Re: How to Make Chicken Cacciatore (was: phonetics by guesswork)

On Thu, Jul 22, 2004 at 05:49:02AM -0700, Philippe Caquant wrote:
> Ah, great. In fact, I didn't know of that word "phone"
> meaning just something like "utterable sound", that's
> why I was using "phoneme" instead. Now I see better.
Right.  If two phones distinguish words in some language, they are both
phonemes in that language; if one can be substituted for the other
without changing the perceived word, they are called "allophones" of a
single phoneme.

As with just about everything else, "some language" really means "some
dialect of some language".  In some French dialects, the distinction
between /e/ and /E/ has been partially neutralized, in others it may
have been completely neutralized, while in others still it may matter
everywhere.

> - X-Sampa refers to many and various languages, so
> clearly one cannot say that X-Sampa handles about
> "phonemes", but about "phones".
(FWIW, "handles about" is nonstandard; "handles" doesn't normally combine
with prepositions.  In this context I would go with "represents".)

Yes, the symbols of the IPA (and therefore X-SAMPA, which is designed to
be a one-to-one mapping from the IPA to an all-ASCII representation)
represent phones.  And between square brackets, that is all they
represent.  However, between slashes, they are being used to represent
phonemes of some language.  Usually, a phoneme represented by some
symbol /X/ has an allophone in some dialect that sounds like the phone
represented by [X], but this is not always the case.  For instance, /r/
is often used to represent the English rhotic even though [r] is rarely
heard in that language; /r/ comes out as [4] or [r\] in most dialects.

As we just learned yesterday, an extra slash means there's an archiphone
involved.

> - nevertheless, every single X-Sampa character should
> (at least in theory) refer to one real phoneme at
> least in one particular natlang, meaning that this
> phone is a distinctive phoneme in at least one known
> language.
Yes, although the reverse is not necessarily true.  There are phonemes that
have no separate X-SAMPA character, for instance /p/ vs. /p_h/ (a
distinction made in many Asian languages).

> - in French, there is an "open o" and a "closed o",
> although this very much depends of the area, of the
> social and cultural level, etc. But I cannot see any
> example where this distinction is relevant in meaning
> (I may be wrong) : so these are simply phones, as far
> as French is concerned. But in some other language,
> these could be phonemes.
Yes.  In dialects of French that don't care about the distinction, the
two o sounds [o] (closed) and [O] (open) are allophones of a single
phoneme (which one would normally spell /o/).

> - in French, there is an "open e" and a "closed e",
> and this is relevant in meaning: "sais" is something
> different from "ses", or "ces", so these are phonemes
> in French (and probably other languages).
Yes, for many French speakers open [E] and closed [e] are two phonemes.
And while I have that distinction in my pronunciation of French
(what's left of it since French class in high school, at any rate),
"sais", "ses", and "ces" all have closed [e] for me.

> So we could see at once that, ex, "mild th" exists in
> English and in Icelandic, but not in French; and "z"
> exists in English and in French, but not in Spanish;
> and yet, those phones being close to each other, if
> you use one instead of another in a particular
> language, it will not be catastrophic, people might
> understand you after all.
Actually, [z] exists in Spanish; it is an allophone of /s/, and shows up
in words like "mismo" /mismo/, which is realized as [mi:zmo] in many
Spanish dialects.  And although "mild (voiced) th" (X-SAMPA [D]) doesn't
exist in Spanish, voiceless "th" (X-SAMPA [T]) is the standard
Castilian pronunciation of |z|, so it should be pretty easy to train
a Castilian Spaniard to pronounce [D].   I wouldn't be surprised if
[D] showed up as an allophone of /T/ in some contexts.

But yes, there's enough redundancy in most natlangs to make replacements
such as [z] for [D] survivable most of the time.  Fortunately. :)

-Marcos

Replies