[Spellyans] reDistribution of <i> and <y>

Michael Everson everson at evertype.com
Wed Jul 23 13:56:11 IST 2008

At 09:40 +0200 2008-07-23, Daniel Prohaska wrote:
>I would like to hear everyone's opinions on the 
>following idea for redistributing <y> and <i> in 
>the SWF. I would write <i> where bother Late and 
>Middle Cornish have /i/ and /i:/, and write <y> 
>~ <e> (in dictionaries <ÿ> ~ <ë>) where Middle 
>Cornish has /I/ and /I:/, but Late Cornish has 
>/e/ and /e:/.

I can't see how this would work. In the first 
place, in stressed syllables, we have [i:] words 
(like <mis> and we have [e:] words (like <res> 
and we have [I] words like <bys>. No problem 
here: we write <i> and <e> and <y> for these. We 
also have in stressed syllables words [i:] and 
[e:] alternate; no problem here: we write <ÿ> and 
<ë> regularly.

Not in dictionaries. Regularly. Nance's "write 
the marks in the dictionary but leave them off in 
plain text" is *not* something to be emulated. We 
know that Revivalists learning from Unified made 
mistakes in vowel length when there was no system 
to mark it; we know they made mistakes 
distinguishing between /y/ and /u/ when there was 
no system to mark it.

Both KS and the SWF correct the first problem 
because they have a system where regular vowel 
length is marked by consonant quality. The SWF 
fails to be able to mark irregular vowel length, 
but KS corrects this failing by using the grave 
<`> for anomalous short vowels and the circumflex 
<^> for anomalous long vowels.

The SWF fails to correct the second problem 
because it cannot distinguish between /y/ and /u/ 
since the graph used is <u>. I do not believe 
that <ou> can be used throughout for [u:] and [U] 
alike; it is absurd to contemplate writing <Louk> 
for 'Luke' [lu:k] and <louck> for 'luck' [lUk]. 
KS corrects this failing by using u-circumflex 
<û> for [u:] <Lûk> and u-grave <ù> for [U] 
<lùck>. (Both KS and the SWF take advantage of 
using <k> and <ck> to mark final vowel length 

I do *not* believe that we can recommend that 
diacritical marks be optional. That "philosophy" 
is childish, and does nothing to help learners or 
fluent readers either. Let's accept that a mature 
orthography for Revived Cornish must accept 
diacritics due to the nature of the language, and 
neither apologize for it nor try to weasel out of 
it by offering them as "optional". To write 
correct Irish, Scottish Gaelic, Welsh, Breton, or 
Cornish, diacritical marks are *required* and 

>SWF <brentin>; RMC /"brentin/, RLC /"brentin/;
>SWF <kegyn>; RMC /"kegin/, RLC /"keg at n/;
>SWF <tir>; RMC /ti:r/, RLC /ti:r/;
>SWF <bys> ~ <bes>; RMC */bI:z/ = [bi:z] ~ [bIz] ~ [beIz] etc., RLC /be:z/;

I think Dan is really mixing up things here, 
because on the one hand he is talking about 
stressed monosyllables and on the other 
unstressed final syllables. These are not the 
same things.

KS writes <tir> and <bÿs>~<bës> in monosyllables. 
We all know that [bI:z] is a failed 
"aspirational" pronunciation of KK, which is 
either realized as [bi:z] with correct vowel 
length or as [bIz] in erroneous hypercorrection. 
[beIz] would, I think, be an English-influenced 
[be:z] and so need not be listed.

But KS is happy with <tir> and <bÿs>~<bës>.

Where Dan has seemed to me to mix things up is in 
his other examples. What does /'brentin/ mean? 
Surely ['brEnt at n] though likelier ['brEnt1n] with 
i-coloured schwa. How does this differ from 
/'kegin/? Surely this is ['kEg at n] though likelier 
['kEg1n] with i-coloured schwa. That is to say, 
the two words are pronounced identically in both 
RMC and RLC -- both with [1], i-coloured schwa as 
in English <wicket> ['wIk1t].

We may have need for <ÿ>~<ë> distinction in some 
short vowels, but not, I think, in short 
unstressed vowels. I don't believe that a 
<kegÿn>~<kegën> distinction is found in Revived 
Cornish -- and if we were to mark this, we would 
certainly be multiplying the use of the diaeresis 
many many many times.

I don't believe that the distinction KK makes 
between <i> and <y> in unstressed final syllables 
is defensible. <-yn> and <-in> in unstressed 
position are pronounced identically. KS writes 
<-yn> for these words and uses final <-in> in 
words where that final syllable is stressed and 
the vowel is long, as in <desin> [dE'zi:n] 
'design', constrin [kOn'stri:n] 'constrains'.

The <-yn>/<-in> distinction in the SWF is indefensible. Here's why:

In GKK 1993, George explains (p. 20) that KK <i> 
/i/ is [i:] when stressed, and "when half-long 
and short, the same but reduced in duration". He 
also explains that when unstressed it is [I]. He 
then explains that KK <y> /I/ is [I] when short, 
and "when half-long and long, the same sound 
extended appropriately". So in 1993 he says 
nothing about unstressed <y>.

In the Gerlyvrik 2005, however, he gives the following table:

<i> when stressed is [i:] [i:\] [i], when unstressed [I]
<y> when stressed is [I:] [I:\] [I], when unstressed [I]

Then he writes <bryntin> ['brIntin] and <kegin> ['ke:\gIn].

What does this mean? That the suffix in the first 
word is *stressed* but in the second 
*unstressed*? Yet George writes both <-in>. How 
is the learner to make any sense of this? And the 
SWF doesn't even write both <-in>: it 
distinguishes them. How? On the basis of the 
Gerlyvrik's phonetic respellings? Why is it 
<brentin> not <bryntin> then?

On this matter I have no confidence in the SWF. 
And if I can't work out how these choices were 
made, what choice has any learner in working out 
when to write one letter and when to write the 
other? I don't believe the AHG decided explicitly 
on this alternation <-yn>/<-in>. I have a feeling 
it has to do with the word "etymology" -- and 
since that concept is left undefined in the 
context of these choices, it is no wonder that 
there are problems.

We have heard that the Partnership has recognized 
a series of edge-cases. For once I am copying the 
authors of the SWF and the CLP officer in charge 
of the process. The problem of the edge cases 
most likely has to do with the word "etymology" 
which was probably introduced into the AHG 
discussions by Andrew Climo, on the basis of work 
we had done on KS.

In that context, we had said we would retain 
"etymological" spellings in words like <colon> 
because in the derived form the unstressed vowel 
was restored as <colonow>. (We needed to describe 
this because the earlier drafts of KS would use 
<-an> throughout for final [@n]. The context here 
is important.) This use of the word 
"etymological" was mine, and it was imprecise. 
Somehow, however, it seems to have been taken to 
mean "as in Ken George's KK reconstructions". As 
such it is a recipe for conflict. Why should 
<tavas> be respelt <taves>? On the basis of 
"etymology", in that the form "should" be 
<taves>? In Revived Cornish, the word has two 
plurals: UC has <tavasow> and we have <tavosow> 
from Tregear which was unknown to Nance.  The 
Welsh is <tafod> pl <tafodau>; the Breton is 
<teod> pl <teodoù>. Old Cornish <tauot>. In the 
texts <tavas> is something like 5 times more 
common than <taves>. On the same "etymological" 
basis as <colon>/<colonow> we might write the 
wholly unattested *<tavos> but of course, neither 
*<tavos> nor *<tavesow> are known.

And from here the whole enterprise brings us to 
the question "Do we trust George's 
reconstructions and etymologies?" And the answer 
is No. We do not. Not everything is wrong. Of 
course not. How could it be? But KK has been 
under sustained academic criticism since it was 
introduced. It is not "safe" enough to serve as 
the base for the distribution of vowels in 
unstressed position. The "edge cases" are not 
really "edge": this is a fairly fundamental 

In any case, to get back to Dan's original 
question... I do not believe that a distinction 
<kegyn>~<kegen> is defensible. KS can write <-yn> 
for final unstressed i-coloured schwa and <-in> 
for final stressed [i:].
Michael Everson * http://www.evertype.com

More information about the Spellyans mailing list