about nasin sitelen kalama:
let’s take pona as an example:
is the first phoneme /p/ a valid mora? no. so if we add the first dot, then it refers to the mora /po/ [total: /po/]
is the next phoneme /n/ a valid mora? yes. but wait, that’s /n/, so is the next phoneme a consonant? no. so if we add a second dot, then it refers to the mora /na/ [total: /po/ /na/]
is the next phoneme a valid mora? we ran out of phonemes! cease analysis!
now let’s take anpa:
is the first phoneme /a/ a valid mora? yes. this mora goes unmarked. [total: /a/]
is the next phoneme /n/ a valid mora? yes. but wait, that’s /n/, so is the next phoneme a consonant? yes. so if we add the first dot, then it refers to the mora /n/ [total: /a/ /n/]
etc.
and finally akesi:
is the first phoneme /a/ a valid mora? yes. this mora goes unmarked. [total: /a/]
is the next phoneme /k/ a valid mora? no. so if we add the first dot, then it refers to the mora /ke/ [total: /a/ /ke/]
etc.
this is how i like to explain the system. it’s not my preference and i enjoy using nasin pu more
(/an/ is actually two morae, in anpa and akesi the initial /a/ is treated exactly the same)
ah true, i think my brain wanted to say “syllable” there instead ahah, will make edits
Yeah, I've noticed 
it was already a little concerning to learn that ni can refer forwards (I must admit I wish that marking that, for example with commas in sitelen Lasina, or changing the direction of the arrow in sitelen pona, or with some sort of non-verbal signal in toki uta (does that work "speech"?), would be mandatory) but [>]
[<] it's nice to at least have some bounds on what references I need to contextually disambiguate between 
another method I'm using is that as you go further away (either forwards or backwards!) from the last reference to something, reference to it becomes exponentially less likely. A half life for referents, basically. It's definitely too simplistic though, for example if we talked about buying a particular puppy half an hour ago and we haven't talked about any other animals since, I'm probably still gonna understand "soweli ni". But if we talked about the neighbor's cat in between, I'm gonna be confused if you use "soweli ni" to refer to the dog without somehow bringing that to attention. So there needs to be some modeling of referents overriding each other as well, but I think that once I go and learn about dynamic semantics, which I intend to at some point, I'll probably get some nice tools for making those sorts of models.
Yeah, I wish
I wonder what the minimal changes you'd need to make to toki pona to make it a loglang would be, specifically if you could do it in a way so that your loglang toki pona is a subset of normal toki pona 
I think that, if you postulate that the 'n' in "nasa" is not a valid mora since it's not a coda n, then it can be explained more simply: if there is no dot, take the first phoneme. Then, for each dot, take as many more phonemes as you need in order to end up with one more mora than you already had.
And with that way of thinking about it, I quite like it
I wonder what the minimal changes you'd need to make to toki pona to make it a loglang would be, specifically if you could do it in a way so that your loglang toki pona is a subset of normal toki pona
Are you familiar with nasin sin Lonuke?
I had heard of it, but hadn't read the document until now. I like it! But it goes the other way than what I had in mind, adding rather than taking away.
You can see every content word as a variable 
I think you can almost do that, but modifiers complicate it: You can see every phrase as its own variable, but then you would lose valuable information that you can infer from inclusion relationships: if we have been talking about both a "soweli laso pona", a "soweli laso ike", and a "soweli jelo", presumably one could refer to one of the two sowelis laso as "soweli laso", and any of the three as "soweli". If those are the only three animals are in the context, I suppose you could refer to the "soweli laso pona" as just "soweli pona" also, but then, a question: descriptively, for toki pona as commonly used, could "soweli pona" also refer to the "soweli jelo"? Assuming it hasn't been established as un-pona. Or would people expect that you meant the "soweli laso pona"?
Indeed, I was considering cases of different referents of ona with different heads.
Another option is modifying ona. You could e.g. say ona jelo to pick out one or more possible referents of ona.
I would say technically yes, but it would be very weird if only one of the soweli was introduced as pona, because that suggests it as a disambiguating trait.
oh absolutely, but I think my point is more so the lack of explicit resumptive pronouns: toki pona is more than capable to handle reference in sentences, but my point was just that it’s not more explicit, which is perfectly fine for a language that really likes contextual inference
in my experience with speaking toki pona, I would say that it gets more intimate with certain features in language that we may often take for granted, such as context, reference, inference, explanation, semantic analogy, metaphor, and more generally just trying to be understood by others, but it doesn’t resolve it in the more explicit and specific ways that loglangs tend to do, but rather it takes typical things that natlangs do and gives you a smaller toolbox to work with, requiring you to innovate more with fewer, less specific words
That's the impression I've been getting! But I'd be interested to see how little you could lose in the process of extracting an unambiguous subset and making its references resolve explicitly et cetera. I think it could be an informative exercise for loglang design, specifically because toki pona relies so heavily on things like context which it's my impression that loglangs have not historically had much to say about: Eberban is the only loglang I know of to explictly handle context, and even that is extremely recent: that feature is only around 3 years old (this is the commit where an explicit context variable was introduced; git bisect is so cool thank you @mia.entropy for having used version control!)!
Validity is a sliding scale :D
I have a hunch that there's a possible analysis of toki pona where there aren't any embedded clauses, despite what "ni:" style constructions look like (analyzing them as just using cataphoric (forwards-referring) ni), in which case there would be no need for a resumptive pronoun
In the course of writing my new post about toki pona noun phrases and formal semantics I generated multiple new questions:
- Would you say that, say, "soweli jelo" and "jelo soweli" mean different things? If so, what's the difference?
- Is "soweli pi jan ala" grammatical? Can it refer to non-people animals? If your worldview is that, say, cows don't have personhood.
- If so, how grammatical is "soweli ala pi jan ala li moku e pipi"? Can it mean that "non-person beasts don't eat bugs"?
- Can "soweli li moku e pipi ala" mean that "cows don't eat bugs"? Or does that need to be "soweli ala li moku e pipi"?
- Can "soweli li moku e pipi ala" mean that "cows eat something other than bugs"? If so, can "soweli ala li moku e pipi ala" mean "cows don't eat anything other than bugs" aka "cows only eat bugs" (this is, if soweli is understood to be cows, what that "soweli li moku e pipi taso" means, right?)?
Yes, they are different. One is a soweli, and in some way related to jelo, while the other is a jelo and in some way related to soweli. To use a not-quite-sufficient analogy in English: "yellow dog" versus "doggy yellow". you coyld for example claim that your house is colored jelo soweli, but you would not say that it is a soweli jelo.
Yes and yes. It can also mean that the soweli is related to (some instance, or a general concept of) jan ala in some other way.
Completely grammatical. It could mean that, if you usr jan for personhood, and assuming you meant a non-soweli.
First is "Cows eat non-bugs." Second is "Non-cows eat bugs." (modulo much wider semantic spaces)
You'd need soweli li moku ala e pipi for what you want.
Yes, no. The second doesn't make any claim what soweli do.
I adsume you meant soweli li moku ala e pipi ala? That would indeed seem to be similar to soweli li moku e pipi taso, but not equivalent. It doesn't claim that soweli eat anything.
Very detailed, thank you! This seems reasonably easy to formalize.
Yes, I'll edit.
EDIT: actually no, I meant what I said, "non-jan soweli". The "soweli ala pi jan ala" was meant to be interpreted as the "does not exist" quantifier.
I was trying to generalize from "jan ala li kama musi" meaning "nobody is going to start having fun", as seen here. Does that meaning work for you? Or would it only be "(non-people / (some/the) non-person) (is/are) going to start having fun"? Either way, another ambiguity to add to my "nasin ale" model
Specifically, the ambiguity between "ala" as a modifier in a pi-less noun phrase or maybe even before pi in a pi-ful noun phrase either complementing the noun phrase or acting as the "does not exist" quantifier.
@_at, if "jan ala li kama musi" can mean "nobody is going to start having fun" for you, can "jan ala pi toki pona li kama musi" mean "no toki-ponists are going to start having fun"?
both of these and “soweli li moku ala e pipi” would work for conveying this meaning, to my understanding. typically, speakers seem to like to bind ala to verbs first, and I presume putting ala elsewhere is more emphatic in nature, at the very least in my mind
yes!
well, I would personally parse this as “no cows eat no bugs” which would logically be the same as “soweli li moku e pipi” (toki pona does not seem to allow negative concord, from my experience)
really this leads to the question of parsing the theoretical “soweli ala li moku ala e pipi ala” which I don’t know for certain how anyone would parse that
So you have "ala" able to mean the "does not exist" quantifier ("no" as in "no cows" / "no bugs"),,,
and able to mean the complement operation ("non-" as in "cows eat non-bugs"); a nasin that embraces the ambiguity! musi pona a :D
In that case I have some more questions not just for you but for everyone!
- What does "soweli ala li moku ala e pipi" mean?
- What does "soweli li moku ala e pipi ala" mean?
- If "soweli li moku e pipi ala" can both mean "cows don't eat bugs" and "cows eat non-bugs" for you, can "soweli ala li moku e pipi ala" mean:
- "no cows eat no bugs"?
- "non-cows eat non-bugs"?
- "no cows eat non-bugs"?
- "non-cows eat no bugs"?
- On a scale "standard > common > rare > weird > wrong", how grammatical is "soweli ala li moku ala e pipi ala"? And optionally also any of the other examples that you wish to comment on; I think this is a useful scale for measuring grammaticality
Oh, I see. Yeah, that reading would require some context for me. E.g. that there is a salient event where some jan participate.
Understanding a bit better what you're going for, I would say all of your meanings are possible, with sufficient context. Where exactly the ala is might only make a real difference in isolation.
All of these are grammatically correct, but I feel that isn't exactly your question?