[Spellyans] voiced <th> and preocclusion

[Owen Cook]

Hey Steve, how are you doing? Please don't take my comments in the
spirit of "shooting you down", but rather as a friendly opening
exchange of ideas.

2008/6/7 Hewitt, Stephen <s.hewitt at unesco.org>:
> Steve Hewitt
>
> Specialist in Breton, where I have a long-standing interest in
> orthographical questions, and have developed my own
> Middle-Breton-based "etymological orthography" which does a much
> better job than even the Interdialectal (Etrerannyezhel) orthography
> of predicting dialect reflexes.

That sounds very interesting! Can one have a look? Is it online anywhere?

> I have a good knowledge of Welsh, and
> some basic Irish.
>
> I am not certain that a phonemic SWF using only authentic, attested
> graphemes, and at the same time catering to several stages of the
> language, is entirely feasible, but I am certainly interested in the
> attempt. That being said, authentic graphemes are not necessarily the
> only solution; Modern Breton works very largely with several basic
> principles which are definitely not traditional <k>, hard <g>, and
> for ZH and OU orthographies <-s-> = /s/. If the Welsh system is
> largely authentic, this is because the traditional orthography
> performed amazingly well, and in many ways constituted a good
> supradialectal norm catering to both North and South Welsh.

Yes, I was very skeptical about it too at first, but I have been
impressed at how well KS performs in practice.

> One question I have tried to ask on Kernowak, but was quickly shot
> down, concerns why medial <th> is thought to have remained
> consistently voiceless, whereas medial (and even initial) <s> and <f>
> are generally agreed to have become voiced. In Breton, the
> corresponding phonemes in <difenn> "defend", <kaseg> "mare", and
> <brezhoneg> "Breton" all became voiced in most dialects. Since the
> grapheme <th> is demonstrably ambiguous for /th/ and /dh/, how can
> anyone tell?

Cornish is extremely fortunate in that it does not have to rely on
<th>, which as you say is ambiguous with respect to voice (as is
yogh). We also have the writings of Edward Lhuyd, who used the Greek
tau for the voiceless phoneme and delta for the voiced version. I
certainly can't recall seeing any instances of delta where we should
expect tau, though any of us with copies or facsimiles of Archaeologia
Britannica to hand can see whether they can spot any.

> With regard to KS, I feel uncomfortable about the strong preference
> for marking pre-occlusion, with the idea that users preferring
> earlier stages could simply ignore it. The more usual solution in
> such cases is to allow divergent pronunciations to be derived from
> earlier forms. For instance, both Icelandic and Faroese have
> widespread pre-occlusion, but neither marks it in the orthography; it
> is automatically derived.

The problem is that in Cornish, deriving pre-occlusion doesn't work
particularly well, especially in compounds and antepenultimae.
Pre-occlusion appears most of the time where it should, but not
always, and sometimes it crops up where it theoretically shouldn't. In
my opinion, the earlier editions of KS were correct in recommending
<bm> and <dn> for all users. However, the latest edition of KS agreed
with the SWF in permitting <mm> and <nn> here as well.

I learned to write KK and pronounce with preocclusion. In theory,
every time you had <mm> or <nn> in a stressed syllable, there should
be preocclusion, and this was the theory I followed. I was quite
heartily disillusioned to find how often this made me pronounce things
wrong.

Furthermore, neither <mm> nor <nn> is going to win any prizes for the
most authentic or traditional grapheme, at least not when written at a
syllable's end, as in 'penn'. The most common traditional spellings
would have been 'pen' or 'pedn'. Given the choice between 'penn' or
'pedn', a true stickler for traditionalism will always opt for 'pedn'.
A silent <d> here should not be unacceptable given that we all live
with silent <g> in 'sign' and 'impugn'...

Oll an gwelha,
~~Owen