Re: A Language built around a novel grammar

Hi!

Harold Ensle writes:
>...
> >After a while of constructing, I *still* arrived at the point where I
> >joined the two open word classes.  Again, I found no way to
> >distinguish them in a good way.
>
> If I understand your meaning here......if one used the word for an
> operator, like "equals" for "=" and then treated it like any other
> word, then the operator would become part of the general lexicon.
> So your point is, why should some word get special treatment?
Exactly.

> >The result is that any lexicon entry
> >may function as either nullary (a 'noun') or unary (equivalent of
> >intransitive verb or case or adposition).  Binary (equiv. of
> >transitive verb) n-ary relations in general are expressed by serial
> >verb construction of 'verbal'-'nomimal' compounds.
>
> I see here that you have minimized the operators completely(?),
> but notice that in the very end, you still had to have three.
I think so, yes, depending on how you count, there are two or three
operations. There are more to make sentences, I'll list them in a
minute.  I think you counted like this:

  1) use a lexicon entry as a nullary (~make a noun)
  2) use a lexicon entry as a unary   (~make on intransitive verb)
  3) combine pairs in serial verb contruction (SVC).

You could better count these as two binary operators:

  1) combine two lexicon entries to a unary-nullary combination,
  2) combine pairs in SVC.

These two are implemented by morphology, without extra morphemes: the
lexicon entries are inflected (well, very regularly changed).

There are more special things, implemented by closed class morphemes:

  - mark the beginning of a (sub)clause (lets mark this with BEGIN)
  - mark an internally headed relative clause
  - mark an externally headed relative clause

Further, there the unary words (~verbs) are optional, meaning you want
to reuse the previous one.  Together with SVC, you can get clauses
like this:

  BEGIN-verb-noun -noun -noun.

That's basically the grammar of this language, I think.

> I found also that there seems to be no way to avoid some
> distinct element that has special behavior quite apart from
> the most generalized word class.
Of course, some way the lexicon entries must be combined to larger
semantical units.  However, there is the possibility of letting the
language be ambiguous, i.e., you could make all operations implicit so
that they must be inferred.  Classical Chinese comes close to this.
It is basically an isolating language that consists only of content
words.  Finding lexical word classes is not very easy.  Being no
engelang, it is not entirely strictly ambiguous in all lexicon
entries, so often you can find restrictions in usage of content words,
but it is very, very interesting to analyse and see how much is left
ambiguous in this *natlang*.

**Henrik