Re: Phonology/orthography sketch

On 2008-05-28 Peter Collier wrote:
 > Oooh, you are awful.
 >
 > Sounds opaque enough to have L2 tourists and
 > schoolchildren scratching their heads and
 > reaching for the aspirin, while being
 > simultaneously logical and clear to the L1s, who
 > can't see what all the fuss is about.

It's actually inspired by Welsh where _i w_ are
both consonants and vowels; then I started playing
with the thought what if Welsh had kept the [G]
sound and decided to spell it the same as /i\/
which it is phonetically close enough to. However
I didn't want to blandly copy Welsh in other
respects -- especially not have a half-assed
/i\/~/@/ contrast, and then /y/ for _u_, as it
probably was in Middle Welsh, and then
consequently an [H] suggested themselves. The
/x/~/X_w/ distinction suggested itself once I had
a /M\/~/w/ distinction, but I rather like it.
    Lastly I'm fond of digraph-less orthographies,
    and ended up assigning a value to all 26
    letters...

Actually there may be some fuss even for L1
speakers such as distinguishing [iG@] from [ji\G].
I guess linguists and high-end dictionaries use
accent marks to distinguish _íyy_ and _iýy_.

 > Is this some new apriori plaything you're
 > crafting?

Seems so, and seems its name is going to be _euir_
or something derived therefrom.

Från: Lars Mathiesen <thorinn@...>
 > This is fun. Since Benct hasn't answered, I'll
 > give it a go...
 >
 > 2008/5/28, Ingmar Roerdinkholder
 > <ingmar.roerdinkholder@...>:
 >>> shouldn't ihtas be ['ixtas]? since above you
 >>> said: 'h' = [x]
 >
 > It is. // // is superphonemic representation,
 > //"i1tas// = /iGtas/ = [ixtas].

Exactly

 >>> and why 'ny' and not 'ng' for [N]?
 >
 > Probably ni/ny/nw are seen as nasal
 > companions to the voiced fricatives. Also,
 > there's no voiced uvular stop to complete an
 > nj/ng/n? series.

Right on both counts.

 >>> and is there a [w] and how is that spelt?
 >
 > "The voiced ficatives have approximant
 > allophones [v\ D j M\ w], occurring mainly
 > between a vowel and a consonant."
 >
 > Like tww [tw"u] I guess. When that isn't
 > [t"uR_w] or possibly [t"uw].

I'm not sure all of these are possible, even. OTOH
at least one of them is possible because i like
the _ii uu ww yy_ graphies to exist. I still have
to decide if the _iii uuu www yyy_ sequences will
exist. I'd guess _Yyy_ is how they'd spell the
sound of retching!

 > The 'mainly' above seems to be hiding a lot, a
 > later section has euir //eu-ir// = /eR_wir/
 > = [ewir] -> [eHir]. Approximant before
 > morpheme boundary?

For now it means mostly that approximants and
fricatives are in free variation except word
initially where fricative is the rule and between
a vowel and a consonant where approximant is the
rule -- i.e. you get phonetic diphthongs. And
before a voiceless obstruent they snap all the way
to voiceless fricatives, as already implied.

 > 2008/5/28, Benct Philip Jonsson
 > <bpj@...>:
 >>> Phonetically there exists [H] or [j\_w] as
 >>> allophones of /j/ next to a rounded vowel or
 >>> /w/, and of /w/ next to /i/. The writing
 >>> system writes this allophonic sound with _u_
 >>> in spite of its occurrence being wholly
 >>> conditioned by adjacent sounds: a word spelled
 >>> _uintou_ could only be /wintoj/; a spelling
 >>> _wintoi_ could not be a distinct word, but
 >>> only an unusual, although phonemically more
 >>> 'correct', spelling of the same word.
 >>> Similarly _au_ or _eu_ could never occur
 >>> without a following conditioning _i u w o_; a
 >>> spelling _euor_ would always represent /ejor/
 >>> and might be derived from a word _ei_.
 >>> Similarly _euir_ would be /ewir/, possibly
 >>> derived from an _ew_.
 >
 > What happens if you have both /i/ and a rounded
 > vowel next to these approximants? Do you get
 > [iju] or [iHu], [uwi] or [uHi], and so on?

It could only be [uHi]/[uj\_wi]

 > Are there any contexts where this could cause
 > the difference between //i// and //w// to be
 > neutralized?

Oh yes, that would be the rule.

 > (If /y/ would trigger [w] > [H] like /i/ does,
 > we'd have tuut [tHyt] for both //tu"yt// and
 > //ti"yt//, and maybe tuut [tyHt] for //t"yut//
 > and //t"yit//).

Right on all counts; I forgot to mention that.
Thanks for reminding me!

FWIW I think //Ci'iC// and //Ci'yC// would surface
as /CjiC/ _CiiC_ and /CHyC/ _CuuC_, but //C'iiC//
and //C'yiCC// would surface as /CiC/ _CiC_ and
/CyC/ _CuC_.

Come to think of it is probably the case that
there exists underlying juxtavocalic //y//, but
that there are rounding/palatalization
assimilation rules which keep the total surface
neutralization. Probably both palatalization and
rounding have rather strong assimilating power so
that e.g. an underlying //'iyu// will actually
surface as /yHy/ _uuu_! Moreover //y'o y'e 'oy
'ey// will surface as [H2 2H], spelled _uo ou_ --
i.e. //'e// and //'o// are neutralized in this
position, and the neutralization product [2] is
felt by L1 speakers to be an automatic allophone
of /o/, or at least the writing system insists on
writing it _o_, the same it insists on writing
the product of //i_^//~//y_^//~//u_^//
neutralization as _u_.

 > Lars

Gott att se dig här igen!


/BP 8^)>
--
Benct Philip Jonsson -- melroch atte melroch dotte se
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  "C'est en vain que nos Josués littéraires crient
  à la langue de s'arrêter; les langues ni le soleil
  ne s'arrêtent plus. Le jour où elles se *fixent*,
  c'est qu'elles meurent."           (Victor Hugo)

Reply