[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

'th' and phonemes




I think I'm getting some idea of the problem from the comments that are
appearing.

There are an infinite variety of sounds possible with the human mouth.
The International Phonetic Alphabet distingusihes over 100 of these sounds, 
and it is recognized as incomplete, using various subscripts and ligatures
to convey subtle variations when these are important in a language.

You, as an English speaker, can hear all of these sounds, called 'phones',
but unless you have been trained to distinguish them, you will almost certainly
classify most of them as varieties of the roughly 44 sounds that English
recognizes as distinct, called 'phonemes'  (not 'phonics' as a commercial
that often plays on a local radio station claims).

To map 100+ sounds to 44 phonemes, obviously, several sounds are mapped as
equivalent to one another in English phonemes.  The distinctions between
them are lost to you - you may not hear them.

Other languages do different mappings, equating different phones to different
sets of phonemes.  Arabic has several flavors of gutteral that English
speakers hear as a single gargle sound - the Lojban 'x'.  (INdeed German has
two sounds equated to 'x', since there is a 'phone' dfierence between the 
consonant in 'ach' and in 'ich').

In most languages, this mapping has some irregularities.  Indeed, there is
one known irregularity in Lojban that we cannot hope to train English speakers
out of, as well as the important ones that we try with some success to teach.
The 'o' in Lojban 'korbi' is not the same sound as the one in 'cunso', and few
English speakers (at least American) will make the latter sound, the official
'o' when it occurs before 'r'.  You will instead make a sound much closer to
that of English "awe".

But of course you won;t hear this, because as a trained English speaker you
hear the 'or' sound as the one resprssent by 'or' in text, and call the
sound an 'o'.  We do the same in Lojban.

On the other hand, we try to train you out of diphthongizing an 'o' as we do
in almost every English occurance of "long o".  If you say "blow", you will
say "blou" in Lojban, not Lojban "blo".  It is importnat for Lojban audio-
visual isomorphism that you not pronounce a letter so that it is understood
as a different >phoneme< by the Lojban listener.

Audiovisual isomorphism means that for every string of text, there is a
unique stream of >phonemes< and stress that is equivalent to that text.  And
for each string of phonemes, there is a unique textual representation for
that string.

Usually, we add the requirement that each phoneme corresponds to one letter
and vice versa, but this is not inherently important as long as you don;t have
an ambiguity that results because a two letter pair can be taken two ways.
Thus, for example, in this more open sense, 'sh' in English doesn't violate
audiovisual isomorphism because (save for the inevitable exceptions in English)
it is always pronounced as the single phoneme /sh/.  (Of course the reverse
isn't true, since we can get the /sh/ phoneme in "nation").  Institute Loglan
now uses 'rr' for the vocalic 'r' in le'avla and this would not violate
audiovisual isomorphism of the broad sense except that they write the same 
sound as a single 'r' in a hyphen like "saircinse".

In Lojban, we have not even gone to this broader version of audiovisual
isomorphism - one phoneme matches one letter.

It is not the goal to match single phones to letters, nor to have letters in
the Lojban alphabet to represent phone distinctions that are not made in
Lojban.  The 't'/'th' distinction is one such non-phonemic distinction.
A more 'acceptable' one is that you can pronounce the Lojban 'r' as anything
from the English 'flap', to the German 'trill' to the French 'uvular r'.
All of these are different phones, and indeed most English speakers here these
as different sounds - but we still identify them as 'r' in English, and
conveniently also in Lojban.;rGe
e!1-~

The phones of the world's languages can be mapped to Lojban, and we did so
in making the gismu. unvoced 'th' was mapped to 't', and voiced 'th' was
mapped to 'd'.  All the varieties we call 'r' in English map to Lojban 'r'.
Both Hindi 'bh' and 'b' map to Lojban 'b'.  The 'ng' of sing and the 'n' of
sin both map to 'n' (English speakers beware trying to pronounce Lojban
"sing" properly - it has a much stronger 'g' sound than you expect, like you
started to say "sing-ging" but choked yourself off at the second 'i').
This does not violate audiovisual isomorphism becuase you and/or a computer
can go both ways without any ambiguity.  You pronounce the 't' as 't' or 'th'
as it suits you because in Lojban THEY ARE THE SAME SOUND.  And you hear
t and th spojken in a Lojban word as Lojban 't' and nothing else.

Does this clarify anything?

lojbab