I've thought some more about how we decide what the noun
meaning of a word must probably be, if we know its verb
meaning, or vice versa. Rene de Saussure came
up with the theory of three semantic categories for root
words in Esperanto, which was later elaborated by
Kalocsay and Waringhien - something like this:

1. substantial (the noun form is the primary meaning, others
are derived from it. adj = related to, made of, this thing;
verb = to be this thing; to be made of this stuff;
to use this tool; or one of several other derivations)

2. quality (the adjective form is the primary meaning, others
are derived from it: noun = this quality in the abstract,
verb = to have this quality)

3. action (the verb form is the primary meaning, others
are derived from it; noun = act of doing this action,
adj = related to the action)

I think recognizing a fourth category (relationship) might be
useful, especially for languages which can derive ad-hoc
prepositions the way Esperanto derives nouns, verbs and
adjectives. I'll get to that later.

There are problems with this theory, at least as it applies
to Esperanto; I'm not sure to what extent they necessarily
apply to other similarly schematic conlangs. Edmund
Grimley-Evans outlines them here:

http://www.rano.org/vortfarado.html

The problem for learners is when words for similar things
are arbitrarily placed in different root categories. For instance,
most words for tools are derived from an action root + the
"ilo" suffix; but a few words for tools have dedicated substantial
root words, and then have verbs derived from them meaning
"to use this tool". It's famously tricky to remember whether
"broso" derives "brosi" and "kombi" derives "kombilo",
or vice versa (I have to look it up almost every time, and I've
been a pretty fluent speaker of Esperanto for six or seven years.)

So, for Konya (and other IALs, or non-IAL conlangs that are
designed to be easy to learn and use) it would make sense to:

1. Define a consistent rule for how verbs and adjectives
derive from substantial roots; which needn't be the same as
the consistent rule for how nouns and adjectives derive from
action roots, etc.

2. Define vocabulary-building principles up front to avoid
problems like broso/kombilo : for instance, "all words for tools
are substantial roots" (probably not a good idea) or
"all words for tools are derived from an action root +
a tool suffix".

In order to come up with a useful small set of
consistent rules for derivation, it might be necessary to
have more than three categories. In particular, relationships
don't seem to fit easily into the above categories.
Some (like "near") can be forced into the quality
category, and some ("mother", "subset") into the
substantial category, but this may may lead to problems.

We could define that for roots signifying a relationship,
the primary meaning is the preposition, and the verb, etc.
are derived from that by some specified rules;
or maybe the verb is primary (to have such a relationship with
the object of the verb), and the preposition and substantive
are derived in a certain way.

I didn't plan this out perfectly well in gjax-zym-byn,
and have consequently had some confusion about how
certain postpositions derived from relationship-roots
should work. For instance, does the postpositionizing of
"subset" result in

X [subset]-[postp] Y = "Y which is a subset of X",
or "Y of which X is a subset"? I finally decided on the former.

A rule like this might work for relationship-words:

1. The prepositional meaning is primary, and means
"the headword of the prepositional phrase is in this
relationship to the object of the preposition".

2. The verb derived from that means "to be in such a relationship
to the object of the verb".

3. The noun derived from that means "an entity which is in
this relationship to some other entity".
[Alternatively: the noun means "this kind of relationship
in the abstract". Either one may have problems.
Bear with me.]

4. The adjective derived from that means "typical of
or appropriate to such a relationship".

I can give rough Esperanto examples (some of which
won't be perfectly standard Esperanto):

1a. Li estas _cxe_ sia hejmo.

2a. Li _cxeas_ sian hejmon. ["cxeesti" is more standard]

3a. Cxu li estas _cxea_?

4a. Diru al la cxeo ke li foriru. [Totally nonstandard;
in normal Esperanto "cxeo" is rare but would probably
mean "being-at-ness" rather than "one who is at" some
place.]


1b. Johano _patr_ Tomaso venos hodiaux.
[Nonstandard; I intend it as "John
who-is-the-father-of Thomas is coming today".]

2b. Johano _patras_ Tomason.
[Rare, but not nonstandard.]

3b. La patra amo malsamas ol la patrina amo.
[Fairly standard.]

4b. Mia patro venos hodiaux.
[Totally normal. In E-o "patro" is the basis and
"patra", "patri" are derived from it.]

#4a seemed fairly awkard. Let's try both series again
using the alternate form of rule 3 (noun means the abstract
relationship rather than an entity who is in that relationship
with another):

4a2. Diru al la cxeulo ke li foriru.
[Pretty normal Esperanto, though "cxeestanto" would be
more typical.]

4b2a. Patro [= patreco] estas grava rilato.
4b2b. Mia _patrulo_ venos hodiaux.
[Nonstandard. But maybe in a new conlang
it would make sense to have most words for persons
derived with one of a few suffixes, as with my
tools example above?]

That seems a little better, but still not quite
satisfactory. Maybe "relationship" is still too broad
a category for one derivation rule to apply to all its
members.

--
Jim Henry
http://www.pobox.com/~jimhenry/review/log.htm