Respecting the Almighty

I found a speech bug the other day.  Speak the sentence "Thanks be to God".  Oops, we don't capitalize 'God'!  Is that disrepectful, or what?

Now say, "Thanks be to Allah".  Try saying other words for the Almighty: Jehovah, Yahweh, Krishna.  Our original training data didn't come with a lot of these words, so the system will probably misrecognize you unless you have been producing enough documents with these words to give us time to train on you.

Note that we always capitalize these other words, but not 'God'. Why?  Is that a bug?  Are we a bunch of atheists? Are we afraid of the reaction from other religions if we don't show them proper respect?

Now try this: "Thanks be to Bill Gates".  No capitalization!?!  Am I going to lose my job?

The problem is that some wordsd (like 'God', or "Bill" or "Gates") are sometimes capitalized and sometimes not. We don't know if you meant to talk about the Christian God (which should be capitalized) or an ancient Greek or Roman god (which is not capitalized).  Try it: "Zeus is a Greek god". Words like Allah and Jehovah (or for that matter, "Jesus Christ") are always upper-case, and we handle them as such.

In situations like this you can often see the problem better if you try typing your sentence into Microsoft Word, which uses (mostly) the same capitalization rules.  Word will not flag a capitalization error if it knows that there is a legitimate lower-case alternative.

Of course, we can always override this rule using a special 'tag' file that specifically marks certain words as to be always-capitalized. But we don't want too many words in that file, because it will slow performance.

So what do you think? Should we make an exception for God?

Published 01 September 06 06:19 by sprague

Comments

# RIch said on September 1, 2006 3:51 PM:
As an avowed atheist, I say screw it :)
# n4cer said on September 3, 2006 4:52 AM:
I think you should really worry more about not capitalizing "Bill Gates". ;-)

In all seriousness, as you point out in your post, there are multiple contexts for usage. General correctness should be favored over ideology. Ideology is best left to the individual end-user (or a specialized SKU -- Some Speech packages have Legal SKUs, MS could have MS Speech Pack: Deity Edition :-)). When the software gets to the point where it can figure out the right usage for each context, then MS can handle it as the issue becomes simply one of language correctness.

In the meantime, you should test the case for capitalization of "God" when used at the beginning of a sentence, though I'm sure that was covered in v. 1.0. :-)
New Comments to this post are disabled

Search

This Blog

Syndication

Page view tracker