wales@ucla-cs.UUCP (11/11/84)
Quite a few languages have writing systems which -- if not completely
phonetic -- are good enough that someone with a fair knowledge of the
language should have very little trouble with spelling.
Some of the "best" writing systems I am aware of is that of Finnish.
There is a very close correspondence in this language between spelling
and pronunciation. I am told that there is one aspect of Finnish pro-
nunciation which is not represented in the writing system -- a phenom-
enon called "aspiration" which occurs at the ends of many words, and
which manifests itself as a gemination (doubling) of the initial con-
sonant of the next word.
Turkish and Hungarian also have good writing systems. I don't know any
juicy details about Turkish, but I have studied some Hungarian. In the
case of Hungarian, there are at least three things which can create
strangenesses in the writing/pronunciation correspondence. Perhaps it
would be worth it to consider these issues, since some of them are ap-
plicable to other languages too:
(1) Several consonant sounds are indicate by digraphs -- for example:
==> "sz": "s" as in English "saint"
("sh" as in "shine" is spelled "s"!!)
==> "zs": "z" as in "azure" (e.g., the name "Zsa Zsa")
As a result, you can sometimes have "ambiguous" combinations in the
middle of a compound word, or a word with a suffix. For example,
the word for "health":
' ' ' '
egesz (healthy) + -seg (-ness) --> egeszseg
(The accent marks indicate vowel length and -- in some cases --
vowel quality -- but not stress. Stress in Hungarian words is
always, always, ALWAYS on the first syllable.)
This quirk could admittedly be solved by a revision of the writing
system. If you know the language, though, you have no problem tell-
ing that the "szs" in the middle of this particular word is "sz"+"s"
and not "s"+"zs" -- so there is no urgently felt need to fix things.
For that matter, if you were going to be picky, you might want to
respell the above word with "ss" (lengthened "sh" sound), since that
is the way the word is often pronounced in common speech. But no
one would really want to do that, because such a respelling would
conceal the origin, structure, and meaning of the word.
(2) In mixed clusters of voiced and unvoiced consonants, all the con-
sonants in the cluster are generally voiced or devoiced according
to the character of the last consonant. For example, the word
"megfelelni" (to answer, suit, conform) is pronounced as if it were
spellet "mekfelelni".
One could make the spelling conform more exactly to the pronuncia-
tion by decreeing, for example, that the extremely common verbal
prefix "meg-" be spelled "mek-" before voiceless consonants -- but
this would create all kinds of problems and drive just about every-
one up the wall. In this particular case, for example, you have to
understand that the "meg-" prefix can be (and, depending on what
else is in the sentence, often is) separated from the main part of
the verb -- so it might sometimes be "meg" and sometimes "mek". It
is much more natural to spell it with "g" all the time, since the
underlying form is "meg" and simply loses the voicing of the final
consonant in certain phonetic environments.
Again, there is really no need to change the spelling -- as long as
the speaker is aware of the structure of the words (which is not too
much to ask in an agglutinative language like Hungarian), there is
no real problem.
(3) The sound "y" (as in English "yes") is sometimes spelled "j", and
sometimes "ly". The "ly" digraph used to represent a palatal "l",
but in modern Hungarian this sound no longer exists.
The existence of two spellings for this one sound does create some
genuine problems for native speakers of Hungarian. I have been
told, however, that the old "ly" sound became a regular "l" in some
rural dialects. Hence, a native speaker can often tell which spell-
ing to use by asking himself or herself whether the word in question
would "sound OK" with an "l" in place of the "j"/"ly" -- if so,
spell it with "ly". Needless to say, this guideline is no help at
all to a non-native student of the language!
Point #2 above (regressive voicing/devoicing of consonant clusters) is
applicable to many other languages as well (German and the Slavic lan-
guages, for example). The thing to remember in all those cases is that
the pronunciation of a given sound in a given morpheme may change in
very predictable ways depending on the environment -- so that it may be
better to stick to a single "underlying form" or "morphophonemic" spell-
ing rather than try to show all the "automatic" variations in all the
different contexts.
--
Rich Wales
UCLA Computer Science Department
3531 Boelter Hall // Los Angeles, CA 90024 // (213) 825-5683
ARPA: wales@UCLA-LOCUS.ARPA
UUCP: ...!{cepu,ihnp4,trwspp,ucbvax}!ucla-cs!wales