Re: Simple grammar, take two

On Sun, 23 Oct 2005 Yann Kiraly wrote:
>
> Hi! I thought about my simple grammar that I posted
> a while back and all of your comments and I've now
> made a new attempt at an easy to use and relatively
> fast to learn grammar.
> This time, there are two grammar rules:
> 1. x y z. => x is in class y which is in class z.
> 2. w si x si y na z. => w and x and y are in class z.
> Here are a few example sentences using English words
> (except for na and si, of course):
> man patient si dog na bite.
> A man who is in class patient and a dog are in class bite.
> A dog bites a man.
> picture si word plural thousand patient na say.
> A picture and a word that is in class plural which is in
> class thousand which is in class patient are in class say.
> A picture says much more than a thousand words.
> What do you think? Any comments or suggestions?
-----------------------------------------------------------------

Hi Yann,

Some comments for you.

A.  Presumably, for all w and x,
 w si x == x si w
(commutativity)

-----------------------------------------------------------------

B. Presumably, for all x, y and z,
 x y z == x na y na z
and
 x na y na z == (x na y) na z == x na (y na z)
(associativity)

You are, in fact, hiding information in your most
compact form
 x y z

-----------------------------------------------------------------

C. Presumably, for all x,
 x x == x na x == x
(stuttering!)

-----------------------------------------------------------------

D. The sentence -
> man patient si dog na bite.
clearly has the same content as -
> dog si man patient na bite.
No?  If so, then your use of various classes for rôles,
such as patient, means word order need not be strictly
determined by your two rules.  Going a little further,
then, with more explicit rôle classes, we could instead
say for total clarity -
> dog agent initiator si man patient na bite.
or equally -
> patient man si agent dog initiator na bite.
or even -
> bite na patient man si agent dog initiator.
Is this where you see your rules leading, or did you
intend to have a strict word order (subject to
commutativity)?

If word order is strict, then of course the last
sentence -
> bite na patient man si agent dog initiator.
would not be grammatical, and you'd need an extra
marker, say "an", so you could say -
> bite an patient man si agent dog initiator.
-----------------------------------------------------------------

E.  In the end, I don't see much difference between
(1) marking different rôles by membership in special
classes such as patient;
and
(2) marking different rôles with affixes or inflection.

Surely (1) would tend, over time, to evolve into (2)?
For example, might -
> dog agent initiator si man patient na bite.
not evolve into something like -
> dog-ag-in si man-pa na-bite.
?

So I would ask you whether you might not manage to
get away without using ANY rôle-marking classes at
all?  Thus -
> man patient si dog na bite.
would become instead -
> man si dog na bite.
This of course requires a precedence rule for
Subject and Object, or for Agent and Patient.

-----------------------------------------------------------------

F. In your second example, the word _plural_ seems
unnecessary given that you have _thousand_.  But
wait, what does this mean? -
> picture si word plural thousand patient na say.
You translate it as -
> A picture says much more than a thousand words.
Surely it says either -
> A picture says as much as a thousand words.
or -
> A picture says as much as thousands of word.
?  This depends on how we associate the terms
> word plural thousand
and whether word order matters (see comments
A & B).

-----------------------------------------------------------------

G.  How will your rules handle complex sentences
where the predicate is another sentence?  For
example -
> I thought that everyone reading and writing
> on all the lists I use could view HTML without
> problems.
-----------------------------------------------------------------

H.  How will your rules handle conditional
sentences?  For example -
> If I thought that everyone reading and writing
> on all the lists I use could view HTML without
> problems, I'd use it like a shot.
-----------------------------------------------------------------

Enough for now! :-)

Regards,
Yahya

--
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.1.361 / Virus Database: 267.12.4/146 - Release Date: 21/10/05