From Http://Members.Aol.Com/Lassailly/Tunuframe.Html wrote:
>=20
> Dans un courrier dat=E9 du 30/05/99 22:03:17 , Charles a =E9crit :
> > In my experiments it seems impossible to have part-of-speech termina=
l
> > vowels and SVO order without eventually gravitating to adjective-nou=
n.
>=20
> it's because you get short of vowels so you have to root-root bon-fisho=
obedoo
> in the end. anyway i think PoS tags should be head like in svo pidgins.
Yeah, http://www.sil.org/silewp/1998/002/SILEWP1998-002.html
> switching adj-noun means that you link inversion and attribution (like =
in
> Chinese and English). in the same time you make a hierarchy ranking
> attribution as secondary, derived word order and i feel it's not accura=
te.
> maybe you can systematize children's language :
> house its door hurt john his head.
That's genitive-noun, which I like, and often leads to adjective-noun
by a process of "grammaticalization", if I got it right.
> dog ADJ it red TOP it nice =3D the red dog is nice
> friend my his sister she nice =3D my sister's friend is nice
> sister my friend she nice =3D (ditto)
> where integrative ADJ makes "it red" attributive like "his sister" is, =
and
> resumptive TOP closes substuff.
> it seems strange, but that how attributive really works in children's h=
eads :
> actor comes first, then attribute.
That works well with adjective-noun, where there is almost a feeling
of pre-limitation and post-attribution of both nouns and verbs.
Attempted example: "My red dog biting your sister is just playing."
> this has nothing to do with verb-noun
> difference. either verb is noun's attribute or noun is verb's attribute
> because it's unaspectivized (that's my favourite rabitting-on subject) =
:
> "gone the car" =3D "the car gone" =3D the car is gone (as my little cou=
sin says :
> "partie, la voiture !").
> that's why i think your "reversive" language is a good basis for childi=
sh
> auxlang.
Well, it has to be easy enough for me, a much tougher task.
> > By using an inverse-transitive voice or doing French-like compoundin=
g,
> > root+preposition+root as in salle-a-manger, it could work well maybe.
> >
> that's basically what i did with "kases" (remember ? ;-)
> but you may try with real prepositions made from real verbs.
I may abandon compounds as unneeded anyway. I like the Tok Pisin
"haus xxx" construction better than what I was thinking about earlier,
letting little prepositions get in the middle of compounds.
As for cases/prepositions, I do like using a real verb instead of a
closed-class particle e.g. "... concerning xxx" vs. "... about xxx".
And it wouldn't be too horrible to try a more consistent VO style,
adjs follow nouns, adverbs follow verbs, etc. The main point for me
is whether there can be a super-productive set of -r- type infixes:
> > > let's say :
> > > i : verb
> > > o : substantive
> > > a : adj =3D attributive to a substantive
> > > e : adv =3D attributive to a verb
> > > -r- : nomen agentis
> > > -k- : noun of action
> > > -s- : genitive
> > > -t- : and (resumptive)
> > >
> > > bone fishi =3D to fish well
> > > bona fishi-r-o =3D the good fisher
> >
> > Hey, I may have to steal this -r- and -k-, at least;
> > what would be the most productive set of these?
> mapping
> like indonesian : around 10. mapping like i would ("kases") : around 50=
(all
> most frequent roles in their main aspects used in derived words). if yo=
u are
> willing to use long words and avoid ditransitives by means of serial ve=
rbs :
> you need 4.
How?! I can see the 50, what are the 4 and the 10? Do you mean "case role=
s"
as in http://www.uoregon.edu/~delancey/papers/bls91.html ??
> funny that Marcos and you both like root+root+suffix pattern.
The lexical information naturally comes out faster than the grammatical,
sorted by topicality. Then the POS vowel is held as long as necessary
until the next word is ready, much prettier than "... uh ...".