Jeff Kell wrote:

> > Wirt's comment reminds me of a story I heard a few years ago about
> > computers and their ability to understand natural languages.
>    [...snip...]
> > The phrase chosen was taken from the New American Standard Bible, Mark
> > 14:38b, "the spirit is willing, but the flesh is weak."  I don't know
> > what the Russian translation was, but when it was translated back to
> > English, it came out as, "the vodka is good, but the meat has gone
> > bad."
>
> One more practical example from the early AI days was:
>
>    "Time flies like an arrow."
>
> Analyzed syntactically, this can result on a few meanings, for example:
>                     noun   verb    prep.     obj.p.
> (a) The passage of <time> <flies> <like> an <arrow>.
>                  adj.    noun                       verb
> (b) The insects "<time> <flies>" have an affinity (<like>) for arrows.
>                                 verb   obj
> (c) When determining airspeed, <time> <flies> in a similar manner
> (<like>) you would measure <an arrow>.
>
> I've always thought this one of the better examples of the difficulty
> imposed on any natural language recognition by the ambiguous nature of
> the syntax and semantics of language.
>
> Jeff Kell <[log in to unmask]>

Jeff, I have to disagree with your statement about "... any natural language
...".

>From my early AI days, the example I remember was "Fruit flies like a
banana". But so do oranges, pineapples, etc. But fruit flies also like to
eat bananas.

The grammar of the natural (ie human) language imposes the syntax and
semantic rules. Grammars are either 'context sensitive' or 'context free'.
The english language is 'context free', and AFAIK is the only 'context free'
natural language. But that's where the fun begins.

Any 'context sensitive' natural language as the host, will work because the
grammar imposes the syntactic and semantic rules, ie you cannot have 2
semantic rules for 1 syntax rule.

The problem is that since english is used as an international business and
'computer' language, it is therefore impossible to implement the full
grammar, and thereby the full vocabulary, in a 'natural language system'
using english as the host. So all the software currently on the market,
using english as the host, is limited for this reason.

But it's also the fun of the english language. Did you hear the one about
...

Kindest Regards
Tom Lang.