The Conversion of Ireland and the Emergence of the Old Irish Language, AD 367–637

The Conversion and the Transition from Primitive to Old Irish c. 367 - c. 637

by John T. Koch Boston Co llege

When giving the Celtic Studies Foundation lecture in Australia in 1993, I inflicted on the audience my notion of a 'central seismic fracture' running through the middle of Celtic studies. On the one side is the ancient Celtic Europe of the pre-Roman Iron Age - Hallstatt and La Time, the Celtic ethnography of Po sidon ius, the ancient Celtic languages, Gauls and Celtiberians, and so on. On the other, we have the familiar Bretons, Welsh, Scots, and Irish who are still very much with us today. The gap is partly defined by geography. In the last centuries BC, Celtic-speaking peoples inhabited regions from central Asia Minor in the south-east, to the Atlantic seaboard of the Hispanic Peninsula in the south-west, to Scotland's Northern Isles in the north and Ireland in the north-west - a range which would extend from the Pacific to Indian Oceans if superimposed onto a map of Australia or from Atlantic to Pacific in North America. On the mediaeval and modern side of the fault line, the Continental Celtic languages have disappeared and the remaining 'Insular Neo-Celts' retain only their narrow territories in the British Isles and Armorican Peninsula. The historical events which may be thought of as the catalyst for this catastrophic transformation are essentially threefold: (1) the rapid expansion of Rome between c. 200 BC and AD 120, (2) the Germanic migrations of c. AD 300-600, rendered possible by the collapse of Roman power in the West, and (3) the spread of Christianity and with it Latin literacy throughout what had been Celtic Western Europe. Ireland participated directly in only the third of these epochal processes. The idea thatI was belabouring in Sydney is that if scholars cannot make valid and meaningful comparisons across this chasm of Late Antiquity-ifthere are no threads of continuity to be traced - then (to use the late C.F.C. Hawkes'words), 'Celtic studies is a nonsense."

In the present connection, the relevant point is that both the gap itself and the theoretical necessity of bridging that gap are nowhere more evident than in the study oflanguage. Taking the particular subject at hand, the Primitive Irish ofthe ogam inscriptions is a strikingly different language from the manuscript Old Irish known from texts of the seventh to ninth centuries. The language of the former corpus is Old Celtic and thus closely resembles the ancient languages of Britain and Gaul.2 It is clearly an 'Old Indo-European language', in that (in the main) it retains the distribution of consonants and vowels that we reconstruct for the proto-language. In this, it broadly resembles other ancient languages of the family, such as Latin, Greek,

and Sanskrit, and conversely disresembles anything one might encounter in the Gaeltacht (or in Wales or Brittany, for that matter) today. The earliest written Old Irish, on the other hand, is recognisable to anyone literate in the modern language as a form of Gaelic. Its syllable structure and consonant pattern, including the morphophonemic mutations and the opposition of 'broad' vs. 'slender' consonants, anticipate the modern language of a millennium late~, but have no evident counterpart in the ancient langliage which preceded it by no more than two-to-three centuries.

Thus, the period in question presents evidence for the most rapid and profound series oflinguistic changes of any two- to three-century span in the history of the language. The phonological innovations assignable to the period are now, in the main, well understood. They have been laid out systematically by such authorities of the previous generation as Kenneth Jackson (1953) and David Greene (1973) and more recently by such capable younger scholars as Damian McManus (1983) and Anthony Harvey (1984). Jackson (1953, 142-43) and McManus (1983, 30-31, 49) offered detailed chronologies with absolute dates.

The thesis of this paper is that the conversion and the transformation ofIrish speech during the period c. AD 367-637 are causally linked as follows. In the final centuries of the pre-Christian Iron Age, Ireland possessed a Q-Celtic language with two markedly different registers. (The sociolinguistic role of registers differs from that of two distinct languages or even dialects of one language within nearby regions. Registers exist in the same time, place, and society and check one another's divergent innovation as poles within a single linguistic system. Separating dialects and languages originate in circumstances of mutual geographic isolation, whereas registers arise when varieties of a given language become associated with specific functional domains within one society.) In late ancient Ireland, the more basic register was the universal colloquial language, a fairly evolved and dialectal form of Celtic speech, out of which Old Irish (the literate vernacular of the seventh to ninth centuries) was subsequently to arise. The second was a far more conservative and less localised learned language, which stood closer from the point of view of linguistic evolution to the fragmentarily attested Old Celtic of Iron Age and Roman mainland Europe and Britain than to Old Irish. This higher register was originally,like the lower, a purely oral medium, but by some point in the fifth century (or perhaps earlier) it

PAGE 40

had come to be used as the basis for inscriptions in both the ogam and Roman scripts. Tills educated language was mastered by only a relatively small minority of the Irish population, including prominently the pagan priesthood called in Old Celtic dru(u)ides, in Old Irish drufd, and the allied professions in the native verbal tradition. In the fifth and sixth centuries, a second religion spread vigorously in Ireland, bringing with it a second learned tradition and a second learned language, namely Latin. At a pivotal early stage in the conversion, many of the leaders of the missionary church were foreigners, mostly British, and would not have been competent in all levels ofthe native language and learning. For a time (the fifth century and probably well into the sixth), the two religions, educational establishments, and learned classes co-existed, perhaps competing in some situations and finding accommodation at others. During the course of the sixth century, Christianity at last effectively overwhelmed any organised native religion, and with that, Latin Christian learning and the Latin language replaced the native educated standard witllln its social domain, that is, it superseded the variety of Old Celtic which is called by linguists 'Primitive Irish'. At tills point, there was no longer a Celtic learned register to act as a check on popular speech, and a number of heretofore substandard tendencies were able to bubble up to surface rapidly. Then, in the mid sixth to mid seventh century, when for its own purposes the unchallenged church chose to cultivate a vernacular literary language alongside Latin, it had only the more evolved register ofthe illiterate monoglot masses upon which to base the new medium. The system of pagan linguistic learning, like organised pagan Celtic learning in general, had faltered during the conversion. The result was the written Old Irish well known to us from texts of the Early Middle Ages.

Problems of Dating The clllef preliminary hurdle in tills inquiry is our imperfect grasp of the absolute chronology of the two chains of events - the transformation of Old Celtic to Old Irish and the Christianisation ofIreland. On both sides, possible starting points may vary by more than a century. Therefore, it will not be possible to evaluate the interaction of the two processes according to absolute chronologies established by external criteria.

As to dating the conversion, any discussion must begin with the famous entry for AD 431 in the chronicle of Prosper of Aquitaine: Ad Scottos in Chris tum credentes ordinatus a papa Caelestino Palladius primus episcopus mittitur 'Palladius is sent by Pope Celestine as the first bishop to the Irish who believe in Christ'. Tills is enough to tell us that there was a sizeable community of Christians in Ireland by 431. But we do not know how long prior to 431 the community had existed. Neither a span of a century or more or a generation or less would necessarily be impossible. Furthermore, the credibility of the traditional chronology for a mid-fifth century career for St. Patrick

CONVERSION & TRANSITION

is not sufficient to assure us that the activities of Patricius peccator (the Briton, author of the Confessio and the Epistola ad milites Corotici) followed the Palladian mission. As is well known, the Irish annals give us two obits for Patrick - 462 and 493. Neither of these is worth expending much effort on (though a tremendous amount of effort has been expended on them), as there is little chance that contemporary Irish annals were being kept in the fifth century. Therefore, both obits are almost surely retrospective insertions, and we have no reason to regard the chronography of either as well informed. Patrick's own writings (wlllch are the only admissible accounts of Ills career) allow us at best tentative inferences concerning Ills dates. He clearly regarded Illmself a bishop, but his writings, to say the least, leave room to doubt that Rome had known of his claimed office or would have recognised it if they had. Thus Prosper'sPaliadius primus episcopus need not prove that Patrick was later. My own preferred Patrician chronology is early. I 'have argued that he was probably taken to Ireland (for the first time at age sixteen) during the barbarica conspiratio of367 -69 and that he had died before the Palladian mission of 431, most probably before Germanus's mission to Britain in 428 (wlllch would be a likely occasion for the Pope to have learned of the Irish Christian community and its circumstances) (Esposito 1956; Koch 1990). But it remains an open question today as to whether Patrick's career followed or preceded that of Palladius, and we cannot at present positively remove any decade of the fifth century out of the possible brackets ofthe Patrician InlSSlon.

On the linguistic side, some recent writers have tended to infer extremely, but indeterminately early dates for writing Irish, both in inscriptions and in manuscripts. In tills vein, readers may encounter a so­called 'emerging orthodoxy' or 'scholarly consensus' for an ogam tradition extending back into the fourth century AD and possibly much earlier , Old Irish texts going back to the later sixth century and possibly the fifth , and an Irish tradition of Latin literacy predating the conversion (Harvey 1987a, Stevenson 1989). While none of these propositions are a priori impossible, we must guard against an impermissible line of argument according to wlllch possibilities, inadequately supported by evidence, are transformed into likelihoods, wlllch then serve as a basis for further extensions of the thesis. It is therefore essential that we begin from a concise statement of what is certainly known concerning the Irish language during the period of the conversion and its immediate aftermath. First, it is certain that there are ogam inscriptions as old as the fifth century. Some may go back to the fourth or earlier, but tills is neither certain nor necessary. It is also certain that Latin texts have been composed and transmitted in Ireland since the time of St. Patrick sometime in the fifth century. The practice may be older, even predating the first missionaries, but we have no earlier evidence for it. It is further certain that the Irish language has been written in its Old Irish form (that is, with syncope

EMANIA 13, 1995

and apocope, Old Celticlenis / kW , t, k / >ch, th, ch,3 and phonemic palatalisation complete and represented) from the mid seventh century. The Old Irish continuum of written texts may go behind that to c. 600 or even earlier , but we lack indisputable proof. For reasons that will become clear below, the Old Irish language (and hence Old Irish texts) could not possibly have existed contemporary with, or prior to, the Patrician mission.

To take an example relevant to the penultimate point, what is today most widely accepted as an Old Iri sh text of the sixth century is St. Columba's elegy Amra Coluimb Chille, attributed to the poet Dallan Forgaill and now often said to have been first written down soon after the honorand's death in 597. On the other hand, the reconstructed textual prototype implied by the extant manuscript witnesses is a redaction of 1007-08. The Middle Irish prefatory legend concerning the sixth-century locus, tempus, persona, and causa scribendi is no earlier (Herbert 1989). Going behind this redaction , no specialist in Old Irish could today credibly claim that any particular linguistic archaism in the Amra could not also have been possible in, say, the later seventh century. Furthermore, the Columban Monastic Federation remained for many centuries a powerful and creative patron , determined in enhancing the cult of its founder (as, e.g., at Whitby in 664 or in Adomnan's Vita Columbae a generation later ); soitcan ha rdly be said that there was neither the will nor the way to produce a 'contemporary' elegy at a later date. The last full edit ion of the Amra was that of Stokes (1899, 1900) who (however wrongly) regarded it a Middle Irish monstrosity and not a specimen of Early Old Irish at all. Only lately have scholars such as Liam Breatnach (1984; 1989) begun to deploy a careful and aggressive scepticism to insufficiently supported early datings of this sort. One may contrast this relative lack of philological rigour with the heated arguments over the authenticity of the roughly contemporary attribution for the Welsh Gododdin , a collection of e legies whose original patrons were probably annihilated by the Northumbrian Angles in 638 (Jackson 1959).

It is beyond the scope of this paper to go through a detailed review of the pre-Christian fifth- and sixth­century dates claimed by Carney (1989) for the composition of the 'Leinster Poems'. It must suffice here to point out that the dating criteria for three in the group Moen 6en, Nuadu Necht, Ende Labraid) were carefully reconsidered by 6 Corrain (1985), leading to a proposal of composition in a post-Conversion milieu in the seventh century. Furthermore, consistent with 6 Corniin's line of argument, we must reconsider Carney's highly interesting and repeatedly quoted list of non-Christian Latin loan-words from these poems. In light of the discussion below, severa l of Carney's words clearly post-date the earliest stratum of Christian borrowings and post-date specifically those assignable to the Patrician mission. Thus, legion < legion-em , Ailpeoin ?< Albion- , and Mercuir < (dies) Mercuri i

PAGE 41

must have been borrowed at a date after the shortening of native Irish long vowels in unstressed syllables; for they have retained long vowels (in the case of the last a Vulgar Latin or Brittonic phonetic long vowel) in post-tonic Old Irish syllables. Drauc < draCiJ and Saturn < (di es) Saturni both show the later type of treatment of Latin intervocal p , t, c; in other words, they were necessarily borrowed a fter Celtic / kw , t, k / in lenis position had become Archaicor Early Old Irish voiceless spirants h (w), 8, X / (Koch 1990). The oldest Christian loans, including the Patrician subgroup, pre-date these changes.

In view of the foregoing, the date range of my sub­title (AD 367-637) must be taken as a highly tentative (some might say desperate) attempt to impose absolute chronological limits (using dates of known landmarks in Irish history) upon two roughly contemporary processes, which possess few !fatable milestones of their own. The higher terminus is provided by the barbaricaconspiratio, a turning point in the unravelling of Roman Britain, also marking a new phase of Irish­British interaction at the onset ofthe Migration Period. No one, to my knowledge, is now arguing that St. Patrick set foot in Ireland prior to this date, and no one is suggesting (at least not with much conviction or evidence) that there had been a sizeable community of Christians in Ireland any earlier. Some of the extant ogams could pre-date AD 367, but most (even of those representing a pre-apocope Primitive Irish) do not.

As to the later terminus, this is the date of the renowned battle of Mag Roth, in which High-king Domnall mac Aedo of the Northern Ui Neill defeated the coalition of Congal Caech of Ulaid and Domnall Brecc of Scottish Dal Riata. It is also the event that the Early Irish grammatical tract Auraicept no nEces ('The Scholars' Primer') presents as the occasion for establishing Old Irish as a learned written vernacular cultivated alongside Latin in the monasteries. By itself, little credence can be attached to the Auraicept's picturesque story of the tradition founder Cenn F aelad sapiens (obit AU 678=679) losing his 'brain of forgetfulness' as a result of massive head trauma at Mag Roth . However, this native foundation legend provides a convenient da te within the realm of possibilities; for on present evidence, 637x679 is by no means impossibly early to start the Old Irish literary continuum, but neither is it provably too late.

Primitive Irish and Old Irish 'Primitive Irish' is the term used for the pre-apocope language reflected in the more archaic inscriptions of Ireland and the Migration Period Irish colonies in western Britain. Thi s term is potentia lly misleading in possibly implying that an 'Irish' language recognisably similar to the living language so called existed already in ancient t imes. One would expect any language called 'Irish' to have alr eady put some appreciable linguistic distance between itself and the Old Celtic of the well known type of names like Vercingetorix or Boudicii. Now, as noted above, the Old Irish ofthe Early

PAGE 42

M;ddle Ages had already turned the corner, so to speak; for example, the Modern Irish personal name Feilim[ (usually still written Feidhlimidh earlier in this century) was Old Irish Feidlimid , i.e. plainly the same name in a not drastically different older spelling. In contrast, Feilimi's ogam genitive occurs on the (sixth-century?) inscription from Thomastown, Co. Kilkenny , as Archaic/Primitive Irish VEDDELLEMETTO (McManus 1991, 108, 116). In fact, in many instances the attested inscriptional forms of Primitive Irish (while starkly more archaic than the Old, M;ddle, and Modern Irish) are identical to the preform that linguists reconstruct for the ancestral language called Proto- or Common Celtic. For example, CVNORIX on the fifth-century Wroxeter inscription (Jackson 1968) is precisely what linguists would reconstruct for a millennium before for the Old Celtic speakers of Hallstatt or La n,ne, and what would be expected as soon as Roman letters were applied to Celtic speech in the final centuries BC, yet it is something quite different from the Old Irish Conri (cf. O'Brien 564), a form post-dating the Wroxeter stone by two centuries at most. Therefore, we must bear in mind that this label covers no more than a late and local variant of the general Old Celtic of Iron Age Europe. Most of the changes which were to transform Old Celtic into the Irish of the Middle Ages and modern period jlad not yet occurred, or at least were not yet revealed in writing.

In terms of structure of the language itself, Old Irish may be said to have emerged when certain phonological changes (noted briefly above) are seen to have operated upon the sound pattern reflected in the more archaic ogam inscriptions . The following are the key innovations.

(1) The strong primary stress which developed on the first syllable of all stressed words (and also secondary stress on the third syllable in most words four syllables or longer and also the fifth in most words six or longer) caused reductive changes in non-initial syllables: (a) shortening of all inherited long vowels in non-initial syllables, (b) reduction and eventual loss of most final and stressless internal syllables (that is the apocope and syncope respectively).

(2) Raising and lowering of the short vowel pairs e, i and 0, u, by a principle of vowel harmony, the height of a short vowel anticipating that of a vowel in the following syllable. Thus, e.g., Old Celtic *wiros 'man' became Late Primitive Irish *[Werah] (> OIr. fer ) by lowering of i so as to harmonise with the following -a­< ·0-. This raising and lowering was necessarily subsequent to the vowel shortening in non-initial syllables (la), because we see it operating on the shortened forms of original long vowels just as on historically short vowels: so, e.g., Old Celtic *rori gos 'over-king' (genitive) gave Late Primitive Irish *[RurYegah] whence Old Irish ruirech, in which (as well as the short 0 in the first syllable being raised to u by the following i) the original long i of the second

CONVERSION & TRANSITION

syllable was first shortened to i, then lowered to e in harmony with the following a < o.

(3) Prior to the syllable losses, the consonants had developed strong and weak phonetic articulations, the strong variant occurring in absolute phrase-initial position and some consonant clusters and the weak articulation between vowels and in other clusters. Originally, this variation had been non-phonemic, which is to say that it did not enter into the system of meaningfully contrastive sounds out of which distinct utterances might be built. Within the Primitive Irish horizon, the strong and weak (also called fortis and lenis or tense and lax) variants were strictly confined to non-overlapping environments and (with a small class of exceptions) could never occur in the same position within a sequence of sounds; thus they could not possibly contrast. However, when the Primitive Irish stressless syllables fell, the distribution of vowels and consonants was drastically altered, so that the old strong and weak variants then regularly occurred in the same phonetic environments. In other words, their contrast became phonemicised, which is to say it became a meaningful feature of the sound system, capable of distinguishing one word from another. For example, in Old Celtic, the phrase 'his house' was 'esio legos and 'her house' *esias legos, which tended to be articulated with a weak I I I in the first, where the consonant stood between vowels, and a strong I I I in the second where it was preceded by a consonant. Nonetheless, this contrast in the two allophonic variants I I I were not at the time required to carry any information, because the contrast of the preceding segments was always present. In Old Irish, the apocope and reduction of unstressed vowels had turned both 'his' and 'her' to a, so that the only contrast was the now phonemic lenition, or lack of it, ofthe 1-; hence a Ihech 'his house' vs. a lech 'her house'. (It is likely that it is part of the phonemicisation process that the lenis values of I I I first came to be articulated as spirant Ie, W /l.

(4) A structurally similar development to the previous is the Old Irish palatalisation. Like many languages, Old Celtic had tended to articulate any given consonant phoneme more towards the front ofthe mouth when it preceded long or short i or e and more towards the back when preceding long or short u, 0 , or a. Originally, the palatal or non-palatal articulation of consonants was automatically conditioned by the flanking vowels and hence non-phonemic. But, when these vowels were themselves redistributed through the syllable losses, Irish developed a phonemic contrast of palatal vs. non­palatal consonants. For example, in Old Celtic the nominative of the word for 'tribe' had been *Ioul li and the dative had been *Iout!, which contrasted by means of a different final vowel phoneme. After the apocope, Old Irish lualh and luailh were distinguished by contrasting final consonant phonemes, non-palatal vs. palatal.

As different as Ogam and Old Irish are, there is no doubt but that the former represents the direct and immediate ancestor of the latter. No newer language

EMANIA 13, 1995

came in from elsewhere in the Celtic world to replace the speech oflater pre-Christian Ireland. For example, the common ogamic MAQQI 'of the son of is the direct precursor of Old Irish (genitive) maie(e) and Modern mie. The genetic identity is revealed by both the velar articulation and gemination of second consonant. It is clearly a Gaelic form as the Gaulish mapo-, Welsh mab, and Pictish Maph( an) are clearly not. Similarly, the common ogamic ANM - 'name' belongs to the variety of Celtic giving Irishainm, and not to that giving Gaulish anuana (nominative/accusative plural), Old Welsh and Old Breton anu, plural enuein, enuen. The disappearance of Primitive Irish and emergence of Old Irish is therefore not a matter of the replacement of one speech community by another, but of transformation, the rapid evolution and reorganisation of a linguistic system constituted (in the main) of an inherited inventory of raw material.

If we take the linguistic testimony of the written records at face value - which would be a naIve attitude - we would have to conclude that the transformation of Primitive to Old Irish had occurred very rapidly indeed. So, for example, we have the famous 'V oteporix' stone from near Carmarthen, whose honorand died c. 500-550. In the ogam, the genitive of his name is given as VOTECORIGAS (McManus 1991, 52) which is purely Primitive Irish. The -T - and -Co, which are in intervocal position, would regularly undergo Celtic consonant lenition and Goidelic spirantic articulation and thus regularly be written Ih and eh in Old as in Modern Irish. The final syllable -AS would have lowered the preceding vowel-I- to e, and that final-AS would then have fallen by apocope. The vowel ofthe second syllable (-E-) would palatalise the preceding consonant (Ih < T) before it fell by syncope. The initial consonant transcribed V- and etymologically Celtic / w-/ would give Old Irish F-. In sum, then, the expected Old Irish reflex is something like * Foilhehirech. Such a form is not in fact attested, but it would hardly disturb current thinking of Celticists if it were to turn up in a text assignable to an Irish scribe at work e. 650. In other words, it seems that all (or at least, virtually all) the sound changes that distinguish Primitive from Old Irish had operated within the intervening century: vowel affection, raising and lowering, apocope and syncope, the consequent phonemicisation of the palatalised and lenited articulations of the consonants. The last of these phonological changes carried momentous changes for morphology as well; for grammatical information that had formerly been carried by syllabic elements (the inflectional endings) was now carried by the alternation of palatal vs . velar quality of final consonants and the radical or mutated morphophonemes ofthe following word-initial sounds.

The only noteworthy innovation that has occurred in the sixth-century Ogam form VOTECORIGAS and that we might expect to be absent from the same at e. 400 or e. 200 (if there were any ogams that early), is that the -C- continues an older / k W / , a phoneme which we find represented by the symbol for Q in the most

PAGE 43

conservativeogams. However, when the velar precedes the round vowel, the neutralisation of its own labial quality is a phonetic triviality, hardly comparable in magnitude to the pivotal Gaelic innovations which were not yet represented. In fact, the same change had apparently already happened in the older Wroxeter inscription, in which the labialisation has been neutralised in the nominative MACVS, but preserved before a non-round front vowel in the genitive MAQVI (McManus 1991,90).

Interpreting the Evidence As to date, the roughly datable evidence of the Wroxeter and Voteporix inscriptions (which could easily be multiplied many times over) may seem at first to imply the continued existence of Primitive Irish into the fifth and early to mid sixth centuries. But in fact, all that these prove is that a tradition of writing Irish in Primitive Irish form existed this late. And as allowed above, though the first ogams m'ay be no older than the fifth century, the tradition could be a century or more older than that. Once a set of orthographic principles is established, it need not be altered to reflect contempor­ary linguistic changes. As it was (and remains) perfectly possible for an educated person to be a writer of Latin though a speaker of Italian, so would it theoretically have been possible (within a stable learned tradition) for speakers of Old Irish to cut Primitive Irish spellings into stones. The relative dearth of abundant and early evidence for Primitive Irish written literature on anything like the scale of classical Latin may make such a possibility appearunlikelyapriori, but negative evidence cannot be decisive on this point.

The Christian Latin Loan-words in Irish Fortunately, the inscriptions are not our only evidence for the state of the Irish language in the period of the conversion. A second sizeable corpus is provided by the Latin loan-words in Irish. Most of these make their first appearance in texts of Old or Middle Irish date. However, loan-words are able to reveal the phonology of both the borrowing and lending language at the time of the loan, and we can see that much of the Latin­derived vocabulary of Irish must have entered the language before the inception of the Old Irish period. Within this large group, we may subdivide earlier and later strata of borrowings that have, respectively, participated in more and fewer of the Old Celtic to Old Irish sound changes listed above. Thus, we can find many Latin loan-words borrowed before syncope (and thus participating in that Irish change), some before / k W , I, k / > eh, Ih, ch, before apocope, before raising and lowering, and so on. In relative, though not absolute terms, this implies a lengthy period of borrowing, as seems naturally quite likely on historical and geographic criteria.

The most populous part of Britain was part of the Roman Empire between AD 43 and 410, and Roman power had reached the English Channel during Julius Caesar's governorship of Gaul in 60-50 BC; therefore,

PAGE 44

it is quite possible that Latin words having to do with trade, diplomacy, and warfare might have entered Ireland through expanding Roman influence in Atlantic Europe long before the first Christian missionary activity. It is also quite possible, even likely, that the numerous documented upheavals in Roman Britain (starting with the resistance movements of Caratacos and Boudlca) resulted in the arrival of influential British displaced persons in Ireland bringing with them a variety of Old Celtic speech already penetrated by Latin. Therefore, a Latin loan-word like Old Irish fin 'wine' < vi num could easily predate the missions of Patrick and Palladius; in fact, it could easily reflect a northward transfer within Old Celtic of a trade word first borrowed in Gaul in the centuries BC.

If we still seek to use the Latin loan-words to say something about the Irish language during the time of the conversion, we cannot assume that all the loans, and especially not all the early loans, date from the Christian period. Instead, we must concern ourselves with the smaller subgroup which surely or probably concerns Christianity. The most archaic of these should be able to tell us at what phonological stage ofIrish the conversion began. Examples follow. Olr. sen 'sign, charm' derives from Latin signum borrowed as Pr.lr. * Isignas l which gave *[sYegnah] by vowel lowering, then * I sY egn I by apocope with phonemicisation of lenition, and finally sen by yet another pre-Old Irish sound change (the vocalisation of g before a resonant with compensatory lengthening of the preceding vowel) (McManus 1982, 61). Olr. /deeh 'layman, warrior' derives from Lat. liiieus borrowed as Pr.lr. * Ilaikas I, which then underwent apocope and lenition (McManus 1982, 61). Olr. deehon 'deacon' derives from Latin di aconus borrowed as Pr.lr. * I diiikonas I, which gave * I diakonas I by shortening of non-initial vowels, then * I deakona(s) I by lowering of short i, then * I elY eaeon I by apocope with phonemicisation of lenition and palatalisation, then finally deehon by syncope (McManus 1982, 62). OJr. baehall '(pastoral) staff, crozier' derives from Lat. baculum borrowed as PrJr. I bakuliil, which gave * I bakola I by lowering of short u , then * I baeol l by apocope with phonemicisation of lenition (McManus 1982,62). Middle Irish ortha 'prayer, incantation, &c.' < Olr. *orthu derives from Vulgar Latin oriitio, which gave Pr.lr. * I oriitiii I , whence * I orati(u)yu l by vowel shortening, * I ora8u I by apocope with lenition, then *orthu by syncope (McManus 1982, 62). (This may be contrasted with Olr. oroit, a later borrowing of the same by way of Brittonic or Brittonicised Latin *orm· .) Olr. eruimther 'priest' derives from a British Latin *premiter or *premiterus, itselffrompresbyter. In Primitive Irish, it became first *kwremiteras, which gave *kwrimitera- by raising of short e to i . It is attested in a post-apocope ogam as genitive QRIMITIR I kWri~ YisYirY I (McManus 1982, 99). Olr. eaille derives from Lat. pallium, borrowed as PrJr. * I kWallia· 1 (McManus 1982, 99). The Olr. saint's name Seehnall derives from Lat. Seeundinus , borrowed as Pr.lr. • I sekundi nas I , which gave • I sekundinas I

CONVERSION & TRANSITION

by vowel shortening, then * I sekundena· 1 by lowering, whence Early Olr. *Seehnen after appocope and syncope, then Seehnall by dissimilation and MacNeill's Law (McManus 1982, 101). Olr.eaileeh '(communion) chalice' derives from Lat. ealie·em, borrowed as Pro Ir. • I kalika- I > eaileeh by apocope with phonemic1enition and palatalisation (McManus 1982, 116). Other borrowings predating Primitive Irish Ikw, t, k l > Archaic Irish voiceless spirants I l w), 8, X I include OJr. baithis < Lat. ba(p)tismus, Olr. Satharn(n) < Saturnus (contrast the later Saturn above), eroch 'cross, gallows' < PrJr. * I krok iil < Lat. crue-em.

Amongst the earliest Christian Latin loanwords are those that date from the period when Primitive Irish still lacked, as had Proto-Celtic, a phoneme I p I . Later, yet before the beginning of the Old Irish period, the language acquired a phoneme p. This occurred either through prolonged contact with suchp-rich languages as Brittonic and Latin or through such internal developments as phonetic [p] arising from [b] and [h] brought together by syncope, e.g. Olr. impu 'around them' < Late Pr.lr. *[iMhi-huh] < Pr.lr. * I ambi-sus I . Olr. axal 'apostle' derives from Lat. apostolus, borrowed as PrJr. *lakwatSolas l , which gave *I axwasol l by apocope with phonemicisation of lenition, followed by syncope; contrast the more common post-phonemic lenition borrowing apstal. Olr. Case 'Easter' derives from Lat. Paseha, borrowed as Pr.lr. * I kWiisk iil.

The combined evidence of the more archaic strata of Latin loan-words is, therefore, that the Christianisation of Ireland began at a stage when there was still a Primitive Irish language of essentially Old Celtic phonology. To view it from another angle, the language into which these loans were taken was in all essentials the same as that represented in the oldest stratum of the ogam inscriptions.

Dating the Christian Loan-words The obstacles to assigning close absolute dates to the archaic Christian loan-words are essentially the same as those canvassed above with regards to the dating of the conversion in general. Thus, as we start weighing hypothetical possibilities, ifPalladius preceded Patrick, then the more archaic loans could belong to the Palladian mission, but possibly not the Patrician. If the relative chronology is to be reversed, then the archaic stratum may include Patrick but possibly notPalladius. It is also possible - no matter in which order the two fIfth-century missionaries came - that the archaic stratum is attributable to both and that the next stratum derives from sixth-century missionaries. Conversely, had their been numerous Christians in Ireland before both Palladius and Patrick, then it might be that the archaic stratum was entirely in place bye. 400.

One set of correspondences helps us part way out of this nearly unlimited assortment of possibilities. As I discussed in Koch 1990, three borrowings ofthe archaic type are most plausibly to be attributed to Patrick's activities. These are as follows: (I)Coirtheeh (the name

EMANIA 13. 1995

of the British slave-raider excommunicated by Patrick) < British ' Coratfeos, borrowed as Pr.lr. * / korotfkas /, giving * / korotikas / by vowel shortening, then * / koruteka- / by rai s ing and lowering, then * / korYi9 Yex. / by apocope with phonemicisation of lenition and palatalisation;4 (2) St. Patrick's oath mode brath 'by God's judgement' (Old Welsh Muin Duiu Braut) < British *mon Dewf brotos, borrowed as PrJr. *mo Dew f brotos, giving * / modYewY brdJ / by apocope with phonemicisation of lenition and palatalisation; (3) OIr. Cothraige and variants as alternative names for Patrick, whose explanation is still most probably as a pre-p borrowing of Patricius (notwithstanding Harvey 1985). We can therefore conclude that these three forms came into Irish between c. 400 and c. 490 and that the rest of the comparable archaic stratum belongs wholly or in part to the fifth century.

Ogam and Old Irish Orthographic Systems A further set of facts to be considered in determining the relative chronology ofthe phonological innovations defining the Old Irish horizon is the writing systems themselves. As noted above, there is no revolutionary break separating the orthographic system in use today from that of the earliest surviving Old Irish texts of the seventh century. The same Roman alphabet has been used continuously. Except for a minority of instances in which words have lost segments through regular sound change, where consonants have vocalised and two syllables have contracted into one, and the spelling has been recently revised to reflect this, Late Modern Irish spellings show the same number of syllables as did the Old Irish. Old Irish spelling reflects the phonemic opposition of palatal vs. non-palatal consonants using the same conventions as indicate the opposition in Modern Irish spelling. Though Old Irish spelling does not show lenition as consistently as does Modern, it does, most often, write the lenitions of c, t, andp as ch, th, and ph, which is the analogical basis for writing -h­after all the 'aspirated' consonants in Modern Irish. From the ninth-century St. Gall glosses onwards, the punctum delens is often used with lenitedfands in Old Irish. Opposed consonant phonemes are also implied in the patterns of doubling unlenited consonants. The principle of morphophonemic lenition is thus acknowledged by the orthographic system as a whole. It is clear from the foregoing details, the Old Irish orthographic system was devised for a language that had a post-apocope and syncope syllable structure and phonemic palatalisation and lenition.

The ogam system contrasts thoroughly. The characters are different of course. Apocope and syncope are not revealed in the more archaic inscriptions. The system has no consistent way of revealing palataIisation and lenition, though in some ogams we seem to see a pattern in which lenited consonants are prone to be written double (in other words, exactly the opposite of the value of double consonants in standard Old Irish) (Harvey 1987b). As a writing system for Old Celtic

PAGE 45

names, ogam was about as good or as poor as was the long-standing and highly stable application of Roman letters to Old Celtic names in Britain and Gaul. There can be little doubt but that the ogam system was devised for a Celtic language which had as yet no phonemic lenition or palatalisation and had not undergone apocope and syncope.

Though we must conclude that the two spelling systems were of distinct origin and were devised at two widely separate chronological horizons ofIrish speech, it would be overstating the case to suppose that the younger system was, upon its arrival, altogether sealed off from the older. (Similarly, though the shift from Romano-Celtic to Neo-Brittonic spelling systems is to be regarded as revolutionary rather than evolutionary, there was never a 'scorched earth layer' between the two, and hybridisation and contamination occasionally occurred') It is apparent fromAuraicept na nEces that once interest in a prestigious written vernacular arose within the Early Irish period·, the residual ogam tradition was thrown into the breech alongside the newer and more mainstream system of writing Old Irish in Roman letters (Ahlqvist 1982). A further interesting point in this connection is that some texts assignable to the seventh century (namely the Wiirzburg prima manus glosses and Tirechan's Collectanea of Patrick's churches) show idiosyncrasies in their spelling of Irish forms. Thus, in the former, post-vocalic b, d, g may stand for / b bY, d dY, g gYI, whereas as the main body of the Wiirzburg glosses would consistently use p, t, c. In the same source, Harvey (1989) has detected a significant minority of examples in which the voiceless spirants are ambiguously represented by t, c alongside the more regular and (even there) more common th and ch. Carney (1978-9, 417-9) drew attention to Tirechan's doublingoflenited consonants in Irish personal names: Connan us, Coimmanus, Cennannus, Roddanus = OIr. Comin, Coiman, Cianan, Ruadan . In these several features, the anomalies of Early Old Irish spelling can be accounted for as transitional holdovers of ogamic principles in the context of a newer spelling system. This newer orthography was in its chief structural principles much the same as the Neo-Brittonic and probably introduced to the Irish by the Britons (cf. Carney 1978-9; Harvey 1989).

To summarise the argument thus far, the earliest Christian loan-words (presumably belonging to the beginning of the conversion), the earliest ogam inscriptions, and the invention of the script itself, all belong to the Primitive Irish horizon. The Patrician mission also belongs to this stage and may be placed in absolute terms within the fifth century. Ogams in Primitive Irish form continued to be produced as late as the earlier-to-mid sixth century. On the other hand, we have written Old Irish by the mid seventh century.

The Neo-Celtic Phenomenon Before attempting to explain how Primitive Irish was rapidly transformed into Old Irish, there is another

PAGE 46

highly relevant fact that must be brought under consideration: closely analogous changes appear to have transformed Brittonic at much the same time. A convenient illustration is provided by returning to the 'Voteporix' stone; for , characteristic of the ogams of Wales, it is a bilingual inscription. In the Roman letters on the stone's front the Romano-British genitive of the name is VOTEPORIGIS, which (like the ogam VOTECORIGAS) must be regarded to be an Old Celtic form. Taking this spelling at face value, there has been as yet no apocope or syncope and thus no phonemicisation of opposed lenited and unlenited consonants. We have also a second contemporary (i.e. early-to-mid sixth-century) attestation of the name of this same individual, namely the Latin vocative Vortepori in Gildas' De exeidio Britanniae. Unlike the Ogam Irish and Romano-British inscriptional forms, Vortepori reveals an early popular ignorance of the name's etymology; thus Latin Vorteporius has lost sight of both of the compound names etyma, Celtic *wo· tekwo· 'protect' and · rj xs 'king'. In this way, Gildas' anticipates the Old Welsh form Guortepir used for this king in the Dyfed pedigree in London, British Library MS, Harley 3859 (Bartrum 1966, 10). It is a significant point that the monastic church father and accomplished Latinist Gildas had the defective Celtic name form, whereas the contemporary stonecutter who wrote the substandard Latin PROTICTORIS had the more correct Celtic. Interestingly, the same individual is named in the Irish pedigree of the Deisi (for this is the origin of Dyfed's dynasty), and the name there is not at all like the * Foithehireeh (nom. * Foithehiri ) posited above, but ratherGartbuir(Bartrum 1966,4), which can be nothing other than a borrowing into Irish of the Old Welsh Guortepir, complete with its mistaken etymology.

That borrowing would be as old as the seventh or eighth century. But we have other Welsh evidence showing clearly that a fully developed Neo-Brittonic (with apocope, syncope, and phonemic lenition) had emerged and was in use as a written form for proper names, at least, by the turn of the sixth and seventh centuries. Spellings of the Neo-Brittonic variety occur in the sequence of the witness lists in the Liber Landauensis beginning in the last decades of the sixth century (Davies 1980; Koch 1985-6). However, these documents are copies (probably at multiple removes), so the exact nature of the original orthographic standard can only be inferred on the basis of an accumulation of examples, which become more abundant from the early seventh century. At the Second Council of Braga in AD 572, one of the signatories was the Galician Breton bishop whose name was recorded as Mahiloe or Mailoe (Thompson 1968; Bernier 1982, 115-17; cf. Sims-Williams 1991, 20-21). This form is wholly consistent with native Neo-Brittonic spelling, and so is likely to reflect the churchman's own rendering of his name, for which the Romano-Celtic spelling would have been Maglaeus; cf. Gildas' Maglocune (Lat. voc.), a spelling of AD 500 x 550. Coinmail (for Archaic Neo­Brittonic Conmail) in the Anglo-Saxon Chronicle entry

CONVERSION & TRANSITION

for 577 may also derive from a contemporary written form (but see Sims-Williams 1990, 245).

Explaining the Neo-Celtic Phenomenon In light of the above, the criterion of logical economy demands that whatever explanation we adopt for the transformation of Old Celtic in Ireland it will hardly do to invoke a different explanation (effectively a coincidence) for the more-or-Iess contemporary and strikingly parallel transformation of the Old Celtic in Britain. We must offer a unitary solution for a unitary Neo-Celtic phenomenon. On this basis we must first reject Jackson's suggestion that the rapid transformation of Brittonic was a by-product of the social upheaval brought about by the Anglo-Saxon 'invasion' of lowland Britain in the fifth and sixth centuries (1953, 690-1; 1954,70). As stated at the top, whereas Britain participated in all three shocks which transformed ancient Celtic Europe into the Insular Neo-Celts (the Roman Conquest, the post-Roman Germanic invasions, and the conversion), Ireland experienced only the last. Whatevertrauma the English settlements mayor may not have caused for Brittonic society, Ireland stayed effectively aloof. Furthermore, Jackson never made it adequately clear why the English settlements in Lowland Britain would precipitate rapid linguistic change in Wales, where an intact society of native kings, lords, commoners, and churchmen remained till Norman times. Likewise, however profound an influence the Romans inflicted upon the Britons between AD 43 and 410, the impact upon the further island had been altogether different and less.

The third epochal change affected both branches of the Insular Celts as a single community. The period of apparent rapid emergence of Neo-Celtic indeed coincides with a key period in Insular Christianity, the age of saints. How is it, then, that the conversion might have the effect of eradicating Old Celtic forms from both Britain and Ireland during the course of the sixth century and replacing it with radically evolved Neo­Celtic vernaculars by the century AD 550-650? (In answering we must bear in mind that neither Insular Celtic subfamily has since undergone a comparable revolutionary change down to the present. )

I must begin to answer that question by explaining that written evidence (which is to say all the evidence we have) is presenting us with an illusion. Notwith­standing Jackson's carefully argued chronologies of sound changes for both Brittonic and Goidelic, it is simply impossible for a language to have evolved as quickly as the evidence seems to imply. At least it is impossible if we think of language as possessing the uniform, mono-dimensional texture that Jackson's models assumed. What is possible, though, is for a language to evolve over a long period of time at a popular colloquial register while evolving at a much slower rate in its educated standard form. This upper register will naturally retain a monopoly on all written documents, including funerary inscriptions. However, late pagan Ireland (and likewise sub-Roman Highland

EMANIA 13. 1995

Britain) were largely oral societies. Most readers will be familiar with the famous statements in Caesar's De bello Gallico 6.14 to the effect that Druids preferred not to commit their teachings to writing, but instead underwent training of up to twenty years memorising oral learning. While no direct evidence of pagan Celtic oral learning could possibly survive , the rote transmission of pagan Celtic doctrine from generation to generation of professionals necessarily implies a learned variety of Celtic speech more unchanging and less dialectal than that of the uneducated tribesfolk. It is Caesar also who tells us that it was the Druids of Gaul who convened annually at a central location in the territory of the Carnutes and who went to Britain for higher training (BG 6.13). Both points strongly suggest that the educated speech of the Druids would exhibit minimal regional variation.

A colloquial language only becomes vulnerable to replacement if an entire society is uprooted or withers away, but an educated standard can be drastically altered or replaced through the far less disruptive displacement of an educated elite or even with an elite's evolutionary reformation of itself (cf. Greene 1971). So, for example, a change of religion (or even the reformation of a continuing religion) can easily have the effect of removing an old sacred language from the curriculum, whether or not the priesthood is replaced, purged, or henceforth recruited from a different class. Witness the declining knowledge of Latin at present, even amongst educated Roman Catholics.

What I am proposing therefore is that until the mid sixth century (or later) both Britain and Ireland retained a learned elite possessing closely related varieties of Old Celtic within their respective educational establishments. In Britain, pagan Celtic tradition had reached a modus vivendi with pagan Rome and Classical Latin as early as the first or second century. But even for Ireland, the current scholarly consensus holds that ogams are an offshoot of later Latin grammatical learning and more specifically ofthe Roman alphabet. However and whenever ogam was invented, one must see the coming of literacy to Britain and Ireland occurring in two stages: first, a pre-Christian or secular wave, compatible with, and duly grafted onto native learning, and second, the less compromising bible­centred literacy of the conversion itself accompanied by the church's monopoly on learning. I am not saying that druidism was necessarily still going strong in Britain, as in Ireland, ate. AD 500; in fact, there is more evidence against such a proposition than there is to favour it. Rather, Late Roman Britain had a mixed tradition - pagan/secular and Christian, Celtic and Roman - and a Romano-British landed aristocracy to support this hybrid . But as part of the transition from Roman Britain to DarkAge Wales, education retreated into the church.

Non-linguists may not immediately grasp the implications of a faltering standard language. The key points may be illustrated well enough with reference to English. Modern English today shares some broadly

PAGE 47

similar features with the Old Celtic of ancient Europe as portrayed here. Both languages spread rapidly from a compact homeland over a wide geographic area. And over this wide area, most teaching and official communication in English is through the medium of an archaisingeducated standard which conceals the degree of divergence and evolution of the local colloquial forms. As a written language, English possesses a conventionalised and old-fashioned spelling system. The retention of such features as the silente, so-called 'long vowels' (which in most dialects became diphthongs centuries ago), long-since vocalised gh's, and so on, could easily give the impression to a historical linguist of some future millennium that the phonology of our speech differed little from that of Chaucer's. Let us then suppose that today's English were completely replaced by another language as a medium of writing and education, though surviving as the everyday spoken language of the people. Something like this did in fact happen during the period of the apparent rapid evolution of English between 1066 and c. 1200. For the sake of our futuristic example, let us say that the language which replaces it this time is not Norman French, but Irish. (I shall leave it to aspiring science fiction writers to invent a plausible scenario for this outcome.) Native English speakers are no longer taught to read and write English; all books and book learning is in Irish. The older learned tradition whose medium was English is, for whatever reason, no longer regarded as legitimate and is therefore abandoned and wholly or largely forgotten. Let us say, then , that at some subsequent generation the intellectual climate changes, and an English literary language is now once more regarded as permissible and desirable to be promoted alongside the formerly monopolistic Irish. Because the tradition of writing English according to the old rules had died (or at least lapsed into serious confusion), a new writing system would have to be created, and this could have no model to base itself upon other than the conventional spelling ofIrish (as I am saying that Irish is the only literary language that had been lately known and cultivated). We may obtain a sample of how the 'Neo-English' implied by this hypothetical fantasy might appear by excerpting an anthologised column by Brian O'Nolan (a.k.a. Flann O'Brien, a .k.a. Myles na gGopaleen).

Aigh no a mean tM ios so LeasaC dat the sUps in thios clos, bhuears a biord, eand dos not sm6c bioeos obh do trabal obh straighcing a meaits. lot ios 56 long soins thi duid an onast daes bhore dat thi tines mainiuilleabar ios de neim obh a Poirtiuguis aiditeatar. [lamhd leaftar] (1968, 263)

[I know a man who is so lazy that he sleeps in his clothes, wears a beard, and does not smoke because ofthe trouble of striking a match. It is so long since he did an honest day's work that he thinks manual labour is the name of a Portuguese agitator. [loud laughter]]

PAGE 48

Now, if we recall our imaginary historical linguist of the distant future and suppose that he had at hand English documents datable to the 1990s alongside O'Nolan-style 'Neo-English' documents datable to c. 2150 and assume that that linguist understood the conventions ofIrish spelling better than he did those of English, then he might conclude that such Middle English to Modem English changes as the Great Vowel Shift had occurred in rapid succession between 1990 and 2150, when in fact they had occurred gradually at a much earlier date. We would also seem to be seeing a very rapid separation of dialects; for a reinvention of written English according to new principles would for the first time reveal the h-less-ness and r-less-ness of the English of England, the advanced diphthongisation of Australia, Southern England, and the American South, and so on.

Such a futuristic scenario is of course enormously farfetched, primarily because the contemporary English tradition of literacy is so strong and widespread. Conversely the tradition of writing the Old Celtic languages had always been relatively slight. The elite recipients of pagan Celtic oral learning had been a small specialist minority. The first waves of Christianity, had these had even a mild democratising effect, might have brought sufficiently large numbers of new clerics from the Irish and British lower classes into being (educated in the Christian Latin tradition but not the Celtic) to swamp the native learned classes and their whole mode of communication without openly defeating them.

On the other hand, it is possible that there was open confrontation between the missionaries and the IGngs' Druids, as Muirchu's Vita Patrieii and so many other Irish saints' lives would have us believe. If so, it is possible that there was hostility by Early Christians to the learned variety of Celtic (as the liturgical language of a rival organised religion) as well to the doctrines embodied in that language. It would not be surprising if the church's evident failure to cultivate vernacular literature in the fifth and earlier sixth centuries coincided with the period when there was still a rival pagan establishment closely linked to the more conservative form of the native language. Only when organised Celtic paganism was dead or moribund was it possible for monastic literati to invent the origin legend of the Irish language found in Auraieept na nEees , according to which Old Irish was a virtual reconstruction ofmanlGnd's perfect pre-Babel language and the full spiritual equal of the tres linguae saerae, Hebrew, Greek, and Latin (Ahlqvist 1982; cf. McManus 1991, 148-9).

Implications for the Literary Tradition As I explained in greater detail in my contribution to the Conference on the Ulster Cycle (Koch 1994), the scenario argued for here envisions a considerable obstacle for the transmission of any learned oral tradition between the pagan (i.e. Primitive Irish) and Early Christian (i.e. Old Irish) periods. One way of

CONVERSION & TRANSrrION

thinlGng of the situation would be the way Jackson thought about the transmission of Early Welsh poetry. According to him, ancient (i .e. pagan-period) Brittonic poetry could not have been transmitted through the sixth century because apocope and syncope would have destroyed the metre (1953, 690-3). This is rather like saying that the younger generation fails to understand the elevated Welsh of the pulpit because it is so much more old-fashioned than their own natural language. In both cases, I think the causality is more nearly the reverse. The youth ofWelsh-spealring Wales find hymns and liturgy incomprehensibly archaic because of the prior operation of social changes which emptied the chapels. Faithful attenders in their pews have not suddenly experienced a latter-day Babel. Applying the analogy of the known to the unknown, the survival of Primitive Irish hymns, poems, and stories was not barred by the chain oflinguistic changes leading to Old Irish , rather, the people when they became Christians stopped listening to old traditional materials because these belonged to the dying social institutions of paganism. And thus cut off from the oral tradition of the pagan past , the language of the converted Irish was then able to evolve rapidly into Old Irish.

Pre-Celtic Substrata The theory as set out thus far does not require that any pre-Celtic language had been spoken in Britain or Ireland on the eve of the conversion. A language may develop polarised registers within a country without the lower register being a creolised version of the higher, that is to say, a derivative of a colonising language transferred to an indigenous population which had formerly spoken a different language or languages. For example, the aboriginal languages have but limited explanatory power in accounting for innovations in worlGng-class Australian English. The popular speech of a linguistically homogeneous population can in this way evolve rapidly according to its own internal dynamic. On the other hand, register variations of the sort proposed for pagan Ireland can indeed be the result of the imperfect transmission of a language from an international elite to a colonised population. The co­existence of standard French and creole in Haiti would be one such example, and it shows how long such substratum effects may persist after the older native language or languages of the under-classes have disappeared. Of course, one need look no farther than the Hiberno-English oflong since non-Gaeltacht areas of the East. So, while pre-Celtic substrata are not required to account for the disparity between Old Celtic and Insular Neo-Celtic, the data is probably somewhat more intelligible with them than without them.

This is not a particularly controversial position. Regarding substrata, Jackson wrote:

... on the one hand ... the Celtic languages, including Irish, are fundamentally Indo­European ones, and hence that their basic

EMANIA 13, 1995

structure is much the same as that of most of the languages of Europe; or at any rate, as early stages of those languages, like Latin and Greek .. . On the other hand, the Celtic languages , and particularly Irish , have developed in other respects in some very strange ways , some of which may possibly be due to the speech-habits of pre-Celtic peoples learning Celtic; 'aspiration' and the other initial mutations are remarkable examples (1969, 10).

And more recently, 6 Corniin:

... Ireland had highly developed and impressive cultures in the Neolithic and Bronze Age, and the incoming Celts, who were never more than a dominant minority amongst a non­Celtic and non-Indo-European majority, were heavily influenced by the societies which they found before them in Ireland. . .. The Irish language is an indigenous realization of Celtic, heavily influenced by the pre-Celtic languages spoken in Ireland and containing an unknown number of words (personal and population names among them) borrowed from these languages (1989, 1-3).

As it stands the preceding passage is liable to be dismissed as a positive pronouncement by a historian on the subjectoflanguage. Nonetheless, it is compatible with (though not in all details necessarily demanded by) the evidence reviewed here.

Various writers have attempted to identifY the genetic filiation ofInsularpre-Celtic. Repeatedly, these scholars have been struck by the numerous Goidelic and Brittonic syntactic affinities for the Hamitic and Semitic languages of North Africa and the Middle East (including Arabic, Berber, and Egyptian), pointing towards the possibility of a pre-Celtic, pre-Indo­European Hamito-Semitic substratum in Britain and Ireland. Ideas along these lines were shared by Celticists ofthe stature of Sir John Morris Jones, Julius Pokorny, and Heinrich Wagner, though even so, the 'Hamito­Semitic Hypothesis' was generally greeted with scepticism. At present, in independent work, the idea is being revived and pursued with more exhaustively detailed and closely controlled methods by Jongeling (1991) and especially by Gensler (1993).

Conclusion In earlier work, I suggested that Britain and Ireland had formed an outer Atlantic Zone of Celtic-speaking Transalpine Europe beginning in the Late Bronze Age, perhaps specifically from the Hekla 3 climatic disaster of1159-1141 BC (Koch 1994; 1991;Mallory and Warner 1988; Baillie 1993, 7-8). From this period, Ireland is seen to be in receipt of cultural trends emanating from the presumed homeland of Celtic speech in Hallstatt West Central Europe. In this view, Old Celtic in Ireland had held the status of an elite international language for many centuries, in fact more than a millennium,

PAGE 49

prior to the conversion. Irish scholars are used to taking ogamic Primitive

Irish as a beginning, but I am proposing here that it is more meaningfully understood as the end, the final remnants of the sacred learned language of the pagan Celts long after this prestige speech had been replaced by Latin and Germanic at the epicentre of Celtic expansion on the Continent. Over the millennium and a half which preceded Ireland's conversion, local and imperfect adaptations ofthis Celtickoine had gradually replaced Ireland's older language(s) at the least evident and least valued levels of society. At that popular register, standard Old Celtic had never obliterated substrate phonetic and syntactic habits . When Christianity and literate Latin finally broke the limits of the Empire to replace the older prestige language on Europe's north-western fringe, this cleared the way for Ireland's previously invisible 'indigenous realisation of Celtic' to become Europe'~ first great literary vernacular.

Notes 1 As he said at the Seventh International Congress of

Celtic Studies at Oxford in 1983.

2 The term 'Old Celtic' as used here covers all the attested and reconstructed languages of the Iron Age and Roman Period. Specific linguistic forms under this label conform to the developments known in the closely related central innovating dialects (Goidelic, Brittonic, and Gaulish) and sometimes show innovations not shared by the more conservative marginal dialects Celtiberian and Lepontic. See further Koch 1992.

3 As discussed in Koch 1990, I agree with Martinet (1952) that lenition on the phonetic level is a Common Celtic or Pan-Celtic phenomenon and therefore most go well back before the period under discussion. The lenis allophones of the voiceless stops had originally been voiceless lerns stops, as they have partly remained in Breton and Welsh. The shift of the voiceless stop series Old Celtic k W , t, k (in lenis position) > Old Irish ch, th, ch is a later and purely Goidelic innovation. It probably dates to the time of the phonemicisation oflenition as a consequence of the Primitive Irish syllable losses. However, this explanation is essentially immaterial to the matters under discussion presently. Therefore, for the sake of simplicity, what I see as a distinct sound change of spirantisation of voiceless lenis stops will here sometimes be covered under the shorthand of 'phonemicisation of lenition', though in fact I think it possible that the Goidelic voiceless lenes might still have had stop articulation as late as the post-apocope, pre-syncope horizon which is sometimes called 'Archaic Irish'.

4 Sims-Williams (1990, 229 n. 48) objected that Coirthech might derive from written Coroticus by what he calls 'mechanical late Hibemicization.' If such a process could, in effect, mimic the historical workings of Irish vowel shortening and lowering, syncope, and t, c > th, ch, then there is no reason to posit chronological strata for Christian loan-words in Irish at all ; instead, we would have e ffecti vely ahistorical sub-groups of words exhibiting greater and lesser degrees of 'mechanical late Hibernicization'. Generally speaking, scholars do not understand the varying phonological features ofthe loan-words in this way. Greater Hibernicisation, like

PAGE5Q

greater assimilation to the sound system of any borrowing language, is most natura lly explained as the earmark of an earlier loan-word. To make a special case (or three different special cases) for the loan-words which pertain to St. Patrick is not a valid line of argumentation.

S For example, the name Clemuis in a Llandaf charter datable to c. 620 (Davies 1979, 155) is obviously derived from Latin Clemens, the regular Old Welsh form of which is Cluim, Archaic Clem.

Bibliography Ahlqvist, A. , 1982 . The Early Irish Lingui st,

Commentationes Humanarum Litterarum 73, Helsinki. Baillie, M. G. L., Dark Ages and Dendrochronology,

Emania 11, 5-12. Bartrum, P. C., 1966. Early Welsh Genealogical Tracts,

Cardiff. Bernier, G., 1982. Les chretientes Qretonnes continentales

depuis les origines jusqu'au ]xeme siecie, Rennes. Breatnach, L., 1984. Canon Law and Secular Law in

Early Ireland: The Significance of Breth a Nemed, Peritia 3,40-44.

---,1989. An Edition of/1mraSenain, in Sages, Saints and Storytellers, ed. D. 0 Corrain, L. Breatnach, K. R. McCone, Maynooth, 7-31.

Carney, J ., 1978-9. Aspects of Archaic Irish, Eigse 1714, 417-35.

- - -, 1989. The Dating of Archaic Irish Verse, in Early Irish Literature - Media and Communication / Miindlichkeit und Schriftlichkeit in der friihen irischen Literatur, ed. S. N. Tranter , H. L. C. Tristram, TUbingen, 39-55.

Davies, W., 1979. LlandaffCharters, Aberystwyth. --, 1980. The Orthography of the Personal Names in

the Charters of Liber Landauensis, Bulletin of the Board of Celtic Studies 38/4, 553-7.

Esposito, M. , 1956. The Patrician Problem and a Possible Solution, Irish H istorical Studies 10.38, 131-55.

Gensler, 0., 1993. A Typological Evaluation of Celtid Hamito-Semitic Syntactic Parallels. Dissertation. University of California at Berkeley.

Greene, D., 1971. Linguistic Considerations in the Dating of Early Welsh Verse, Studia Celtica 6, 1-11.

---, 1973. The Growth of Palatalization in Irish , Transactions of the Philological Society, 127-3!i.

Harvey, A., 1985. The Significance ofCothraige, Eriu 36, 1-9.

---, 1987a. Early Literacy in Ireland: The Evidence of Ogam, Cambridge Medieual Celtic Studies 14, 1-15.

---, 1987b. The Ogam Inscriptions and their Geminate Consonant Symbols, Eriu 38, 45-71 .

- - -, 1989. Some Significant Points of Early Insular Orthography, in Sages, Saints and Storytellers, ed. D. 6 Comiin, L. Breatnach, K. R. McCone, Maynooth, 56-66.

Herbert, M., 1989. The Preface to Amra Co!uim Cille, in Sages, Saints and Storytellers, ed. D. 0 Corrain, L. Breatnach, K. R. McCone, Maynooth, 67-75.

Jackson, K. H., 1953. Language and History in Early Britain, Edinburgh.

--, 1954. The British Language During the Period of the English Settlements, in Studies in Early British History, Cambridge, 61-82.

--, 1959. Edinburgh and the Anglian Occupa tion of Lothian, in The Anglo-Saxons: Studies in Some Aspects of their History and Culture, ed. P. Clemoes, London, 35-42.

---, 1968. A La te Inscription from Wroxeter, The Antiquaries Journal 48, 296-300.

CONVERSION & TRANSITION

- - , 1969. The Irish Language and the Languages oJ the World, in A View of the Irish Language, ed. B. 0 Cuiv, Dublin, 1-10.

Jongeling, K., 1991. The Character of the (Non Indo­European?) Substratum of the Insular Celtic Languages. Abs~ract. In the program of IX' Congres International d'Etudes Celtiques, Paris.

Koch, J. T ., 1985-6. When was Welsh Literature First Written Down?, Studia Celtica 20/21, 43-66.

- - -, 1990. *Cothairche, Esposito's Theory, and Neo­Celtic Lenition, in Britain 400-600, Language and Hi story , ed. A. Bammesberger , A. Wollmann, Heidelberg, 179-202.

--, 1991.Eriu,Alba, and Letha: When Was a Language Ancestral to Gaelic First Spoken in Ireland?, Emania 9,17-27.

---, 1992. 'Gallo-Brittonic' vs. 'Insular Celtic': The Inte r-relations hips of the Celtic Languages Reconsidered, in Bretagne et pays celtiques: langues, histoires, civilisations - melanges offerts a la memoire de Leon Fleuriot 1923-1987, ed. Gw. Le Menn, J .-Y. Le Moing, Saint-Brieuc and Rennes, 471-96.

--, 1994. Windows on the Iron Age, 1964-1994, in Ulidia: Proceedings of the First International Conference on the Ulster Cycle of Tales , ed. J. P. Mallory and G. Stockman, Belfast, 229-237.

McManus, D. , 1982. The Latin Loan-words in Irish. Dissertation. Trinity College Dublin.

---, 1983. A Chronology of the Latin Loan-words in Early Irish , Eriu 34, 21-71.

--, 1991. A Guide to Ogam, Maynooth. Mallory, J . P ., and R. B. Warner, 1988. The Date of

Haughey's Fort, Emania 5, 36-40. Morris Jones, J., 1900. Pre-Aryan Syntax in Insular

Celtic, in J . Rhis and D. Brynmor-Jones, The Welsh People , London, 617-41.

Na Gopaleen, M. (Brian O'Nolan , a.k.a. Flann O'Brien), 1968. The Best of Myles, Harmondsworth.

O'Brien, M. A. (ed.) 1962. Corpus GenealogiarumHiberniae , i, Dublin. o Corrain, D., 1985. Irish Origin Legends and Genealogy:

Recurrent Aetiologies, in History and Heroic Tales: A Symposium, ed. T. Nyberg, 1. Pi~, P. M. S~rensen, A. Trommer, Odense, 51-96.

- - - , 1989. Prehistoric and Early Christian Ireland, in The Oxford Illustrated History of Ireland, ed. R. F. Foster, 1-52.

Pokorny, J., 1960. The Pre-Celtic Inhabitants of Ireland, Celtica 5, 229-40.

Sims-Williams, P. , 1990. Dating the Transition to Neo­Brittonic: Phonology and History, 400-600, in Britain 400-600, LanguageandHistory, ed. A. Bammesberger, A. Wollmann, Heidelberg, 217-62.

--, 1991. The Emergence of Old Welsh, Cornish and Breton Orthography, 600-800: The Evidence of Archaic Old Welsh, Bulletin of the Board of Celtic Studies 38, 20-86.

Stevenson, J ., 1989. The Beginnings of Literacy in Ireland, Proceedings of the Royal Irish Academy 89/ C/6, 127-65.

Stokes, W., 1899, 1900. The Bodleian Amra Choluimb Chille,Revue Celtique 20, 31-5, 132-83, 248-89, 400-37; 21, 133-6.

Thompson, E. A. , 1968. Britonia, in Christianity in Britain, 300-700, ed. M. W. Barley and R. P. C. Hanson , Leicester, 201-5.

Wagner, H., 1959. Das Verbum in den Sprachen der britischen Inseln, TUbingen.