Geoffrey Sampson


[LOGO]

Stephen Oppenheimer and the Language Prehistory of Britain

My page on the birth of the English language gives the consensus account of how English came to exist as a language separate from the other European languages to which it is related. Barring minor details, this account has been uncontroversial for a very long time.

Now, though, Stephen Oppenheimer’s book The Origins of the British (2006) has made a splash by arguing in effect that the whole story is radically wrong.

Oppenheimer’s book is about more than language origins, and I have no qualifications to assess some of the other things he writes about. His claims about the beginning of English are (to my mind) wildly mistaken. But a lot of people are reading Oppenheimer, interviewing him for television, and so forth. So, rather than ignoring his book, for the benefit of visitors to my site who are interested in this kind of controversy, I shall explain as briefly as I can what is wrong with Oppenheimer’s argument.

Overall, Oppenheimer is making the following claims about British prehistory:

  1. If we forget about the very recent (post-Second-World-War) waves of immigration, then wherever we look in the British Isles, most of the ancestral bloodlines of present-day inhabitants go back to people who were already here in the Neolithic period – say, six thousand years ago. The well-known Iron Age and later “invasions”, such as the coming of the Anglo-Saxons, were more like the Norman Conquest – smallish élite groups arrived who sometimes had large cultural impacts, but never amounted to more than a tiny percentage of the subsequent bloodlines in any region.

  2. The genetic division between what we think of as the “Celtic” west and north of the British Isles and the “English” south-east itself dates back to the Neolithic – it is not the result of late-comers expelling or killing off inhabitants in one part of the territory.

  3. The Celts originated not in Central Europe as standardly believed, but in the Spain–Southern France region.

  4. When the Celts came to the British Isles, they occupied only the traditionally Celtic western and northern areas; England was never inhabited by Celtic-speakers.

  5. The inhabitants of England spoke a Germanic language long before the Romans arrived, and it was this language which evolved in due course into English – the invasions from the Continent at the end of the Roman period did not have much impact on the local language, except for introducing some Scandinavian influence.

The first three of these claims are based on statistical analyses of DNA data, which is outside my professional expertise. It is not Oppenheimer’s professional expertise either (he was a career paediatrician, and has turned to this stuff on a hobby basis); but he has evidently gone into genetics much more deeply than I have ever tried to do.

(It seems a pity that he does not do more to explain the basis of this part of his work to readers like me. He has many maps showing gradients of frequency of particular genetic markers across Europe – an easy concept to grasp – on which he superimposes directional arrows, giving dates for the respective genetic spreads. But he never tells us how the directions or the dates are worked out. The arrows do not always go from areas of high concentration of a marker to areas of lower incidence; sometimes they go the other way.)

I begin to feel quite sceptical when Oppenheimer comes to claim (4), which contradicts the standard belief that by the Roman period the whole of Britain was Celtic-speaking. Point (4), if true, would explain the strange lack of Celtic vocabulary borrowed into English. But England does have Celtic place-names – not a huge number, but why are there any if England was not once inhabited by Celtic speakers? The best-known, but far from the only, examples are the various English rivers called Avon, which is obviously the Celtic word for “river” – afon in modern Welsh. (In Welsh the letter F is pronounced as a V sound.)

It isn’t just place-names, either. For instance, the chieftain who foolishly invited Hengist and Horsa to settle in Kent, in the far south-east corner of the island, had a name – Vortigern – which makes sense in Celtic: it is modern Welsh Gwr-theyrn, “overlord”. The later kings of what is now the Welsh county of Radnorshire believed that they were his descendants. Odd, if he actually had no relationship with the Welsh. (Particularly since, after what he did, Vortigern was not exactly a figure with whom Celts would be proud to invent a fictitious connexion.)

In fact Oppenheimer seems to have been a victim of every scholar’s nightmare: a work of scholarship that refutes his ideas was published just as he sent his own manuscript in to his publisher. Sportingly, Oppenheimer prints a map derived from Patrick Sims-Williams’s Ancient Celtic Place-Names in Europe and Asia Minor (2006), which examines place-names that are known to have been current in Roman times, and shows how the proportions of them which derive from Celtic languages vary across Europe. Among modern place-names, obviously the Celtic proportion is massively higher in Celtic areas such as Wales than in England. In the Roman period, surprisingly, it was the other way round. Within Britain, one of the two main concentrations of Celtic-derived names was the area between London and the Wash – squarely within the English heartland. The proportion in what is now Wales was lower than in almost any part of England.

Of course, if Oppenheimer’s point (4) is wrong, then his point (2) is suspect too. It would imply that when English replaced Celtic speech in “England”, English spread just through the territory which had been settled from the east in the Neolithic age, thousands of years earlier, and stopped when it got to the border of the territory which had been settled from the west. But why would a language spread be affected by settlement patterns of millennia in the past? The conventional idea is that the Germanic tribesmen spread from the east in the Dark Ages, bearing their language with them, and stopped when they reached the agriculturally less valuable uplands of the west and north. The same event accounted for both language difference and genetic boundary; no coincidence required.

When we come to Oppenheimer’s claim (5), that English split off from the other Germanic languages before they divided into North Germanic (Scandinavian), West Germanic, and the now-extinct East Germanic branch: here we really are in la-la land, I fear. Linguists have well-worked-out methods (with which Oppenheimer does not pretend to be familiar) for establishing the shape of “family trees” of related languages, where there are adequate data to apply the methods; for the Germanic languages the data are rich. To quote Sally Thomason’s online response to Oppenheimer’s book, “the linkage of English with West Germanic – through its closest relation, Frisian, and then the also closely-related Dutch and Low German – is absolutely solid.” Oppenheimer’s idea that English might have separated from other Germanic languages before even Norwegian and Dutch had diverged really is not seriously tenable.

Part of the problem is that when Oppenheimer quotes research on linguistic analysis, instead of using the work of people knowledgeable about language he depends heavily on ill-informed studies of language by geneticists. Oppenheimer repeated cites as authoritative a 2003 paper by Peter Forster and Alfred Toth, which to a professional linguist reads as an embarrassment rather than a serious scientific contribution. Rather than justify this statement here, let me simply refer to an online critique by my late, much-missed colleague Larry Trask. (See also his follow-up.) Larry summarizes his view by saying that Forster and Toth “have committed every schoolboy howler I can think of”. Strong words, but justified in my view.

Oppenheimer is impatient with standard methods in linguistics because they do not yield dates for events in language history, in the way that archaeologists’ radiocarbon technique dates organic physical remains. He thinks we linguists naïvely “misunderstand the meaning of, or are unable to accept, scientific uncertainty”, whereas archaeologists understand that it is necessary and scientifically respectable to accept a degree of imprecision (“plus or minus X years”) in carbon dating.

It isn’t that we cannot get our heads round scientific inexactness! It’s that we know full well that language change does not have any degree at all of the mathematical constancy of a physical process like radioactive decay. Language is human behaviour. Expecting language change to proceed at a fixed rate always and everywhere would be like expecting all human beings to utter a fixed number of words per day. Some people are naturally talkative, others taciturn; sometimes people have abundant opportunity for chat, at other times their situation may be solitary. Rates of language change are like that too.

I could say more about shortcomings in Oppenheimer’s treatment of British linguistic prehistory, but this is perhaps enough. His book is interesting – it is not the kind of awful tosh that one blushes to see in print, like the Forster and Toth paper. But what he says about the origin of English is not true. English as a language separate from other Germanic languages does not have a history longer than 1600 years or so. There is little real doubt that before England became “England”, its inhabitants were Celtic speakers.

On the other hand, for all I know Oppenheimer could be correct to believe that the Celts came to the British Isles from Spain.



Geoffrey Sampson

last changed 11 Jun 2007