488881 - "judgement" not in dictionary (SCOWL-based dictionary should include common variants)

Status: RESOLVED FIXED

(Core :: Spelling checker, defect)

Description

[Jason Orendorff [:jorendorff]]

The Merriam-Webster Dictionary of English Usage has this to say:

> Both ways to spell this word have been in use for centuries. ...
> Most dictionaries now show both spellings. Usage commentators generally
> allow that both are acceptable.... [B]oth spellings are in reputable use
> on both sides of the Atlantic....

Americans tend to write 'judgment' but both spellings are universally considered ok.  Flagging the variant as wrong seems unnecessarily pedantic.

Comment 1

[Jason Orendorff [:jorendorff]]

Attachment: v1

I noticed a few curious things about the dictionary file:

1.  The number at the top is not the number of entries in the file.  This patch doesn't change the number, following what the previous patch to this file did (see bug 418076).

2.  The list is not entirely in a consistent order.  It's close to strcmp order (that is, ordered case-sensitively by ASCII codes) but three words (ChatZilla, MySpell, and SeaMonkey) are misplaced.  Contrast the placement of DeGeneres, FedEx, and JavaScript.  I don't know if this matters.

Comment 2

[Ryan VanderMeulen [:RyanVM]]

Jason, please submit a bug upstream so this gets fixed by Kevin as well.

Comment 3

[Ryan VanderMeulen [:RyanVM]]

Also, mscott is long gone. I'm not sure who should be reviewing the patch, but it's not him.

Comment 4

[Kevin Atkinson]

"judgement" and "acknowledgement" are defiantly variants with judgment and acknowledgment being the preferred form for American English.  Both words are in the variant (level 1) list in scowl. I do not include variants in the dictionary in order to promote consistent spelling.  I make exceptions for variants when there is no agreed upon preferred form and the two spellings are different enough that it is unlikely to confuse the two (which generally means they are pronounced differently).

I might consider including variants such as judgement if there in a consensus among Open Office developers that this is the right thing to do.  But here is not the place to discuss this.

Comment 5

[Jason Orendorff [:jorendorff]]

Thanks for the response.  Where is the right place to discuss this?  Should I file a bug against Open Office?

I'd be OK with Open Office and Firefox having different policies on this.  Most of us Firefox people are "defiantly variants" ourselves...

Comment 6

[Kevin Atkinson]

Oops, I thought this was an Open Office bug.  Sorry.

Maybe here is the place to discuss this, what makes you say "Firefox people are "defiantly variants" ourselves"?

Comment 7

[Jason Orendorff [:jorendorff]]

To clarify my comment about policy: I think there's a sensible division of responsibilities between the compilation of an accurate, descriptive word list (like SCOWL) that captures these nuances; and the decision of to what degree we want to use Firefox to promote consistent spelling.

Our current policy seems to be "no variants".  We underline "blonde", for example.  This is a word that gets 18 million Google hits.  If I search the Web for "blonds"--not something I ordinarily do while at work--the first thing Google says is: "Did you mean: blondes".

I don't know how en-US.dic is compiled, but I suspect we can produce a word list that includes common variants, and that would probably be in line with the user experience we're after.

Comment 8

[Jason Orendorff [:jorendorff]]

The "defiantly variants" remark was a lame joke.

In any case, Kevin, thank you for your work on SCOWL.  I've used it before, recreationally.

Comment 9

[Kevin Atkinson]

For the record: blonde is a word I would consider adding since it fits by 
blonde is something I would consider adding since it is different enough from
blond.  I suggest you file a bug report at
https://sourceforge.net/tracker/?func=browse&group_id=10079&atid=1014602 on
that so I don't forgot about it.

Comment 10

[Jason Orendorff [:jorendorff]]

Comment on attachment 374095 [details] [diff] [review]
v1

Withdrawing this patch.

I read extensions/spellcheck/locales/en-US/hunspell/README.txt and README_mozilla.txt and things make a little more sense now.

It should be fairly easy to hack together a script to generate a SCOWL-based dictionary that includes the common variants.  I'll attempt this next time I'm waiting for a build.

Comment 11

[Matt Caywood]

I'm fixing the identified problems "judgement" "blonde" etc. in Bug 479334.

The idea of autogenerating variants is worth a try. If you could generate it as a separate wordlist, that might make it easier to maintain when upgrading upstream dictionaries.