RE: CE Starter grammar

[John.Velman@HSC.com]

Graham,

I've started this message over, hopefully to save confusion.


You said:

GH2>>>    You've interpreted me correctly. At this stage, I can't
see what is gained by rolling the direct & indirect objects up with the
verb. Did you get an intelligible version of the hierarchy I was proposing?
I see considerable advantage with keeping the "noun" parts, with their
adjectives, etc., separate from the "verb" parts with their adverbs, etc.

JRV response:

There is a long and broad history of expressing English syntax in this way.
There are a lot of parsers out there that develop this kind of syntax tree
(and many are freely available).  And, its not a lot different from the way
I learned to diagram sentences in school a long long time ago.  This
terminology and structure are used by Attempto Controlled English, (sorry,
I misspelled this as "ATEMPO" in my last message) as well.  Like it or not,
this is a pretty good way to represent English syntax.  Perhaps there are
better ways, but a lot of material exists that can be re-used for SUO-CE if
we stick to this.

I seem to have a legible copy of your hierarchy, and have "looked at" it.
I certainly haven't absorbed it.  I'm not totally afraid of something new,
but haven't had time to really understand and possibly critique it.
It does seem to depart significantly from anything I am aware of.  I still
plead to being a non-expert, but I've done some dabbling in language
processing and theory and haven't run across anything like this.   If you
have any references along these lines I'd be grateful.  If not, I hesitate
to ask for more explanation, since you are probably as busy as the rest of
us;  however, if this is a novel approach to English syntax, you are
certainly going to have to defend it extensively sooner or later.  :-)

One advantage of the NP -- VP description is that it permits a fairly
compact recursive definition of phrases in a phrase grammar. In something
like BN notation we have something like this (Items in parentheses are
optional) (the BNF notation lines start with a . as requested):
.
.DeclarativeSentence:: NounPhrase VerbPhrase
.NounPhrase ::  (Determiner) N1 (Prepositional_phrase)
.N1         ::  Adj N1
.           ::  Noun
.VerbPhrase ::  Verb (NounPhrase) (Prepositional_phrase)
.                                  (NounPhrase)
.                                  (Prepositional_phrase)
.           ::  Verb Adjective
.           ::  Verb 'to' VerbPhrase
.
.Prepositional_phrase ::
.
...

and so on.  Well, this is something of an oversimplification, but
by and large, this sort of thing works fairly well.  It fails most
often on general English text because the English text is not
"controlled."

Now I imagine that:
     1) the basic language for SUO will be KIF;
     2) an SUO-CE should have an unambiguous translation to and from KIF.

Work that has been done previously in NLP to translate this type of phrase
grammar into first order logic can be a great help.  Translating from KIF
to CE should then be a piece of cake.  (Well, I've been wrong about many
pieces of cake in the past, for that matter).


GH2>>> When you get to it, I will be interested in seeing the problems
you see.

JRV: No problem.  I only hope I get to do something more concrete than
scrible on scratch paper during odd moments pretty soon.

If you haven't done so, I highly recommend having a look at Attempto
Controled English, see references at John Sowa's site:
http://www.bestweb.net/~sowa/misc/ace.htm.   I've also run accross a recent
reference that may be helpful: "From Plain English to Controlled English,"
at http://www.comp.mq.edu.au/~rolfs/

Best,

John V.