Irish vs Welsh conservatism

[Otto Kretschmer]

Which of them is more conservative/closer to Proto Celtic?

[Linguoboy]

Which of them is more conservative/closer to Proto Celtic?

Yes.

[Jonlang]

Which of them is more conservative/closer to Proto Celtic?

Neither. Proto-Celtic isn't recognisable in either Irish or Welsh unless you're familiar with many of the sound changes which occur to allow you to visualise their relationships to each other. (i.e. Proto-Celtic *makʷ(kʷ)os > Welsh mab and Irish mac.

Why do you want to know which is closer to Proto-Celtic?

[Otto Kretschmer]

Curiosity. In many linguistic families certain languages are more conservative/archaic than other. Just wanted to know if this is also the case among Celtic langugaes.

[KathTheDragon]

In many linguistic families certain languages are more conservative/archaic than other.

I'm pretty sure this is more of a folk story among historical linguists than anything particularly insightful. Among contemporary related languages, they've all undergone significant innovation and all have archaisms. The only sensical notion of "language X is more archaic than language Y" is when X is significantly older than Y, but this isn't a terribly interesting idea - it's simply obvious that it's the case.

[Vijay]

How strong is the case for Insular Celtic? I.e. are Goidelic and Brythonic languages more closely related to each other than to other Celtic languages?

[Linguoboy]

How strong is the case for Insular Celtic? I.e. are Goidelic and Brythonic languages more closely related to each other than to other Celtic languages?

I feel like it can't ever be particularly conclusive given how incomplete the attestation of Continental Celtic languages is.

[Vijay]

How strong is the case for Insular Celtic? I.e. are Goidelic and Brythonic languages more closely related to each other than to other Celtic languages?

I feel like it can't ever be particularly conclusive given how incomplete the attestation of Continental Celtic languages is.

Ahh, okay! That makes sense. Thanks!

[hwhatting]

AFAIK, the little we have in old attestations of Brythonic from Roman times (names in sources, inscriptions) is indistinguishable from Gaulish, so it looks like the similarities between the Insular Celtic languages (lenition, initial mutations, loss of unstressed syllables) seem to be due to parallel / areal developments. In tree models based on old isoglosses, I have seen groupings like Goidelic vs. Brythonic plus Gaulish vs. Celtiberian.

[Nortaneous]

In many linguistic families certain languages are more conservative/archaic than other.

I'm pretty sure this is more of a folk story among historical linguists than anything particularly insightful.

it's not. english vs icelandic, mandarin vs japhug, buyang vs thai, etc. for there to be no vague ordering of archaism whatsoever, wouldn't diachronic losses of information have to have a constant distribution over spacetime? which they don't due to branching, areal convergence, sociolinguistic factors, etc.

vague orderings of archaism are very visible in the east asian convergence area, since the areal pressure (which different languages follow or resist to different extents) is toward losing a lot of segmental information to compression

[Moose-tache]

Welsh and Irish are subject to a lot of similar sound changes regarding loss of final syllables, intervocalic lenition, etc. Irish has managed to hold on to more nominal and verbal morphology, but the magnitude of the sound changes isn't that different. Welsh was just unlucky I guess. Unless you like analytic morphology, in which case it was very lucky. Neither one really jumps out at me as "more conservative."

[fusijui]

Like, one is Tea Party conservative, the other's Chicago School conservative? (Pardon the Americanism)

[Linguoboy]

Welsh and Irish are subject to a lot of similar sound changes regarding loss of final syllables, intervocalic lenition, etc. Irish has managed to hold on to more nominal and verbal morphology, but the magnitude of the sound changes isn't that different.

Even that's arguable, depending on which varieties you look at. Written Irish retains the genitive but in spoken dialects it's often restricted to fossilised forms. Welsh analytic verbal paradigms tend to retain distinct forms for each person-number combination where Irish paradigms are more analytic. For instance, compare the preterite inflections for "speak" in the colloquial standards of each language:

Welsh
siarades i
siaradest ti
siaradodd e/o/hi
siaradon ni
siaradoch chi
siaradon nhw

Irish
labhair mé
labhair tú
labhair sé/sí
labhraíomar
labhair sibh
labhraíodar

(Connemara dialect--arguable the most vigourous of the traditional spoken varieties--takes this even further, replacing labhraíomar with labhair muid and labhraíodar with labhair siad. Conversely, some NW Welsh varieties replace the entire analytic preterite with ddaru + pron + VN, i.e. ddaru fi siarad, ddaru di siarad, etc.)

[WeepingElf]

It is anyone's guess why the Insular Celtic languages drifted so far from the "common IE" typological profile, and in similar ways. There doesn't seem to have been a "Proto-Insular Celtic" distinct from Proto-Celtic, which probably means that we are dealing with a Sprachbund or the influence of a common substratum here. However, the substratum theory now has a bad reputation, after all that nonsense about a Semitic substratum and the failure of substratum theories in Romance historical linguistics, so most Celticists tend to assume that internal factors were at work here. Yet, of course, given that Paleolithic Continuity is utter bullfrogs, there were languages in the British Isles before Celtic spread there, and we do not know what those languages were like, so it may have been those languages that are "to blame" here.

[WeepingElf]

Let me add some more thoughts to the post above.

Many people assume that Celtic was brought to the British Isles by the Bell Beaker people about 2500 BC, but that seems too early - the degree of similarity between the known Celtic languages suggests that Proto-Celtic is to be dated approximately in the 1500-1000 BC range, which would make the Urnfield culture the most likely candidate for Proto-Celtic. That leaves the question what else the Bell Beaker people spoke - perhaps a now lost, extinct branch of IE related to Anatolian (both Anatolian and the Bell Beaker people seem to originate in a south(west)ern offshoot of the Yamnaya culture in the Lower Danube area) - alas, we don't know.

Also, it seems as if the changes that led to the aberrant typology of the Insular Celtic languages happened rather late, as the oldest Ogham inscriptions and what little we have from Roman-Era British do not yet show them. At such a late date, the idea that they were effected by a substratum seems unlikely and an internal cause more plausible. However, it may have been the case that there was a tendency to head-initial word order and some sort of allophonic lenition of intervocalic consonants that is not shown in the writing earlier on, and that could have been a substratum effect. Note that Western Romance, spoken where Continental Celtic had been spoken before, shows an intervocalic lenition very similar to the Brythonic one, with the main difference being that in Western Romance, it never operated across word boundaries - thus, no initial mutations - as it did in Brythonic (and the different lenition in Goidelic). That we see no initial mutations in Continental Celtic is thus unsurprising: they had not evolved in Insular Celtic by that time, either; but if any Continental Celtic languages had survived, they would perhaps show lenitions of the Western Romance type, i.e. without initial mutations.

But to get to the original question: Irish is a bit more conservative than Welsh, as it did not undergo the *kw > p change, and has retained noun cases, although in a much eroded form - but in Welsh, they are utterly gone!

[Linguoboy]

But Irish developed phonemic palatalisation in the entire consonant series plus a fortis/non-fortis distinction in nasals and laterals. And is there any PIE precedent for the distinct (often suppletive) dependent verb forms of Irish?

See, this is why I think questions of “Which variety is more conservative?” are something of a mug’s game. Maybe there’s a clear answer when one of the varieties really hasn’t innovated much. But when they’ve all innovated quite a bit, it comes down to a quite arbitrary ranking of features. (Why rank q>p, which affects relatively few items in the overall lexicon, higher than palatalisation, which affects every single one of them without exception? Just because it’s earlier?)

[WeepingElf]

Sure, the *kw > p thing is overrated, and the pervasive Irish palatalization is a massive innovation that has no counterpart in Welsh (but it helped preserving case distinctions in Irish even though the endings themselves are mostly lost). Well, asking which of the two is more conservative is moot, in both the British and the American sense - both have departed so far from Proto-Celtic that the question cannot be definitely answered, and it doesn't really matter.

[Jonlang]

Welsh and Irish are subject to a lot of similar sound changes regarding loss of final syllables, intervocalic lenition, etc. Irish has managed to hold on to more nominal and verbal morphology, but the magnitude of the sound changes isn't that different.

Even that's arguable, depending on which varieties you look at. Written Irish retains the genitive but in spoken dialects it's often restricted to fossilised forms. Welsh analytic verbal paradigms tend to retain distinct forms for each person-number combination where Irish paradigms are more analytic. For instance, compare the preterite inflections for "speak" in the colloquial standards of each language:

Welsh
siarades i
siaradi di
siaradodd e/o/hi
siaradon ni
siaradoch chi
siaradon nhw

(Connemara dialect--arguable the most vigourous of the traditional spoken varieties--takes this even further, replacing labhraíomar with labhair muid and labhraíodar with labhair siad. Conversely, some NW Welsh varieties replace the entire analytic preterite with ddaru + pron + VN, i.e. ddaru fi siarad, ddaru di siarad, etc.)

Siaradi di is the future, siaradaist ti / siaradest ti is the preterite.

Living in North East Wales I am yet to encounter ddaru (Preterite III) in the wild, everyone I speak to uses the gwneud preterite (or Preterite II), which is annoying. But, ddaru always takes mi and ti: ddaru mi weld, ddaru ti weld, etc.

Sure, the *kw > p thing is overrated

How can a sound change be overrated?

the pervasive Irish palatalization is a massive innovation that has no counterpart in Welsh (but it helped preserving case distinctions in Irish even though the endings themselves are mostly lost).

This has reminded me of something I've been meaning to ask about: are the Irish broad consonants actually velarised or are they just "not palatalised"?

[WeepingElf]

Sure, the *kw > p thing is overrated

How can a sound change be overrated?

By considering it more pervasive or relevant for the classification than it actually is. An example of an overrated sound change is the satem/centum division in IE, which earlier IEists interpreted as a pair of primary branchess of IE (i.e. all satem languages were considered more closely related to each other than to any centum languages, and vice versa).

[Linguoboy]

Siaradi di is the future, siaradaist ti / siaradest ti is the preterite.

Good catch. Diolch, y mêt i!

But, ddaru always takes mi and ti: ddaru mi weld, ddaru ti weld, etc.

You may be right about ddaru ti but I've definitely come across ddaru fi from fluent native speakers.

This has reminded me of something I've been meaning to ask about: are the Irish broad consonants actually velarised or are they just "not palatalised"?

In traditional Irish, they are velarised bzw. labialised. You can hear this quite clearly in the speech of native speakers.

In non-traditional Irish, the palatal/non-palatal distinction seems to be mostly lost. Palatalisation is sometimes realised as yodicisation, e.g. [ˈbjoː] for beo. Dentals are affricated, e.g. [ˈʧæŋ(ɡ)ə] for teanga. The lack of palatalisation has no particular coarticulation.