Offline: Registration request for new subtags for Portuguese orthographies

Luc Pardon lucp at skopos.be
Thu Mar 26 11:52:13 CET 2015


On 24-03-15 23:47, Kent Karlsson wrote:
> 
> Michael Everson wrote:
> 
>> This is exactly wrong. A user in BR or a user in PT might know exactly what
>> features they prefer. The problem is that this subtag on its own is an
>> umbrella for all the options, and no writer of Portuguese wants his text to
>> wander randomly through the options. This is precisely why I said that
>> "ao1990" is practically identical to a raw "pt", because it's a collection
>> of features which have been used in Portuguese from time to time. What, "fato"
>> and "facto" should just be identical in the spell-checker? That's really not
>> how users expect spell-checkers to work.
> 
> I agree.
> 
> /Kent K

I disagree. If/when I adopt a spelling convention that has a word list
that allows both, I fully expect a conforming spell checker to accept
both and flag neither.

It is true that I would not want my writing to "wander randomly through
the options", but I still am the master of my own writing, am I not?
Yes, I would probably choose one of the options and stick to that
choice, maybe in all my writing or maybe in one particular text only
(depending on the context, the target audience, or whatever), but no, I
would not expect a spell checker to enforce such consistency. If you
want that, you need a style checker, not a spell checker. Re-reading
what you wrote can be helpful too.

We had exactly the same situation in Dutch. From 1996 up until 2005 (but
don't nail me on the dates), the official spelling allowed both
"cultuur" and "kultuur" (for culture), "copie" and "kopie" (for copy)
etc. To most spell checkers both forms were identical - as in fact they
were, since both were in the official dictionary. Any writer of Dutch
would know which one to select when writing for publication in newspaper
A, and which one for newspaper B, and which one for school homework etc.
We didn't expect "anti-wandering protection" from our writing tools, no
more than we'd expect them to assist us in selecting the proper subset
of all the words available in the dictionary for use in a particular
context (legal proceedings, tabloid, sports magazine, astrophysical
journal...). Quite to the contrary: we wanted - and needed - the ability
to "wander through the options", just not randomly.

So, if you say that "this is really not how users expect spell-checkers
to work", I dare say that you are referring to a subset of all users
that most certainly does not include writers of Dutch. Are you sure that
your subset includes users of Portuguese? Are you sure that yours is not
simply a knee-jerk reaction of a native speaker of a language (such as
English) that has exactly one "right" spelling for each and every word?
I am asking, not to be offensive, but because I fail to understand why
anybody, who has actively been writing in any language that allows
multiple spellings, would expect a spell checker to flag valid ones as
invalid. To me, such a spell checker would be broken.

In any case, the fact that ao1990 is "an umbrella for [several] options"
should be no reason to reject it. The set of options seems to be
well-defined, not open-ended, and the spelling is being used in
practice. Some people may not like that, but refusing it a proper subtag
is not going to make it go away. What you will make go away is the
original requester, and other folks who need to distinguish ao1990 from
other stuff. By refusing them a means to tag within BCP47, you are
forcing them to invent their own subtag, thereby defeating the very
purpose of the registry that you're supposed to nurse.

Luc Pardon


More information about the Ietf-languages mailing list