I consider "that sounds weird" to be on one dimension of grammaticality that goes from from "totally normal" to "entirely wrong", and "people wouldn't generally say that" to be on another dimension that goes from "anyone would say/understand understand that" to "nobody would say/understand that"; the two correlate so I decided to merge them into one scale for convenience. On that scale, "common" is supposed to mean something along the lines of "it might sound a little off to some but basically everyone would say/understand that" and "weird" is supposed to mean "I mean you can say that but nobody would, really. And even if they could puzzle out the parsing, you'd confuse people".
Basically, trying to un-binarize "grammatically correct" a little
I suppose "adjective phrases exist" is too strong of a claim. I should have said "proves that, conditional on you thinking that toki pona has adjectives, adjective phrases exist", because what I mean is that it shows that phrases get their category from their head.
Ah! Then you refered to an entirely different thing than what I thought.
While I do agree that the underlying lexical word classes exist, I don't think it is necessarily implied by the example.
Consider e.g. kala jan and jan kala: the former could refer to a fish you bought from a fisherman, and the second to that fisherman, and not vice versa. (They could also both refer to a mermaid or something, but I'm not sure that's relevant.)
Interesting! I wonder if you can model this as follows: "normal" content words (not ale: it can be a quantifier. not ala, because whether or not it can be a quantifier, it can be complementation. probably not any number?) must be adjectives when being modifiers. So it's "kala(N) jan(N→A)", and "jan(N) kala(N→A)". That would still allow adjectives to commute: "kala(N) jan(N→A) loje(A)" = "kala(N) loje(A) jan(N→A)" for "the red fish I bought from the fishmonger" (does that meaning work? and are "kala jan loje" and "kala loje jan" equal?). Does that work?
I think that theory would imply that "sitelen sitelen" is fine but "sitelen sitelen sitelen" is iffy for semantics reasons but "sitelen pi sitelen sitelen" should be fine, how are those predictions?
Ah, I didn't mean it in either such a limited sense as "writy writing" nor as broad of a sense as "this can do anything reduplication does", I meant it only in the sense that the adjective and noun form can have different enough meanings that applying adjective "sitelen" to the noun "sitelen" is meaningful. Such as the hieroglyphic(A) writing-system(N) "sitelen sitelen".
But now I am wondering: can modifiers narrow their meaning contextually? Such as a "meaning which is both spiritual and like a gas": "kon kon kon"? I hypothesize that they cannot, i.e. that they have their generally broad meanings while acting as modifiers, and then only the final phrase is narrowed from context (which would make "kon kon kon" sound weird because it's a semantically superfluous thing to do, while "sitelen sitelen" is fine because it's a form of writing/drawing that is itself writing/drawing related), but I would be intrigued to be wrong!
I think "sitelen sitelen" doesn't really work as intended. (It is fossilized enough that people use it but) I would not agree that generally modification with the same word as the head makes them necessarily refer to different subsets of the semantic space. All sitelen is necessarily sitelen-related. I feel like this approach is using an interpretation of semantic spaces that requires translation into different English categories and thus is invalid, to me at least.
I'm now wondering whether this is the case, but I feel like it isn't. Derivational patterns already make sure that e.g. (A->N)(A) and (N)(N->A) are basically tautological.
Two strikes against "sitelen sitelen"! Any "sitelen sitelen" defenders?
That would be a simpler semantic model, indeed.
I think the "modifiers can be contextually narrowed, not just entire phrases" theory can give similar results without requiring that, allow me to try: since "sitelen(N) sitelen(A)" (not writing "N→A" because this argument works regardless of derivation, it just needs you to believe that sitelen(N) and sitelen(A) can be distinct, as illustrated by "jan kala" vs "kala jan") would be redundant if sitelen(A) had its broadest possible meaning, pragmatically sitelen(A) must have been meant in a narrower meaning among those that it can take, which are not necessarily the same meanings as those that sitelen(N) can take, which makes it meaningful to apply the narrowed sitelen(A) to sitelen(N).
For the record: I am not making this argument to convince you of that theory (I certainly don't want to believe it), only to convince you that it could explain it non-krokodilily (mixing english and esperanto derivational morphology: fun!)
personally, I don’t really find modifying a head with the same word that the head is e.g. sitelen sitelen to… mean anything unique outside of when the two words are used for different referents. toki pona seems to be rather indifferent to reduplication, and I would say much the same
…but, there’s the monkey wrench of lili lili
hmm.... especially "barely tiny" is interesting to me, I didn't realize it could have that adverbial meaning
EDIT: question, can you modify "lili lili" further? "lili lili pona", say? How about "lili pona lili"? Can you do "lili lili pi pona", "pi" with just one modifier on the right? I'm considering if reduplication could be analyzed as a lexical(ized) thing.
Also, preliminary decision: for my nasin, I will use "ala X" as the \neg\exists quantifier, not "X ala", to save "X ala" for unambiguous non-X. "ale X" can be \forall, right?
I know what the rules say but since "lili lili" is already a strange case, I wanted to see if that strangeness would make the rules more flexible. To be honest I forgot what the "pi" example was supposed to test, but "lili lili pona" vs "lili pona lili" was supposed to test whether "lili lili" is lexicalized. It doesn't appear so! What are your thoughts on "lili lili lili"? Or "sitelen sitelen lili lili" vs "sitelen sitelen pi lili lili" vs "sitelen lili sitelen lili"/"sitelen lili lili sitelen"? This is super helpful for figuring out what algebraic properties modification needs to have or not have :D
