Celtic historical linguistics

[WeepingElf]

In this thread about a Celtic conlang project, some questions of Celtic historical linguistics arose for which I am creating this thread. The relevant questions are:

1. What would a modern Continental Celtic language look like? Would it resemble an Insular Celtic language, or not?
2. Which factors caused the Insular Celtic languages to diverge typologically from the "Common IE" type the way they did?
3. When was Proto-Celtic spoken, and when and how did it spread, especially to the British Isles?

The first question has a short answer: We don't know. The long answer is: The problem is that with any comparison of Insular and Continental Celtic, we are comparing differing time stages. The Continental Celtic languages were all dead and gone by 600 AD, when the oldest preserved literature in Insular Celtic was composed - and already shows the characteristic features of Insular Celtic like initial mutations, VSO word order, absolute vs. conjunct verbs and "conjugated prepositions". We do have fragments of Insular Celtic languages from the times when Continental Celtic languages were alive (Ogham inscriptions for Goidelic, and a few curse tablets, graffiti and other informal writing for Brythonic), and they resemble Continental Celtic languages. So the typological changes of Insular Celtic apparently happened during the "Dark Age" from 400 to 600 AD from which we have no Insular Celtic documents; and it is not entirely inconceivable that the same changes would have happened to Continental Celtic as well. But they did not happen in the Western Romance languages that supplanted them, even though they underwent a similar lenition of post-vocalic consonants, with the key difference that it did not transgress word boundaries. So it seems likely that a modern Continental Celtic language would be more like a Western Romance language than an Insular Celtic language in this regard.

This brings us to the second question. It all happened, as said above, between 400 and 600 AD. That's a pretty short time for such a profound change, but it actually rests on just two sound changes: 1. Lenition of post-vocalic consonants, leading to initial mutations. 2. Loss of final syllables, causing the morphosyntactic changes such as loss of case inflection in Brythonic (in Goidelic, traces of the old case endings remained by the workings of palatalization and vowel umlaut). The initial mutations arose from that in early Insular Celtic, words within a phrase were phonetically run together, such that post-vocalic lenition operated across word boundaries within phrases, resulting in initial mutations. The "conjugated prepositions" have a similar origin. The absolute vs. conjunct distinction in verb inflection arose from a combination of VSO word order, separable prefixes and Wackernagel's Law (clause-level clitics attaching to the first word in the clause, also observed in many other IE languages). So if the verb had no separable prefix, clitics attached to it, protecting the verb ending from loss of final syllables: this is the absolute form. But if it had a separable prefix, the clitic went between the prefix and the verb, whose ending was then eroded by the loss of final syllable: this is the conjunct form. Thus. all we need as "input" are VSO word order and a phonetic running together of phrases, both of which may have happened earlier than 400 AD. Not much of a substratum influence, it seems.

All this also connects with the third question, the age of Celtic. Some scholars maintain the opinion that Proto-Celtic was the language of the Bell Beaker culture around 2500 BC, but I think that is too early. The degree of similarity between the Celtic languages around 1 AD points at a Proto-Celtic somewhen between 1500 and 1000 BC, which suggests that it was the language of the northern branch of the Urnfield culture (the southern branch would have spoken Proto-Italic, which appears to be of a similar age). The spread of Celtic to the British Isles and to the Iberian Peninsula may have been later still, perhaps as late as 500 BC. Also, the Old European Hydronymy, which pretty much coincides in area with the Bell Beaker culture, seems to be from an IE language which had merged PIE *o with *a, a change that did not happen in Celtic, so these river names cannot be Celtic but must come from a different branch of IE (if the Old European Hydronymy isn't just spurious, of course).

[Salmoneus]

I'd add a couple of things:



- it's not certain that the big changes took place between the 4th and 6th centuries. Around this time there was a shift from writing in ogham to writing in the latin script, which necessarily required a total respelling of everything. If you add together the big coincidence of a complete respelling happening at exactly the same time as a massive language shift, and also the known conservative nature of the early inscriptions (memorials written by a caste of literate religious professionals), it seems plausible to suggest that the writing of the later Primitive Irish inscriptions may simply have been fossilised - that is, that the druids just wrote the way they'd learnt to write and ignored deviations in the 'corrupted' vernacular of their day, until a change in writing system forced them to abandon the old conventions. This could widen the window for the primitive > old irish grammatical transition by several centuries.

- regarding the origin of mutation, we should probably link this to a supposed Proto-Indo-European (and certainly Proto-Italo-Celtic (or areal northwestern PIE for those who don't believe in PIC)) feature, which is the phonological non-independence of particles. It seems likely that many particles lacked independent stress and were phonologically bound to neighbouring 'strong' words. We see this, for instance, in the (often separable) binding of adpositions to verbs in many languages, and the binding of things like -que productively and -met and -pte unproductively in Latin (and still of -se in Irish). We see it more widely in early Latin poetic scansion, and suggested by early Roman descriptions of Latin. But it had ended centuries before Western Romance became a thing. In Insular Celtic, however, it probably continued to be a thing for longer. And if you combine this with intervocalic lenition, mutations make sense: the lenitive contexts can bridge word boundaries because, when one word is a weak particle, there IS no phonological word boundary. This difference can explain why seemingly the same change produced mutation in Irish and not in French.

[note also that lenition still DOES cross word boundaries between article and noun in Sardinian, in many ways the most conservative Romlang. It's just that it's predictable and synchronic, unlike in Celtic where it's become fossilised and phonemic]

[Esneirra973]

All this also connects with the third question, the age of Celtic. Some scholars maintain the opinion that Proto-Celtic was the language of the Bell Beaker culture around 2500 BC, but I think that is too early. The degree of similarity between the Celtic languages around 1 AD points at a Proto-Celtic somewhen between 1500 and 1000 BC, which suggests that it was the language of the northern branch of the Urnfield culture (the southern branch would have spoken Proto-Italic, which appears to be of a similar age). The spread of Celtic to the British Isles and to the Iberian Peninsula may have been later still, perhaps as late as 500 BC. Also, the Old European Hydronymy, which pretty much coincides in area with the Bell Beaker culture, seems to be from an IE language which had merged PIE *o with *a, a change that did not happen in Celtic, so these river names cannot be Celtic but must come from a different branch of IE (if the Old European Hydronymy isn't just spurious, of course).

I already touched upon my other thoughts in the previous thread, but the exact age of the Celtic language family and its arrival in the British Isles are important to consider. I mentioned a archaeological study I had found which focuses on genetic inflow from Bronze Age Continental Europe to the British Isles. One archaeological sample came from a man (I21306) with an Indo-European haplogroup (R-P312) who likely lived between 2200 BC and 1400 BC. This raises interesting implications, and allows for one of three possibilities.

• He represents an earlier Celtic migration into Britain that took place during the Bronze Age.

• He represents an Indo-European migration predating the arrival of Celts into Britain.

• He represents a pre-Indo-European population whose language likely influenced the Insular Celtic languages (and perhaps adjacent languages in Continental Europe as well, depending on the size and scope of the language family.)

Personally, I am inclined to agree with Salmoneus in the sense that we have no way of knowing that the features Insular Celtic languages are known for (consonant mutations, etc.) were innovated within 200 years. Rather, it could be the byproduct of phonological processes that have existed in the Italo-Celtic languages for millennia. As a Spanish speaker, I am well aware of how the pronunciation of words can be affected by phonological boundaries. For example, I could say [buh.ko] “I find”, but then [lo βuh.ko] “I find it”, with the labial plosive being weakened to a labial fricative or approximant.

This phonological feature has certainly existed in Romance languages for well over a millenium, but it has not resulted in consonant mutations akin to the Insular Celtic languages. For whatever reason, Italic languages lost the conditions necessary to evolve productice consonant mutations. This could be due to an older substrate, but who really knows? The fluidity of word boundaries likely allowed for consonant mutations and inflected prepositions & particles within Insular Celtic, but this trait was lost among their Italic counterparts in continental Europe. If the continental Celtic languages (such as Gaulish, Gallaecian, etc.) had survived, it is possible that they could have developed consonant mutations just as their living counterparts did. Alternatively, they might have been influenced by the Romance languages of Western Europe, never developing these features as a result. In this case, they would more resemble Spanish, French, and Portuguese in terms of phonology and grammar than they would to Welsh, Irish, and Breton. Both are real possibilities, but I am inclined to think that consonant mutations were a phenomenon unique to Insular Celtic, and the farther one goes from the British Isles, the less likelihood there is of a language evolving these features. After all, no languages in continental Europe today (with the exception of Breton) demonstrate productive consonant mutations to this extent.

Ultimately, I believe that Insular Celtic languages evolved differently from their Italic counterparts in Europe simply because of the archaic phonological processes mentioned by Salmoneus. The reason why these processes had a much more extensive impact on the evolution of Insular Celtic languages is largely unknown. It could very well be due to a pre-existing substrate in the British Isles that made Celtic speakers in the region more inclined to retain these processes. The reverse could also be true. Perhaps a linguistic substrate across the English Channel influenced the languages of Western Europe to drop these features, preventing them from developing consonant mutations in the first place.

[Nortaneous]

In addition to intervocalic lenition there's syntactic doubling. In Tuscan, la casa [laˈhasa] but a casa [akˈkasa]. If Tuscan had lost /l/ from the article like Portuguese, initial mutation would be contrastive.