Re: Can realism be retro-fitted?

Benct Philip Jonsson wrote:

> - i| > i
> - i( > y /i\, @/
> - e| > oe /Oi/, wy /Ui/
> - e( > e
> - u| > i
> - u( > w /u/
> - o| > u /u\/ > /i\/
> - o( > o
> - a| > aw /au/, ô /o:/
> - a( > a
Here's a recent (tentative) summary of how the pre-Tirelat vowels might
have developed:

a, a:, o > a
ai > a:
i, u > y
ui > y:
i:, e: > i
ei > i:
e > ə, o
je > e
eu > e:, ə:
au > o:
o:, u: > u
ou > u:

I'm not quite satisfied with the whole system, but it gives me something
to work with. I note that short and long /i/ match the Welsh development.

The /je/ > /e/ development is part of an explanation why /Ne/ doesn't
occur: /Nj/ merged with /nj/, so the original /Nje/ ended up as /ne/
instead of /Ne/. In other contexts, /e/ ended up as /@/, except after
bilabial consonants, when it changed to /o/.

> I remember you saying that _ai_ is the most common Tirelat
> diphthong. It should not be too hard to derive it from _e|_:
> the Welsh developments are similar to the French _e| > ei >
> oi > oe/ui > we > wE > wa_, which should be compared to the
> rather much simpler German development _ei > ai_. To get a
> preponderance of /a i u/ you can have a| -- a( and o( -- u(
> merge, Of course you can also have an original Q|( to create
> more instances of /o/.
Well, I have /o:/ and /u:/ merging, which is pretty similar.

> The second is the English Great Vowel Shift, particularly in
> its northern form. The chains were as follows:
>
> The front vowels developed like this in both the south and
> the north:
>
> - a: > E: > e: > i: > @i
>
> The E: > e: > i: part later repeated itself, the /E:/ from
> Middle English /a:/ pushing on to /e:/ (which later became
> /eI/ in the south) and the /e:/ from M.E. /E:/ merging with
> the /i:/ from M.E. /e:/ (the meat-meet merger). The /@i/
> from M.E. /i:/ went on to /aI/, pushing M.E. /ui/ (in French
> loans, cf. above!) before it into /Ai/ and eventually /Oi/.
>
> For the individual vowels the trajectories were:
>
> - a: > E: > e: > eI
> - E: > e: > i:
> - e: > i:
> - i: > Ii > (ei >) @i > aI
> - ui > @i > Ai > Qi > Oi
>
> I doubt that the /ei/ stage for M.E. /i:/ is necessary: a
> direct [I] > [@] development is surely possible. That 16th
> and 17th century writers respelled the diphthong as _ei_ is
> not conclusive, since before the foot-strut split they had
> no good grapheme for [@], or _ei_ spelled [I\i].
>
>
> Back vowels went like this in the south:
>
> - O: > o: > u: > @u
>
> Since there was no M.E. /Q:/ the back vowel space wasn't
> overcrowded, and there was no secondary O: > o: > u:.
> However a secondary /O:/ later developed from M.E. /au/,
> probably by way of [Q:].
>
> But in the north the back vowels went like this :
>
> - O: > o: > 8: (u: remained unaffected!)
>
> The new /8:/ didn't last long: it became /u\/ in Scotland
> and /Ie/ > /I@/ in the north of England. That's why a "good
> house" is a [gu\d hus] in Scotland. In tirelat /i\/ can
> descend either from earlier /o:/ or earlier /u:/ depending
> on whether you choose an o: > 8: > u\ > i\ or an o: > u:, u:
>  > \i trajectory. Scots notably later lost its length
> distinction by changing the distinction between DRESS and
> FACE into an /E/ <> /e/ distinction. In Tirelat short vowels
> may merge with each other as /@/ or with their corresponding
> longs in different patterns to get the right weightings of
> the vowel phonemes -- except original /a(/ which should stay
> /a/. Original /a:/ could merge either with original /a(/ or
> intermediate /E/. If you have a lot of /@/ a merger of all
> or some original short vowels except /a(/ would not be
> unrealistic.
/@/ is not especially common, and it occurs both long and short. Both
/y/ and /@/ are relatively less common than the other vowels. Of course,
the actual frequency is something that may change once I establish how
they developed and start creating new vocabulary with a historical
background.

I don't know if I want anything as massive as the Great Vowel Shift in
the recent development of Tirelat vowels, but it might be appropriate
for an older stage of the language. I've settled on pretty much a
standard 5-vowel system (plus length) for the immediate precursor of
Tirelat, although I could just as easily go with something having 3 or 4
vowels, or 7 vowels without distinctive length, if I could get it to
work out.