Pádraic Moran writes for RTÉ about an interesting find:
It’s not often that medieval Irish manuscripts make the news – and it’s all the more unusual when they feature on Dutch national media. Last October, the Dutch national newspaper NRC Handelsblad carried a report that a new word of Old Dutch had been discovered in an Irish manuscript. […] The story has its origins in PhD research carried out at the University of Leiden by Peter-Alexander Kerkhof. He published some of his ideas on a blog dedicated to Dutch studies, where he cited an early Irish text known as “O’Mulconry’s Glossary”. This is found in a manuscript in Trinity College, Dublin, dating from 1572, now bound as part of the Yellow Book of Lecan. The name “O’Mulconry’s Glossary” was assigned by the great (and often controversial) Celticist Whitley Stokes (1830–1901), although the text’s title is really De origine scoticae linguae (“On the origin of the Irish language”). Despite being copied into a 16th-century manuscript, the language is very ancient and coherent with Irish of the early eighth or possibly even seventh century, putting it among the earliest compositions in the Irish language.
De origine scoticae linguae does exactly what it purports to. After a remarkable prologue which claims that the Irish language derives from Hebrew, Latin and Greek (and that the Irish people are descended from Greeks!), it discusses the origins of about 880 mostly Irish words, deriving them from Latin, Greek, and Hebrew, as well as Welsh and Norse. Modern scholars would not now accept most of these derivations, but nonetheless this represents one of the earliest milestones in the study of the Irish language and the beginnings of Celtic linguistics.
Entry number 183 in De origine scoticae linguae reads Blinnauga .i. dallsúilech in linga galleorum “Blinnauga, i.e. blind-eyed, in the language of the Gall.” Kerkhof took the Latin word galleorum to refer to the Gauls, whose territory in ancient times extended as far north as the Rhine, and whose population included some Germanic speakers. The first part of the glossary headword, Blinn-, certainly corresponds to the Germanic word blind, explained as dall in Irish (nn and nd are frequently interchanged in Irish manuscripts). Kerkhof took -auga as an error for Old Dutch augo “eye” (referenced by Irish -súilech “-eyed” in the text), and reasoned that some early Dutch speakers must have come to Ireland as exchange students to study in its famous monastic schools.
But there is another explanation. Gall in Irish can mean “foreigner” and especially “Scandinavian” (as in the place name Fingall). The term galleorum might therefore equally be a Hiberno-Latin hybrid, referring instead to the Old Norse language. This alternative is more congruent for several reasons. Firstly, language contacts between the Norse and the native Irish are very well documented. Indeed, many Irish words in current use were borrowed from Norse […]. Secondly, other Norse words crop up in the same glossary. […] Finally, auga does not need any correction as it is already a perfectly good Old Norse word. We find it elsewhere in similar compounds, for example vindauga “wind-eye”, borrowed into Irish as fuinneóg and into English as window.
The earliest recorded alliance between Irish and Norse was in 850, so we must assume that by then both groups were at least on talking terms. Recent research by this author has shown that the parts of the glossary in which all of these Norse words occur make up a latter addition to the original text, inserted in the late-ninth or early-tenth century.
The term Blindauga sounds like a Scandinavian nickname, like that of Thorsteinn Krókauga “hook-eye” or Sigurðr Kýrauga “cow-eye” (both noted by Peter Schrijver). Who Blind-eye might have been is unknown, but we owe the survival of his name today to the interest of an early Irish linguist.
Thanks, Trevor!
Is Fingall a place name? I suppose it might be, but it’s more often a personal name – one of the fair foreigners, as opposed to the Dubhgall, dark foreigners. Not that I have anything against ‘blinnauga’ being Norse.
Fingall:
[Ir. finn gall, fair foreigner].
Name once used for the portions of Co. Dublin north of the Liffey; gives the title of earl to the Plunkett family. There is also a Lough Fingal, having no Fenian associations, in south Co. Galway, 4 miles ne of Kinvarra.
The name Fingal (with one L this time) was revived in 1985 when co. Dublin was split in three administrative counties. There was a long established “Fingallians” sports club in the area. The other two new counties were originally Belgard and Rathdown but those names, though also historic, lacked public support and were replaced with the clunkier “South Dublin” and “Dún Laoghaire–Rathdown”
If it’s really an eighth-century composition, the Welsh bits would be interesting; the corpus of Old Welsh is tiny.
(A good bit of poetry has survived from that period, but because there was no break in the literary tradition it’s been steadily reworked into Middle Welsh by successive copyists.)
Norse seems vastly more likely than Dutch for the source of blinnauga. “Gall” immediately made me think of
https://en.wikipedia.org/wiki/Cogad_G%C3%A1edel_re_Gallaib
The Latin title of the book confuses me: why is Irish referred to as lingua scotica there? Was that the usual name of the Irish language then?
Kerkhof’s Twitter is interesting for Dutch speakers. I really like it when scientists communicate in an accessible way on their research. He also made this movie about how old Dutch sounded: https://www.youtube.com/watch?v=0hVdz9gyGX4.
Was that the usual name of the Irish language then?
Yes, it was. Scottice is indeed the usual mediaeval Latin for “in the Irish language.”
“Scotland” gets its name from the Scots (i.e. Irish) of Dalriada.
https://en.wikipedia.org/wiki/D%C3%A1l_Riata
In Old English, “Scot” means “Irishman” (or, to be a bit less tendentious, “Gael.”)
As late as James IV’s time, “Scottish”, applied to language, meant “Gaelic” (as opposed to what is nowadays called “Scots”, which was then simply called “English.”)
Scottice is indeed the usual mediaeval Latin name for the Irish language.
Scottish Gaelic is, of course, Erse.
The Scots came from Ireland, but I have a feeling that the Irish might have come from Scotland at some earlier point…
He also made this movie about how old Dutch sounded
That’s a delightful clip (he talks quite naturally, not with that stilted “I’m speaking a reconstructed language” feel so many scholars have), and I love the verb klonk!
I have a feeling that the Irish might have come from Scotland at some earlier point…
We were just hearing the other day about Goidels coming from Lloegr (if you believe it) …
http://languagehat.com/indo-european-and-the-yamnaya/#comment-4001703
I associate “Erse” used for Irish Gaelic with fictional caricature English twits.
The Cruthin are attractive to…em…. speculative historians of Irish-Scottish migrations
“You can kiss my royal Irish arse [~ Erse, certainly] — Joyce, somewhere.
From “Ulysses”
https://en.m.wikisource.org/wiki/Page:Ulysses,_1922.djvu/144
I first became aware of The New Netherland Institute through Shorto’s history of early Dutch Manhattan, The Island at the Center of the World. As I recall, the archival materials of early Dutch America — vast, little perused, hardly translated — are a valuable source for old Dutch, with much to offer to the lexicographers.
Yes, it was. Scottice is indeed the usual mediaeval Latin for “in the Irish language.”
“Scotland” gets its name from the Scots (i.e. Irish) of Dalriada.
Thanks, David.
Irish might have come from Scotland at some earlier point…
According to Peter Shrijver, the first speakers of Irish arrived in Ireland from what is now Northern England sometime in the 1st century AD.
Scottice is indeed the usual mediaeval Latin name for the Irish language.
The Irish Benedictine Monks who showed up in medieval Vienna founded a church in the 12th century, which is of course called the “Schottenkirche”.
According to Peter Shrijver, the first speakers of Irish arrived in Ireland from what is now Northern England sometime in the 1st century AD.
That’s… startling. Do you have a link/cite? My vague understanding is expressed by this paragraph from Wikipedia:
His “Language contact and the origin of Germanic languages” has a chapter on origin of Irish:
The earliest datable linguistic development that was not shared between Irish and British is the development of the Proto-Celtic diphthong *ai to *ɛ̄ (as in English bed but long), which affected British Celtic but not Irish, probably at some point during the later first century AD at the earliest.75 Before this happened, Irish and British Celtic were not just mutually comprehensible dialects; they were indistinguishable from one another.
…
So, to all intents and purposes, in the first century AD Irish and British Celtic were one single undifferentiated language. This is highly relevant to us if we wish to determine when Irish first arrived in Ireland. If Irish had been geographically isolated from British Celtic for any length of time before the first century AD, one would expect to find at least some early differences between them, although it is impossible to say how many: language does not change at a particular rate comparable to a molecular clock, and periods of little change follow short bursts of rapid change. It seems safe to say, however, that any scenario that has Irish arriving in Ireland before, say, 1000 BC, is impossible for linguistic reasons: more than a thousand years of relative isolation seeing no linguistic change whatsoever simply strains our credulity. Beyond that, dating becomes more difficult, but it seems safe to say that an Irish arrival in Ireland close to or in the first century AD is much easier to unite with the linguistic evidence than an arrival around, say, 500-1000 BC.
There is much more there. I hope DM can summarize, it gets very technical and I don’t feel adequate to the task.
Doesn’t the argument equally well imply that all of Insular Celtic must have arrived recently in the British Isles? An allegedly more-or-less entirely undifferentiated Insular Celtic covering practically all of Britain as late as the first century …
more than a thousand years of relative isolation seeing no linguistic change whatsoever simply strains our credulity
“Whatsoever” is doing a lot of work there.
Well, the quoted text does not argue against the possibility that Britain was (re)settled from Ireland in the first century AD, but I supposed the Romans would have noticed if that was the case.
Indeed.
It seems to be time for me to bring out my Dravidian-origins theory.
Seems to me all of the above could be consistent with a long-standing settlement of Ireland by people likely speaking some sort of Celtic, and almost certainly some Indo-European tongue, who were overrun by people speaking an early form of Irish in the first century AD.
*Is* Ireland geographically isolated? I suppose so, if Great Britain is by the Channel, but later on Sruth na Maoile* was as much a highway as a barrier.
*I can’t think of its English name, if it has one – the narrow stretch between Kintyre and Antrim.
Variably the Irish Channel or the North Channel. In Scots, jocularly the Sheuch /ʃ(j)ux/ ‘trench, furrow, gutter, passageway, gullet, nape’ (why ‘nape’? )
On this thread-
http://languagehat.com/urchin/
-I had brought up Dahl’s work on the spread of (Proto-)Old Norse: his core (and to me most convincing) argument is that the vast surface area and low population density of the pre-Modern Nordic-speaking area, combined with the fact that all attested North Germanic languages/dialects (no exceptions!) could be traced back to pre-Old Norse unproblematically, made it likely that the ancestral language had spread comparatively recently over the whole area: as he pointed out, the alternative -that diffusion could explain the spread of so many linguistic features so uniformly across such a vast, sparsely-populated area- is inherently implausible.
Peter Schrijver’s argument for a comparatively recent spread of Old Irish is very similar, and takes a similar fact as a starting point: the fact that Early (Ogham)/Old Irish is unambiguously ancestral to ALL later attested varieties of Goidelic. It is likewise very likely that Old Irish spread from some geographically restricted Urheimat at a comparatively recent date. Considering how radically its phonology and morphosyntax were transformed between the fifth and ninth centuries AD, diffusion is even more improbable as an explanation than it is in the case of Old Norse. A similar argument applies to the spread of Brythonic and to the earlier uniform “Insular Celtic” ancestral to both.
(David Eddyshaw: my own take on “Insular Celtic” is that its spread across Roman Britain, annihilating earlier linguistic/dialectal diversity, was probably a post-Roman conquest phenomenon. The reason for this belief is simple: such places as pre-Roman Italy and Iberia were linguistically phenomenally diverse, and Britain before the Roman conquest was politically disunited: on first principles we would assume extreme ethnolinguistic diversity, not uniformity, on the territory of what was to become Roman Britain).
What I find intriguing is that the spread of North Germanic, Brythonic and Goidelic all seem to have taken place at approximately the same historical period (late Roman imperial times/Early Dark ages). Other language spreads took place at the same time in the vicinity of the Roman Empire (Basque, Berber, Albanian, West Germanic). While the role of the Roman Empire in the spread of Latin (later Romance) is undeniable (just as undeniable, but not as widely known, is its role in the spread of Greek in the Eastern half of the Mediterranean), I suspect that the Roman periphery/borderlands were more generally conducive to *language spreads in general* in late Roman/Early Medieval times. One thing which is striking in this connection is that in several instances (Basque, Albanian, Brythonic, West Germanic, possibly Berber) the expanding language did so *at the expense* of Romance varieties.
Thoughts, anyone?
@Etienne
Is it possible that the Ogham/ Old Irish was an artificial written prestige language, like Classical Latin or Sanskrit, and so the lack of linguistic diversity in it does not imply a corresponding lack of diversity in the sermo vulgaris?
Then you’d expect it to have been more regular. It’s rough and splintered like Vedic, not smooth like Sanskrit.
Etienne, it seems to me that the two halves of your post contradict one another. If Latin and Greek ate the related and unrelated languages during its spread, why couldn’t Old Irish and Old Norse have done the same? We know that Common Brittonic ate Pictish, for instance, whereas we know nothing about any Celtic or other varieties spoken in Ireland before St. Patrick. A rapid spread of a southeastern variety powered by missionary Christianity might have made Irish speech uniform in a very short time.
John Cowan: I am not sure I understand what contradiction you are referring to: I agree, Old Irish and Old Norse (and others) must have expanded and eliminated their closest relatives, just like Latin and Greek, yielding the linguistically uniform landscapes which subsequently started to diversify.
What I was writing/arguing against above is the tendency to regard the non-Latin languages spoken in the Roman borderlands in late Imperial/ Early Medieval times as reflecting the pre-Roman linguistic landscape: it is tacitly (and sometimes explicitly) assumed that in Roman times all the inhabitants of Ireland spoke a direct ancestor of Old Irish, all of the inhabitants (minus Punic and Romance speakers) of North Africa Proto-Berber or a direct ancestor thereof, all of the inhabitants of Denmark, Southern Norway and Southern Sweden a direct ancestor of Old Norse, perhaps Proto-Germanic itself…and like Schrijver and Dahl, I think this is unlikely in the extreme.
The borderlands of a economically prosperous, politically stable empire can provide a rich environment for trade and room for significant personal mobility. Both physical mobility and longer-distance commerce naturally lead to a delocalization of culture. One effect of this could be linguistic leveling at the margins of the Roman Empire, even among ethnic groups who were politically independent and remained largely outside the boundaries of the Empire.
I did; it feels like just a month or two ago, and someone linked to that thread just a few days ago. It’s bedtime (3 am), so I’m not going to look for it now.
Sea of Moyle is the narrow part of the North Channel and the eponym of the 1972–2015 local government district of Moyle in Northern Ireland.
“North Channel” is unsatisfactory on two grounds: (1) like “Northern Ireland”, it uses North instead of Northeast; (2) it is Irish-centric: to Scots it would be the South[west] Channel. The name “Irish Channel” was sometimes given to the whole of the waters between Ireland and Great Britain, ie the Irish Sea, North Channel, and St George’s Channel. Have Scots repurposed it since?
In the nineteenth century the North Channel was sometimes “St Patrick’s Channel”, to complement “St George’s Channel” at the far end of the Irish Sea. “George’s Channel” is also unsatisfactory, apparently a late English imposition of their patron; more suitable saints than George would include David, patron of Wales, or Kieran or Maedoc, who actually crossed the channel from Wales to Ireland.
The North Channel makes good sense from a seaman’s point of view. It’s the northern exit from the Irish Sea and opposed to a *South Channel at the far end.
Furthermore, it doesn’t make sense to expect names to reflect exact compass bearings; that’s not how things work.
True, but geographical names usually do reflect some logic. The investigation of names is mostly about identifying the logic.
Of course; what I’m saying is that it doesn’t make sense to say “North Channel” is unsatisfactory because it uses North instead of Northeast. In Massachusetts, Easthampton is southwest of Northampton; there are historical reasons for this, as there are for everything, but to say it’s “unsatisfactory” would be to misunderstand how language works.
One more for the Norse etymology. Ogham’s razor.
Is there any way to connect Old Irish Gall and Germanic wal-?
Ogham’s razor.
Made me laugh!
In Southern Scandinavia, water connected and forest separated. But Øresund (or even Storebælt) are not as wide as the Irish Sea, and more sheltered from Atlantic weather.
I suspect that the Roman periphery/borderlands were more generally conducive to *language spreads in general* in late Roman/Early Medieval times … Thoughts, anyone?
Reminds me of Johanna Nichols’ recent thesis that the spread of Indo-Iranian “catalysed” also the original west–east expansion of Uralic (as fur traders).
In Southern Scandinavia, water connected and forest separated.
Water connects pretty much everywhere, even across big chunks of the Pacific. “Samoan/Hawaiian/Maori is essentially bad Hawaiian/Maori/Samoan.” Of course not all cultures are seafaring. Native Australians had a monopoly on Australia for 40Ky, because nobody else knew how to make a living and even Australians didn’t have enough surplus to go seafaring. (The most extreme example of a non-tariff barrier to entry ever imagined.)
John Cowan: Actually, there is good evidence that Australia was not isolated for 40 000 years: the dingo has been called Australia’s first introduced species, but it was introduced some 5000-10 000 years ago. I think it was Johanna Nichols who suggested that many of (or all?) the ancestors of known Australian languages may have expanded into Australia comparatively recently from present-day New Guinea and Indonesia (before the expansion of Proto-Austronesian, obviously). Perhaps all Australian languages known today descend from the language(s) of whatever group(s) introduced the dingo into Australia.
Brett: I like your point on the stability to be found in the borderland of a stable empire. And this makes me wonder: the expansion of various non-Romance languages on the borderlands/in the isolated areas of the Roman Empire (Brythonic, Goidelic, West Germanic, North Germanic, Albanian, Basque, Berber) may have something in common, in terms of the social and economic dynamics driving their expansion, with the expansion of various non-Sinitic languages/proto-languages in the borderlands of China (The Tai family, Vietnamese, Burmese all expanded centuries after the unification of China, and all are heavily Sinitic-influenced in both vocabulary and structure: I wonder whether the expansion of Mongolian and Tibetan, comparatively unscathed by Sinitic influence though they are, might not be due to similar factors).
Interestingly, just in the same way that several of these languages of the Roman imperial boondocks appear to have expanded at the expense of Romance, I recently ran into a claim (if someone could provide the exact reference, that would be appreciated!) that Vietnamese owes much of its Sinitic-like structure and vocabulary to a Sinitic substratum -in other words, Vietnamese expanded at the expense (in part) of Chinese.
A comparative linguistic study of Romance and Sinitic on the one hand, and of the linguistic neighbors of each Imperial language and what sort of interactions/exchanges took place in each language area (both Sinitic and Romance contributed a great many lexemes relating to writing/literacy, commerce and religion in both instances, for example) would make for a very interesting series of articles or even more interesting book.
Of course, the number of outraged nationalists generated by such diachronic work today would make the research project a rather, err, lively enterprise. I know this first-hand: a well-known female writer from Kabylia, who at first was delighted to realize I knew something about Berber languages in North Africa, very nearly slapped me (and certainly never spoke to me again) when I actually was so impertinent as to suggest that the spoken language of Kabylia in Roman times was probably not the ancestor of present-day Kabyle Berber (and she did not even let me present my arguments. Why yes, I was young and hopelessly naive, how did you guess?)
Yes; we discussed the quite convincing paper here a few years ago.
Schrijver thinks that Welsh is a result of Romance speakers switching to Celtic.
Similarly, Albanian is the result of Romance speakers shifting to Illyrian, and Maltese is the result of Romance speakers shifting to Arabic.
Speakers of majority prestige languages completely switch to minority languages all the time, as we see everywhere in modern times, alas. (It makes nation-building so hard, damn it!)
Greek is the result of Romance speakers shifting to Hellenic.
Welsh is a result of Romance speakers switching to Celtic.
Well, David E’s satire notwithstanding, I think it is fair to say that Welsh and the other Brythonic languages would not be what they are today without a great many Romance (or Latin) speakers switching to Old British. There is simply too much Latin substrate to think otherwise. The words for body parts (save pen ‘head’) are Latin, and so are a bunch of of ordinary-life words from achub ‘save, rescue; redeem, deliver’ < occupare to barf ‘beard’ < barba to cumul ‘cloud’ < cumulus all the way to torf ‘crowd’, twrf ‘disturbance’, and twrw ‘clamor’, all < turbus which could mean any of those things.
By comparison, Old English has a mere two dozen words inherited from Proto-Germanic but borrowed there from Latin, plus a few non-ecclesiastical words brought in with Christianity, of which fork < furca is one of the few that survived: scamol ‘chair, stool < scamellus, for example, lost out within OE Itself to the ancestors of stool, bench, settle.
Of course, Latin speakers switched to Celtic exactly because Latin stopped being prestige language in Britain in 5th century AD.
According to Schrijver, Latin speakers in the highland zone of Britain were impoverished refugees from lowland zone overrun by Anglo-Saxons.
Lacking political power and prestige they had no choice, but to switch to language of their new Celtic overlords.
The words for body parts (save pen ‘head’) are Latin
Troed, llaw, bys, calon, dant, llygad, clust, ysgwydd, glin, ffrwyn, gwallt, tafod …
their new Celtic overlords
I, for one, welcome …
There is simply too much Latin substrate to think otherwise.
Gotcha! Petitio principii, as we say in Welsh!
There actually is an awful lot of Latin vocabulary in Albanian, including, for example, the words for “king” and “hundred” (neither of which are Latin loans in Welsh, as it happens.)
Come to that, there is an awful lot of French vocabulary in English, but this is not usually attributed to a French substrate.
I’ve no political axe to grind in this at all: I’m perfectly happy to imagine my ancestors speaking British Latin (and indeed, some of them surely did.) I’m just wary of just-so stories about substrate populations, conjured into being to “explain” linguistic features, but curiously difficult to identify historically (like the supposed Brythonic substrate population that lurked secretly into the Middle Ages in the bush in Lloegr and snuck in by night to give Middle English its verbal system. Thank you, John McWhorter …)
[Wps! Ffroen, not ffrwyn. “Ffrwyn” is from Latin, of course. Nasty foreign things. We used to steer horses by gentle blandishments, not like those brutal Romans.]
David Eddyshaw, John Cowan, SFReader: I think all three of you are making a common assumption which to my mind is unjustified: namely, that if a given language (Albanian, Welsh…) is Romance-influenced and has been claimed to have expanded at the expense of a Romance variety, then the Romance influence must derive from said Romance substrate.
In fact, in the case of Albanian it is the phonological mismatch between most borrowed Romance morphemes of Albanian on the one hand and the Romance place-names located in present-day Albanian-speaking Europe on the other which drove Norbert Jokl to claim that Albanian had expanded (from further North, where it had been in contact with a Romanian-like Romance variety which had earlier supplied it with a majority of its Romance loans) at the expense of a Romance variety spoken along the coast of present-day Albania (this Romance variety being the source of the Romance place-names of Albania). Meaning that this extinct Romance variety (called “Labeatic” by one scholar, after a major lake in the area) did NOT contribute most of the Romance morphemes found in Albanian.
A similar situation can be observed in the case of Breton: this language indubitably expanded at the expense of a Romance variety spoken in Brittany earlier (again, the evidence of place-names is quite conclusive), but the bulk of the Romance elements in Breton (of any time period) are either gallicisms (deriving from dialects of French spoken further East) or earlier loans from British Latin (shared with Cornish, Welsh or both). The Romance variety Breton replaced seems to have contributed little if anything to Breton.
David Eddyshaw (and any other interested readers, assuming there are any): my September 25, 7:41 PM comment at this thread-
http://languagehat.com/how-did-latin-become-a-dead-language/
-answers your point above about speakers of prestige majority languages shifting to minority languages. It also refers back to another thread where I had given some scholarly references on Brythonic, Basque and Albanian expanding at the expense of Romance.
Direct link to the comment.
I would guess the features in question are either due to a Celtic substrate in Anglian OE, which is underdocumented compared to Saxon OE with its Romance substrate; or perhaps due to a later Celtic substrate in, say, Cumberland, from where they slowly spread south together with all the Norse-influenced northern features & words in the Late Middle Ages.
Actually, that’s not a bad thought. After all, those prissy Southerners eventually gave up their silly affected 3rd sg -eth in favour of our manly northern -s. They could have learnt from us how to conjugate periphrastically like real men do too. ‘Appen. There’s nowt like t’ Hen Ogledd for talking proper, like.
Where does that -s come from, actually?
Tocharian.
(Tocharian A, obviously. But you knew that.)
David E.: prissy Southerners
I think the preferred term is ‘Southern pansies’.
Come to that, where did you Scandinavians get your 3rd sg -r? Eh?
i won’t even try to summarize the “romance substrate” debate about yiddish – speaking of mountains that walk or stumble – except to say that manaster ramer has raised the excellent point that it seems odd to assume people in the 10thC rhineland whose descendents speak a germanic language weren’t speaking frankish…
David E.: where did you Scandinavians get your 3rd sg -r? Eh?
I’ve never seen a good explanation for that, and even the new 4-volume Norsk Språkhistorie brushes over it. I’ve been thinking that it started as allophonic voicing of 2sg -s and 3sg -þ before unstressed vowels, but maybe not. What Norsk Språkhistorie does say is that -s > -z > -R in nouns must have happened before final devoicing, which in turn happened before syncope. It also mentions 2sg -s > -R, but only implicitly, in an example of vowel breaking. Furthermore, it says that syncope led to a lot of analogical levelling and restructuring based on mis- and reinterpretation (presumably of now obscure allophonic correspondences -[my conjecture]). The allophones of -R (on its way from sibillance to -r) and -đ in a weak final position must have been identical in some environments, or overlapping in a non-transparent way.
That, or IKEA.
(I was going to go through the comment and cut&paste italicizers, but I forgot. I also meant to find an actual ð.)
Troed, llaw, bys, calon, dant, llygad, clust, ysgwydd, glin, ffrwyn, gwallt, tafod …
Concedo.
Gotcha! Petitio principii, as we say in Welsh!
Distinguo. It’s hypothetical substrate languages that have that problem, where “from our defective material [we] cannot devise any etymology fit to print” (JRRT). But Latin is an actual language which has actually contributed to Welsh. Why then call it a substrate? Because the vocabulary is, so I am told, generally similar in semantic range to vocabulary from unquestioned substrate languages: various Native languages in American English, Tai languages in Chinese, the Slovene substrate in Carinthian German, etc. etc. Of course, what weirdens the matter is that Welsh also has a Latin superstrate.
Where does that -s come from, actually?
The best story I know is that the 2sg -st and the 3sg -eth merged in Northern English as -es. This may have been by analogy with the same merger in Scandinavian. In the end, N.Eng. got -es in all positions except immediately adjacent to I, we, you, they (the Northern Subject Rule); Scand. wound up with all persons and numbers identical.
It’s pretty easy – Vulgar Latin speaking commoners learned more prestigious Welsh of the warrior elite who in turn learned more prestigious Bible Latin of the Church.
So you get both Latin substrate and Latin superstrate at the same time.
On English 3rd sg -s, I wonder whether the s of “is” might have something to do with it?
Doesn’t look too plausible, admittedly. On the other hand, the regular Welsh 3rd sg past ending -odd (unique to Welsh within Brythonic, and an innovation during the Middle Welsh period) is supposed to have been generalised from a highly irregular remnant group of ablauting preterites with about three members altogether, via a process of reanalysing the ablauted stem as an ending. If that‘s possible (and I’ve never seen any better alternative explanation), then anything is …
@de
Is or was there a sibilant in final t (maybe after particular vowels) in “Northumbrian” English? You get this in Hiberno English and I think also Welsh. A final palatal t can be very weak and hard to distinguish from an s in speech.
The original English ending won’t have been /t/, though, but /θ/.
Word-final -Vt is pretty marginal in Welsh, turning up (I think) pretty much only in loanwords and the preposition at, which I think is probably a back-formation from the conjugated forms ataf etc.
Written final -t in Middle Welsh represents /d/ (or, in really old texts, /ð/.)
The idea is that there was a Latin superstrate during the Empire, and a Latin substrate afterwards. The superstrate provided most of the loanwords, including those that that distinguish Latin /a/ from /a׃/ (as a vs. o like in native Celtic vocabulary); the substrate provided the grammatical influence (the pluperfect in particular) and probably the loans in the basic vocabulary.
(That said, “arm” and “leg” make perfect sense to me as superstrate loans: the distinction of “hand” vs. “arm” and “foot” vs. “leg” was imported, and the missing words with it. That said, that also makes perfect sense as a substrate phenomenon…)
The German -̈er plural, common and still growing, comes from the OHG -ir of the PIE *s-stems, a class that in OHG had only nine members left.
The German -̈er plural, common and still growing, comes from the OHG -ir of the PIE *s-stems, a class that in OHG had only nine members left.
Another example for an ending spreading from a small class of verbs is the 1st sg. “-m” in some Slavic languages. In Proto-Slavic it was limited to a couple of verbs, but these were the basic and high-frequency verbs “be”, “have”, “eat”, and “give”. From there it spread to become the 1st sg ending of the big class of a-verbs in Polish and the almost exclusive 1st sg ending in FYLOSC.
I suppose there is an analogy within Welsh (and the rest of Brythonic) in the way the original u-stem plural has become the default plural ending for nouns (though still a default among a good many alternative formations), on account of its distinctiveness and relative simplicity.