On Fri, Jun 15, 2012 at 1:20 PM, Alex Fink <[log in to unmask]> wrote:
> Well, starting at "sounds" is, charitably, the naive way about it.  If you think that phonemes are real, then what argument is there not to start at phonemes?  You'll have to learn their realisational rules to speak the language; it's no extra burden to have to rely on that knowledge in reading it.
> (Or are spectrograms the optimal writing system?)

Sorry, I just wanted to use the smallest (or more general) measure
possible. Maybe one woudl like to use a symbol for each formant a
phoneme uses. Who knows?

> >4. Complex, but yet rule-driven: The same as 3, but the context to be
> >considered is far larger than the one in 3. There are several exceptions.
> >English belongs here. One can even create a rule for "gh" pronunciation (If
> >at the end of a word, preceded by V+"u", but the consonant preceding V+"u"
> >is either "c","l", "r" or "t", then it is pronounced /f/. On the other
> >hand, if the consonant preceding V+"u" is either "d","n" or "th", it is
> >silent.);
> I think it's very bad descriptive practice to consider that the "rule", as opposed to just saying that -ugh has multiple readings and it's _lexically_ determined which one to use.  The initial consonant has nothing to do with it historically, nor anything in the minds of current speakers, as far as I can tell.
> If you're going to try to use formal properties, what tells you that _slough_ 'marshy region' is /slaU/ but _slough_ 'cast off skin' is /slVf/ -- or what tells you which of those should count as the exception?  If the initial consonant was criterial, how was it that _furlough_ shifted from /-lAf/ to /-loU/, with that ‹l› there favouring /f/?  Etc.

I understand your point. I just wanted to say that English is somewhat
predictable. <l> surely is not favouring /f/, but it is a way of
creating these rules. There are only two possible ways of seeing this:
Either English spelling system is fair enough and rules can be
inferred from it (that is what some people here in this thread said),
or it is unpredictable and there is no way of deducing the
pronunciation of a given word (Sorry for the ones in the future that
will have to learn lots of neologisms that will appear from now on).

It is the same for fury x bury. One would expect "ury" to be
pronounced as /jʊrɪ/, but if the consonant preceding it is <b>, then
you have "bury", so "ury" pronunciation would become /ɛrɪ/.

> >5. Most complex: No rule can be created. It happens with semanto-syllabic
> >scripts, like the ideogrammic Chinese writing.
> No present natural language has an "ideographic scripts" as its current main script (though certain forms of proto-writing were ideographic).  Chinese writing, like any other, writes not abstract ideas but Chinese words.  There are of course some complications, many-to-one and one-to-many correspondences, the fact that the script has ignored the internal divergence of the Chinese language family, etc. -- but when are there not?  The modern term for such scripts is "logographic".

I used the wrong word. It should be logographic. Sorry.

> And "no rule"?  Again, a big table is a rule.  It doesn't capture all the correlations which could be extracted about e.g. the modern phonetics, but you could say your rule is this:
> ‹一› is read _yi1_, but _yao1_ is acceptable in a list of digits;
> ‹丁› is read _ding1_;
> ‹七› is read _qi1_;
> etc. for thirty thousand lines.
> Consider the development of the Yi script as an indication that there isn't really a bright line between this and a syllabary:

What I tried to say in my last e-mail is that this "big table" can be
applyable for English (or to any other language), but one would have
to create a new row each time a new word is created.
Chinese would have a finite table (unless new characters start being
created). On the other hand, English and other languages with
non-logographic scripts and with unpredictable pronunciation would
require an infinite one.

> Alex