Re: Romanization of Reduced Vowels

At 12:37 9.12.1998 +0100, Kristian Jensen wrote:
>Sinulat ni Matt Pearson (na Mat Pilson siespe ia):
>
>>Kristian Jensen skrev (sinulat):
>>
>>>All right... I think I'm beginning to understand this now.
>>>Thanks to you. I believe that all these vowel sounds are all
>>>conditioned variants of /@/. So you'd suggest using one symbol.
>>>But the question is, what symbol? It seems really contrived to
>>>me to write for instance a word that sounds like [mw@j] and
>>>[mj@n] as "maway" and "mayan" respectively.
>>
>>What's contrived about it?  Looks good to me.  Or how about
>>"mewey" and "meyen"?
>>
>Actually, "mewey" and "meyen" looks a lot better. Not for Khmer, but
>for Boreanesian - my reform of Lumanesian. Thanks. All of a sudden,
>this seems quite natural.
>
>Regards,
>-Kristian- 8-)
I often use accent marks to differentiate vowel qualities, differently in
different languages, but there is is a tendency towards what I call "the
accent sign hierarchy":

^   circumflex  MOST
/   acute       |
\   grave       |
"   umlaut      V
    no mark     LEAST

The actual parameter of the hierarchy may be length, stress, reduction to
schwa (rather non-reduction, since the MOST schwa-like sound will end up at
the bottom of the scale!) or any suitable combination of these.  In
practice I will modsify the system so that the more frequent sounds get the
simplest mark (with no mark being the simplest of all): if there is only a
three-way contrast ^ and " will not be used; in particular this affects the
use of " vs. no mark, where it is always the least frequent class of sounds
that gets marked by ".  Another case of adaptation in a conlang of mine /e
was high mid [e], \e was low mid [E], while ^e was used in cases of
neutralization of this distinction, plain e was [@] and "e was lowered [@],
which was the realization of \e before r and very infrequent.  In another
lang ^ indicated that a word had lost an intervocalic consonant (usually
h), / marked a long stressed and \ a short stressed syllable.  At another
instance I actually used ^ to indicate nasalization!  Not infrequently I
have used a scheme like this:

        FRONT                           BACK
UNROUNDED       ROUNDED         UNROUNDED       ROUNDED

/i              "u              "i              /u
\i              ^u              ^i              \u
/e              "o              "e              /o
\e              ^o              ^e              \o
"a                                      /a
^a                                      \a

The sign to be dismissed as redundant in this system would normally be \,
since the values of the plain vowels if ^ or " were left out would be a bit
odd, and I simply think / looks better than \.

Ray Brown once pointed out to me that the system:

^       long stressed vowel
/       long unstressed vowel
\       short stressed vowel
no mark short unstressed vowel

was more logical than my previous system where ^ was long UNstressed and /
long stressed, since ^ might be considered a combination of / (length) and
\ (stress).  It also happens to agree better with the implied hierarchy.
I'm still grateful to you, Ray :), although I've not found a satisfactory
replacement for " as indicating short OR long unstressed vowel :(.

If I can't or don't want to use accent marks I'm not adverse to using ^ ~ `
as letters in their own right, with % & @ as their respective capitals.
Usually ^ and ` would be schwa-like vowels, while ~ often is an extra
nasal, usually velar, but more frequently I have used ~ as a _mark_ of
nasalization, with ~e etc. as nasalized vowels, ~n as palatal vowel and ~g
as a velar vowel [I know that ~j for palatal nasal would be more
consistent, but ~n _is_ easier t figure out! :)].  Often ` plays the role
of palatalization diacritic (`s="sh" etc.), leaving ^ with the capital @ or
& as schwa.  If the lang has got retroflex vowels, in which case ` marks
these and ^ marks palatalization -- then poor schwa gets decreed to be a
retroflex vowel!

My most cherished conlang Funus uses no diacritical marks, writing its
vowels as:

        i       v       u
                y
        e       a       o

It also uses g for velar nasal and q for /g/, which may seems odd for some,
but I definitely don't like q for /N/!

Nuff rambled,

/BP

    ----------------------------------------------------
    <bpj@...>

        Solitudinem faciunt pacem appellant!
                                            (Tacitus)