Re: Optimum number of symbols

On Thu, 23 May 2002 15:06:24 -0400, David G. Durand <dgd@...> wrote:

>At 6:28 PM -0400 5/22/02, Mike S. wrote:
>>The problem is that the compellingness found in this argument
>>proceeding from the German example does not extend well to
>>English or French; written English's overbearing insistence
>>on morphemics and written French's overbearing insistence on
>>marking inflections long since disappeared from speech, while
>>useful in some ways, are clearly not worth their cost in terms
>>of learnability or ease of usage, or at least I think very few
>>people would argue otherwise.  I personally think both of these
>>systems are atrocious, and this atrociousness stems from their
>>*non*phonemic charcteristics that, on balance, are not
>>enhancements in the least.
>
>Actually, I think that many people would argue otherwise -- this is a
>contributing factor in why these systems have not been reformed. For
>an educated speaker of English, the morphemic representations in
>English give useful information about the structure of unknown or
>infrequently encountered words. Even in English, regular rules can
>explain the pronunciation of most words.
>
>The morphemic representation also strikes a compromise in the
>representation of English dialects, in which vowel quality is highly
>variable. The writing system  provides distinct spellings for
>distinctions that are not universally present. This enhances written
>communication in English, and acts as a unifying force on the
>English-speaking (and writing) community. English dialects vary not
>just in phonetic inventory, but phonemic inventory, so even a "pure"
>phonemic system would introduce spelling variations.
>
>Finally, I'd like to expose an issue that you haven't discussed,
>which is that you probably don't just want a system that is phonemic
>(each significant graphical unit represents a phoneme) but one in
>which each phoneme has only one written representation.
>
>Because my wife is Greek, I've been learning a phonemic orthography
>for Demotic Greek. It has a fairly simple 5-vowel system, that writes
>simple vowels in 13 different ways. There are also a number of
>variant ways to write consonants, and letter (like Upsilon) that
>represent either a consonant, or a vowel, or that modify the
>pronunciation of another letter, depending on context.
>
>The phonetics of Demotic are also interesting, in that there are a
>lot of rules of palatalization that make the letter->phone
>correspondence complex. For a native speaker, presumably, these rules
>are relatively transparent.
>
>But back to the main point, I think that English spelling has
>contributed to the unity of the language across the world, and
>prevented, or at least retarded, the development of a range of
>English-derived national (or regional) languages. As an English
>speaker, this is in my self-interest.

I'd like to preface my remarks with an URL:
http://www.arts.uwa.edu.au/LingWWW/LIN101-2001/Eng-Spelling_Poem.html

In regards to my objection to English orthography, you seem to
be making a case for the morphemic principle against the case
for the phonemic principle.  Aside from that, I just wanted to
emphasize that, regardless of the outcome of the phonemics/morphemics
debate, it must be admitted that in English there are hundreds
of words whose spelling simply can't be justified.

The morphemic principle has indeed historically provided a great
service insofar as it provided an orthographic bridge to the
classical world, but as Thomas Wier pointed out, with the decline
of classical education, this benefit has greatly attentuated
in recent years.  In addition to that, the morphemic principle
serves to distinguish homonyms, and to graphically unify allomorphs
which have diverged primarily due to the Great Vowel Shift.
I can't dispute the benefits of these functions, but I certainly
can call attention to the fact that they do not come without a
significant cost: namely the additional tax on ease of learning.
And as I have pointed out, the whole idea that we should insist
on burdening writing with the task of disambiguating language
in ways that speech does not, instead of maintaining the dead
simplicity of a pure phonemic approach, is really not self-evident.
Accommodating dialects, to be sure, is yet another matter.

Aside from these attenuated benefits of morphemics, the only
justifications for our crazy spelling system are the ones I have
already mentioned: accessibility to the existing corpus (the most
massive problem), the training of the entire population in the
existing system, and just plain conservative sentiment.

Interesting, after you mention a couple problems that would face
spelling reform, you immediately name the solution that I myself
would implement if I had a say in any spelling reform.  The two
problems are maintaining the already-mentioned morphemic principle,
and accommodating the dialectal variations (most notably the rhotic
variations).

These problems can be both be accommodated in the spelling system
in almost exactly the manner you mention: by creating a regularized
version of what I call an "allographic" system.  The operating
principle of an allographic system is that, like a pure phonemic
system, the spelling would render one phonemic value (or one
per dialect), but unlike a pure phonemic system, there may be
several ways to spell any given phoneme.  This is achieved
by assigning several graphemes and diagraphs ("allographs") to
each phoneme.  The mapping between allograph and phoneme does
not have to correspond exactly across dialects; each dialect
can interpret the allographs as it wishes.  Furthermore, it will
be possible in most cases to maintain graphical distinctions
between morphemes as we do now.  However, we would have the
opportunity to rid our language of hundreds of unforgivables
like "bury", "rough", "debt", "women", etc., ad nauseam.

Thus, the two concerns you have, morphemic recognition and
world unity of the language, are fully compatible with a spelling
reform.  Moreover, the system I have in mind would be maximally
readable to the existing population, given the design criteria.
The benefits, I feel, would be enormous: an *immense* increase
in ease of learning for both children and non-native speakers.

I will endeavor to provide the details of my plan in a future post.



>>  >IMO no type of writing is per_se superior to another - it depends on
the
>>>language and, to some extent, what you're writing.  The Linear B
syllabary
>>>appears to have been a satisfactory way of recording early Greek for
>>>clerical purposes; but few would deny the later alphabet gives less
>>>ambiguous representation and is, for Greek, more suited for general &
>>>literary contexts.  Nevertheless, it is notable that, altho the alphabet
>>>had been adopted by the Greeks as early as the 8th cent BC, it continued
to
>>>be written in a syllabary in Cyprus as late as the 2nd cent BC.
>>>
>>>But a language with a simpler phonology than ancient Greek, and with a
>>>proponderence of open syllables might well be written more efficiently
in a
>>>syllabary.
>>>
>>>At any rate, as this interesting thread continues, I am coming more and
>>>more to favor the idea of a syllabary for BrSc  :)
>>>
>>>Ray.
>>
>>I agree that the details of an optimum writing system will
>>tend to vary among languages, but I do have to question whether
>>we are really compelled to apply automatically this "nothing is
>>superior to anything else" concept everywhere, no matter how
>>difficult or inefficient a system appears to be.
>
>Sure, but.
>
>Another factoid for the mix: subtitles for movies in Chinese and
>Japanese are much more complete than those in alphabetic writing
>systems, because reading speed for logographic scripts is higher in
>terms of words-per-time-unit. The inarguable difficulty of learning
>such a system does have some payoff. Further, even more than in the
>case of English, the Chinese writing system unifies a set of _very_
>divergent dialects, that would are mutually unintelligible at the
>phonemic level.
>
>The point is that the criteria for optimality are so divergent,
>across cultures and languages, that even objective facts are
>susceptible of differing interpretation.
I agree that you have to have particular criteria in mind when
making a judgement.  But there only a small handful of criteria
that crop up regularly in these discussions: ease of learning,
readability, writeability.  In addition to these, I have offered
universality (see posts to Christophe yesterday, Raymond today).

Sometimes I am tempted to believe what you just mentioned regarding
logographic script being fastest to read.  I am also tempted to
believe this may well be the most important criterion; you only
need to learn once, but you read numerous times. In a constructed
language, you can ensure that each morpheme is phonemically unique.
Then you can construct characters for each morpheme based on their
sounds.  I wonder if this indeed would not be the ideal approach
towards a writing system for a highly regularized conlang.



>We are also hampered from any true evaluation on these matters
>because we can't evaluate writing system types across languages very
>easily, because it would be immoral to train children to fluency in a
>non-standard writing system _only_. Second-writing system learning is
>likely to be very different. And even if we wanted to experiment on
>adult learners, development of fluency is difficult and
>time-consuming; practically, we're unlikely to be able to perform
>these kinds of experiment.
Agreed.

The only way I think we might be able to get something close
to objective is if we can locate persons fully bilingual from
childhood and ask them their *opinion*, on, for instance, which
language do they find is it easier to read in.  If you ask enough
people you can see if any patterns emerge statistically.

The second best option is seek those who have become fluent
as adults.



>>It is clear to me that scripts arose through a centuries-long
>>evolutionary process, in which enhancements were gradually added.
>>The development of consonant-only syllabary (if I may call it that)
>>represented a major step in this process;  the Greek innovation
>>to add vowel letters to their alphabet represented another.
>>Both these cases, IMO, marked a definite objective improvement
>>over the older systems.  This is not to do not deny the fact
>>that the older systems were adequate;  I do claim however
>>it is possible to make relative utilitarian judgements
>>in certain areas.
>
>In at the case of some semitic languages, the writing of vowels,
>while commonly taught in schools, is not used in practice, evidence
>that this innovation has not paid off in some communicative
>environments.
I am hearing from just about everyone that these vowels just didn't
matter much.  Interesting and unusual.



>To really answer these questions would require controlled (and
>difficult) experiments that have simply not been performed. Some of
>those experiments, are IMHO immoral to perform, just as it would be
>immoral to raise a child in an L1 environment designed to test the
>reality of linguistic universals of various sorts.
>
>   -- David

Regards

Reply