Hi Haukur,

>But you never know. I am told that Chinese discards the difference between its
>tones when sung. Similarily one could imagine that a language might discard
>the difference
>between short and long sounds (if the difference is really only in the
>quantity) in singing.
>Any thoughts on this, Yi-Qing?

Well, yes and no. Melodies don't reflect the Chinese tones in any way that
I can pick out, but from what I've *heard* about people who write Chinese
lyrics to given melodies, the tones are always a consideration when you're
trying to fit words to the tune. You try to choose words whose tones are
somewhat compatible to that tune. You'll hear a lot of glides from one
note to the other in traditional Chinese music, instead of straight quick
transitions from one level note to another as is usual in Western music - I
think this might have its source in the many contour tones of Chinese. I
have a feeling that most Chinese now could sing most words to most tunes
and not feel too weird about it, so maybe this is not as important as I'm
making it sound. To summarise my state of confusion, I think tones matter
in singing but I'm not sure. :-)

How does Icelandic manage the difference between long and short vowels in
singing? (I guess pre-aspiration takes care of the consonants.) Japanese
has long and short vowels and consonants, and I'm really intrigued by what
they do in the few songs I've followed with lyrics. Short vowels can get
very long in singing. But long vowels almost always get two different
notes, or if they're on the same note, the second "mora" of that vowel gets
a little extra punch so that even this ignorant foreigner can can often
tell that it's there (often it's the same deal with diphthongs, which I
guess is why the Japanese insist they haven't got diphthongs). Long
consonants are not pronounced the way they are in ordinary speech, but they
are indicated with, of all things, a preceding glottal stop: yatte becomes
[ya?ate]. I would not necessarily expect any other language to use these
strategies; I think it happens in Japanese because there's such a strict
rule that every mora in a word takes up the same amount of time (loosely
defined, a mora is one (consonant+)vowel(+nasal) group, or just one nasal,
or one long consonant. So there are 2 morae in ha-na 'flower', ha-a 'yes',
ha-i 'yes', and ha-n 'half').

This message feels SO off-topic! :-)

E-Ching