LINGUIST List 2.217

Sunday, 12 May 1991

Disc: Phonology and Orthography

Editor for this issue: <>

Directory

1. Mike Hammond, Possible Words of a Language
2. , Re: Phonology and Orthography
3. , rule/list dichotomy
4. John Coleman, RE: Phonology and Orthography
5. , Letter names
6. , Data in Linguistics

Harry Bochner responds to my contention that "determining what the
'possible words' of a language are is a linguistic question and
determining what the occurring words are is largely a nonlinguistic
question, hinging on history, technology, etc."

He maintains that he is interested in characterizing the class of
"acceptable" words, not the set of "occurring" words. I think we may
be agreeing here, but we may not. If by "acceptable", we mean words
that the grammar accepts as well-formed, then we agree. If by
"acceptable", we mean what syntacticians mean by "acceptable", as
opposed to "grammatical", then we're interested in different things.

I take the syntactic use of "acceptable" to refer to sentences that
are parsable, though not necessarily well-formed with respect to the
grammar. As a linguist, I am more interested in the words that the
grammar says are well-formed, rather in the words that the parser can
figure out.

Harry's second point that the distinction between "rule-generated"
and "completely idiosyncratic" breaks down if we allow rules to apply
to only a single form is well-taken I think. I think we gotta eschew
that kind of analysis.

mike hammond

Richard Ogden writes:

> Phonology needn't have anything at all to say about spelling. It is there
> as an abstraction from the phonetics - and the spoken language should
> be the prime source of enquiry, not the written one...

I did not say that orthography should be a prime source of enquiry, although
there is no obvious reason why it should be unimportant. After all, Ogden
himself sees a connection between spelling and phonetics. We could explore
that issue further--e.g. what he thinks about the question of representing
so-called allophonic variation such as aspiration and vowel nasalization in
an idealized English orthography. I like to think that such questions might
drive him to think more highly of a phonemic-orthographic correspondence.
And, although Ogden didn't use the "I-word", I do believe that most of his
objections to my position have been based solely on his intuitions about the
nature of phonological theory. (Touche, Richard! :-)

> The connection with spelling is by-the-by. Perhaps phonemics has something
> to say about spelling but phonemics is certainly not the whole of 
> phonology (despite the impression one might get from reading most 
> books on phonology). Many linguists abandoned the phoneme long ago -
> is Rick WOycik seriously saying that non-segmental, non-phonemic
> phonological theories don't say much about phonology just because they
> are independent of the writing system?

I agree that phonemics is not the whole of phonology and disagree that most
books on phonology say as much. In fact, that would be rather strange in a 
linguistic milieu where phonemic theory is essentially dead. In fact, I was 
not aware that there was such a thing as a phonological theory that denied the
existence of segments. Proponents of such a theory would indeed find the
existence of alphabetic writing to be a puzzle! (Well, maybe they are not very
curious folks. Who knows? :-) BTW, I would be interested to know what 
explanation Richard would give for his consistent misspelling of my name. As
a phonemicist, I offer the following explanation: He knows enough about
Polish spelling to realize that the "j" letter corresponds to the /y/ phoneme.
This phonemic correspondence causes him to substitute the English "y" letter,
which corresponds to the /y/ phoneme in English spelling. How would a
non-segmental theory explain the phenomenon? Or would it even bother? (My
name is pronounced /wojIk/ in English and /vuyCik/ in Polish.)

Lesli LaRocco questions my use of the expression "prototypical alphabet" on
the grounds that most alphabets are, in fact, borrowed. I was using the term
"prototypical" more in the sense of so-called prototype theory, but a better
term might have been "ideal alphabet". I think that most speakers would like
their alphabetic writing to have that primitive correspondence. We have all 
met English speakers who ponder whether or not they should pronounce the "w" in
"sword" or the "p" in "psychology". Discussions of spelling reform tend to
bring out the ugly truth about our orthographic ideals. Such discussions only
make sense in the context of an implicit phonemic theory--under the mistaken
assumption that we all can agree on the ideal phonemic inventory for English.

 -Rick Wojcik (rwojcikatc.boeing.com)

Let me further reply to Mike Hammond in re rule vs. list. In a
paper delivered at the recent phonology conference at the
University of Illinois, Morris Halle argued for the non-exclusivity
of rule vs. list representation. He noted that, at least for
morphology, there are suffixes whose addition (and effect on words)
are totally rule-governed, but that words with those suffixes must
nevertheless be listed in the lexicon *with* the words. Thus,
although -ian is a regular adjective-forming suffix (cf Bostonian),
it must also somehow be blocked from applying to London (*Londonian).
Similarly for *Shakespearic (cf existing Homeric). His suggestion
was that speakers extract rules, but nonetheless store the existing
forms separately. Similarly, notice that `homicide' is the *act* of
killing a man, but `insecticide' is the *substance* used for
killing insects. Thus, both by distribution and by semantic
considerations, derivational morphology must, at least in part,
be both listed in the lexicon (to account for these problems) and
represented as rule-governed (to account for (partial) productivity.)
 Geoff Nathan <ga3662siucvmb>
Southern Illinois University at Carbondale

Rick Wojcik writes:
> If we accept that the prototypical alphabet is in one-to-one corespondence
> to phonemes...
I would endorse Lesli LaRocco's remark about alphabets often being borrowed,
and add that given the non-uniqueness of phonemic analyses of a language,
any one-to-one correspondence between THE "prototypical alphabet" and 
some set of phonemes is going to be rather elusive, isn't it?

--- John Coleman

Bob Hoberman's remark about the Turkish letter name with a velar
fricative (a voiced one) is interesting, but it is interesting to
note that this is NOT the sound that is represented by the 
letter in question in so-called standard Turkish. Hence, I suspect
dialect interference. 

More important perhaps is the point is that in the context of
discussing Baudouin's theory of the phoneme, it does not make
sense to dismiss certain examples as "outside the system", since
he believed that all of automatic phonology was anatomically
determined!

Those who have been following the recent discussions of the history
and present state of phonology may have been puzzled by the recent
exchanges regarding the name of the Russian letter commonly
transliterated as 'y'. In particular, by the fact that Rick
Wojcik, who is an old friend of mine and knows perfectly well that
I know Russian rather well, would not accept my statement about
the name being 'y', until he had received independent confirmation.

However, what Rick was doing was in fact exactly the right sort of
thing, and I would like to commend him for this, since it is
something that we do all too rarely. Unlike many other sciences,
linguistics almost never bothers about reproducibility of results,
and all of us know the consequences. Rather than relying on your recollection
of what someone told you a year ago in the men's room at the LSA about
a language they may or may not know much about (or on one's so-calle
d intuitions, which is perhaps worse), we should all do as Wojcik
has done, and demand independent expert testimony on every point 
of fact.

[End Linguist List, Vol. 2, No. 217]