Re: Expansion/Contraction

Albert ARIBAUD wrote:
Mok-Kong Shen wrote:

In the context of their table "varies" means not only "measured, and
known to vary" but also "can be either expansion or contraction", for
otherwise they would be able to at least say " > 0 %" or " < 0 %", I
would think.

This is plausible, although I personally would not venture so far as to support the idea that Omnilingua meant this.
If you believe that it is highly improbable that they have
succeeded in dealing with the difficulty of comparion I pointed out (in
having a correct comparison based on "perceptually" equivalent
"readability" of source and target text pairs -- that would require much
indepth research by cognitive scientists I suppose), then their
statement of "varies" would thus be simply a false claim. For with a
theoretically questionable apparatus one can't do any scientific
measurements.

I think you take a quite highly scientific approach to what I believe is simply a measurement, not scientific in the sense of 'modelled, tried and proven' but in the sense of 'measured over a body of samples large enough for the result to be useable in predicting future cases', under which sense the statement "varies" simply means the range of measured ratios for Chinese / English is too large to usefully help predict ratios.

Our discussion has lasted too long in my view for the passive
readers of the group. So allow me to make a "last" attempt to
render clear my point. For the translation of e.g. English vs.
French, the measument of space expansion/contraction is without
technical problems, since one uses the same font size. However,
for the case English vs. Chinese, a big problem arises. For, if
the English text is with the font of what you now see in E-Mail,
what should the font size of the corresponding Chinese text be?
If a Chinese character is displayed of the size of an English
character here, then one could hardly read it with the eye. This
is the problem of "readability" I mentioned. For a fair comparison,
the "readability" should be "equivalent". But how does one know
that the chosen font size of the Chinese character is o.k. in
this sense? At least I am not aware of any result of scientific
studies in that matter todate. I suppose someone of Omnilingua
just used one particular English font and had chosen a corresponding
Chinese font based on his "personal" judgement. Now, for the
English font, if a given one is comfortably "readable", then a
20% smaller one is somehow more difficult but usually also
"readable" nonetheless (and a 20% larger one is definitely
"readable"). The same of course applies to the Chinese font.
So, suppose with the Chinese font and English font chosen by
Omnilingua one finds in a certain case, say, a 10% contraction
from Chinese to English, then using a 20% smaller Chinese font
would lead to a result of 10% expansion! (I am grossing over,
for purpose of argument, the space between the lines.) Now
which result is the right one? One is contraction and the other
is expansion, i.e. just the opposite! Therefore such kind of
comparison doesn't make much sense from the very beginning,
until the problem of determining the size of the Chinese font
that corresponds to a given English font on the basis of
"equivalent" readability is adequately solved.

To avoid misunderstanding: Certainly different pieces of source texts lead to (more or less) different amounts of expansion/contraction
in the case of, say, English vs. French. (The figures given in their
table are certainly statistically mean values.) If the term "varies"
"were" interpreted in this sense, i.e. as "non-constantness",
then the the claim of "varies" would be (trivially) true, albeit
without any scientific value.

Thnaks,

M. K. Shen
.

Relevant Pages