Re: S7 grammar in a nutshell (long)

Hi!

And Rosta <a.rosta@...> writes:
> True of natlangs, too, isn't it? (The Livagian name for Livagian
> is "our language", 'nostratic'.)
Yes, probably. :-)

> Somebody, perhaps you, opined in a recent message that if heads are
> marked for valency and version, then there is an exponential
> increase in complexity with extra cases, which is very true, but SVC
> does not strike me as a true reduction in cases: it's more like a
> reduction in complexity by switching to dependent marking.  (If
> that's unclear, I can try to restate it.)
I think I understand you.

But I need one way to be able to any number of adjuncts.  Using SVC
seems the most obvious and simple structure.  Using cases (or
adpositions) is equivalent to a fusing of a verb+argument.  What's the
alternative that totally eliminates complexity without limiting
expressiveness?

> > But no means of shifting.
>
> I see that as part of 'Phase 2' of the design process.
Yes.  I never dared to introduce a means of breaking my word order
principles, however.  I simply couldn't do it.  It would have doubled
the number of grammar rules.  At least.  I couldn't do it. :-)))

> is getting it to be user-friendly.
S2?  User-friendly?  Hmmm.  It is so ambiguous -- even the function
words are optional.  A parser for it would probably explode at a
sentence length of 2 words. :-)

> For meanings that are fully compositional, their lexical expression
> doesn't have to be listed in the lexicon, whether the mode of
> lexical expression is syntactic or morphological. Words with
> meanings that are less than fully compositional must be listed in
> the lexicon regardless of whether they have any synchronic
> morphological structure.
Well, ok.  But I think I like rules in order to be able to be
consistent about non-trivial composition.  This way, I don't
unconsciously do it like in my L1.  Further, if rules are visible, I
suppose I'd learn the lexicon entries easier.

If someone does not like morphological rules during language
aquisition, I promise to mark the corresponding chapter in the book so
that it can be skipped and the lexicon be learned without being
bothered with rules. :-)

> To put it another way, it is pointless to have rules generating
> structures whose meaning is at all idiomatic.
Hmm, no, I don't think so.  Fully compositional vs. idiomatic is not
binary.  There are degrees in between.  An idiomatic entry is still
composed, it is not a totally new word, so it is good to see the
structure how it is composed, although the composite is used in a more
idiomatic/specific/general way.

> > As to tense affixes: I don't know how to further clarify my point,
>...
> Do have another go at clarifying -- for I don't see why expressing
> tense by means of an affix rather than a separate word should have
> a bearing on its idiomaticity. Okay, you might say that you
> allow idiomaticity in morphology but not in syntax -- fair enough.
> But why would you want tense to be idiomatic?
Hmm.  First of all, S7 is meant to be polysynthetic.  Sticking to that
principle, it should have many, many affixes. :-) S2, OTOH, has
lexical entries for tense.

I also like to have tense affixes at word level.  E.g. I want to say
'ex-showman' in the same way as 'it was a snowman'.  Being a
polysynthetic language, I need to get into the words to do it.

It should be noted that S7 tenses are normal stems.  Lexicon entries.
The affixes are agglutinated to words in the same way as, say,
incorporated arguments.

Case is a bit different.  Case affixes have to reduce to one vowel
group in order to force the use of evidentials on the phonology level.
This was a design goal from the very beginning.  Still, they are also
transparently related to lexical entries.  This becomes overt when a
heavy-shift is performed on an adjunct, since the original lexical
entry must then be used.

**Henrik