Move back one step  Move forward one step  List of themes 

    The Celtic background

   Divisions within Celtic
   Indo-European affiliation
   Insular Celtic
   Present-day languages
   Mutation in the Celtic languages

Knowledge of the Celts in pre-history is derived from (i) references to them in the works of classical authors (the earliest is Herodotus, 5th century BC, from whom comes the term ‘Celt’: Greek Keltoi ‘Celts’, later Latin Celtae) and (ii) archaeological remains (Schlette, 1979: 13-43; Laing, 1979: 1-14). For the latter one can consult the chapter ‘Ethnogenesis: Who were the Celts?’ (Renfrew 1987: 211-249) in which Renfrew gives a very broad overview of the supposed distribution and movement of the Celts since their appearance in history. He furthermore touches on the question of the spread of the Celts to Britain which he does not see as consisting of identifiable migrations but successive waves over a very long period starting perhaps as early as 2000 BC with the so-called Beaker People.

There is an identifiable culture known after the location Hallstatt in Austria. This was early Iron Age (c 800-450 BC), though other authors (not just Renfrew) see in the preceding Bronze Age Urnfield culture, and perhaps in the tumulus (‘earth mound’) culture in central Europe north of the Alps, the first appearance of the Celts in an area roughly from the Rhineland across Bavaria to Bohemia. The late Iron Age is represented by the La Tène (c 450-100 BC) stratum of Celtic culture named after a site in Switzerland.

The coming of the Celts to Britain is difficult to date and can be placed in any period from a distant 2000 BC when the Bronze Age Beaker Folk came to Britain to a more recent 500 BC when the Iron Age people arrived in successive waves (Dillon and Chadwick, 1967: 4). The last distinct wave of immigration is that of the Belgae in the first century BC (Caesar mentions that they crossed from northern Gaul to Britain). This gives the following picture for Britain.

0) Pre-Iron Age settlers ?
1) Hallstatt stratum 500 BC ->
2) La Tène stratum 300 BC ->
3) Invasions of the Belgae 100 BC
4) Immigration from Gaul on Roman subjugation 58-50 BC

Divisions within Celtic

Before discussing the development of the Celtic languages it is necessary to introduce some terms. Proto-Celtic is the stage at which Celtic separated from the remaining Indo-European dialects. Unlike Germanic there is no single major linguistic change which marks the initial stage of Celtic as a separate branch, for instance there is no equivalent in Celtic to the Germanic sound shift. However, there are a number of features which are common to all Celtic languages and which are assumed to be inherited from the earliest stage of the branch. The most prominent of these features is the loss of IE *p in all positions except adjacent to a homorganic obstruent (*t, n, s Hamp 1951: 230). This involved the weakening and final deletion of the labial plosive. The labial stop may have been articulatorily weak anyway (Sommerfelt 1962: 347). In this context, one can also refer to loss of *p elsewhere in Indo-European, e.g. in Armenian, where it disappeared as part of a series of shifts from stop to fricative roughly on the lines of the Germanic sound shift. However, there was probably a general leniting quality to early Celtic anyway (the ultimate source of later lenition) which would have encompassed the labial plosive. Recall that Indo-European *b, which would have formed a plosive pair with *p and thus added stability to the latter, is a sound which probably did not exist anyway. Adhering to the glottalic theory (Hopper, 1973; Gramkrelidze and Ivanov, 1983), i.e. that *b would have been an ejective, does not result in a labial plosive pair, but may account typologically for the lack of *b which is interpreted as /p’/, i.e. not /b/ (Greenberg 1970: 127; Gamkrelidze and Ivanov 1973: 154f.).

Continental Celtic (Eska and Evans 1993: 26-43) is a term which is given to Celtic as spoken chiefly in Gaul and on the Iberian Peninsula (Tovar 1961: 76-90). From the few inscriptions one can say that the language, which was probably introduced to the Iberian peninsula from 850 BC onwards (Tovar 1961: 78), was more archaic than later forms of the Celtic languages. One of the main features is that the language retained IE labio-velars and did not lose IE *p. Both Tovar (1961: 80) and Lejeune (1955) are of the opinion that lenition existed in Celtiberian although the evidence is scanty.

The Botorrita Inscription – the major attestation of Celtic on the Iberian Peninsula – on four bronze plaques (I-IV, but III is in Latin) found near Zaragoza in Aragon in northern Spain. The plaques were discovered in 1970, 1979 and 1994. They are dated to the first century BC.

The term ‘Gaulish’ refers to that variety of Continental Celtic which is attested from inscriptions up to the first few centuries AD. For comparative Celtic studies Gaulish has a certain referential value as it illustrates a stage of the language which is prior to both forms of Celtic (P- and Q-Celtic) which developed in the British Isles. At this stage the common language was still inflectionally complex as the Gaulish inscriptions attest (Schmidt 1957; Gray 1944).

Literature (general, see relevant sections of the following books)


Literature (specifically on Celt-Iberian)



The arrival of Celts in Britain can be assumed to have taken place at the very latest by the 5th or 4th century BC. By the end of the 2nd century BC the Romans had begun to expand into Gaul, something which led to the suppression of Celtic. The scant linguistic remains of Gaulish (Fowkes, 1940; Gray, 1944) are largely onomastic: inscriptions as well as references in classical sources.


Two Gaulish inscriptions illustrating remnants of this language. The one on the left is from Vaison-la-Romaine in Provence and the one on the right from Chamalières in the départment of Puy-de-Dôme.

Linguistic evidence for continental Celtic is slight but there is enough of it to realise that the language forms spoken on the mainland of Europe still retained much of the morphology which it had inherited from IE.

Gaulish Old Irish Welsh meaning
uxellos uasal uchel ‘high, noble’
vindos find gwynn ‘fair, beautiful’
nertomaros nertmar nerthfawr ‘strong, powerful’

It is clear from just a few forms that adjectival and nominal endings were present in Gaulish, clusters like /ks/ and /nd/ still existed and internal voiced stops had not yet been fricativised, or at least this was not so established a feature of Gaulish for it to be orthographically recognised. Gaulish still maintained many of the inherited inflectional endings, particularly in the nominal area, e.g. for the nominative singular of nouns as evidenced in the following forms.

Gaulish Irish Welsh meaning
/-os/ cattos cat cath ‘cat’
/-on/ dúnon dún din ‘castle’

Relative chronology of syncope and apocope

It would appear that the syncope and apocope which occurred in Celtic in the pre-written period followed lenition as the intervocalic environment for the latter must have been available. Consider the following forms.

Gaulish Old Irish Modern Irish meaning
nertomaros nertmar neartmhar ‘strong’

The adjectival ending -mhar shows lenition which historically can only occur if the /m/, from which the /v/ (nowadays written mh) derives, was originally in intervocalic position. That is, the lenition must have occurred before the preceding -o- was lost by syncope. Note that Irish, along with Scottish Gaelic, has resisted the development of /z/ despite the many intervocalic occurrences of /s/.

Cluster simplification

This can be seen as part of the general tendency to reduce word forms phonetically, a tendency which can be taken to be connected with the strong initial stress accent which led to a natural weakening of unstressed syllables.

Gaulish Old Irish Welsh meaning
uxellos (x = /ks/) úasal uchel (ch = /x/) ‘noble’
vindos find, later N+stop > NN Mod. Irish: fionn ‘fair’


Indo-European affiliation

All the languages just mentioned belong to a branch of Indo-European known as Celtic. Its relationship to other branches is unclear, formerly scholars thought – on certain morphological grounds – that there was an earlier unity between Italic and Celtic (see Kortlandt 2007 for a re-assessment of this view). On a firmer footing is the location of the Celts. There are two archaeologically defined cultures which are associated with the Celts in the latter half of the first millennium BC, the earlier Hallstatt and the somewhat later La Tène culture (see above).


One can safely say that the La Tène Celts were located in central Europe in a band stretching from eastern France across to approximately present-day south-west Poland. Onomastic evidence helps in determining this, for instance the names of the river Rhine ( < Celtic *re:nos < IE *reinos, Schmidt 1986: 206) and Isar and names of regions like Bohemia (Böhmen, the area of the Boii, the wood-dwellers, Chadwick 1971: 52; Krahe 1954: 123) are Celtic in origin. Not all the hitherto accepted Celtic origins for place names can be upheld, however. Vennemann (1994: 275) sees Isar, for instance, as Old European (his own, not Krahe’s, which he takes to be pre-Indo-European, agglutinative in structure and hence identifiable vis à vis the later subgroups of Indo-European) with Is-. This he sees as cognate with Basque stem iz- ‘water’.

In approximately the 6th century BC the Celts probably began a period of expansion. They moved in different directions. There are references to them in western and central Turkey, in the historical province of Galatia (Dressler 1967), best known for St.Paul’s epistle to the church there, and of course the Celts were in Italy and sacked Rome in 390 BC. Another thrust of the Celts was to the west and north. One section moved into the Iberian peninsula and is responsible for Celtiberian, recorded in a number of inscriptions. The group which moved north, north-west occupied the centre and north of France – historically Gaul – and then crossed the English channel to Britain.

Insular Celtic

By Insular Celtic one means that form of Celtic spoken on the British Isles. It falls into two groups, P-Celtic or Brythonic and Q-Celtic or Goedelic. The latter branch is confined to Ireland, the Isle of Man (now extinct) and Scotland where Irish immigrants moved in the early centuries AD bringing their language with them. Breton is a form of Brythonic which arose due to emigration to Brittany by speakers of Celtic in the south-west of Britain as a consequence of the Germanic invasions which set in in earnest as of the mid-fifth century AD (there may well have been survivals of Gaulish in Brittany as well, a view supported by Breton scholars like Léon Fleuriot).

P- and Q-Celtic

These designations derive from the treatment of original /kw/ in Brythonic and Goedelic (Schmidt 1993). In the latter the inherited sound is retained whereas in Brythonic and Gaulish /p/, a consonant originally lost in all Celtic languages, was regained by the shift of /kw/ to /p/ as in Old Welsh map (> Modern Welsh mab) and Irish mac /-k/ ‘son’; Modern Welsh penn /pɛn/ and Modern Irish ceann /kjɑ:n/ ‘head’. Irish later reintroduces the bilabial stop via Latin loans like pian ‘pain’ or póg ‘kiss’ ( < [osculum] pacis) and still later via Anglo-Norman loan-words like píosa ‘piece’ or pláta ‘plate’. Note that the closing of a labial element to a stop had a precedent in the shifting of IE gw to /b/ in Celtic, cf. Old Irish ben ‘woman’ < IE *gwena; compare Old Irish béo ‘alive’ and Latin vivos (Thurneysen 1946: 117).

Hamp (1958: 211) assumes that p and k were allophones of each other at an early stage. However, he offers little evidence for his view beyond his own assertion of it. It would have been useful if he had produced some present-day case or one clearly attested historically instance in which p and k are allophonic, i.e. non-distinctive variants of each other.

MacManus (1984: 186f.) sees the main body of early Latin loans (the so-called Cothrige words after an early form of the name Patricius) before 500 AD as showing the shift from p to k (puteus > cuithe ‘well’, planta > cland ‘children’, the latter probably via another variety of Celtic as it shares the meaning ‘children’ with Brythonic, O’Rahilly (1957)). After the mid 6th century the Pátraic loans enter the language without any shift of labial to velar place of articulation. There would appear to be a shift from f to s judging by words such as fenestra to senester, later replaced by the Scandinavian word vindauga (as in English) which resulted in present-day fuinneog ‘window’.

Following Sarauw (1900), MacManus (1984: 179) maintains that Latin p was replaced by kw in Irish as long as a labialized version of the velar stop existed in the language. By the Old Irish period kw had been simplified to k so that there was no native labial or labialised voiceless stop hence Latin p was retained in later loanwords. With those loans in the Primitive Old Irish period (before 500 AD) Latin t and k appear as /θ/ and /x/ respectively. Later loans have voiced stops instead of fricatives for dental and velar stops: paiter /-d-/ < pater ‘father’ and póc /-g/ < pacis ‘kiss’.

O’Rahilly (1957: 80f.) maintains that the naturalisation of *p was facilitated by the fact that in the 5th century a Hiberno-Brythonic type of dialect was still spoken in Ireland which, being Brythonic, would have had *p.

Present-day languages

The Celtic languages today comprise six languages with greater or lesser degrees of vitality. These fall into two main groups traditionally known as Brythonic or Brittonic (P-Celtic in type) and Goidelic from the Goídil, modern Gaels (Q-Celtic in type).

P-Celtic Q-Celtic
Welsh, Cornish, Breton Irish, Scottish Gaelic, Manx

The distinction between P- and Q-Celtic is based on the realisation of words with inherited IE /k(w)-/. In the Q-Celtic branch this was retained whereas in the P-Celtic branch it shifted to /p/.

Irish Welsh meaning
ceann pen ‘head’
mac mab ( < /map/) ‘son’
ceathair pedwar ‘four’ (IE *qetwar)

The distinction between the two main types of Celtic already existed on the continent. Celtiberian like Irish is Q-Celtic whereas Gaulish and Welsh are P-Celtic.

Mutation in the Celtic languages

One of the features of all Celtic languages is that the initial consonants of words change their form under certain grammatical conditions. The changes are (i) lenition, a weakening of articulation, specifically a shift of stop to fricative or of voiceless to voiced fricative. (ii) nasalisation, the prefixation of a nasal before vowels and voiced consonants or the voicing of voiceless ones. Here are a few examples from Modern Irish and Modern Welsh.

Irish cat ‘cat’ Welsh
a chat /ə xat/ ‘his cat’ ei bib /i bib/ ‘his pipe’ ( < pib )
a cat /ə kat/ ‘her cat’
a gcat /ə gat/ ‘their cat’
ei fara /i varə/ ‘his bread’ ( < bara )
ei chorff /i xorf/ ‘her body’ ( < corff )

In all the Celtic languages the reaction to the gradual decay of the inherited inflections of Indo-European was to functionalise the phonetic lenition present in each language (probably at around the 5th century AD in Britain and Ireland, i.e. in P- and Q-Celtic. Jackson, 1953: 561 assumes the second half of the fifth century).

What is curious here is that each language group adopted the same solution which, seen typologically, is not a very obvious reaction to inflectional attrition. Given this situation, one is justified in assuming that the seeds of lenition, the weakening of consonants, was already present in the continental forms of Celtic. Indeed authorities like Jackson (1953: 546) would seem to assume that in Continental Celtic there was a systemic distinction between geminate and simplex consonants and this developed into the opposition non-lenited # lenited later with the demise of distinctive length for consonants. The geminates occurred in absolute initial position (strong syllable onset) and internally where they derived from previous clusters, e.g. -mm- from -sm-. Where the cluster was still present, e.g. lt or χt, no lenition is later observable. Evidence is present for the fricativisation of labials on the continent, e.g. in that of /m/ to /v, w/. Tovar (1961: 79ff.) provides instances and calls this lenition (in the simple sense of a shift from stop to fricative). Furthermore, he would seem to subsume under this process the very early loss of *p which is the defining feature of early Celtic (cf. Irish athair, Latin pater).


Ball, Martin J. and James Fife 1993. The Celtic languages. London: Routledge.

Buck, Carl Darling 1988 [1949] A Dictionary of Selected Synonyms in the Principal Indo-European Languages. Chicago: University of Chicago Press.

Chadwick, Nora 1971. The Celts. Harmondsworth: Penguin.

Dillon, Myles and Nora Chadwick 1972. The Celtic realms. 2nd edition. London: Weidenfeld and Nicolson.

Dressler, Wolfgang 1967. ‘Galatisches’, in: Meid ed. 147-154.

Eska, Joseph F. and D. Ellis Evans 1993. ‘Continental Celtic’, Ball and Fife (eds), pp. 26-63.

Evans, D.Ellis (ed.) 1986. Proceedings of the 7th International Congress of Celtic Studies. Oxford: University Press.

Fowkes, R.A. 1940. ‘The phonology of Gaulish’, Language 16: 285-299.

Gamkrelidze, Th. and V. Ivanov 1973. ‘Sprachtypologie und die Rekonstruktion der gemeinindogermanischen Verschlüsse’, Phonetica 27: 150-156.

Gray, H. L. 1944. ‘Mutation in Gaulish’, Language 20: 223-30.

Greenberg, Joseph 1970. ‘Some generalizations concerning glottalic consonants, especially implosives’, International Journal of American Linguistics 36: 123-145.

Hamp, Eric 1951. ‘Morphophonemes of the Keltic mutations’, Language 27: 230-247.

Hamp, Eric 1958. ‘Consonant allophones in Proto-Keltic’, Lochlann 1: 209-217.

Hopper, Paul 1973. ‘Glottalized and murmured occlusives in Indo-European’, Glossa 7: 141-166.

Jackson, Kenneth 1953. Language and history in early Britain. Edinburgh: University Press.

Kortlandt, Federik 2007. Italo-Celtic Origins and Prehistoric Development of the Irish Language. Amsterdam: Rodopi.

Krahe, Hans 1954. Sprache und Vorzeit. Europäische Vorgeschichte nach dem Zeugnis der Sprache. Heidelberg: Quelle und Meyer.

Laing, Lloyd 1979.Celtic Britain. London: Routledge and Kegan Paul.

Lejeune, M. 1955 Celtiberica. Salamanca: University of Salamanca.

MacManus, Damian 1984. ‘The so-called Cothrige and Pátraic strata of Latin loan-words in early Irish’, in Ní Chatháin and Richter eds, pp. 179-196.

Mallory, J. P. and D. Q. Adams 2006. The Oxford Introduction to Proto-Indo-European and the Proto-Indo-European World. Oxford: University Press.

Meid, Wolfgang (ed.) 1967. Beiträge zur Indogermanistik und Keltologie Julius Pokorny zum 80. Geburtstag gewidmet. Innsbruck: University Press. Innsbrucker Beiträge zur Kulturwissenschaft 13.

Ní Chatháin, Próinséas and Michael Richter (eds) 1984. Irland und Europa. Stuttgart: Klett-Cotta.

O’Rahilly, Thomas F. 1932. Irish Dialects Past and Present. Dublin: Browne & Nolan.

O’Rahilly, Thomas F. 1935. ‘The Goidels and their predecessors’, Proceedings of the British Academy The Sir John Rhys Memorial Lecture, pp. 323-372.

O’Rahilly, Thomas F. 1971 [1957]. The two Patricks. Dublin: Dublin Institute for Advanced Studies.

Renfrew, Colin 1987. Archaeology and language. The Puzzle of Indo-European origins. London: Jonathan Cape.

Sarauw, Christoph 1900. Irske studier. Kopenhagen.

Schlette, Friedrich 1979. Kelten zwischen Alesia und Pergamon. Berlin: Urania Verlag.

Schmidt, Karl-Horst 1957. Die Komposition in gallischen Personennamen. Tübingen: Niemeyer.

Schmidt, Karl-Horst 1986. ‘The Celtic languages in their European context’, in: Evans ed. 199-221.

Schmidt, Karl-Horst 1993. ‘Insular Celtic: P- and Q-Celtic’, in Ball and Fife (eds) 64-98.

Sommerfelt, Alf 1962. Diachronic and synchronic aspects of language. (The Hague: Mouton).

Thurneysen, Rudolf 1946. A grammar of Old Irish. Dublin: Institute for Advanced Studies.

Tovar, Antonio 1961. The ancient languages of Spain and Portugal. New York: Vanni.

Vennemann, Theo 1994. ‘Linguistic reconstruction in the context of European prehistory’, Transactions of the Philological Society. 92.2: 215-284.