I'll attempt to answer WRT my exolang project May29. One of the ideas is to rely much less on metaphors, unlike natural languages.

On Sun, 14 Jun 2015 13:51:32 -0700, Jeffrey Brown <[log in to unmask]> wrote:
>As an example: It is not sufficient to merely create a lexical entry for
>the preposition “in” for a conlang. For a physical context, “in the house”
>or “in the box”, this works fine, but where the trouble occurs is in the
>extensions to abstract contexts. Does “in” extend to temporal durations?
>Can you say “in Tuesday”?

"Tuesday" would be a predicate, used as an adjunct; "while" is implied by the default aspect, so "in" would be superfluous (and there are no prepositions, anyway).

>Is “in” used for abstract containment: “in the king’s speech”, or “in Kant’s philosophy”?

The language lacks abstracts of that sort.

>Or membership: “in the army”?

There's a predicate for membership: "He's in the army." => "person army member".

>Or fandom: “in the Red Sox” (compare, “he is really into the Red Sox”)?

English is really weird.

>And similarly, it is inadequate to merely create a grammatical entry for a
>morpheme. It is not enough, for example, to specify an opposing pair of
>nominal affixes meaning “quick” versus “slow” – for how does that extend to
>abstract concepts? Is beauty quick or slow? What about liberty? Or trade?

There's a sort of framework WRT the morphology of scalar predicates; the root specifies only the scale, not the "direction". The suffixes -ma and -ki determine that, with -ma used when the quality is "more" than normal and -ki when it's "less" than normal. For each scale, I have to decide which is which, e.g. "hot" and "large" are -ma while "cold" and "small" are -ki. This is easy when there's a distinct 0-end to the scale, but "hungry" isn't.

>This is why cognitive frameworks are important: If you *don’t* think about
>them, you will end up importing your own cognitive framework, from you own
>native language, into your conlang.
>--> For an artlang: This importation implies that the speakers of your
>conlang conceptualize their world the way *you* do, the way speakers of
>your native language do. You haven’t fully integrated your conlang with
>your conculture.

I think the May29 grammar guarantees that the framework can be imported only on an item-by-item basis.

BTW I also think that clause chains, which are perfectly naturalistic, can "open the door" to getting away from SAE-clones.

>There is a real language spoken by a people in the Amazon river basin for
>which some of the verbs are marked for “upstream” and “downstream” – in the
>phrase “to paddle a canoe” the verb form is different depending on whether
>you are going against or with the current. This is all I remember about
>this language, not even its name.
>Let’s suppose we create a conlang based on this language. We’ll call it
>Xanthic, the language of the Xanth people. In Xanthic, the riparian
>morphemes (upstream/downstream) are obligatory on all verbs, not merely
>river-oriented verbs. (Lest this turn into an argument about morphemes, I
>am using the term loosely, and the morphology may be expressed by an affix,
>it may be fused with other morphemes in an inflection, it may be irregular
>for some verbs, it may be a null marking, whatever.)
>And all this is only *one* aspect of the cognitive framework of Xanthic.

Xanth??? "Yukon tundra stammer eater?" :)