E-mail List Archives

Re: accessible documents and translation


From: Jukka K. Korpela
Date: Mar 31, 2010 1:54PM

Shawn Henry wrote:

> Several things that make web pages accessible also make them easier
> to translate.

I'd like to add two points:

If you use as simple language as possible and avoid ambiguities (these are
key factors in practical accessibility and affect virtually all visitors),
your text will be much easier to translate. This often means using common
phrases and using words consistently instead of fancy, "creative" choices of
words and variation in usage. Such usage increases the possibilities of
using translation memory successfully.

If you use "fluid design" and do not unnecessarily restrict the amount of
space available for various pieces of text, your pages are not just more
accessible (as users can change font size without breaking the layout) but
also more translatable. It is not uncommon to see a piece of text become 3
or 4 times longer in translation*). If the text has fixed width in pixels,
there will often be serious problems.

*) Translated text can be longer than the original due to structural
differences of languages, mere length of words, or different conventions on
saying things. Moreover, e.g. if the target language uses Cyrillic alphabet,
the average width of letters (for the same font size) tends to be
essentially larger than that of Latin letters, since Cyrillics lack common
narrow letters like "i" and "l".

Oh, I have third point too, a rather technical one, and one that was perhaps
included in the point you mentioned, "Reduce redesign and translation time
and skills needed by using standard markup - -". My point is that "standard"
markup that conforms to published specifications or is at least well-formed
(in the XML sense), which is often mentioned as an accessibility factor,
though its real impact there is not that big, can be a major factor in
modern translation systems. Those systems often operated on marked-up text,
e.g. extracting fragments of a document in XHTML format, possibly adding
something to it, and submitting it to a translator using an XML enabled
tool. This lets the translator operate on the textual content only, without
worrying about markup, but having access to the markup, in case it might
help in correct interpretation of the original text. Any well-formedness
error might simply crash this process.

Yucca, http://www.cs.tut.fi/~jkorpela/