Re: Phonologies

(Before I begin,  when I hit "reply" to this e-mail, it brought up your personal
address, James, and not the CONLANG address.  I forget what this is a symptom
of.)
I wrote, and then James wrote:
<<David Peterson mentioned in his explanation of his first conlang that:         
  1.) Though it had a large "phonology", it wasn't a phonology at all: It was
just a list of sounds. Being rather new at this, I don't understand the
difference. So, could someone explain what a phonology *is* and what it's
supposed to *do*?>>
Actually, this is presupposing something here.  When I say "phonology", I mean "natural
phonology", and what I mean by "natural phonology", is a phonology like that of
an natural language, and there don't happen to be any natural languages on
Earth where you can learn to speak the language (or read it aloud) just by
knowing how to produce a list of sounds.  Also, there are no languages where
every phoneme corresponds to exactly one phone (all explain this later, if it's
vague now).
*HOWEVER*, I think it might be safe to assume that an *ideal* phonology would, in fact,
be a list of sounds.  So, let's take Hawaiian, since its phonology is
relatively small.  In Hawaiian you have the phonemes:
Vowels: /a/, /e/, /i/, /o/, /u/ (long and short)Consonants: /p/, /k/, /?/, /m/, /n/, /l/, /w/, /h/
Ideally, if you learned how to pronounce each of these sounds, you should be able to
speak Hawaiian with a perfect Hawaiian accent.  However, there is phonological
variation in Hawaiian.
So, for example, in Hawaiian you have long and short vowels.  The ideal situation
would be that for a short /e/ you get [e] and for a long /e:/ you get [e:].
 This isn't the case, though.  In reality, short /e/ shows up as [E] and long
/e:/ as [e:].  Most of the time.  In some dialects, though, short /e/ can
surface as [i], while long /e:/ can surface as [ej] or even [e].  And, also,
there is a chance that short /e/ can show up as [e], as well.  So, to take an
inventory, the two phonemes /e/ and /e:/ have the following distribution:
/e/ = [E], [e], [i]/e:/ = [e:], [e], [ej]
Ideally, there should be some way to explain the variation above, and I can give
somewhat of an account for short /e/, at least.  The phoneme /e/...
1.) Shows up as [i] word-finally (in some dialects),2.) Shows up as [e] in
stressed position (again, in some dialects),3.) Shows up as [E] elsewhere (in
most dialects, though some have [e] sometimes).
So it is predictable where you get each sound, to a certain extent.  If we
wanted to rewrite the inventory of Hawaiian as a list of sounds, though, the
vowels might have to be rewritten (and this is only considering /e/ and /e:/)
as...
Vowels: /a/, /a:/, /i/, /i:/, /o/, /o:/, /u/, /u:/, /e/, /e:/, /E/, /ej/
Then you'd have to go back and respell all Hawaiian words with /e/ in them.  Plus,
you would lose a distinction between, say, "haole" = "non-Hawaiian (pej.)", and
"haoli" = "happy".  This might be a good thing for some speakers, but not for
others.  Also, a word like "henehene" would become "henEhenE" for some
speakers, and "henehene" for others, and possibly even "henEheni" for others.
Using this last example in particular, what you're doing by creating a list of
sounds, rather than a list of phonemes, is forcing a distinct mental
representation for each sound.  So even though "henEheni" is a reduplication
of "henE" (or "heni"), you wouldn't be able to tell, since you would have a
different phoneme on the last syllable.
I think what I'm getting at here is if you have a list of sounds, each of the
sounds becomes a phoneme, which means, in effect, that a sound can never
change, because a sound change would be equivalent to a phoneme change.
Why do sounds need to change in a conlang?  Well, they don't *need* to, but the
language won't look realistic, because languages change over time, and there
are no perfect phonologies.  When might you want a perfect phonology?  Well,
if you're creating a proto language, you might.  If you're creating a language
spoken by non-humans, or some type of perfect creature, you also might want a
perfect phonology.  Basically, whenever the state of your language is planned
to be finite, then a perfect phonology is the way to go.  The result, however,
might be unrealistic, simply because it'd be hard for a human to use.  Imagine
a language that made the following distinctions:
ankaaNkaaJkaamka
Though it's by no means impossible to produce these words, when used in speech, it
becomes very hard for a listener to tell the difference between, for example,
/anka/ and /aNka/.  This is why nasal place assimilation is such a common
sound change.  If, however, you had a list of sounds, and had those four nasal
phonemes, you would predict that these four words could pop up, and could be
distinguished by speakers and listeners (and, in fact, this is *exactly* what
happened in Megdevi).
One place where you can make such distinctions, though, is in writing.  So just
because you have a phonological system that neutralizes certain distinctions
doesn't mean you can't have an orthography, for example, that does.  Take my
language, Zhyler.  In Zhyler there are two nasal phonemes: /n/ and /m/.  /m/
never assimilates in place to a following consonant, so you can get:
amkaamtaamSaampa
/n/, however, does.  So, if you were to replace the /m/ with an /n/ in the
examples above, you'd get:
aNkaantaaJSaampa
You don't want to say that the nasal in these words are four separate phonemes,
though: It's all the same phoneme that's just changing in particular,
predictable environments.  *However*, in the orthography of Zhyler, each of
these words would be spelled differently.
Looking over this, I'm not sure if I answered your question very directly (or at all).
 I think the main point is that a speaker of a language (and a listener) will
not attach meaning to a sound, but to a phoneme.  A given phoneme may have
only one phonetic realization (like /p/ in Hawaiian), but it may also have more
(like /e/, in Hawaiian).  Just because it has more than one sound associated
with it, though, doesn't mean you want to associate meaning to those particular
sounds.  So, even though Hawaiian has the sounds /e/ and /E/, you wouldn't
want to say that a speaker of Hawaiian would say that /kEni/ and /keni/ are
different.  To a Hawaiian speaker, these would map onto the same word: "keni"
(or "kene").  In English, though, you might say that the first would be the
name "Kenny" and the second "caney" (relating to canes?).  There are some
extra phonetic differences when going from these examples to English, but the
point is that in English we do attach special meaning t!
 o the difference between /e/ and /E/, whereas Hawaiian speakers don't.  This
is the type of thing you want to capture in a realistic phonology.  If you
want to get a better idea of what different phonologies do, you might start
looking at them, or looking at the phonologies of the languages you know.
I hope I gave you at least an idea of the issue at stake.  I'll leave it to
others to be more explicit (I'm kind of tired right now; not at my best).
-David

Reply