Re: Some problems with the phonology and orthography of Funus (my conlang)

At 8:30 am +0200 28/10/98, B.Philip.Jonsson wrote:
>Friends, Glossomaniacs, Conlangers,
[CONSONANTS]
>My conlang Funus has tri-cononantal roots C-C-C like the Semitic languages,
[snip - I like the cluster rules]
>/nl/ -> [(n)dl]; any sequence of two nasals turns into a nasally released
>voiced stop at the point of articulation of the second nasal, e.g.
>/m|n|N+N/ -> [?gN] (whose Romanization {'qg} I'm perversely fond of :).
So is /g/ really written as {q} and /N/ as {g} ?  Or is the Romanization
{'gq}.  I've never been overfond of the convention {q} = /N/ which occurs
in several conlangs (Does any natlang use this convention?) but {q} = /g/
seems strange.

[VOWELS]
........
>        (phonemic)          (phonetic)
>HIGH   /i   @   o   u      [i   @   o   u
>LOW     e   a   O   U/      e   &   A   ^]
>
>(where LOW are thought to be not only lowered but also de-rounded wher
>applicable.)
>
>The problem with extending the vocalism is that I want the lang to be
>possible to write with the A-Z alphabet plus apostrophe.  Currently the
>only non-used letters are C J V.   If I introduce /@/ I will change the
>current Y into J and use Y for /@/, or Y for /e/ and E for /@/.
Yep - I see no problem with {j} = /j/, and would use {y} = /@/ or /^/
rather than /e/.

> I can
>accept V for /U/, but C for /O/ feels somehow harder...
Yes, it does somewhat :)

> I might introduce
>a constraint allowing /o/ and /O/ only as the second vowel of a "pattern",
>thus making them allophones able to share one spelling, still giving me a
>larger set of vowel distinctions: I E Y A O U V, but only three more VV
>patterns than the 5-V system unless I introduce diphthongs.
I don't like that solution - it upsets the neatness of your pattern above.

>Maybe I could
>allow myself to use umlauts?:
>
>        (Romanization)      (phonetic)
>HIGH   {i  "o   o   u      [i   @   o   u
>LOW     e  "a   a  "u}      e   &   A   ^]
>
>But this makes the language less elegant to write in plain ASCII.
It does - and the use of "o and "u for the sounds above is somewhat odd;  I
think this is to be avoided if possible.

>Another
>possibility is to spell:
>
>        (Romanization)      (phonetic)
>HIGH   {i   a   o   u      [i   @   o   u
>LOW     e   av  ov  uv}     e   &   A   ^]
>
>But this is a rather great departure from Roman spelling traditions: the
>values of the graphies wouldn't be even remotely "guessable"; besides I
>wouldn't like to use digraphs for what isn't digraphs in the native script.
I agree on all three points.

The problem it seems to me is that we need 8 vowel symbols & the Romans
bequeathed us only 6: a,e,i,o,u,y.  (As you know, was used only as a vowel
by the Romans.)

We need two other symbols; fairly obviously one of them will have to be {v}
or {w}.  Because of my familiarity with Welsh I'm quite used to the use of
{w} as a _vowel_; if you adopt that, then /w/ will have to be written {v}
which I do not see as a problem.

I understand that {v} is used in the Romanization of some Amerindian
languages for sounds that are similar to [^].  Although I find this use
strange, it does have natlang precedence and probably to most people
unfamiliar with Welsh it is no less odd than using {w} as a vowel.  (Does
any other natlang do this?)

Since Funus has /w/ clearly we cannot use both {v} & {w} as vowel symbols,
since one of these is needed to spell /w/.

So we need an eight symbol.  I have come across {x} for /@/ and {q} for /O/
- but only in conlangs & I don't like either convention; besides Funus
Romanization will need both these symbols as consonants.  The only solution
must be to extend the Roman alphabet, as many natlangs have done.  But if
you want to use plain ASCII (by which, I assume you mean ASCII 32 throught
to ASCII 217) then our choice is somewhat restricted; possible, it seems to
me, are: &, @, ^, # and ~.  The last is so suggestive of some nasal sound
that I think we can discount it immediately.

& - actually included in the Speedwords alphabet and in the 1951 dictionary
it is placed by Dutton before {a}.  It could be used to represent [&] or
possibly [@]; but the natlang use as an abbreviation for Latin 'et' = "and"
is probably too strong &, indeed, even in Speedwords it is used only in the
words & [&nd] = 'and' and &e ['&nde:] = etcetera.  [Note to RJP - *NO*
criticism of Speedwords is indended here]

@ - abbreviation for 'at' which is 'a' in many Romancelangs.  Therefore, we
could use it for some sort of A-sound, i.e. in Funus for [&] or [A]; these
seem preferable to me than the email IPA 'asciification' [@].

^ - could, of course, be used for [^]  :)  This seems to me preferable to
the use of {@} above.  Ideally we'd want the actual IPA symbol with Greek
lambda as the upper case variant; but ^ is a possible 'ASCII compromise'.
BUT whether we use {w} = [u] & {u} = [^] or we use {v} = [^], this use of
{^} is redundant.  We could, of course, use {^} = [@], but I'm not sold on
this.

# - might be usable for, say, [@] - but little to support its use.  And the
Amrican usage of # = 'number' is likely to become more widespread.
Therefore, I think this should be avoided.

Thus of the possible symbols, {@} is probably the least objectionable.  It
can be interpreted as 'modified {a}' and is thus probably best for [&].
Better still, probably, would be to use the <ae> ligature for the sound (if
only everyone could read F and f !), with the convention that {@} is an
acceptable variant of this.

Thus I suggest either:
        (Romanization)      (phonetic)
HIGH   {i   y   o   w      [i   @   o   u
LOW     e   @   a   u}      e   &   A   ^]

or:
        (Romanization)      (phonetic)
HIGH   {i   y   o   u      [i   @   o   u
LOW     e   @   a   v}      e   &   A   ^]

It will be no surprise to learn that I prefer the former ;)

Ray.