Benjamins, 31-63. Gover, J.E.B., A. Mawer and F.M. Stenton, with A. Bonner. 1934. The place-names of Surrey. Cambridge: Cambridge University Press (SEPN vol. 11).

25

Gover, J.E.B., A. Mawer and F.M. Stenton. 1939. The place-names of Wiltshire. Cambridge: Cambridge University Press (SEPN vol. 16). Higham, Nicholas. 1992. Rome, Britain and the Anglo-Saxons. London: Seaby. Higham, Nicholas. 1994. The English conquest: Gildas and Britain the fifth century. Manchester: Manchester University Press. Higham, Nicholas. 2002a. The Anglo-Saxon/British interface: history and ideology. In Filppula et al. (eds), 29-46. Higham, Nicholas. 2002b. King Arthur: myth-making and history. London: Routledge. Hogg, A.H.A. 1964. The survival of Romano-British place-names in south Britain. Antiquity 38, 296-9. Hogg, Richard M. (ed.) 1992. The Cambridge history of the English language, vol. I: The beginnings to 1066. Cambridge: Cambridge University Press. Horovitz, David and Richard Coates. 2000. Gnosall, Staffordshire, and the Middle English word <genow>. In Coates and Breeze, 184-92. Jackson, Kenneth H. 1953. Language and history in early Britain. Edinburgh: Edinburgh University Press. Jackson, Kenneth H. 1962. The Celtic languages during the Viking period. In B. Cuv (ed.) Proceedings of the International Congress of Celtic Studies. Dublin: Institute for Advanced Studies, 3-11. Johanson, Lars. 2002. Contact-induced change in a code-copying framework. In Mari C. Jones and Edith Esch (eds) Language change: the interplay of internal, external and extra-linguistic factors. Berlin and New York: Mouton de Gruyter, 285-313. Kastovsky, Dieter. 1992. Semantics and vocabulary. In Richard M. Hogg (ed.) The Cambridge history of the English language, vol. I: The beginnings to 1066. Cambridge: Cambridge University Press, 290-408. Keys, David. 1999. Catastrophe! An investigation into the origins of the modern world. London: Century. Klemola, Juhani. 2000. The origins of the Northern Subject Rule: a case of early contact? In Tristram, Hildegard L.C. (ed.) The Celtic Englishes, vol. II. Heidelberg: Universittsverlag Carl Winter, 329-46. Klemola, Juhani. 2002. Periphrastic DO: dialectal distribution and origins. In Filppula et al. (eds), 199-210. Kristensson, Gillis. 2000. The place-name Owermoigne (Dorset). Notes and Queries 245, 5-6. 26

Kristensson, Gillis. 2001. The place-name Owermoigne again. Notes and Queries 246, 102-3. Laker, Stephen. 2002. An explanation for the changes kw-, hw- > w in the English dialects. In Filppula et al. (eds), 183-98. Lambert, Pierre-Yves. 1997. La langue gauloise. Paris: Librairie Gungaud. Leeds, Winifred. 1972. Herefordshire speech: the south-west midland dialect as spoken in Herefordshire and its environs. Privately published. McMahon, April M.S. 1994. Understanding language change. Cambridge: Cambridge University Press. Matthews, Keith J. 2001. What's in a name? Britons, Angles, ethnicity and material culture from the fourth to the seventh centuries. The Heroic Age 4, web-site http://www.mun.ca/mst/heroicage/issues/4/Matthews.html. Moravcsik, Edith. 1978. Language contact. In Joseph Greenberg, Charles A. Ferguson and Edith Moravcsik (eds) Universals of human language, vol. I. Stanford: Stanford University Press, 93-123. Nicolaisen, W.F.H. 1976. Scottish place-names. London: Batsford. Oftedal, Magne. 1961. On the frequency of Norse loanwords in Scottish Gaelic. Scottish Gaelic Studies 9, 116-27. Owen, Arthur E.B. and Richard Coates. forthcoming. Traiectus/Tric/Skegness: a Lindsey name explained. Lincolnshire History and Archaeology. Padel, O[liver] J. 1985. Cornish place-name elements, 2 vols. in 1. Nottingham: English Place-Name Society (SEPN vol. 56/57). Parry-Williams, T[homas] E. 1923. The English element in Welsh: a study of English loan-words in Welsh. London: The Honourable Society of Cymmrodorion. Parsons, David N. 1997. British *Caratcos, Old English Cerdic. Cambrian Medieval Celtic Studies 33, 1-8. Parsons, David N., and Tania Styles. 2000. The vocabulary of English place-names, fasc. II, bracecster. Nottingham: Centre for English Name-Studies. Pokorny, Julius. 1923. Review of Frster 1921 [in Erschienene Schriften]. Zeitschrift fr celtische Philologie 14, 298-9. Preusler, Walther. 1956. Keltischer Einfluss im Englischen. Revue des Langues vivantes 22, 32250. Redin, Mats. 1919. Studies on uncompounded personal names in Old English. Uppsala: Berling. [Dissertation for the degree of PhD.]

27

Schrijver, Peter. 1995. Studies in British Celtic historical phonology. Amsterdam: Rodopi (Leiden Studies in Indo-European 5). Schrijver, Peter. 2002. The rise and fall of British Latin: evidence from English and Brittonic. In Filppula et al. (eds), 87-110. Serjeantson, Mary S. 1935. A history of foreigh words in English. London: Routledge and Kegan Paul. Smith, A.H. 1928. The place-names of the North Riding of Yorkshire. Cambridge: Cambridge University Press (EPNS vol. 5). Stevenson, W.H. (ed.) 1904. Asser's life of King Alfred. Oxford: Clarendon Press. [New edn by Dorothy M. Whitelock (1959).] Sweet, Henry (ed.). 1885. The oldest English texts. London: Early English Text Society (old series 83). Tengstrand, Erik. 1940. A contribution to the study of genitival composition in Old English placenames. Uppsala: Almqvist and Wiksell (Nomina Germanica 7). Thomas, R.J. (ed.) 1950-. Geiriadur Prifysgol Cymru. Cardiff: University of Wales Press. Thomason, Sarah Grey and Terrence Kaufman. 1988. Language contact, creolization and genetic linguistics. Berkeley: University of California Press. Trask, R.L. 1996. Historical lingusitics. London: Arnold. Trask, R.L. 1997. The history of Basque. London: Routledge. Tristram, Hildegard L.C. 1999. How Celtic is Standard English? St Petersburg: Nauka. Tristram, Hildegard L.C. 2002. Attrition of inflections in English and Welsh. In Filppula et al. (eds), 111-49. Vennemann genannt Nierfeld. Theo. 2000 English as a Celtic language: Atlantic influences from above and from below. In Hildegard L.C. Tristram (ed.) The Celtic Englishes, vol. II. Heidelberg: Carl Winter (Anglistische Forschungen 286), 399-406. Vennemann genannt Nierfeld, Theo. 2002. Semitic > Celtic > English: the transitivity of language contact. In Filppula et al. (eds), 295-330. Wacher, John. 1995. The towns of Roman Britain (2nd edn). London: Routledge. Wacher, John. 1998. Roman Britain (new edn). Stroud: Sutton Publishing. Wakelin, Martyn. 1975. Language and history in Cornwall. Leicester: Leicester University Press.

28

Ward-Perkins, Bryan. 2000. Why did the Anglo-Saxons not become more British? English Historical Review 115 (462), 513-33. Watson, William J. 1926. History of the Celtic place-names of Scotland. Edinburgh: Blackwood. Watts, V.E. 2004. Cambridge dictionary of English place-names. Cambridge: Cambridge University Press. Weyhe, H. 1905. Beitrge zur westgermanischen Grammatik. Beitrge zur Geschichte der deutschen Sprache und Litteratur 30, 55-111. White, David L. 2002. Explaining the innovations of Middle English: what, where and why? In Filppula et al. (eds), 153-74. Winter, Werner. 1973. Areal linguistics: some general considerations. In Thomas A. Sebeok (ed.) Current trends in linguistics 11: diachronic, areal and typological linguistics. The Hague: Mouton, 135-48. Yokose, Hiroyuki. 2001. Aboriginal words in Australian English. Web-site www.kasei.ac.jp/library/kiyou/2001/13.yokose.pdf. Bears page-numbers 169-80; place of publication unknown. Yorke, Barbara. 1995. Wessex in the early Middle Ages. London: Leicester University Press. Yule, Henry and A.C. Burnell. 1903. Hobson-Jobson: a glossary of colloquial Anglo-Indian words [etc.] (3rd edn by William Crooke). London: Routledge and Kegan Paul. Zrate, Mikel. 1976. Influencias del vascuence en la lengua castellana. Bilbao: Editorial La Gran Enciclopedia Vasca.

29