Newsgroups: sci.lang
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!uunet!bcstec!charlotte!snake!rwojcik
From: rwojcik@atc.boeing.com (Richard Wojcik)
Subject: Re: languages with phonetic alphabets?
Message-ID: <1995Mar31.161929.8903@grace.rt.cs.boeing.com>
Sender: usenet@grace.rt.cs.boeing.com (User that posts news)
Reply-To: rwojcik@atc.boeing.com
Organization: Research & Technology
References: <D69sFw.9r6@spss.com>
Date: Fri, 31 Mar 1995 16:19:29 GMT
Lines: 50

In article 9r6@spss.com, markrose@spss.com (Mark Rosenfelder) writes:
>In article <D64o7r.Jts@eskimo.com>, Richard Wojcik <rickw@eskimo.com> wrote:
>>At this point, we need to consider the nature of phonemic mergers and how
>>they ought to be represented.  What is the optimal way of representing
>>phonemic overlap in alphabetic writing?  Sanskrit happens to use a more
>>shallow representation, but most of the world's writing systems seem to
>>prefer a deeper representation.  Hence, we do not tend to represent vowel
>>reduction in our formal spelling system, although it is the cause of many
>>spelling errors, both in English and Russian.  If you were to represent
>>Russian "with" in "s Borisom" with a "z" letter, you would be more accurate
>>from a phonemic point of view, but you would impede morphological
>>recognition.  
>
>Why would this be more accurate from a phonemic point of view?  Are you
>saying that the Russian preposition is sometimes /s/, sometimes /z/?
>Why not say that it's always /s/, but is realized as [s] or [z] depending
>on the phonetic context?

That is one possibility, and it is certainly what early phonemicists such as Baudouin
and Sapir did, not to mention the later Moscow School phonologists.  Indeed, I tried
to convey this point, perhaps too subtly, with the following passage.  This is the
material that you deleted from the above paragraph.

    So it is best to keep the writing system the way it is, but
    it further removes the script from being a pure correspondence with
    phonemes, at least in a superficial sense.  (Few modern linguists follow
    Baudouin's original use of the term "phoneme" to mean the most abstract
    phonemic representation of the sound in a morpheme.)

In other words, I was using "phoneme" in the more modern sense to represent
a superficial sound categorization, rather than the sound that speakers might
assign to the basic phonological representation of a word.

>BTW, do Russian speakers ever write <3> for an /s/ realized as [z]?

Good question.  I suspect that children do.  English-learning children tend to make
some interesting errors with phonemic neutralization in their language.

---
+--------------------------------------------------------------+
| Richard H. Wojcik               |  Boeing Computer Services  |
| Senior Principal Scientist      |  P.O. Box 24346, MS 7L-43  |
| Natural Language Processing     |  Seattle, WA 98124-0346    |
|=================================|  Phone: (206) 865-3844     |
| Opinions expressed above are    |  Fax:   (206) 865-2965     |
| not those of The Boeing Company.|  Email: rwojcik@boeing.com |
+--------------------------------------------------------------+



