Re: Linguistics journals

[HACKER G N <c9709244@ALINGA.NEWCASTLE.EDU.AU>]

On Thu, 23 Oct 1997, Logical Language Group wrote:

> >Cyc can disambiguate natural language very effectively simply by using
> >commonsense knowledge about the world, which is stored in its knowledge
> >base in CycL. Take the two sentences:
> >
> >John saw the airplanes flying over Zurich.
> >John saw the mountains flying over Zurich.
> >
> >They differ only in ONE WORD, but that word affects the grammar of the
> >rest of the sentence.
>
> You are talking to a community that sees and relishes the ambiguity in those
> English sent4ences.  that there are conventional interpretations of reach with
> different grammars, does not mean that the two sentnces NECESSARILY have
> different grammars.

Well, I buy your skepticism, but there is no reason to interpret those
sentences as having grammars different from the default unless you are
actually INTENDING to have fun with their meanings. Lots of plays on words
are based on this.

>
> It is nice to have an interlanguage that can store the differences between
> the three verbal interpretations of "Time flies like an arrow".  But I do not
> think it sound that the language automatically stores/assumes one of those
 based
> on conventional semantics.  That should be the job of the translator engine's
> semantic analyzer.

It is. CycL is only there to store knowledge. The Cyc semantic analyser
does the job of interpreting the sentence's grammar.

Geoff