[Spellyans] SWF (t) and Maga web site

Daniel Prohaska daniel at ryan-prohaska.com
Fri Aug 10 15:43:47 BST 2012

On Aug 9, 2012, at 10:24 PM, Michael Everson wrote:

> On 9 Aug 2012, at 15:14, Daniel Prohaska wrote:
>>> *Piw rather than pyw,
>> I respect your sentiment, but there are some things which aren't so clear cut. The graphs ‹i› and ‹y› and were interchangeably used in traditional Cornish and it is only sensible to assign specific functions or positions to ‹i› and ‹y› in Revived Cornish. You have done this for KS, and the SWF has done it, too, albeit differently. Since in traditional Cornish ‹y› and ‹i› were used interchangeably I have no problem with writing ‹iw›. 
> It is true that we distinguish these vowels in monosyllables as in ‹gwin› [ɡwiːn] and ‹jyn› [dʒɪn], and that the traditional language did not divide the letters according to this function. (Please do not quibble about my using KS spellings here.)

Dear Michael, 
Not quibbling. 
So, you have assigned a new function to the letters ‹y› and ‹i› in KS. This is good and very useful. It is not authentic in the strictest sense of the word. 

> Nevertheless we do not distinguish ‹pyw› 'who' and ‹pyw› 'owns' because there is no pronunciation difference between them: [piʊ], [piʊ]. 

I don't understand… The word for 'owns' is given as ‹pew› in "Desky Kernowek"… where do you get ‹pyw› from?
Where this word is concerned, it should definitely contain ‹ew›, not ‹yw›. ALL attestations show ‹ew› (or ‹ev› etc.). There is no occurrence of ‹yw›. This word was mistakenly spelt **piw in KK and is also wrongly assigned to the alternating ‹yw› ~ ‹ew› class in the SWF. It should be ‹pew› only. 

As for SWF ‹piw› 'who' or KS ‹pyw›… they essentially mean the same. It's just a convention whether you choose to spell them with ‹iw› or ‹yw›. Though the SWF isn't consistent about this yet, I wish to propose to formulate a rule for the review that "stable" /i/, i.e. word which have /i/ both in the MC and LC based varieties write ‹i›, while ‹y› is written in SWF/M ('Middle' here, not 'main form') where /i/ alternates with /e/, e.g. the ‹bys/bes›-words or ‹byw/bew›-words…; ‹ew› again would be used for stable /ew/ where both SWF/M and SWF/L show /ew/. 
Furthermore Lhuyd writes ‹piụ›, so a spelling with ‹i› has an historical precedent. 

Also, as we can see from Lhuyd's spelling of the ‹ew› and ‹aw› diphthongs, the nucleus of the diphthong long, so I phonologically treat them as VVw rather than Vw, as Nicholas does. 

Only BK, TH/SA and CW spell ‹piw› with ‹ew› or ‹eu›, but this was at a time, the second half of the 16th century and the beginning of the 17th century where there distinction between /iu/ and /eu/ in English had been lost and ‹ew› and ‹yw› could mean the same, and since ‹yw› was rarely used in English, the Cornish scribes were no doubt influenced by English and prompted to write the more frequent ‹ew›. In Late Cornish, after the breakdown of the Cornish scribal tradition the ‹i›-nucleus of the diphthong "resurfaces". The ‹u›-spellings also mean /iw/, as they do elsewhere, e.g. in diphthongised ‹du›... 

So, I don't believe your above statement to be correct that ‹piw› 'who' and ‹pew› 'owns' are pronounced identically, I believe we have /i:w/ : /e:w/ here just as we have /i:g/ and /e:g/ in ‹kig› 'meat' and ‹keg› 'cook'. 

>>> *melin rather than melyn,
>> The SWF has the rule of the etymological vowel. I don't think you will contest that Brythonic borrowed this word from Latin ‹molina› with */i/ rather than */I/. Since traditional Cornish wrote ‹y› and ‹i› interchangeably I don't see a huge issue, also considering the fact that the recommendations for pronunciation of the SWF specifically say that ‹i› and ‹y› are to be pronounced the same in unstressed syllables. 
> It is an issue because this "rule of the etymological vowel" for ‹i› and ‹y› serves no useful function whatsoever.

It did to a certain extent in KK which had to indicate the quality of the vowel when the unstressed /I/ vowel shifted into stressed position when, i.e. a suffix was added, e.g. 

‹kegin› => ‹keginys›

This was important for Ken George's phonology because phonetically such words would be [ˈkegɪn] => [kɛˈgiˑnɪs]. This explains why Ken George chose to distinguish between ‹y› and ‹i› orthographically. If he had chosen to write more phonetically he would have to have chosen ‹kegyn› ~ ‹keginys›. 

It's the same principle why you in KS write ‹tir› ~ ‹tiryow›… for [tiːr] ~ [ˈtɪɹjɔʊ]. You could just as well write KS **tyryow for the /i/ is short, but because the simplex ‹tir› has a long vowel, the orthographic choice of writing ‹i› is carried over into the plural. 

The same applies to the SWF. You have ‹conin› 'rabbit' because when stressed you get /i/ (cf. Pryce ‹kynîngen› 'rabbit skin'). 

I'm not entirely happy with this feature of the SWF, but there appears to be no other way than to decide on a case to case basis, when to spell ‹y› and where ‹i› according to what reappears when the unstressed vowel in the simplex appears as stressed with an added suffix, but basically it's the same principle as with KS ‹tir› ~ ‹tiryow›. 

Where I definitely agree with you is in the cases where there is no alternation with a stressed  vowel, i.e. in the verbal and prepositional pronoun endings. I see no reason why, e.g. vb.adj. ending ‹-ys› and 3.sg.pret. ending ‹-is› should be distinguished. 

So what I wish to propose for R13 is not so much to keep the "etymological vowel", but to look at how the pattern can be used in the revived language, i.e. if you have a word like ‹conin› with a "stable" /i/ in stressed reflexes such as ‹conina› or ‹coningen› then the word is spelt with ‹i›. If the reflexes have the alternating ‹y› ~ ‹e› pattern then SWF/M should have ‹y› and SWF/L ‹e›.  

> It simply invites learners to make a spelling error.

Who cares. Spelling errors in Cornish are the least of our worries, I think...

> Now in words like ‹colon› [ˈkɔlən], pl ‹colodnow› [kəˈlɔdoʊ]

I take it you meant [kəˈlɔdnoʊ]?

> we preserve a distinct written vowel because that vowel reappears under stress. And so for ‹melyn› [ˈmɛlən], ‹melynyow› [məˈlɪnjoʊ]. But I don't know of any words where in this pattern (polysyllables in -in or -yn) exists where [iː] returns under stress. In the pattern which we have observed, [iː] is always long, as in ‹despît› [dəˈspiːt], ‹despîtya› [dəˈspiːtjə]. 

I realise what you do, and the SWF basically does the same, without the diacritic length marker in ‹despit› which makes the pronunciation less predictable than KS where the long vowel before ‹p t k› in concerned. I would like to see this remedied but I doubt this will be possible because of the dislike of diacritics of many current Cornish users. As I said above there is room to move away from the strict Georgian etymological vowel, to a workable and applicable system within the context of the SWF, but more often than not, they coincide. 

> The only reason for writing SWF ‹colyn› 'puppy' and SWF ‹melin› 'mill' is that Ken George thought it was cool to know that one word has the origin of Common Celtic *koligno- and the other word has the origin of Latin molîna.

As I explained above, this wasn't the only consideration. It was the exact reverse from what you do with ‹tir› and ‹tiryow› where the simplex has [iː] but the suffixed form has [ɪ] usually spelt KS ‹y›, but here written ‹i› because of its being a derivation of ‹tir›. See above for why KK also writes thus…

(Just to be clear, I'm not defending KK here, just explaining it.) 

> But who the hell cares?

See above. Despite its many shortcomings and mistakes KK appeared simpler in word recognition and this is obviously what many people went for. Not that I agree with this, but that may be one of the reasons why KK had such an uptake when the Kesva and the Cowethas changed from UC to KK. 

> Most learners of Cornish don't have any Latin, and nobody can learn Common Celtic because it's entirely reconstructed. Now, etymology helped me learn Danish very quickly, because when I started I had English and Old English and German. I could apply sound changes in my head and got good results. I don't think that etymology is much of a help for learners of Cornish or Breton or Welsh -- well, that's not true, there are many things about some consonants and vowwls which can be rather helpful. Perhaps it would be more precise to say that I don't think that the etymological vowels of **these unstressed final syllables** helps anybody in the least bit.  

As I said, it's not just about the etymology alone, but also which vowel is written in derived forms. I'm sure there's plenty of room for improvement! 

>>> *menedh rather than meneth,
>> The spelling ‹menedh› is SWF/L, alternatively ‹mena›. The SWF/M has ‹menydh›. This spelling follows Lhuyd who was the only one to make distinction between the phonemes /D/ and /T/ in Cornish. He has several spellings with ‹dh›. 
>>> *genev rather than genef,
>> Again, Lhuyd has several spellings with final ‹v›. 
>>> *orthiv rather than orthyf
>> Lhuyd and Pryce have ‹orthiv›.
>>> for example, are not neutral. Nor are they authentic.
>> They cannot be said to be inauthentic in the light of the attestations. 
> Taking these three together, Dan, I have to say to you again that yes, it can be argued (as you always do) that there is some room for doubt if you take Lhuyd's spellings at face value.

I don't take Lhuyd's transcriptions at face value and I don't necessarily believe there wasn't a stressed/voiced : unstressed/unvoiced phonotactic relationship in traditional Cornish. However, the textual spellings don't give us very much to go on. 

The plosives are different from the fricatives in so far as there never was a clear cut voiced : voiceless opposition in the first place. The texts show that even in stressed monosyllabic words there was no opposition between /k/ : /g/ or /p/ : /b/. The fricatives, however, have this opposition as you can see in ‹kôth› (Lh) for /θ/ and ‹bêdh› (Lh) for /ð/. The native writers always wrote ‹th› for both /θ/ and /ð/; so from their writings we cannot determine when, where or how the two were distinguished, but it seems plausible to assume that the distinction was valid throughout the development of Cornish as a living language. The cognates of C ‹coth› are W ‹coth› 'old man' and B ‹kozh› 'old', as well as Gaulish ‹cottos› preserving the original Celtic *kott-o-s. By the same token original British /ð/ continues to be distinguished in C ‹bedh› 'grave, tomb', cf. W ‹bedd› and B ‹bez›, going back to CC *bed-o-. 

Yet the Middle Cornish scribes wrote ‹coth› and ‹beth› with no way for us of knowing whether they heard or pronounced /θ/ or /ð/ and in which context. We have only Lhuyd to tell us that these two consonants were distinguished. 

The same holds true for the contrast between /θ/ and /ð/ in unstressed syllables. There is no way of knowing from the native writers which of these two phonemes was intended. In medial position PA sometimes wrote ‹ȝ› to indicate /ð/, so here we have independent proof that the etymon ‹menydh› 'hill' contained /ð/ at least in medial/intervocalic position, cf. PA.(170) ‹menyȝyow› 'hills', corroborated by Lhuyd's ‹menedhiou›. 
In this word Lhuyd writes the simplex with ‹dh› also. 

So even if Nicholas' theory of Cornish phonotactics after the prosodic shift seems nice an neat and ties the MC spellings (which don't distinguish word/final /f/ : /v/ and /θ/ : /ð/) it remains a theory only. Nothing more, nothing less. 

For us the relevant question is how to treat this in Revived Cornish. Why is distinguishing /θ/ : /ð/in ‹coth› and ‹bedh› any more authentic than distinguishing /θ/ and /ð/in ‹dalleth› and ‹menydh›? Just because it fits Nicholas' theory on Cornish phonotactics? If that is the case, then I'd prefer to err on the side of caution. Rather than following one man's theory, I'd write ‹th› and ‹dh› consistently, that is as it has already been chosen to distinguish /θ/ and /ð/ in RC at all. 

The principle of ‹menydh› pl. ‹menydhyow› can be regarded as the same with ‹tir› pl. ‹tiryow›. ‹i› is written in ‹tiryow› because the simplex has ‹tir›. The same rule can apply for ‹menydh› which has ‹dh› because the derived form has ‹menydhyow›. With ‹menydh› and ‹nowydh› you can chose to devoice them in speech if that is what you think the correct pronunciation is. 

It's really not that difficult.

> But then we come to the area of language structure and practical phonotactics.

I see no evidence that this was the case in traditional Cornish. Yes, the unstressed/unvoiced consonants are more frequent in the stops, but not universal and there's no way of knowing /θ/ : /ð/, both written ‹th› and /f/ and /v/ both written ‹f› or ‹ff› in final position. 

> We have offered -- based on our observation of what is in the texts -- a phonological model of Cornish where final consonants are very often voiced in stressed monosyllables, but which devoice in unstressed final position.

Final devoicing can have other reasons as well, which has been theorized by others. I'm not on anyone's side here. I'm not saying this particular linguist is right and the other is wrong. I want to find a sensible and consistent way of spelling RC. and in so doing accommodate the greatest number of users. I feel in the case of /θ/ and /ð/ this is best achieved by distinguishing /θ/ and /ð/ consistently as ‹th› and ‹dh› even in final unstressed position, as any final devoicing can be done by rule. Final voicing of ‹nowyth› can on the other hand not be predicted because the learner wouldn't know why to voice in ‹nowyth› but not in ‹dalleth›. So the best way would be to write ‹nowydh› and ‹dalleth› and allow for the speaker to decide whether to pronounce /ˈnɔwɪθ/ or /ˈnɔwɪð/.

Again, I see no reason in the argument that ‹dh› in ‹bedh› is any more or less authentic than ‹dh› in ‹nowydh›. 

> [ɡ] rag, teg, kig
> but
> [k] marhak, carrek, uthyk
> [b] mab, glëb, me a dÿb
> but
> [p] methêwnep, morrep, modryp
> [ð] sedh, bëdh, scoodh, rudh
> but
> [θ] gorseth, meneth, nowyth, arlùth
> [v] gwrav, nev, gov, cuv
> but
> [f] manaf, genef, orthyf (often drops to Ø)
> [z] bos, tas, in mes (original tad)
> but (at least earlier)
> [s] gwelys, gweles, myternes (original gwelet)
> We also think that the influence of final voiced ‹s› in English plurals is the cause of the general voicing of  final ‹s› in later Cornish. 

I find this highly unlikely. If Cornish writers identified the Cornish ending with English ending they would have spelt them accordingly, with ‹s›. No, I believe they wrote ‹z› because they pronounced and heard [z]. 

There's no way of knowing if MC distinguished /s/ : /z/, /f/ : /v/ or /θ/ : /ð/, in final unstressed position, but it seems odd that if in MC the general rule was that the stressed/voiced : unstressed/unvoiced phonotactic relationship was operational and that after the MC period in LC ‹bedh› and ‹nowydh›, as well as ‹coth› and ‹dalleth› "resurface". If indeed MC had had */beð/, */koð/ and */ˈnowɪθ/, */ˈdal(ː)əθ/, why do their show their expected sounds in LC? 

Yes, one possible explanation is that Lhuyd substituted the Welsh distribution.
The other explanation is that Cornish retained the original values all along. 

Whatever it may be, neither can be ruled out, nor proven.

> This is a structured pattern, which fits the data.

Partially, and very dependent upon dismissing evidence from Lhuyd. 

> It is coherent, easily explained, and indeed, easily taught.

Nice and neat in historical linguistics always makes me a little suspicious, especially in this muddle of scribal spellings, prosodic shifts and Cornish-writing Welshmen of the 18th century….

> Do you have a different description for these pairs, Dan?

Yes, I do. I've explained them many times, as you know. 

> If so, shouldn't you wish to explain it to us?

Yes, I have done. Many times, as you know. 

> Or have you not thought about the phonotactics of Cornish?

Yes, I have. 

> It's true that Lhuyd writes -v in ‹a hanav›, ‹a yuhav›, ‹dredhev›, ‹genev/ᵹenev›, ‹orthiv›, ‹ragov›, ‹ụarnav›, ‹olav›, but this -v is **not found in traditional Cornish** (though -ff and -f and Ø are) and so the most likely reason for this is the influence of Welsh. He heard "gene vy" and the thought it was like Welsh "genef vy". 

It's true that there is even less dependable evidence concerning /f/ : /v/ than there is for /θ/ : /ð/. This is why I would like to propose for R13 an umbrella graph ‹f› that more or less works like ‹s›. We would spell ‹f› word-initially and word-finally with the optional pronunciations [f] ~ [v], but spell ‹v› where both 'dialects' of Cornish agree on /v/, i.e. ‹ev›, but ‹bedhaf›, in essence what you have in KS, but with the rule-based option of pronouncing [f] or [v] for words/position in doubt or where one variety has [f] and the other [v]. SWF ‹cavos› could then be spelt SWF13 ‹cafos› meaning [ˈkafəz] or [ˈkavəz]. 

> It's true that Lhuyd writes -dh or -ꝺ sometimes -- but he also writes -th and -ꞇ, and the texts all write either -th or -ȝ. So the most likely reason for this is the influence of Welsh. 

Dismissing Lhuyd's evidence, though I'm not saying you shouldn't view his work critically, doesn't really help, because what you find in MC doesn't tell you much and leaves only speculation without any kind of proof. 

> For your view to be taken seriously, shouldn't you wish to produce a coherent argument which takes the linguistic system into account?

Yes, I should.

> You'd have to have a convincing reason why we have g/k and b/p (and probably z/s) on the one hand but ð/ð and v/v on the other.

Yes, as mentioned above, I have. 

> Because so far, Dan, while we have shown a coherent and systematic linguistic structure, you have just said "it's doubtful because of what's in Lhuyd, and so we should have dh and v throughout".

But that's precisely what we have! You have a coherent theory for which there cannot be any proof and I have several theories that I believe may be possible and simply wonder what the best way is to treat these doubtful cases in RC. I'm not saying you're wrong, nor do I favour any other particular theory, but when it comes to the point where several theories are unprovable yet imaginable and fit the data, as I believe both theories here do, then I need to find out how to accommodate these theories and include them in written and spoken Revived Cornish. 

You on the other hand seem to be so convinced that your theory is correct, that you wish to impose it on the spelling of RC regardless of lack of absolute and convincing proof. This is what I don't want to go along with. I don't believe it is inclusive, I don't believe it is sensible. And I don't believe it is very respectful of the theories other linguists working on Cornish have where this problematic issue is concerned. 

As Nicholas has written in "Clappya Kernowek" - along the lines of, if he were to jump-start the Revival he would probably choose not to distinguish ‹th› and ‹dh› and write only ‹th›. I can relate to this and this is legitimate position. But he didn't jump-start the Revival and decided to distinguish the two sounds. 

Originally in the first edition of Cornish today, his UCR spellings did include words that had ‹dh› in final unstressed position. But by the time of "Clappya Kernowek" and his dictionary, the final unstressed ‹dh› had changed to ‹th›. 

I would like to explain ‹dh› as an umbrella graph in the context of the SWF: those who believe Nicholas' theory that final unstressed /ð/ was unvoiced to [θ] in final unstressed position can do this by rule, whereas those who believe final unstressed /ð/ was variously pronounced either [θ] or [ð] according to the phonological environment can do so as well, while those who believe final unstressed /ð/ was always [ð], can do so, too. Thus we have a spelling that caters to all tastes and doesn't prescribe one man's theory saying that final unstressed /ð/ was always unvoiced. 

> That's not a structural argument.

From the point of view of orthographic design it is. 

> And infeed, it would only approach being a convincing argument if you were to list **all** of the words in -dh and -ꝺ and -th and -ꞇ in Lhuyd, with a following word in instances of words within a sentence or phrase.

I know, and this would take a lot of time, as I haven't got a single searchable computer database of attested Cornish spellings. In my dictionary I'm currently in the process of putting such a database together so in time it'll be easier to search. Perhaps then I will write something along the lines. I will prepare a short statement though within the R13 proposals. 

> Really, there aren't all that many. To know what the implications of Lhuyd are (since there is doubt because of his Welsh) the only way to deal with it is exhaustively.

You are mistaken, in my opinion, to dismiss Lhuyds distinction of ‹th› and ‹dh› by assuming analogy with Welsh. Since the other evidence is silent, Lhuyd is all we have in this question. We can choose to ignore him, choose to view him critically or choose to dismiss him. I view him critically. With the evidence available I feel I have to entertain the possibility that final unstressed /ð/ was indeed distinguished, at least occasionally from final unstressed /θ/ and that I cannot in full conviction endorse Nichoals' theory that final unstressed /ð/ was always unvoiced. 

> Now, we know that the reason KK has -i is that Ken George was very impressed by Welsh and Breton. Ken George said as much in his Linguistic Advisor submission to Albert and Ben and Trond. He said:
>> I advise against <-y>. Welsh and Breton both write <-i>.
> The reason the SWF (via KK) has -dh and -v is that Welsh has them. 

Welsh has -v?

> You suggested that perhaps there could be a rule for people who say [ˈɡɔɹsəθ] that they could learn to devoice -dh in some positions.


> And you said, I believe, that there was evidence in Lhuyd that some of these words were voiced in intervocalic position. I don't think there is much evidence of that, and you haven't produced a complete analysis with exhaustive examples (the sort of thing Nicholas does), but even so. 

No I haven't, just a few examples from JCH. Though I wouldn't go so far as to endorse the theory that we have a voiced : unvoiced alternation similar to Breton which is what Ken George proposes. 

> The easier rule,

Says you...

> the safer rule,

says you….

> the rule that is more certain in terms of Traditional orthography, is to spell -th and -f as the texts do (for then we know we are not erring), and to allow these sounds to be voiced in intervocalic position.

No, it is not, because it is no easier, safer or more authentic to distinguish final stressed /ð/ and final stressed /θ/ while deciding not to do so in final unstressed position. The evidence for the stressed distinction doesn't weigh much heavier than for the unstressed distinction. The native MC writers didn't distinguish them either. 

So the safest bet in my opinion would be not to distinguish ‹th› and ‹dh› at all and write only ‹th›. But this option isn't practical for the non-linguistically inclined average learner of Cornish, who wishes to be able to predict the pronunciation of /ð/ and /θ/ by the spelling. That leaves us with distinguishing the two consistently. 

> Most people including KK users whom I hear speak say -əθ or -ɛθ for these words. Maybe they voice them intervocalically. But they don't in absolute auslaut. 

True, but hardly relevant to my argument, as I said that ‹dh› is to be viewed as an umbrella graph for [θ] and [ð] by the learner/speaker according to which school of though s/he may wish to follow. 

> (We don't have to worry about -dh and -v in stressed monosyllables because we all agree that they were voiced.)

But the evidence for this distinction in the traditional texts is also lacking. So why treat the two differently?

More information about the Spellyans mailing list