Origins of Welsh, influence on English, age of Arthur

I got a (to me) really fascinating and provocative linguistics book for Christmas (Peter Schrijver Language Contact and the Origins of the Germanic Languages) - ostensibly about the origins of the Germanic languages, but which actually had an awful lot to say about the influence of (pre-proto-Welsh) British Celtic on the early development of English, of Late Spoken Latin on the early development of Welsh, and on the relationship between British Celtic and very early Irish. It turns out that the author, Peter Schrijver, is actually a Celticist by background, so all this Celtic content should perhaps not have surprised me; but his conclusions about language change also have implications for how we think the change from Roman Britain to English and Welsh kingdoms happened (the age of Arthur and the origin of Welsh, more or less).

(Tl;dr on the Germanic languages: Schrijver concludes that Anglo-Saxon is basically North Sea Germanic with a strong Celtic accent; High German is West Germanic with a Romance accent; Dutch is West Germanic with a dash of North Sea and a strong Picard/Walloon accent; and Germanic as a whole is Indo-European with the accent of some unknown speakers who also strongly influenced Saami.)

Anyway: some of Schrijver’s conclusions support, and others conflict with various other ideas & arguments I’d come across recently about the end of the Empire and post-Roman Britain, and I thought I’d try and share some of them here with anyone who might be interested. (I’m looking at you, @siaronjames, but possibly not at anyone else. Warning: ridiculously long post.)

The first source I’m thinking of is Guy Halsall Barbarian Migrations and the Roman West, 376-568. This is a big, fairly serious historical book, which I read about a year ago, so not much of the detail has stuck with me – and, as the title implies, it’s about the picture acrosss the whole of the Western Roman Empire, not just Britain. However, the big points that have stuck are: the artefacts we see as defining someone’s material culture don’t necessarily tell us anything about their ethnicity or even their politics – culture could change within someone’s lifetime, depending on their circumstances – for example, burials with grave-goods become much more common in more unsettled periods, for some reason, so finding an inhumation with “Germanic” grave-goods doesn’t necessarily tell you that you’ve found a barbarian chief or queen. Leaders of tribes on the borders of the Empire won prestige for being able to distribute to their followers Roman goods - won by raiding, by trade, or by periods of service in the Roman army; as such, so long as Rome kept a close eye on its Western frontiers (e.g. when Trier in Germany was one of the capitals of the Empire), the culture and prestige of the tribes on its borders was actually very much dependent on the Empire, rather than in opposition to it. Cultural traits and bling went both ways, though: just as a Goth, after serving for years in the legions, might settle down somewhere in the Empire as a “Roman veteran”, so also some legions recruited so many Germanic soldiers that being, say, “Gothic”, and sporting “barbarian” versions of Roman bling (e.g. belt clasps and other metal gear) became so much a part of the culture that legionaries from elsewhere in the Empire would consider themselves “Gothic” while they were in the army (and go back to being Roman when they retired). There isn’t really any evidence for big, important dynasties of barbarian kings: what happened was more that administrative failures led to the Roman state failing to feed and pay its Gothic legions, at which point they very much had to make shift for themselves, and declared their generals to be kings: the barbarian states that emerged at the end of empire were very much a product of the late Empire, much more than something coming from outside of it.

This matches up with what I remember of Max Adams The First Kingdom: Britain in the age of Arthur. This is packaged as much more of a trade, popular history paperback than the Guy Halsall, but I have to say that I found it unfortunately hard to get through, and although I read it after the Guy Halsall, again, not much detail has stuck. However, my main memories of it are: there is no archaeological evidence for some great “Anglo-Saxon Conquest” at a specific 5th-century date, with the migration of large numbers of individuals and complete chaos and social breakdown in post-Roman Britain. For a start, some of the culturally “Saxon” material is earlier – presumably from Roman troops of Germanic origin (or culture) stationed in Britain; it seems likely that such people would have regarded themselves (and been regarded) as just another part of the fabric of Roman Britain, rather than as invading heroes or barbarians (take your pick). The archeological evidence that we have is basically very, very piecemeal – in some areas there was a clear continuation of Roman-style culture (Chedworth Roman villa in Gloucestershire had a new mosaic floor laid in the 5th century, which remained in use long enough for the middle of the room to show more wear and tear than the edges - Stunning Mosaic Found in England Shows Some Lived in Luxury During 'Dark Ages' | Smart News| Smithsonian Magazine) and high-status imports (e.g. at Tintagel), while other areas seem to have moved over to a more “Saxon” material culture (but with no evidence that that transition was particularly militaristic or traumatic); in still other areas, there may have been more dislocation and conflict. (In terms of cultural crossover, we mustn’t forget that the West Saxon royal line was traced back to someone called Cerdic, which looks suspiciously like an Anglo-Saxon version of Caredog or caredig, and that the first Old English poem quoted by Bede was supposedly composed by someone called Cædmon, which looks like a version of the name Cadfan.)

This is very much backed up by something quoted by John Koch in his book The Gododdin of Aneirin : text and context from Dark-Age North Britain (uploaded by the author and freely available online at The Gododdin of Aneirin : text and context from Dark-Age North Britain : Koch, John T : Free Download, Borrow, and Streaming : Internet Archive). Talking about Yeavering, in Northumbria, which is thought to have been a royal seat of the Anglo-Saxon kingdom of Bernicia, he says:
“Brian Hope-Taylor, Yeavering’s excavator, has offered a new view of Anglo-Celtic relations in the North. . . at Yeavering, Anglo-Saxon authority is seen from the very outset to have respected native tradition, and the circumstance seems to preclude any fundamental disturbance of the native population. The British who survived, were so trustworthy that from the start the alien power could set up its halls on open ground in their midst, and were fundamentally contributive to the development of a hybridizing culture. Thus where actual or latent hostility would be expected to have continued longest and left its strongest mark, there is not so much as a hint of any discord. The local picture, at this point of the Central Zone of Bernicia, is of a harmonious relationship between the native population and a minute, governing Anglo-Saxon élite, itself susceptible and responsive to formative influences from its British environment. All the evidences would be consistent with the proposition that what is at issue was an English overlordship which, from a very early stage, had been found mutually convenient and congenial.”

John Koch is a senior research fellow at the Centre for Advanced Welsh and Celtic Studies, University of Wales, with a very distinguished career; he is currently working on a grammar of Old Welsh and, although some of his conclusions are not universally accepted (for example, he is a leading proponent of the theory of “Celtic from the West”), he is clearly a hugely knowledgable and well-respected scholar. What he tries to do in his book on the Gododdin is to give us the historical context for the poem, as well as a reconstructed text of what the oldest parts of the poem might have looked like when it was first composed and (he argues) written down, in the 6th century. (The Gododdin is, for those who don’t know it, amongst other things, the source of our earliest reference in Welsh to King Arthur.) Linguistically, he has probably been successful: I read a response by another scholar who disagreed with some of his historical conclusions, who nonetheless was happy to grant that, linguistically, we were in safe hands with Koch (The Heroic Age: The Forum/GODODIN Revisited). As such, I’m afraid that only the occasional word jumps out at you after 1500 years: most of the 6th-century Welsh is not merely incomprehensible, but even unrecognisable; fortunately, he also provides an English translation. (There is a version of the original text but in modern Welsh spelling, published by Gwasg Gomer, if you’re interested - However, it his interpretation of the historical context which I found fascinating: far from being the heroic defeat of the Men of the (Old) North in a heroic battle against the invading Saxons, as it is usually seen, he sees it as a conflict between two largely Celtic kingdoms, dating from before Bernicia was even Saxon: at this period, he sees the picture as a much more mixed one, in which it was perfectly possible in different contexts that Britons could have ruled Anglo-Saxons and led them into battle, as well as (by implication) vice-versa. “The implicit assumption is that the struggles recollected in the poetry attributed to Neirin and Taliessin must have involved monoethnic forces with nationalist objectives… In fact, we have no evidence (historical or poetic) that Urbgen’s war against the Bernician Angle Theodoric was anything more than that. And the latter was far from being in any sense ‘the king of the English’. The concept itself would have been precocious in the 570s.” And in fact, one of the few Anglo-Saxon names to be found in the poem appears in connection with one of the leaders of the Gododdin forces: Yrfai, lord of Eidyn – son of Wulfstan. In other words, at this point in time, there are small states with competing interests coming onto conflict with each other, but there simply is no ethnic narrative of heroic conquest and valiant defence – that’s something that gets read into it later, after the days of Penda of Mercia’s 7th-century alliance with Powys, Gwynedd, and Strathclyde, when everything finally began to be seen in terms of Welsh vs English. (I am reminded of the claim by the late duchess of Medina Sidonia that her forebear, Guzmán el Bueno, hero of the Spanish reconquista, may actually have been Muslim. All the most clear-cut battle lines seem to get drawn up centuries after the fact.)

So after all this historical and literary reading I finally got onto the Peter Schrijver, and very interesting indeed it turned out to be. Schrijver’s interest is in how contact between different languages transforms one or both of them. He starts from the point that, in general, borrowing (of vocabulary) and substrate influence (influence of speaking something as a second language on, for example, pronunciation and syntax) tend to go in different directions. That is to say, if I am an Englishman living in Wales, I might pepper my speech with references to, say, the Eisteddfod, where that is a technical term I have knowingly borrowed into my first language for a specific concept, from a source which is, at best, a second language for me; if, on the other hand, I struggle to distinguish between pan and pryd, or persist in pronouncing Abersoch as “Abba-sock”, the influence is unconscious, and from my first language onto my second: borrowings tend to go from L2 → L1, because L2 has more prestige in a particular context; substrate influence tends to go L1 → L2, because of imperfect control of the target, second language. Additionally, if there is a large group of people who all share the same substrate influence, it may come to be seen as just a different variety of that language, rather than a bad or faulty variety – think, for example, of people speaking English with any given Welsh accent: it may be that their ancestors had a poor command of English, but their present-day descendants may be wholly bilingual, or even di-gymraeg; the accent is now, simply, an accepted areal way of speaking English, and definitely not an imperfect copy of an English accent.

So Schrijver is trying to look for clear evidence of how the languages spoken in Britain at the time of the “Anglo-Saxon Conquest” affected Old English; and in order to do so, he goes into quite some depth about the probable 4th-5th century pronunciation of Late Spoken Latin in Britain; of Lowland British Celtic (mostly influenced English); of Highland British Celtic (forerunner of Welsh, Cumbric, Cornish, and Breton); and, surprisingly, of Old Irish. Again, Peter Schrijver is clearly a knowledgable scholar, and is actually a Celticist not a Germanicist by background: he is professor of Celtic Languages at Utrecht University, and has had work on Old Irish published by Maynooth University (home of David Stifter, a key figure in Old Irish, and author of the easiest - for certain values of “easiest” - introductory textbook on it); but again, some of his conclusions are surprising to me and probably a wee bit controversial. Frankly, I don’t have the technical knowledge to critically assess them myself, and I’d love to see a take-down of them by someone with the relevant skill-set to disagree with him in an informed way, so that I could then maybe try to make some judgment as to their relative merits; but, even so, they’re intriguing.

Essentially, what he says is that many of the fairly dramatic sound-changes that mark the transition from Common British Celtic to the earliest Welsh (softenings of consonants, loss of Latin case-endings) are very similar to changes happening in the earliest stages of the Romance languages in general, and French in particular – think of something like French chèvre from Latin capra, alongside Welsh gafr from *gabros. He sees this as substrate inluence (i.e. from people changing language), separate from and later than all the borrowings of Latin vocabulary into what ultimately became Welsh. In other words, it would imply a large influx of people speaking Late Spoken Latin, or Celtic with a strong Latin accent, into the areas that later gave rise to Welsh and Cornish – presumably educated, middle-to-upper class Romano Brits, fleeing as refugees from areas where the advent of the Anglo-Saxons was leading to turmoil, having to switch to Celtic in order to integrate into a more Celtic-speaking area. (“The Latinization of the British Celtic sound-system but not of the lexicon strongly indicates that it resulted from low-status speakers of Latin rapidly shifting to high-status Celtic and in the process retaining a Latin accent but avoiding the use of Latin words.This is a reversal of the situation in previous centuries, when Celtic speakers shifted to Latin.”)

However, he sees no real evidence for influence from Highland British Celtic (proto-Welsh and ur-Cornish) on Anglo-Saxon: for instance, their vowel-systems don’t show any of the features that made Old English so different from its closest Germanic relatives. Instead, he thinks that the influence came from Lowland British Celtic, or from people speaking Latin with a strong Lowland Celtic accent – presumably a less well-educated, more working-class cohort, who were absorbed culturally and politically into the “Anglo-Saxon” population. But to get some idea of what the featuires of this poorly-evidenced Lowland British accent might be, he winds up taking a surprising excursus into the origins of Irish.

Now, it is a fact that there is actually no consensus on the origins of Irish. It is clear that it is a Celtic language, and so related to Welsh and Gaulish and Lepontic and Celtiberian and so forth. It is also a fact that it is spoken in “these islands” (avoiding the phrase “the British Isles”, which sounds to many Irish too much like Britain/England still trying to lay claim to the island of Ireland); and it is true that when British Celtic changed inherited kw sounds to p (Welsh pwy, pan, pump vs Latin quis, quando, quinque), Irish kept kw before eventually dropping the w (Old Irish cía, cuin, cóic). So the place of Irish within the Celtic family tree is unclear: some scholars lump it together with British as “Insular Celtic”, despite the kw/p difference, while others see the consonant change as fundamental, and divide Celtic into Q-Celtic and P-Celtic, which they see as arriving in these islands in separate waves of migration; and all of this family tree business has an effect on and is affected by the question of when Irish arrived here.

As to that, in another interesting paper (‘Ériu, Alba, Letha: When Was a Language Ancestral to Gaelic First Spoken in Ireland?’ (PDF) ‘Ériu, Alba, Letha: When Was a Language Ances¬tral to Gaelic First Spoken in Ireland?’, Emania ix (1991 [‘Focus on the Origins of the Irish’]) 17–27. | John Koch -, John Koch sets out the competing positions before giving us his own take on it: essentially, the earliest possible date is c.4500 BCE, even before the Beaker People; while “at the low extreme, the Association of Professional Irish Archaeologists […] failed to rule out any horizon before the fifth century A.D.,” which gives us a fully 5000-year window to wrestle with. John Koch himself plumps for around 1000BCE, but Schrijver argues from the lack of different dialects in written Old Irish (unlike the earliest English, say) that Irish must have been only comparatively recently introduced to Ireland. In fact, given that Irish too underwent a huge transformation in its phonology in the years 500-600CE, he argues that it was possibly introduced to the island of Ireland no earlier than about the first century of our era; that in the first century it would have been broadly identical to British Celtic, apart from the whole p/kw thing; and that it expanded and transformed in Ireland due to large numbers of speakers of an unknown pre-Irish (probably non-Indo-European) language changing over to speaking Irish (and transforming it in the process). All of this is a bit mind-boggling, and maybe requires a bit of unpacking. The first claim, that Irish reached Ireland only in the first century, seems hugely surprising; but what he is claiming on linguistic grounds is congruent with the fact that the archaeologists wouldn’t rule out anything right up until St Patrick’s era. (Schrijver suggests that the language could have been brought there by Britons fleeing the Roman Conquest, and cites links between archaeological finds on Lambay Island, near Dublin, and the British kingdom of Brigantia in what later became Northumbria.) Secondly, the idea that P-Celtic vs Q-Celtic isn’t a major branch in the family tree, and wouldn’t hamper mutual comprehensibility, seems a big claim: but in Italic languages there were P-Italic as well as Q-Italic languages (Oscan pis, pid for Latin quis, quid); in Polynesian languages, Hawai’ian has a k where most other related languages have t - so the Hawai’ian name for Tahiti is Kahiki, and a Hawai’ian kahuna corresponds to a Maori tohunga - except on the island of Ni’ihau, where they speak a dialect of Hawai’ian with a t (and swap to k when they’re talking to people from other islands, or reading from the Hawai’ian translation of the Bible). Even in English, if someone says “but” with a Cockney accent, we don’t hear the glottal stop at the back of the mouth and think they’re saying “buck” or “bug” - we just know that it’s a different way of saying “but”. Presumably, if the P-Celtic/Q-Celtic distinction was that minor back in the day, it was probably a cause for taking the mickey out of someone’s accent, rather than a cause of complete incomprehension. Certainly, when Patrick’s name (Patricias) was first borrowed into Irish in the fifth century, the P got turned into a Q as *Kwatrikias (later ending up as Cothraige - but compare the Hawai’ian version of “David”, Kawika); it got borrowed from British (or from British Latin) again in the sixth century (by which time Irish had got used to Latin p sounds) as Pádraig, which gives an idea of just how much Irish had changed in only a century or so. Finally, the claim that some other language was spoken in Ireland before Irish is not, in itself, revolutionary (see Goidelic substrate hypothesis - Wikipedia); however, Schrijver draws attention to the fact that some of the possible substrate loan-words in Irish begin with p (including peta, possibly the ultimate source of English “pet” = anifail anwes). Given the fact that the word is peta and not **ceta, and comparing it with Cothraige and Pádraig, he concludes that the unknown substrate language must have been spoken in Ireland right up until at least about the year 500CE, and that it was its speakers that shifted to and so transformed Irish in that period.

Anyway: after all that, he feels he has made a case for comparing the Primitive/Old Irish sound-system (as a proxy for Lowland British Celtic) with that of Old English; and he definitely makes a case for there being some striking resemblances on just those points where Old English differs significantly from its nearest Germanic relatives.

Putting all this together, it’s a little bit difficult to get a single, clear narrative. The historians and archaeologists seem pretty clear on the idea that the “Anglo-Saxon Conquest” was more likely a very piecemeal affair, that may have largely consisted of native Britons coming to culturally identify with a numerically fairly small Anglo-Saxon elite; and we can see this in the same context as the Roman “Gothic” regiments becoming actual Gothic kingdoms - possibly speculating that an Anglo-Saxon hegemony might arise in a context where they were, essentially, professional soldiers, here in Britain, at a time when the military leadership of professional soldiers was an important thing. Names like Cædmon and Cerdic seem consistent with this picture of culture-shift rather than conquest; and Ylfai son of Wulfstan, Lord of Eidyn, suggests that in the early period culture-shift was more of a two way street. Only later on do we get the heroic ethno-nationalist narratives of “our forefathers arrived here in only three ships, led by two brothers called, um, ‘Horse’ and, er, ‘Stallion’, m’kay?” versus “the great tyrant betrayed us to the scroungers of Thanet.” But what seems difficult to integrate with this picture is Schrijver’s claim that the upper-class Romano-British refugees were numerous enough that their weirdo Latinate accent so transformed Highland British Celtic: I can understand the wealthy and the magistrates finding themselves rather dispossessed in areas that had decided “barbarian” professional soldiery was the way to go, but for those refugees to have been sufficiently numerous (despite their newly lowered social status) as to completely transform the language seems rather to imply much more significant dislocation and chaos in Lowland Britain than the archaeologists’ and historians’ picture would imply.

Of course, none if this is remotely relevant to things like how we support and promote Welsh in the twenty-first century, or even to how we got to the present situation with things like the scandal of the Blue Books; but I did find it all quite interesting, nonetheless – and felt moved to share (at length!).


…but fascinating! :nerd_face:


Agree, Siaron! Very much a “thank you for reading all of that so I don’t have to” moment. Really interesting, @RichardBuck - thank you!

1 Like

Absolutely fascinating! There is sooo much I want to read about and so little time to do it! :joy:
Diolch yn fawr for the overview!

1 Like

Thanks for this post. I will have to make some time to read it all later. I recently came across a paper by Schrijver talking about Celtic influence on English. But it is on JSTOR which is really only available to Universities so, I can’t access it. So, I will look into this book. :+1:

I think you can have a limited access to JSTOR - I’ve used it for a couple of things (because like you I really missed being able to access all these things after finishing my postgraduate degree!) See: JSTOR: Registration (I put “independent” for institution and “independent researcher” for role, and it seemed perfectly happy with that…)

Also, you might be interested in, which has a load of pdfs available. I’ve found it a little bit spammy at times (always pushing their bulk-buy option, and emailing me with suggestions all the time) but I always click on “no thanks”, and have a filter set up to send their emails through to a separate folder so that I can scan them for any interesting titbits in my own time.


I hadn’t realised that about JSTOR! One other thing is that if you see a pdf of an article anywhere that you can’t access, I have often seen academics online say that they’re perfectly happy to email copies to interested individuals. So if you can’t read it, but can find their institutional email address, that can be worth a try.


Thanks for that info Sara. I’ve heard nefarious things about They sell your info and track you etc. As bad as Google.


Thanks for raising that - important for people to know. (I access with my work address and work IP so I’m not too worried personally, but glad now I didn’t join with my personal details!)