I was going to write a post about UKIP’s rise in support in England, but it’s difficult to get that interested. Following English politics is getting increasingly like following French politics or Greek politics – it’s all very interesting if you’re into that sort of thing, but few people in Scotland are. So I’m going to waffle on about something else that few people are interested in, at least outside of Scotland, the origins of Gaelic. I’ll get back to slagging off politicians the morra. Meanwhile – time for a bit of culture.
The other day I was asked about a recent archaelogical theory that the traditional account that Gaelic sailed over from Ireland with Fergus Mòr MacErc wasn’t true, and that Scottish Gaelic evolved out of the Celtic dialects spoken in Argyll and Kintyre and was never imported from Ireland. The truth is more complicated than either the theories of archaelogists (seriously – don’t ask an archaeologist about language … the only people with a worse understanding are geneticists) or the old stories of conquering Gaelic kings. But first a wee bit of background about Celtic languages is necessary.
There is a widespread belief that a community will only change its language if a new language is imposed upon it. This belief is based in our recent historical experience of language replacement, where state institutions and agencies actively discouraged Celtic speakers from using their languages, and in European colonies native languages were displaced by colonial settlement. In modern societies, there are institutions which reach right into the domestic sphere, state education was often used as a tool to spread one language and discourage another.
But one language replaced another in ancient tribal societies too, and since it was believed that a community could only replace its everyday language with a different language if there was some means of imposing the new language, then language replacement in ancient times was thought of in terms of large scale population movements. In the absence of state institutions like an educational system, the theory went that there needed to be a large body of settlers to enforce their language on a conquered population. But this is not true, and is even less true where the languages in question are all very closely related to one another, as Gaelic, Pictish and Brittonic were.
In fact language replacement can occur without any population movement at all, or more commonly with population shifts affecting only a tiny number of elite individuals – such a small number that it would not show up in DNA tests carried out in the descendant population many generations later, and would not show up in archaeological evidence as a foreign invasion.
The other issue which is important when discussing the origins of Gaelic is that the older theory that there was an ancient division of Celtic into the Q-Celtic Goidelic and the P-Celtic Brittonic/Gaulish is not supported by the linguistic evidence. When you strip out all the linguistic changes known to have occurred in Insular Celtic which are contemporary with or later than the Roman occupation of southern Britain, there’s really very little left to distinguish Iron Age Goidelic from Iron Age Brittonic. The changes which occurred from Roman times onwards were massive and profound, and led to the two language groups becoming very different from one another, but in Iron Age times the great majority of those differences didn’t exist and the Celtic dialects of the British Isles – and right across much of Europe – were very similar to one another.
So here’s my own view of the origins of Gaelic, but bear in mind that I’m merely an armchair linguist, not a proper academic. What follows is the linguistic version of “I’m not a doctor but I’ve read some medical books.”
It’s better to think of Celtic language during the late Iron Age as a dialect complex without any sharp boundaries between adjacent dialects. Instead the Celtic dialect of one district merged imperceptibly into the dialect of the neighbouring districts, and on the eve of the Roman invasion of Britain the dialects spoken in the west of Ireland and northern Scotland were linked to the Celtic dialects of southern England via a series of intermediate dialects.
There would have been dialects displaying features associated with Goidelic, others displaying Brittonic features, and no doubt many others which had both (or neither) in varying proportions. In fact, given the way that sound changes typically progress throughout a language community, it’s likely that there were once Celtic dialects which were simultaneously “Q-Celtic” and “P-Celtic”. There is in fact some evidence for this from Gaul. Gaulish was a “P-Celtic” language, but in some inscriptions there are a few words and names where the original Q is preserved. The Gaulish name for the river Seine was Sequani, not “Sepani” as you might expect from a P-Celtic language like Gaulish.
Here’s a wee explanation as the idea of a Celtic dialect that minded both its Ps and Qs is a bit unusual for modern people.
In ancient Celtic there was a phoneme traditionally transcribed Q, it was pronounced similar to English qu in queen, which is phonetically /kwin/. However in English, the w follows the k, whereas in ancient Celtic the k and w were pronounced simultaneously. Technically this Celtic sound is called a labialised velar stop, and in many languages which have it, it tends to shift to p, a sound which takes less energy, effort and muscular coordination to produce. This is especially likely to happen in languages which don’t otherwise have a P consonant, like early Celtic.
Brittonic is characterised by this shift of older q > p – as for example in the number five, which was something like *qenqe in Proto-Celtic. In Brittonic q shifted to p, giving *penpe, and later changes produced the modern Welsh pump. *qennos ‘head’ likewise became penn in Brittonic. (I’ve simplified the examples for ease of discussion.) It’s this characteristic which leads to Brittonic and its daughters Welsh, Cumbric, Cornish and Breton, being classified as “P-Celtic”. Gaulish, Galatian, and a number of other Celtic varieties of ancient Europe were likewise “P-Celtic”.
However it’s likely that when the sound change first occurred, it only occurred in certain phonetic environments, this is usually the case with sound changes. So it’s possible that there could have been Celtic varieties where for example q shifted to p only at the ends of syllables or internally, but q was preserved initially. In this dialect then *qenqe would appear as *qempe. Alternatively p could have replaced q in some words but not all, as seems to have happened in Gaulish. This could happen because there was no other p in Celtic languages, so whether you pronounced q as kw or as p made no real difference. It was simply a case of having a different accent, it would not affect intelligibility.
Meanwhile in “Goidelic” dialects, the sound change didn’t happen at all. In these Celtic varieties, Q was preserved. Q was also preserved in the Celtiberian language of Iberia, but this does not mean that Goidelic and Celtiberian have a particularly close relationship within Celtic. All it means is that in both varieties, the sound change failed to occur.
In a similar way, ancient Germanic had a sound th, preserved in English, Scots and Icelandic (where it is written þ). In Dutch and German, this sound has been replaced by d, whereas in Continental Scandinavian it has been replaced by t. So we have the series English thing, Scots thing, Icelandic þing, German Ding, Dutch ding, Norwegian Danish and Swedish ting, and we could speak of TH-Germanic, D-Germanic and T-Germanic. But this does not mean English and Scots are most closely related to Icelandic within Germanic. Icelandic is more closely related to Norwegian and Danish. You require positive evidence of common innovations in order to demonstrate that two languages are more closely related to each other, the simple failure of a sound change to occur is not evidence of a close relationship. There are no common innovations shared by Goidelic and Celtiberian, so they are not particularly closely related to one another within the Celtic family.
However Goidelic varieties do possess distinctive sound changes of their own. The clusters nt, nk and nq shifted to d, g and gw in Goidelic. (Later q and gw simplified to k and g.) In these dialects *qenqe shifted to *qegwe, and via later sound changes ended up as coig. The same change can be seen by comparing the Gaelic word deud ‘tooth’ with its Welsh equivalent dant, in ancient Celtic the word was something like *dent, the nt has become d in Gaelic but is preserved in Welsh. This typically Goidelic sound change is absent from Celtiberian, which preserved original nt, nk and nq.
It is theoretically possible there were Celtic dialects which had both sound changes. In these, *qenqe would end up as *pebe, although no evidence of any such Celtic dialect has survived. There were certainly dialects where neither change had occurred – like Celtiberian. In fact there could have been dialects with any permutation of the various early sound changes which would later characterise Goidelic and Brittonic. These dialects were not P-Celtic or Q-Celtic in the sense we understand the terms, but were transitional between P-Celtic and Q-Celtic.
One of the most striking things about the earliest Goidelic – in the form of Ogam inscriptions and the earliest attestations in manuscripts – is that it displays no significant dialectal variation yet these inscriptions and glosses predate the establishment of the highly prestigious Old Irish literary language. This can only really be explained if Goidelic had spread at a relatively recent date from an original centre – which was almost certainly somewhere in Ireland, the centre of Goidelic culture.
The spread of Goidelic may be associated with the cultural and political upheavals which took place in Ireland as the country responded to the Roman occupation of southern Britain and the estrangement this created between Celtic tribes on either side of the Irish sea. It did not spread by population movements or conquests, but rather by speakers of transitional dialects adopting prestigious features associated with Goidelic. Celtic dialects in Ireland which had “P-Celtic” features lost these features, as their speakers dropped them in favour of “Q-Celtic” features. The transitional dialects then became thoroughly Goidelicised.
Meanwhile a prestigious form of Pictish and a prestigious form of Romanised Brittonic were also spreading in the same way, and likewise absorbing the transitional dialects adjacent to them. Eventually the increasing spread of these three varieties caused them to meet up, instead of a series of dialects merging imperceptibly, there were now three sharply distinguished Celtic languages.
I suspect that people in Argyll may have originally spoken one of these transitional varieties of Celtic. The local Celtic dialect contained some features associated with Brittonic varieties spoken further south, and with Pictish varieties spoken to the east, and also with Goidelic varieties spoken in Ireland, and quite possibly features unique to itself. As Argyll came firmly into the orbit of Irish cultural influence due to the political and cultural upheavals associated with the Roman occupation of southern Britain, Celtic speakers in Argyll would have come to regard themselves as Gaels and participated in the formation of a Gaelic social identity. As this occurred, local dialect speakers looked increasingly to Ireland for their cultural and linguistic models, and would have adopted features of pronunciation, grammar and vocabulary associated with Goidelic and dropped those associated with Brittonic or Pictish.
Argyll Celtic speakers would have stigmatised dialect features which they associated with Pictish or Brittonic, and came to regard Goidelic dialect features as prestigious models to be copied. Over the course of several generations the local dialect approximated ever more closely to Goidelic. By the time that written records appear it was a fully fledged Gaelic variety. Gaelic did come from Ireland originally, but it didn’t require a full scale invasion from Ireland into Argyll in order to Gaelicise the local variety of Celtic. By the time Fergus sailed over with his 40 ships, the local people were probably already Gaelic speaking.
Intriguingly, the formation of a Gaelic social identity seems to have occurred relatively late during the period of Roman occupation of southern Britain. The Word Gàidheal is actually a loan word from Brittonic, and moreover one which displays a relatively late Brittonic sound change. Etymologically the word comes from Brittonic goedel and originally meant something like “barbarian, savage”. Goedel is literally, “one from the forests”. It’s based on the Brittonic form of the Welsh word for trees gwydd, which has a native Gaelic equivalent in fiodh. Both these words descend from an ancient Celtic word something like *widos, but in later Goidelic the w turned into f, whereas in Brittonic it shifted to g. It’s this g form we seen in the name Gàidheal, so it can only have been borrowed into Gaelic after this sound change had taken place.
And here’s where it gets interesting, because the same sound change also occurred in Latin words starting with v (which was pronounced w in Classical Latin) which were borrowed into Brittonic during the Roman occupation. So the sound change can only have happened after these Latin words had been borrowed into Brittonic, and therefore it must date to the late part of the Roman period at the earliest. The formation of a Gaelic social identity probably dates to the same period.
The Roman occupation estranged the Celtic speakers of Ireland and Scotland politically and culturally from those in Roman Britain. Slave raiding was commonplace, and linguistic features which were associated with one community or the other would have been either stigmatised or regarded as prestigious. In Ireland, “P-Celtic” features would come to be associated with low status Brittonic slaves – like St Patrick. In Roman Britain the opposite occurred.
Under such social conditions, speakers tend to emphasise the distinctions between their speech varieties. Goidelic became the prestige variety of Celtic in Ireland and in districts where Irish cultural influence was strong because it contained the maximal differentiation from features associated with the speech of the Romano Britons. The new Gaels were concerned not to sound like Britons – and vice versa. And neither wanted to sound like a Pict.
Throughout the known history of Pictland, cultural and political influence from Ireland was extremely strong, and this led to the eventual replacement of Pictish by Gaelic throughout almost all of the previous range of Pictish. Pictish doubtless displayed strong Gaelic influence from the very beginning, but presumably at the time of the Roman invasion Celtic varieties in Pictland had already experienced the change of q > p. This led to Pictish becoming distinct from other Celtic varieties, but since the surviving evidence for Pictish is so meagre, it’s almost impossible to discern what its exact characteristics were. All that can be said for certain is that it was extremely close to Brittonic, and like Brittonic displayed p for original q.
So Scottish Gaelic is both an indigenous development of Celtic in Argyll, and also represents the effect of Irish cultural influence on Celtic speakers in the West of Scotland and the settlement of small numbers of high status people. There were certainly movements of groups of people in both directions across the Irish channel, but there is no need to insist that Gaelic can only have become established in Scotland thanks to a full scale invasion of Irish Gaels.