Wiktionary talk:About Proto-Celtic

Notation

I'm pretty sure that when I first created Proto-Celtic entries, the diphthongs ended in u and i and the letter j was used instead of y. Why was this changed? —CodeCat 20:44, 1 November 2014 (UTC)

I'm pretty sure no modern English-language sources use "j" for Proto-Celtic, though some German-language sources still may. Diphthongs end in w and y to match both Matasović 2009 and PIE entries. —Aɴɢʀ (talk) 21:09, 1 November 2014 (UTC)

But for PIE there's a good reason to use w and y; it's to show ablaut and the fact that they were phonologically consonants. That doesn't apply for Proto-Celtic, *ai was one phoneme and not two. —CodeCat 21:13, 1 November 2014 (UTC)

Really I just wrote down the w/y forms because looking through the existing Proto-Celtic entries that contained diphthongs, far more of them were written with w/y than with u/i. I don't know who wrote them that way; I may have written some of them but I don't think I did most of them. I don't really care which way we write them as long as we're consistent, and moving the few that used i/u is considerably less work than moving all the ones that use y/w. Ablaut is a very bad reason to use y and w in PIE, though, since ablaut is more clearly seen in "ei ~ oi ~ i" than in "ey ~ oy ~ i". —Aɴɢʀ (talk) 21:34, 1 November 2014 (UTC)

I looked through some of the older entries. Apparently User:Victar renamed them all, then started adding more in "his" format. I think they should be moved back. —CodeCat 21:40, 1 November 2014 (UTC)

First off, there were only a few Proto-Celtic entries to begin with before I added the majority, maybe 2 or 3 with u/i. Secondly, w/y and u/i were used interchangeable in entries, largely English etymologies. I went with w/y because it's what used in the most recent published literature, namely Matasović (2009), which is also the source of most of the reconstructions. To use u/i is to actually go against the cited source. Victar (talk) 22:44, 1 November 2014 (UTC)

We're not required to follow Matasović's notational conventions to the letter. I still prefer u/i. —CodeCat 23:13, 1 November 2014 (UTC)

We're not required but I see no reason to break from it just because it isn't to your preference. Victar (talk) 23:24, 1 November 2014 (UTC)

But the same applies to you too doesn't it? —CodeCat 23:36, 1 November 2014 (UTC)

The difference is I'm not pushing any personal preference. I was simply adding new entries using the convention of the source of my entries, which is the most recent and complete reconstructed dictionary of Proto-Celtic. Victar (talk) 23:51, 1 November 2014 (UTC)

I checked some other sources; Kim McCone and Peter Schrijver both use i/u. And we don't follow Matasovic for ɸ, since he uses f. Consistency with PIE is important too, though, in my opinion; if we say Proto-Celtic *oinos came from PIE *óynos it will look like more changed between PIE and PC than actually did. —Aɴɢʀ (talk) 23:57, 1 November 2014 (UTC)

But we use u/i with most other PIE descendants. Germanic, Italic, Balto-Slavic, and Hellenic. —CodeCat 00:52, 2 November 2014 (UTC)

As I said, as long as we're consistent I really don't care one way or the other. It is a matter of zero importance. Both approaches have their advantages and disadvantages. If there were a way to do it online, I'd be willing to settle it by coin toss. —Aɴɢʀ (talk) 00:55, 2 November 2014 (UTC)

I just don't see enough justification to change things based on your personal preference. Maybe when a new dictionary is published with i/u we can revisit this, but for now, the consensus appears to keep it as is. Victar (talk) 01:18, 2 November 2014 (UTC)

All I see is no consensus for either state of affairs. You want it one way, I want it another, and Angr is indifferent. So we need more opinions. —CodeCat 01:55, 2 November 2014 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ So who else works on Proto-Celtic that we can ping? User:Word dewd544 and User:Ivan Štambuk, perhaps? —Aɴɢʀ (talk) 16:11, 2 November 2014 (UTC)

I don't work on Proto-Celtic entries. However, I'm in favor the most widely used reconstruction system, whatever that is. --Ivan Štambuk (talk) 17:01, 2 November 2014 (UTC)

There isn't one though. At least not any more than for PIE. In PIE some sources use ey/oy, some use ei/oi, some use y/w, some use i̯/u̯, some use ḱ/ǵ, some use k̂/ĝ, some use bh/dh/gh, some use bʰ/dʰ/gʰ, and other things like that. For Proto-Celtic, some sources use f, some use ɸ, and there probably are sources that use oi while others use oy. My own preference is for oi, because this was a single diphthong phoneme in Proto-Celtic, whereas it was a vowel followed by a consonant in PIE. I would also prefer f for ease of typing, but I don't feel as strongly about it. —CodeCat 17:07, 2 November 2014 (UTC)

I agree; there isn't one "most widely used reconstruction system". Matasovic uses f and ow/oy; McCone and Schrijver use ɸ and ou/oi; German authors tend to use ou̯ and oi̯. But as far as I can tell, no author follows CodeCat's suggestion of using one system for PIE and another for PC; everyone uses either ow/oy for both PIE and PC, or ou/oi for both PIE and PC, or ou̯/oi̯ for both PIE and PC. And seriously I don't think any reader, not even a reader with advanced knowledge of historical linguistics, is going to look at oy and think "that's a vowel followed by a consonant" and then look at oi and think "that's a monophonemic diphthong". —Aɴɢʀ (talk) 17:49, 2 November 2014 (UTC)

Can we please move back to using oi/ou and such? All other reconstructed PIE descendant languages already use this notation, except for Proto-Indo-Iranian. Most importantly, Proto-Italic uses it, but also Proto-Balto-Slavic, Proto-Hellenic and Proto-Germanic. I find the current notation nonsensical for a language that didn't have syllabic sonorants the way PIE does. The y in the diphthong oy is not a consonant in Proto-Celtic, it's the second part of a diphthong. This is different from PIE and plenty of attested non-PIE languages such as Slavic and Turkic where it is a consonant, and thus writing it with the consonant letter is appropriate. —CodeCat 22:15, 9 October 2016 (UTC)

I still don't mind one way or the other. @Victar? —Aɴɢʀ (talk) 15:16, 10 October 2016 (UTC)

I still disagree. I think they should be inline with the sourced materials. --Victar (talk) 15:31, 10 October 2016 (UTC)

So you think we should also be using f instead of ɸ wherever Matasovic uses it, and use different notations for our entries when different sources have different notations? —CodeCat 15:32, 10 October 2016 (UTC)

As I said above, different sources have different conventions. McCone and Schrijver do use oi/ou, as well as ɸ. We definitely shouldn't be following different conventions based on whichever source one editor happens to be using; we need conventions of our own that we stick to. —Aɴɢʀ (talk) 15:43, 10 October 2016 (UTC)

McCone and Schrijver are 20-30 years old and are very limited in comparison. We wouldn't even use many of the PIE reconstructions. Matasovic is the most comprehensive and up-to-date source, and is thus what is used to cite the entries themselves. --Victar (talk) 18:43, 10 October 2016 (UTC)

I'm pretty sure they're both still publishing. Matasovic is fine as far as he goes, but there's a lot of stuff (e.g. regarding inflections) that he simply doesn't cover, nor his he infallible. We are perfectly entitled to add words that aren't even in his dictionary, provided they're sourced to other reputable scholars. —Aɴɢʀ (talk) 20:53, 10 October 2016 (UTC)

I don't think either has revised their phonology of Proto-Celtic since. Matasovic certainly isn't infallible; he isn't some supreme being. =P I'm just saying that when most entries are added, they're based on, if not taken directly from, his dictionary. The only alternative, I know, is from the University of Wales, and it's older, isn't peer reviewed, and isn't published. If you rather use ai/oi, then I'll concede to the majority, but I just don't find the change warranted enough. --Victar (talk) 22:24, 10 October 2016 (UTC)

What do @JohnC5, Anglom think about this? —CodeCat 22:48, 10 October 2016 (UTC)

ɸ and ou/oi. —JohnC5 23:50, 10 October 2016 (UTC)

Hah, you need to elaborate more than that (person I only heard of this month). --Victar (talk) 00:02, 11 October 2016 (UTC)

JohnC5 is a longstanding and trustworthy editor of PIE and Celtic entries, among other things. That you haven't seen him before says more about you, to be honest. —CodeCat 00:17, 11 October 2016 (UTC)

I don't work at all in the PIE entries, but if I'm not mistaken, JohnC5, with all due respect, has never created a PCelt entry (from what I can see), or edited a entry I've created (until this month). So you can understand my bewilderment. --Victar (talk) 00:22, 11 October 2016 (UTC)

Thanks for the defense, Code. Here some edits I've made: diff, diff, diff. I'll admit that Celtic has not been an strong interest of mine until recently, but this is partially to do with the lack of infrastructure (templates) and activity in Proto-Celtic until recently. I do frequently reconstruct PC lemmata in PIE entries with the hope that others will come and do the dirty work of fleshing out the PC entries (I do the same with Italic, Germanic, Hellenic, Balto-Slavic, etc.).

As to the question at hand, I find it singularly frustrating having to look up whether a particular protolanguage on Wiktionary uses ai or aj every time I point to an entry. The argument about the PC's diphthongs vs PIE vowels followed by resonants I find quite cogent and thus support ai. ɸ vs. f is less important to me.

I think Matasović is most useful as an aggregator of reflexes and PC lemmata and less useful as an etymologist. I have found his PIE reconstructions to be often incomplete, speculative, or totally out of line with other reputable scholars. —JohnC5 01:50, 11 October 2016 (UTC)

Same as JohnC5. I wouldn't mind using ⟨f⟩ if /ɸ/ had shifted to it as it did in Germanic, but ⟨ɸ⟩ seems fine. Anglom (talk) 00:38, 12 October 2016 (UTC)

In my personal experience, I tend to skip over an unfamiliar letter when reading, even if the letter is far from unknown to me, it's just "out of place" as it doesn't look like any Latin letter. My instinct to recognise the letter and sound it out in my head is not as strong. In this case, it helps, because the letter went silent later. The same thing also helps with the yers in Slavic. —CodeCat 01:03, 12 October 2016 (UTC)

I would absolutely support changing entries from /ɸ/ to /f/. --Victar (talk) 18:43, 10 October 2016 (UTC)

Well, I think the majority has it. CodeCat, could you change over the Gaulish descendants too? Probably need to run a bot to fix all the etymology links in the various descendant entries. --Victar (talk) 02:15, 12 October 2016 (UTC)

I'll edit the About page and start moving pages now. —Aɴɢʀ (talk) 09:51, 18 October 2016 (UTC)

The only one I'm not moving is *wewlos since PC didn't have a diphthong *eu. Matasović says maybe this is a reduplication that occurred after the change of ew > ow was complete in PC, but since the word has no known cognates outside Celtic anyway, I don't see why it couldn't be reconstructed *weblos. —Aɴɢʀ (talk) 12:24, 18 October 2016 (UTC)

• All existing PC pages, and the links to them, now use āu/āi/au/ai/ou/oi. There are probably lots of red links still using the w/y spellings, though. —Aɴɢʀ (talk) 14:00, 18 October 2016 (UTC)

Notation revisited 2019

Is there someone reason why we're using ɸ and not φ, as in all PC sources that use phi? @Rua --{{victar|talk}} 22:54, 26 August 2019 (UTC)

Because it's the IPA symbol. —Rua (mew) 08:08, 27 August 2019 (UTC)

@Rua: Phonetic transcription ≠ orthographic transcription -- we're not writing *ɸlaːnos either. If I Google search for *φlānos, the standard form of transcription, I don't get *ɸlānos. Unless there is any good reason we're not using φ, I'm going to push for moving those entries to be in line with standard transcription. @Mahagaja, JohnC5 --{{victar|talk}} 16:20, 27 August 2019 (UTC)

I don't care, so long as the symbols are consistent. —*i̯óh₁n̥C^[5] 18:44, 27 August 2019 (UTC)

@Victar, JohnC5: I don't care either, but I wonder what makes Victar so sure that all PC sources that use phi use the Greek character rather than the IPA character. In most fonts, they look identical in print, don't they? —Mahāgaja · talk 17:44, 31 August 2019 (UTC)

@Mahagaja: not in their cursive form, which is distinctive, as seen in {{R:cel:KPV}}, but also many resources are unicode encoded these days, like {{R:ine:HCHIEL}} and Stifter (2012) Old Celtic Languages^[1]. --{{victar|talk}} 18:02, 31 August 2019 (UTC)

@DTLHS is this something you could do with your bot? @Erutuon put these together. --{{victar|talk}} 23:09, 13 September 2019 (UTC)

Yes, if you're sure that's what you want. DTLHS (talk) 23:18, 13 September 2019 (UTC)

@DTLHS: Unless anyone else has some last minute objections, please. --{{victar|talk}} 00:35, 14 September 2019 (UTC)

Victar alerted me to this by asking me to search for instances of ɸ in link templates, and my opinion as an outsider is that the published sources ought to use ɸ, which Unicode categorizes as Latin, rather than the Greek character φ. But I don't have any authority and I guess using the Greek character is a good choice practically to allow our entries to be found, since apparently search engines don't treat ɸ as equivalent to φ. — Eru·tuon 22:00, 13 September 2019 (UTC)

We use θ even though it's the same letter used for Greek. --{{victar|talk}} 23:14, 13 September 2019 (UTC)

But that's different; Unicode doesn't provide a separate IPA theta from the Greek theta, but it does provide separate phis for IPA and Greek. Anyway, I don't care if we shift, as long as the redirects remain so that people can search for either character. —Mahāgaja · talk 08:15, 14 September 2019 (UTC)

Actually, scratch that. I do think we should continue using the IPA phi, since it's defined as a Latin letter and Proto-Celtic is written in the Latin alphabet, but we should add redirects from forms using the Greek phi, so that people searching using that letter will still find the entries. —Mahāgaja · talk 08:17, 14 September 2019 (UTC)

That's not a solution. That may work for page titles if you manually create Greek phi versions of each page, but even then, the pages would probably suffer in Google search rankings for just being redirects, if Google indexes them at all. It wouldn't help in the slightest for links within pages. -- {{victar|talk}} 13:52, 14 September 2019 (UTC)

@Erutuon, Victar, Mahagaja So could you summarize what the consensus is and what you want me to do, if anything? DTLHS (talk) 04:05, 19 September 2019 (UTC)

At this point I think the consensus is there isn't a consensus. —Mahāgaja · talk 11:05, 19 September 2019 (UTC)

@Mahagaja: You're the only person left objecting. Can you give an updated rational as to why? --{{victar|talk}} 13:26, 19 September 2019 (UTC)

Because Unicode defines ɸ as a Latin letter, and we write Proto-Celtic in the Latin alphabet. And Wikimedia projects do try to use Unicode characters correctly, even when most other websites don't. For example, we make a concerted effort to use ʻ (U+02BB modifier letter turned comma) and ʼ (U+02BC modifier letter apostrophe) correctly, even when other websites use ‘ (U+2018 left single quotation mark) and ’ (U+2019 right single quotation mark) or even ' (U+0027 apostrophe) for them. I seem to recall a lengthy discussion some years back about not using the modifier letters that look like superscript letters for entries where some of the letters are usually written superscript, e.g. French 1er not 1ᵉʳ, because that would be using modifier letters in a way other than they were intended to be used. And ɸ is intended to be used in Latin-alphabet contexts (primarily but not exclusively the IPA), so we should use it there. —Mahāgaja · talk 13:51, 19 September 2019 (UTC)

@Mahagaja: And as I pointed out, we also use both θ and β, which are encoded in Unicode as Greek letters, not Latin, so your objection to using Greek within Latin text falls rather flat. ɸ was created with the intention to be used in IPA transcriptions, so saying we're using it as it was "intended to be used" is also not true. I also doubt ₁ ₂ ₃ are being used as "intended", but here we are in PIE using them. --{{victar|talk}} 14:51, 19 September 2019 (UTC)

@Victar: And as I pointed out, Unicode hasn't provided a Latin theta or a Latin beta, so we have no choice there. But where we do have a choice, we should take it. If there were a Latin theta or a Latin beta in Unicode, I would be arguing for their use too. And, though this is difficult to prove, I strongly suspect the use of phi in Proto-Celtic reconstructions is derived from the use of phi in IPA, not the use of phi in Greek. —Mahāgaja · talk 14:58, 19 September 2019 (UTC)

@Mahagaja: You are wrong in that suspicion. See my links above that employ Unicode Greek phi in their PC reconstructions. Greek phi and IPA phi are very distinct in their lowercase form. I've yet to find a single PC reconstruction that use IPA phi. --{{victar|talk}} 15:07, 19 September 2019 (UTC)

Vote for changing ɸ to φ in Proto-Celtic

• Support: If φ is the standard the academic community has chosen to go with, we should follow suit, per both the academic and technical reasons voiced above. --{{victar|talk}} 14:51, 19 September 2019 (UTC)
• Oppose: I'm far from convinced that the scientific community has explicitly chosen to go with φ (Greek phi) as opposed to ɸ (Latin phi) in the first place (i.e. where Greek phi is used in other works, it's probably out of ignorance or apathy rather than an explicit rejection of Latin phi), nor am I convinced that using Greek phi in a Latin-alphabet context is consistent with the principles of Unicode. —Mahāgaja · talk 15:03, 19 September 2019 (UTC)

  I've listed sources above that employ Unicode Greek phi in their PC reconstruction, which are distinctive in their lowercase form. I challenge you to demonstrate a single source that uses IPA phi in their PC reconstructions. --{{victar|talk}} 15:11, 19 September 2019 (UTC)

  I didn't say they don't do it; I said they don't do it deliberately. They pick the Greek phi because they don't care about the difference, not because they've made a conscious decision to avoid the Latin phi. —Mahāgaja · talk 15:14, 19 September 2019 (UTC)

  If you're using modern Unicode, you have a choice and they made that choice. You're also choosing to make the choice to use IPA phi, which has major technical disadvantages. --{{victar|talk}} 15:17, 19 September 2019 (UTC)

Nominative singular of dental stems

@Angr, Anglom, JohnC5 As of right now, we have a bunch of dental stems with -ts and another bunch with -ss. We should presumably settle on one type and use it everwhere. If I understand the ending -ss correctly, it is the result of assimilation ts > ss, which also occurred in the middle of words, and has parallels in Latin and Greek. However, can we be sure that the gemination was contrastive and/or reconstructable? Is the assimilation itself even reconstructable word-finally, or is it just assumed by analogy with the medial outcome? In Germanic, the original situation is unrecoverable, so we settled for just sticking -z/-s onto the stem as we can, at least, be sure of the ending. —CodeCat 21:56, 22 September 2016 (UTC)

I'm not sure to what extent a distinctively geminate -ss is provable for PC. Even if the nominative singular of such nouns is attestable in some Continental Celtic language, it's probably spelled with just a single s. But in other cases, e.g. *rīxs, we do show the sound change that a consonant undergoes before the s, so I feel like moving *neɸūss to *neɸūts might entail moving *rīxs to *rīgs, which is a step I'm not sure I'm ready to take. —Aɴɢʀ (talk) 11:18, 23 September 2016 (UTC)

I agree with you there, but what about *neɸūs? —CodeCat 14:45, 23 September 2016 (UTC)

I can live with that. (Sorry for the late answer, I only just noticed your response.) —Aɴɢʀ (talk) 15:17, 10 October 2016 (UTC)

I think I'm fine with either -s or -ss. We can't tell which and -ss seems a bit unlikely, but it does have the advantage of immediately revealing that the word is a dental stem. —CodeCat 13:50, 18 October 2016 (UTC)

*-sk(y)os, *-sk(y)ā suffix in Proto-Celtic

What can we say about the noun suffix *-sk(y)os, *-sk(y)ā in PCelt? Does it derive from some possessive PIE *-sḱo- suffix? Was it still functioning in PCelt in some capacity? @Angr @CodeCat @JohnC5 @Anglom --Victar (talk) 04:04, 17 October 2016 (UTC)

I can't think of a word that has it. Our PIE entry is at *-iskos, which would give -esc in Old Irish, -easc in Modern Irish, and masculine -ysg/feminine -esg in Welsh. I'm unaware of any such suffix. Maybe there's something in Continental Celtic, though. —Aɴɢʀ (talk) 07:09, 17 October 2016 (UTC)

Words in PCelt? Here are a few: *udenskyos, *talskos, *ruskos, *ganskyos, (possibily) *ɸēskos. --Victar (talk) 08:01, 17 October 2016 (UTC)

*-sḱéti is the only suffix in PIE that I know of containing -sḱ-. Celtic could have formed a secondary suffix from PIE adjectives in *-ḱós (e.g. *h₂yuh₁n̥ḱós) with sigmatic stems or by some outcome of *-kḱ- > *-ksk- > *-sk-, but I'm not familiar with a productive -sḱ- suffix in PIE adjectives, though I could be wrong. —JohnC5 17:30, 17 October 2016 (UTC)

It's possible that some originate from the durative/iterative imperfective verb suffix *-sḱé-, as in Proto-Celtic *miskos (“mixed, confused”) (< *mikskos) < Proto-Indo-European *miḱ-sḱé- (“to mix”), but it's harder with words from noun roots like Proto-Celtic *udenskyos (“water”). --Victar (talk) 19:25, 17 October 2016 (UTC)

Here are my best guesses for PCelt reconstructions based on a *-sḱé- durative/iterative imperfective verb to a deverbal noun:

• *udenskyos (“water”) < *udén-sḱ-yos (“stream water, spring water”, literally “(water) of a stream, spring”) < *udén-sḱé- (“to stream-out, to spring-forth (continuously)”) < *wódr̥ (“water”)
• *talskos (“hulled grain”) < *tl̥k-sḱ-os (“pounded material”) < Proto-Indo-European *tl̥k-sḱé- (“to pound, strike (repeatedly)”), from *telk- (“to thrust, strike”)
• *ganskyos (“plough, branch (for ploughing)”) < *ḱn̥k-sḱ-yos (“plough”, literally “(implement) of ploughing”) < Proto-Indo-European *ḱn̥k-sḱé- (“to plough”), from *ḱonk- (“branch”)
• *ruskos (“(harvested) bark, beehive”) < *h₃rewk-sḱ-os (“something harvested”) < Proto-Indo-European *h₃rewk-sḱé- (“to harvest”), from *h₃rewk- (“to dig up”)

--Victar (talk) 11:02, 18 October 2016 (UTC)

*Koimogenan

@Angr, Victar What declension class does -an represent here? There's none that I know of other than the neuter -men-stems (like *anman), but this word lacks the m. —CodeCat 13:59, 18 October 2016 (UTC)--Victar (talk) 20:48, 20 October 2016 (UTC)

I was wondering that too when I moved this page earlier today. Matasović says *genan n (whence Old Irish gein n (“birth, conception”)) is an archaic neuter n-stem from *ǵenh₁en, but I have trouble believing a masculine name would keep that declension. As ever with Proto-Celtic personal names, I also have to wonder what evidence there is that this name is of Proto-Celtic date. If it isn't attested in Brythonic or Continental Celtic, it may have been coined only in Old Irish. —Aɴɢʀ (talk) 14:09, 18 October 2016 (UTC)

This just shows how much of a bint you are. UtherPendrogn (talk) 14:27, 18 October 2016 (UTC)

I think this is an issue with many of our Proto-Celtic proper noun entries currently. I doubt all of them existed in Proto-Celtic already. I think a good criterium would be if a name is found in both continental and insular Celtic. —CodeCat 14:32, 18 October 2016 (UTC)

I.E words should only exist if CodeCat approves of them. UtherPendrogn (talk) 14:40, 18 October 2016 (UTC)

I have no problem merging this into the OI entry, along with the various others I created with no bilateral cognates to their respective root entries. I'll probably need help with the Brythonic forms though. -- Victar (talk)

Does Cóemgen#Etymology look fine, given we're deleting Reconstruction:Proto-Celtic/Koimogenan? Ideally, I could have just used {{compound}} or something, but neither have OI entries. --Victar (talk) 01:39, 19 October 2016 (UTC)

Looks fine to me, until someone gets around to creating entries for cóem and gein. —Aɴɢʀ (talk) 07:01, 19 October 2016 (UTC)

I was wondering, should I create the entries like *Adyantumāros, or should I give the reconstruction on an entry of the attested form? And if I did the latter, the Gaulish attestations are really Gallo-Latin, so should they belong to Latin, with Latin declensions? --Victar (talk) 05:52, 20 October 2016 (UTC)

I've actually always been hesitant in creating Gaulish entries because I'm never sure on how enter them. For instance, we have an entry for the attested boduus. We also have the attested prefix form boduo-, so we can rather confidently conclude the form boduos. Now is that a reconstruction? To take it a step farther, can we assume that the -uo- is -wo-, simply confined to a limited foreign alphabet, and conclude the form bodwos? Is that a reconstruction? --Victar (talk) 06:14, 20 October 2016 (UTC)

Yeah, I tend to avoid Gaulish too, for precisely the same reason. It can be hard to tell the difference between a word attested as Gaulish and a Gaulish loanword in Latin, especially when it comes to proper nouns like personal names and place names. My inclination is to focus on what's attested and put as much as possible in an etymology section, so rather than creating Gaulish *Adyantumāros, it's probably better to create Latin Adietumarus with an etymology section that says something like "From Gaulish, from Proto-Celtic *X + *Y + *Z". As for boduus, is it attested within a Gaulish text, or as a Gaulish word mentioned in a Latin text? —Aɴɢʀ (talk) 08:25, 20 October 2016 (UTC)

Thanks. boduus and boduo- are both from names, from what I understand. --Victar (talk) 14:49, 20 October 2016 (UTC)

Then again, I'd create entries for what's attested and put everything else in etymologies. —Aɴɢʀ (talk) 17:37, 20 October 2016 (UTC)

I actually feel more comfortable creating reconstructed Gaulish entries than labeling a Latin interpretation of a Gaulish name as Latin, and applying Latin inflection to it. That way I can say hey, this is out best guess, take it as you will. Here are some moves I've done: Katurīxs, Arianrhod, Brangen. Let me know if you see any problems with them before I do the same for the rest I've entered in PCelt. Thanks! --Victar (talk) 20:48, 20 October 2016 (UTC)

They appear in Latin texts, and would presumably be inflected as Latin words. For example, if any of the names appeared in a context that called for a Latin dative, I have no doubt that the Latin ending for the dative would be used, not the Gaulish one. —CodeCat 19:22, 26 October 2016 (UTC)

I was wondering if nouns and adjectives count as attested cognates? For instance Gaulish Segomaros is perfect cognate for OI segmar ("strong, vigorous"). Is this a pattern of OI forming nouns and adjectives from personal names? --Victar (talk) 03:52, 21 October 2016 (UTC)

If anything, it's the other way around: Gaulish using an ordinary adjective as a personal name. —Aɴɢʀ (talk) 13:38, 21 October 2016 (UTC)

Ah, I see -mar is a denoninal adjective suffix in OI, so segmar is actually unrelated. I'll look into the others I noticed. --Victar (talk) 14:31, 21 October 2016 (UTC)

@CodeCat, Angr, Anglom I've moved all the PCelt personal name entries I created to their highest common form with attested descendants. There are several that UtherPendrogn created which still need moving. --Victar (talk) 18:50, 26 October 2016 (UTC)

@Victar: Thanks for this! I also appreciate your marking the redirects for deletion so as to remove any confusion. —JohnC5 18:53, 26 October 2016 (UTC)

No problem. I'm a prolific deleter of my mistakes. ;) --Victar (talk) 19:09, 26 October 2016 (UTC)

Would you mind checking Uther's too? They do need to be done, even if none of us wants to deal with him. —CodeCat 18:58, 26 October 2016 (UTC)

I'd be happy to, but some just deleting, like Reconstruction:Proto-Celtic/Akkalonos and Reconstruction:Proto-Brythonic/Akkalon. --Victar (talk) 19:09, 26 October 2016 (UTC)

Accalon is attested. Don't suppose you give a fuck. UtherPendrogn (talk) 19:47, 26 October 2016 (UTC)

Attested? Attested in what, Proto-Brythonic, Proto-Celtic? All I see is the Welsh form. --Victar (talk) 19:58, 26 October 2016 (UTC)

You know full well reconstructed languages aren't attested. I've formed a comparative reconstruxisction which is correct. Sorry you don't like that. Oh wait, I'm not. UtherPendrogn (talk) 20:22, 26 October 2016 (UTC)

We have attested forms of Proto-Brythonic and even pre-Proto-Brythonic. Your reconstruction isn't of concern; the issue is reconstructing given names that may have never existed. --Victar (talk) 20:41, 26 October 2016 (UTC)

What you've done is taken a name attested in one single language and reconstructed it all the way back to Proto-Celtic, but there's no evidence it was in use as a Proto-Celtic name. It's like taking the modern English name River and constructing a Latin first name *Riparius from it, regardless of the fact that Riparius wasn't a name in Latin. It's bad linguistics. —Aɴɢʀ (talk) 21:00, 26 October 2016 (UTC)

Proto-Celtic *Akkalonos would have given Welsh *A(c)halon anyway, as voiceless geminates become voiceless fricatives (ie. *katt- > *kaθ > cath). Anglom (talk) 22:39, 26 October 2016 (UTC)

Proposal to change PC notation ē to ei

@Caoimhin ceallach @Djkcel @Mahagaja @Silmethule @Erisceres I would like to change all instances of Proto-Celtic *ē to *ei. There's quite a few reasons to make this change:

• More convenient to type, obviously.
• Makes e-grade vs. zero grade ablaut more transparent (ei-i), which really shows in s-subjunctives and derivatives in *-om where such root ablaut alternations appear in multiple roots.
• Projecting *ē to Proto-Celtic requires the assumption that the monophthongization happened before even Celtiberian split off, which is not widely accepted at all. Stifter, Villar, Schmidt and Wodtko reject total pre-Celtiberian monophthongization outright, while I have seen Irslinger, Schrijver, KPV, Prosper and Koch notate their Proto-Celtic forms with *ei as if the monophthongization had not happened.

— Ceso femmuin mbolgaig mbung, mellohi! (投稿) 20:51, 3 September 2024 (UTC)

@Caoimhin ceallach, Djkcel, Mahagaja, Exarchus, Silmethule, Erisceres pinging you again in case my ping chain above didn't process (it's buggy, we all know). — Ceso femmuin mbolgaig mbung, mellohi! (投稿) 20:54, 3 September 2024 (UTC)

* Support. I'm always for clarity and consistency. DJ K-Çel (contribs ~ talk) 21:00, 3 September 2024 (UTC)

No objection from me. It also provides a parallel with *ou, which we do not transcribe *ō. Also, since the Proto-Brythonic reflex is itself a diphthong (*uɨ), I've often found myself wondering if there was any monophthongization in Proto-Celtic at all. The monophthongization in Gaulish and Goidelic could have happened independently. —Mahāgaja · talk 21:08, 3 September 2024 (UTC)

I don't object. The fact that most scholars use PC *ei is important for me. Is McCone the only one who prefers *ē? (He argues in Relative Chronology I.3.6 that Celtiberian <ei> represented /eː/ or /i:/.)

What about *ē < *en/_s? I don't think we currently consider this a PC development (judging for the gen.sg. at Proto-Celtic *anman), but an acc.pl. like Celtiberian tekametinas makes this likely actually. —Caoimhin ceallach (talk) 22:30, 3 September 2024 (UTC)

Matasović uses ē as well. How does tekametinas make the case for *ē < *en/_s? If anything, it looks like it has *ās < *ans. —Mahāgaja · talk 07:22, 4 September 2024 (UTC)

The assumption is that it's the same sound law: *Vn > *V̄/_s. —Caoimhin ceallach (talk) 09:59, 4 September 2024 (UTC)

@Caoimhin ceallach: The fact that Old Irish lés < *ɸlenstus and Middle Irish fés < *wensos both have é rather than ía before a broad consonant suggests that ē < *e/_ns did not merge with PC *ei. The fact that fés still has its s also suggests that the loss of n before s with compensatory lengthening occurred after lenition of intervocalic s and therefore probably didn't happen until Proto-Insular Celtic. In addition, Old Irish fíal, píast, ríagol, síans and srían all show that Latin ē in loanwords did merge with PC *ei, suggesting that *ens > *ēs didn't happen until after the incorporation of these loans. —Mahāgaja · talk 17:34, 28 September 2024 (UTC)

The sound change I mentioned only applies at the end a word *Vn > *V̄/_s#. Sorry for the omission. You're right though about yours, the result doesn't merge with *ē from *ei, but is conventionally written *ε̄. —Caoimhin ceallach (talk) 18:37, 28 September 2024 (UTC)

I deliberately looked for cases in the first syllable because at the end of a word, there's no way to tell whether they merged or not. —Mahāgaja · talk 18:51, 28 September 2024 (UTC)