As most of you know, I have long opposed UEB for use in the United States
and, not surprisingly, I still do so. Just to be complete, I am a
sighted
retired computational scientist who became interested in braille software
development back in 2000. Because of this interest I was lucky enough to
meet John Boyer back then. I would estimate that I have spent the
equivalent of at least two working years on UEB issues: studying UEB in
great detail, trying to help various organizations who were opposing
it, and
carefully documenting my objections on my website. I realize that the
current situation is unlikely to change but I would like to respond to
a few
of the comments in this thread.
As a programmer for more than 40 years I find it essential when I read
technical material that includes software fragments that these
fragments are
identical to what would appear in working software that incorporates
those
fragments. That is, I should whenever I wish, be able to copy and paste
these software fragments directly into actual computer code. I find it
impossible to comprehend why braille-using programmers wouldn't prefer to
learn, read, or write computer code using the same characters that
sighted
people do. I've seen programs written by persons whose native language is
French or another non-English language which uses a similar alphabet
to English (ASCII)
and while the comments are often in their native language, they do not
translate the keywords of the programming language to their native
language.
(It's possible that there are cases I'm unaware of where there are
forms of
programming languages that use other alphabets where compiling requires
backtranslation.)
On a similar note it is my impression that many braille-using adults
prefer
to directly enter print or computer code using either a standard
keyboard or
the default 8-dot computer braille table built into their braille display
and have no trouble switching back and forth from computer braille to
six-dot contracted braille.
Since the deficiencies of UEB math have been well-domented elsewhere I
have
just this one statement. As a computational mathematician I have been
continually impressed with how Nemeth math represents the true nature of
mathematics in a way that I've not seen any other braille system come
close
to.
Next I'd like to make two technical comments about translating and
back-translating. First, both of these processes are technically
speaking
examples of parsing. Those of you with an advanced computer science
background are likely aware that the standard techniques long used for
lexical analysis and parsing are quite different from the way these
processes are handled in table-based braille software. I understand that
the use of tables is intended to make it possible for the same engine to
translate according to numerous braille systems and I've observed that
the
popularity of this feature is a primary reason for the widespread
adoption
of liblouis. However, I wouldn't be surprised if some of the advances in
parsing lead to new approaches to braille software in the future.
The second comment is specific to backtranslating UEB. My
understanding is
that there is a "mathematical" proof that the prefix-root nature of UEB
makes it possible to automate fully correct backtranslation of UEB.
(It may
be that the use of shortform contractions violates this; I'm not
sure. If
so, one would need to scan for shortforms first.)
However, my concern is not whether correct UEB can be automatically
backtranslated, my concern is whether human-produced UEB, which is
likely to
contain various errors, can be automatically backtranslated. It is
important to realize that it is the presence of extra rules more than the
elimination of a few problematic contractions that makes accurate UEB
backtranslation potentially automatable. For example, as was pointed
out in
an earlier post on this thread, UEB allows a leading period (full
stop) if
the item is preceded by a Grade 1 indicator. In other words, the real
question is whether a UEB backtranslator can localize braille errors
just as
a compiler can generally find all the mistakes in a piece of code without
crashing.
Finally, congratulations and best wishes to everyone who has been and/or
still is working so hard to make liblouis a success. This is truly an
impressive project of worldwide importance!
Sincerely,
Susan Jolly
www.dotlessbraille.org
For a description of the software, to download it and links to
project pages go to http://www.abilitiessoft.com
