On Mon, 28 Jan 2019 21:10:19 -0500
"Mark E. Shoulson via Unicode" <unicode@unicode.org> wrote:

> On 1/28/19 3:58 PM, Richard Wordingham via Unicode wrote:
> > Interestingly, bringing this word breaker into line with TUS in the
> > UK may well be in breach of the Equality Act 2010.
> >
> > Richard.  
> 
> OK, I've got to ask: how would that be?  How would this impinge on 
> anyone's equality on the basis of "age, disability, gender
> reassignment, marriage and civil partnership, pregnancy and
> maternity, race, religion or belief, sex, and sexual orientation"?
> (quote from WP)

The most relevant clauses are 9(1), 9(4), 19(2), 29(5) and 29(7).

The change would restrict Thais' access to the provision of a service.
The service provided is to allow one to use a persistent, correctable
spell-checking system for one's native language.  Firefox and
LibreOffice provide this service.  Of course, one may have to supply
the spell-checking databases oneself.  Withdrawing this service for
some ethnic groups would be breach of the law.

By persistent, I means that corrections to the spell-checking remain
when the text is revisited.  For English plain-text, the easy
correction is to remove false positives by adding the word to
'personal dictionaries'.   The difficult correction, not always
possible, is to remove the word from the spell-checker's word list.

For scriptio continua scripts, line_break=complex_context in UCD terms,
there is the additional problem that word-breaking is not infrequently
wrong, even for Thai in Thai script.  (Recent loanwords into Thai can
be a nightmare.  So is Pali in Thai script, though Pali spell-checking
has its own issues.)  Line-breaking can be corrected with WJ and ZWSP.
At present, word-breaking can currently be corrected by inserting these
characters, and then spelling can be negotiated - the visible
characters are non-negotiable. The changes in the text will persist in
plain text. If WJ ceases to be treated as joining words, then the
service of a persistent, *correctable* spell-checking system is lost.

Now, one defence to the denial of the service would be that it would be
unreasonably difficult to allow users to solve the problem of
word-breaks in the wrong place.  However, if one is already providing
that service, that defence cannot be applied to subsequently denying
the service.

Richard.

Reply via email to