Even the handling of misspellings is problematical.  Back in the heyday
of CAI in the 80s I ran directly into this problem while developing ESL
materials.  It was easy to tell if the answer was right.  It matched the
desired spelling exactly.  What was not so easy was how to tell if what
the student meant was right, but misspelled.  Misspelled answers and
wrong answers aren't necessarily the same thing.  I played around with a
probabilistic approach to this that worked most of the time.  It was
based on probability tables for letters by position by length of word,
up to 12-letter words.  Where it broke down was on misspellings like
"*recieve", where the probability of the "ie" sequence was higher than
that of the "ei" sequence.  Vagaries of English spelling like that I
couldn't fix probabilistically.  It takes intelligence, not an
algorithm, even a probabilistic one.

Herb

-----Original Message-----
From: Assembly for the Teaching of English Grammar
[mailto:[log in to unmask]] On Behalf Of Craig Hancock
Sent: Thursday, February 15, 2007 11:27 AM
To: [log in to unmask]
Subject: Re: question on why "unwinnable" is not a word

   Another way to say this is that the computer will never understand
human language in the way we understand "understand' to mean. It can
"recognize" forms ("recognize" here is just a metaphor.) In the case of
spelling, it simply measures the form of a word against a list of forms
in its memeory, and if it finds an exact match, it "accepts" the word.
It has purely formal ways to recognize the presence of a passive, but
no ability to understand the complex conversation about the uses of
passives, or the rhetorical nuances brought into play by its use.
   Human language is deeply tied to human experience, and the computer
has
never "lived" in that way. It's a machine.

Craig

 Paul,
>
> I take pretty much your position.  As we've seen frequently in our
> responses to "Is this grammatical" questions, the answer very often
> involves pragmatic conditions, that is, the decisions are not
deductive
> but at the very least inductive.  Software doesn't handle induction or
> vast knowledge-based systems well, and probably won't for some time.
>
> Herb
>
>
> -----Original Message-----
> From: Assembly for the Teaching of English Grammar on behalf of Paul
E.
> Doniger
> Sent: Wed 2/14/2007 1:09 PM
> To: [log in to unmask]
> Subject: Re: question on why "unwinnable" is not a word
>
> This conversation makes me wonder what the status of grammar check
> software is today. Regarding MS Word grammar check, I don't let my
> students use it because it often "corrects" grammar that needs no
> correcting, and sometimes it even ruins a perfectly fine sentence. It
> seems to me that it has a great deal of difficulty with extended
sentences
> and with singular / plural issues (and it doesn't like ANY passive
> constructions, either).
>
> Are there any new developments in software design on the horizon? I
have
> often though that computers can never fully understand human grammars.
>
> Paul D.
>
>
> ----- Original Message ----
> From: "Stahlke, Herbert F.W." <[log in to unmask]>
> To: [log in to unmask]
> Sent: Wednesday, February 14, 2007 12:50:23 PM
> Subject: Re: question on why "unwinnable" is not a word
>
>
> Natalie,
>
> My guess is that the spellcheck dictionary in your version of Word
doesn't
> include it.  I tried it in Word 2007, and it was accepted.  I don't
> attribute too much subtlety or analytical thought to Word spelling and
> grammar checkers.
>
> Herb
>
> -----Original Message-----
> From: Assembly for the Teaching of English Grammar on behalf of
Natalie
> Gerber
> Sent: Wed 2/14/2007 12:11 PM
> To: [log in to unmask]
> Subject: Re: question on why "unwinnable" is not a word
>
> Hi folks--
>
> A colleague asked me why "unwinnable" is marked by Microsoft Word as
an
> unacceptable word form, and I wasn't sure. Do you have an answer?
>
> She had wondered whether it is "because the prefix 'un' connotes the
> undoing of something as opposed to the 'not' doing something (as in
'not
> winnable')?" Yet it strikes me that, however awkward the word form is,
> there are many counterexamples, such as "unachievable," that negate
this
> reason.
>
> Thanks for your help,
>
> Natalie Gerber
> SUNY Fredonia
>
> To join or leave this LISTSERV list, please visit the list's web
interface
> at:
>      http://listserv.muohio.edu/archives/ateg.html
> and select "Join or leave the list"
>
> Visit ATEG's web site at http://ateg.org/
>
> To join or leave this LISTSERV list, please visit the list's web
interface
> at:
>      http://listserv.muohio.edu/archives/ateg.html
> and select "Join or leave the list"
>
> Visit ATEG's web site at http://ateg.org/
>
> To join or leave this LISTSERV list, please visit the list's web
interface
> at:
>      http://listserv.muohio.edu/archives/ateg.html
> and select "Join or leave the list"
>
> Visit ATEG's web site at http://ateg.org/
>
> To join or leave this LISTSERV list, please visit the list's web
interface
> at:
>      http://listserv.muohio.edu/archives/ateg.html
> and select "Join or leave the list"
>
> Visit ATEG's web site at http://ateg.org/
>

To join or leave this LISTSERV list, please visit the list's web
interface at:
     http://listserv.muohio.edu/archives/ateg.html
and select "Join or leave the list"

Visit ATEG's web site at http://ateg.org/

To join or leave this LISTSERV list, please visit the list's web interface at:
     http://listserv.muohio.edu/archives/ateg.html
and select "Join or leave the list"

Visit ATEG's web site at http://ateg.org/