Tristan Mc Leay wrote:

> > then again, there's finnish, which has frontness-harmony, with the  > 
>vowels [y], [2], [&] as front and [u], [o] and [A] as back. the
> > vowels [i] and [e] are neutral. yet how would an infant
> > acquiring the language pick up on this? there is a somewhat weak
> > featural way of describing the difference, i guess.

>it seems to be more than just a lexical-type process
>like noun genders, it's rather a neutralisation of phonemes in
>unstressed syllables. As I understand it, Finns don't even
>hear/produce the difference between [y] and [u] there e.g. "olymia"
>is for most people pronounced ["olumpia], not *["olympia]

Schwaaaaa? Of course we do hear the difference - it's just that eg.
/olympiA/ or /pAst2roidA/ ("to pasteurize") sound *wrong* to our
harmony-attuned ears, so most people assimilate them. I can even
provide a few minimal pairs of harmonic native word vs. non-harmonic
loanword:

/myonia/ <> /muonia/
muon|PART. <> food provisions|PL|PART

/pyro/ <> /puro/
fire- <> creek

/s2rkkA/ <> /sorkkA/
(a part of Helsinki; colloquial) <> cloven hoof

/pyyt2n/ <> /pyyton/ <> /puuton/
hazelgrouseless <> python <> treeless

"Olumpia" just happens to be one of the more common assimilations, probably 
because sports reporters tend to concentrate on speed more than accurate 
pronounciation. Most if not all Finns are perfectly capable of producing 
this sort of a distinction if they want to. In my experience, the difficulty 
is the greatest with two same-height vowels separated by one consonant 
(directly adjacent it looks so bizarre that people tend put more effort into 
pronounciation then.)

...And there're also the few native words which do violate the harmony 
rules. Most just add back suffixes to neutral roots (a special case of which 
are the 1P and 2P singular pronouns which end with <ä> in the nominative but 
have <a> in its stead all inflected forms); then there's at least 
<tällainen> ("this kind of") and <hyla> ("low-lactose"> which were 
originally compound words but have been
shortened until somewhat non-transparent.
OTOH, /A/ and /&/ are in some dialects starting to merge into /a/ when 
unstressed, and this AFAIK would eliminate practically all of these 
exceptions.


>No doubt we can blame the absence of /M 7/ in stressed syllables for
>the neutral behavior of /i e/. So if Finns don't pick up on/use [+/-
>front] in unstressed syllables, and if there is no [-back +high
>-rounded +short] phoneme, then [+high -rounded +short] has
>no choice but to equal /i/, whereas [+low -rounded +short], which
>remains ambiguous, is given a final interpretation based on the
>syllable's stressed vowel.
>
>--
>Tristan.

/i e/ are certainly not backed to [M 7] in back-vowel words, not even [i\ 
@]. There IS some backing in back-vowel words, even when stressed, but it's 
much subtler. Furthermore, [M i\ 7 @] would actually be perceived as /y y 2 
2 - kinda weird, isn't it? :)

John Vertical