LINGUIST List 2.245

Monday, 27 May 1991

Disc: Phonology and Orthography (Part 2)

Editor for this issue: <>

Directory

1. Dr M Sebba, letter names
2. Margaret Fleck, Syllabaries and "syllabaries"
3. Paul Saka, formalism, 2d try

I have two observations to add on English letter names - just to add to
the database, as I'm not sure what all this may prove.

1. 
Reading out the English alphabet, for non-rhotic speakers like me,
/r/ always ends in /r/ (the linking /r/): /a:r es ti:/ etc.; but when
referring to the letter, e.g. in spelling out my own name, I can
include the /r/ in the letter name for emphasis: /em ei a:r kei/;
otherwise, the letter name is /a:/. Normally, of course, I wouldn't
pronounce an /r/ before a /k/ or a word boundary.

2.
For many speakers of English in the North of England, the name of /h/
is "haitch", with initial /h/. I suspect that for these same speakers,
word-initial /h/ is rare or absent, and that the letter name may be
the only word with initial /h/ for them. Since word initial /h/ is a
famous linguistic variable in Britain, I think this shows, if it shows
nothing else, that phonology is not free of sociolinguistic
considerations - as if it weren't complicated enough already!

Mark Sebba
Dept. of Linguistics
University of Lancaster, Lancaster LA1 4YT, England
Telephone (0524) 65201 ext. 2241 (W) (0524) 69223 (H)
Fax: (0524) 843085
e-mail: eia023uk.ac.lancaster.central1

To John Coleman:

Having had personal experience with writing systems that are largely
ideographic and largely syllabic, making a big fuss about the
difference between the Arabic and Latin alphabets looks like
nit-picking. I don't see that it matters much whether you put the
vowel symbols on top of the consonants, between them, or merged with
them in a (semi-)regular way. It's still about 30 distinct symbols to
memorize and a lot easier to learn than the alternatives. 

In real syllable or demi-syllable based systems, like Akkadian,
Japanese, and (I believe) Cherokee, there is a distinct symbol for
each syllable or half-syllable and there are few regularities in the
structure of these symbols. If the language allows all (C)V(C)
syllables and has 5 vowels and 20 consonants, that means 2205 symbols
in a pure syllabary and 205 if you represent CVC syllables as CV-VC.
If you assume some CVC symbols (as in Akkadian), 300 symbols might be
a good back-of-the-envelope figure for a syllabary for an average
language. It is a major pain in the neck to learn such writing
systems.

In the Near East, the main trend seems to be replacement of
ideographic (e.g. Egyptian and Sumerian) and syllabic (e.g. Akkadian,
Hittite) systems by alphabetic or near-alphabetic ones (e.g. Greek,
Arabic). This pattern holds within both the Semitic languages
(Akkadian dead, replaced by e.g. Arabic) and the Indo-European ones
(Hittite dead, replaced by e.g. Greek).

Near-total deletion of vowel indications is a rather different matter.
It is made possible in some Semitic languages because much of the main
root structure is consonantal. Dltng vwls n nglsh mks t dffclt t
ndrstnd. I would be surprized to find (does anyone know?) the same
wholesale omission of vowel indications in non-Semitic languages that
have borrowed e.g. the Arabic alphabet. Even within the Semitic
family, users of Akkadian felt it necessary to indicate vowel type on
an obligatory basis. I don't see that deleting vowel indications
makes an alphabet into a syllabary.

Margaret Fleck (fleckrobots.oxford.ac.uk)

 In LINGUIST 2.191, John Coleman quotes my earlier message:
> If [an] algortihm is formal in Manaster-Ramer's "form" sense
> (in the sense that its operations are defined over shapes),
> then the algorithm is [or tends to be] also formal in the "rigorous"
> sense: there is much less room for disagreement about the SHAPE of
> an object than there is about the INTERPRETATION of an object.

 With this Coleman takes issue. Some of his remarks are unclear
to me, but the nub seems to be this:
> in the absence of agreement about the interpretation of the forms
> which a formal theory proposes, there is plenty of room for disagreement
> about what the shape of objects in the theory should be.
This, to my mind, contains a couple of related misunderstandings.
 
 First, I was describing what the relation IS between rigor and
attention to shape. I had absolutely nothing prescriptive to say about
SHOULDS or SHOULD NOTS.
 My second point is best understood by means of an example.
Let's consider formal rule 1 and informal rule 2:
 (1) [+nasal] --> [alpha velar]/___[alpha velar]
 (2) Nasals assimilate to the succeeding sound.
Rule (1) is nothing more than an abbreviation for the following:
 Replace every sequence of "[+nasal][+velar]" with "[+nasal][+velar]".
 [+velar]
 Replace every sequence of "[+nasal][-velar]" with "[+nasal][-velar]".
 [-velar]

Granted, in order for theory (1) to have any empirical significance,
shapes like "[+nasal]" need ultimately to be interpreted. BUT, in order to
calculate the logical consequences of theory (1), you can simply push the
symbol "[+nasal]" around, as if it had no meaning, in accordance
with the given rules. In contrast, you cannot draw consequences from
(2) by means of purely syntactic manipulation. 

 (For more on algorithms and their interpretation, I highly
recommend:
 Lewis Carroll's "What the Tortoise Said to Achilles", Mind 1897.
 John Haugeland's _Artifical Intelligence_, MIT, 1985.)

 It is important to distinguish what a theory IS 
from what a theory is ABOUT. I have been talking about the former;
it would seem that Coleman was sometimes talking
about the latter.