in my last post on this subject, I said
>>>>>>>
If one were setting out to do language description in an inflected
language one would first have to come up with a rule notation which
does not have built in L-R (Left Right) characteristics. One would
want to formulate a rule notation that would allow one to represent
in an elegant fashion the word order flexibility of an inflected
language. So far I have seen no form of notation that solves this
problem.
>>>>>>>>
This statement probably didn't make sense to some people so I
would like to illustrate it and show a possible solution to the
problem.
Lets start out by writing a rule for a simple NT Greek verb-cluster.
For the sake or argument let's postulate a verb-cluster that
contains the following elements: a finite verb, a noun (accusative)
and a prepositional phrase (preposition + noun). First lets write the
rule first using an L-R (Left Right) notation and show the weaknesses
of this notation.
verb-cluster ----> verb(finite) noun(acc) preposition noun(gen)
This is an L-R rule. This rule will only match the elements of
verb-cluster in a single order as they are read from left to right.
What we need is a means of marking this rule so that certain
elements are *word order independent* and other elements are
*word order dependent*.
Lets make a small modification to our rule writing syntax. Lets
introduce the bracket marker [ ... ] to indicate word order
independence. Any elements enclosed in brackets are *word order
independent*. Lets introduce the marker < ... > to indicate *word
order dependence*. Any elements enclosed in < ... > are word order
dependent. Now lets rewrite our simple rule using these markers.
verb-cluster ----> [verb(finite) noun(acc) <preposition noun(gen)>]
This rule allows the elements verb(finite) and noun(acc) to appear in
any order. The prepositional phrase is set off with < ... > indicating
that the word order in this cluster is fixed, but the phrase itself,
being an element in the higher level construct can appear in any
order relative to the verb(finite) and noun(acc).
This solution to the L-R rule problem is not perfect. It runs into
some problems when you try to write rules for complex constructs.
It does however serve to illustrate the problem that I was
addressing.
Clay Bartholomew
Three Tree Point
Postscript:
I am not a Gramcord guru. Is there a way of writing word order
independent rules like this in Gramcord?