Pretty Little Girls School

[janKipo]

One of James Cooke Brown's more ingenious ploys was the description of how Loglan (his baby) disambiguated the phrase "pretty little girls school" (I can no longer remember or reconstruct the Loglan (or later Lojban) version (what I get, bilti cmalu nirli ckola, is probably a mixture of and wrong for both languages -- my usual state). The nearest simple tp parallel would be (reading left grouping for right and the head for all complexes) tomo meli lili pona, (((AB)C)D). How many of the remaining possibilities can we say unambiguously (or even ambiguously) in tp as now set up -- without changing word order, for now?

[janMato]

"pretty little girls school"

I'm not following where you were going on the word order constraint.

tomo sona pi meli lili pi pona lukin. There is a school. It has (the qualities of) little girls and those girls are good looking. (But I suppose the pona could modify the tomo as well)

tomo sona pi meli lili pona. There is a school. It has (the qualities of) good (somehow), little girls.

And going off in a different direction and embedding it into a sentence (not all sentence fragments work equally well in sentences, imho)
mi wile e ni: jan meli lili mi li tawa e ni: tomo sona li tawa meli lili li tawa meli pona lukin. the girls are pretty and small. The school isn't.

mi wile e ni: jan meli lili mi li pali lon ni: tomo sona. tomo li tawa jan meli. jan li lili li pona lukin. the girls are pretty and small. The school isn't.

I guess the next step is to work on "little schools", "The school of one girl vs the school of two girls", "The school is little and it's littleness is remarkable."

[jan Josan]

I'm bad at this, but this is what I got:

Searching for the phrase I found at this website:

4.2) ta melbi cmalu
nixli ckule
That is-a-((pretty type-of little)
type-of girl) type-of school.
That is a school for girls who are beautifully small.

4.3) ta melbi cmalu nixli bo ckule
That is-a-(pretty type-of little) (girl type-of school).
That is a girls' school which is beautifully small.

4.4) ta melbi cmalu bo nixli
ckule
That is-a-(pretty type-of (little type-of girl))
type-of school.
That is a school for small girls who are beautiful.

4.5) ta melbi cmalu bo
nixli bo ckule
That is-a-pretty type-of (little type-of
(girl type-of school)).
That is a small school for girls which is beautiful.

so in TP:
4.2) tomo pi meli pi lili pona (((A)B)CD)
4.3)tomo meli pi lili pona ((AB)CD)
4.4)tomo pi meli lili pona (A)BCD))
4.5)tomo meli lili pona (((AB)C)D)

[janKipo]

tawa jan Mato:
The classic approach is using bracketing to get the basic idea concrete (this was set up to invite you, jan Mato o, in), so you have -- taking into account the differences in your base form -- ((AB)C) or (A(BC)) (where A,B, and C are all internally complex). Your expansion is the latter.
I'm not sure why you use "has the quality of" here, "belongs to" or "teaches' seem the natural connections between schools and little girls (or girls, generally, in fact). The second case (relisting the components) (A ((BC)D)) (or ((A (BC))D) (not mentioned).
The third (embedded -- the first chunk is better as tomo (sona) ni:) and is expanded out of the tight structure (probably a good idea for clarity but not quite the present issue) (A ((BC) & (BD))) (something JCB didn't do, though Lojbanists added it later, to make the whole really complex). 'meli pi pona lukin' btw, not "apparently good girls".
Next case grammatically awkward: not clear what the tomo sona does where it is -- it should be before the ni, I suppose. It collapses to (A (B (C&D))) (tp needs to think about en -- and presumably anu and ala -- in these right groupings -- add that to the list, along with the "approximately" about.")
I don't see the one girl /two girl problem (though lili does mean "few."
tawa jan Josan:
nice, though cheating a bit by checking out the Lojban.
I read the tp as giving different parentheses, given the rules about pi, but the matchup for the Lojban cases is correct, except for the ambiguous one, which the Lojban doesn't give (a beautiful school for little girls)
4.2 (A(B(CD)))
4.3 ((AB)(CD))
4.4 (A((BC)D)) or ((A(BC))D)
Lojban avoids this ambiguity by using a strictly binary regrouper (infix not prefix to boot) and a complicated set of rules (I remember the Loglan ones as being simpler originally, but I suppose they got worse as stranger cases came along). tp has no device for avoiding this problem, other than "Don't do that" and rewriting:tomo pona pi meli lili or so (which may not mean quite the same thing -- but that is later detail). Even Lojban has to introduce another device to handle the final case (but it can't be stated all, not merely as one possibility in an ambiguous phrase, in Lojban with only the binary infixes; it takes parentheses, or at least prefixes: melbi ke cmalu nixli ckule [ke'e]) Nothin looks like a plausible tp device for this case; but the, disambiguation is not a tp thing -- "Context will decide and paraphrase is always best any how>"

[jan Josan]

Sorry for the cheat, but not knowing Lojban I didn't understand the question at first.
My mistake on the parentheses (never having had a linguistics class), but now I think I see how yours work. I'm glad the matchup still looks good, because after just now rereading Pije's lesson on "pi", I'm not sure what means what anymore.
I don't understand how 4.4 (tomo pi meli lili pona) could be grouped two ways. I would think the (beautiful school for little girls) would have to be a different order, right? (tomo pona pi meli lili). But then, I guess this is what we are always trying to do with ala and mute--sticking them at the end to negate or pluralize the head noun.

so with jan Mato's more exacting 'tomo sona' and 'pona lukin' can we still make it work?

4.2) tomo sona pi meli pi lili pi pona lukin. (school for girls who are beautifully small)
4.3) tomo sona meli pi lili pi pona lukin. (girls' school which is beautifully small)
4.4) tomo sona pi meli lili pi pona lukin. (school for small girls who are beautiful)
4.5) tomo sona meli lili pi pona lukin (small school for girls which is beautiful)

As for context will decide, meli mi li toki e ni: "lon ilo ni la, toki ale pi jan lili pi pona lukin li monsuta."

[janKipo]

tenpo mute la, meli pi mije pi toki pona li sona ala e pona pi toki pona. a!

The problem is that pi just says the next two words at least are grouped together (boed in Lojban), but not where the right end of that grouping is. (In Loglan we tried with a strictly binary prefix grouping for a while, but it left us with long strings of the prefix in some cases (the hard case was A pp BCD, I think -- not the worst possible, of course).

The problem with going to the more literal reading, ignoring that "girl" is already meli lili, is that each old block is now complex and so needs more pis to cling together -- and more pis means harder reading (and also introduces possibilities that break up the blocks). Your 4.4 is still ambiguous in the same way, since where the pi phrase fits is as uncertain as it was before, with a single word. The suggestion to use pi pi before meli to get your meaning got rejected early on -- it takes more forethought than can be reasonably expected and it is ugly. Back to paraphrases.

So, are the five groupings so far all that are possible using only modification (introducing conjunctions makes for a new set of possibilities, though not really of groupings)?

[janMato]

Carefully stepping off the toki pona track, onto the languages in general track, why not have verbal parenthesis?

foo pretty foo little foo girls school bar bar bar.
where foo = ( and bar = )

Or an even more unnatural, put he binding order in verbally

pretty two little three girls one school --> ((pretty little) (girls school))

If that is the number of symbols that it takes to encode the information with zero loss in the algebraic form, then, I think we'll need that many in the linguistic form. For speed and economy in number of words, I think human syntax is lossy when it turns thought into words-- the listener will never get one meaning but will always have to select from a menu of options.

I read an essay recently where the author recounts giving an autistic guy, who was good at languages but profoundly disabled at everything else, some sample fake language problems. Some syntax rules were inspired by real languages and some where inspired by math (say requiring an inflection based on the number of words in a phrase). The guy had no problem with the former and flunked the later. I suspect that we're tilting at windmills trying to get the menu of options down to one, at best we can seek out styles of writing that keep the menu short, or failing that, with one or two good options and all the other's being preposterous.

I shot an elephant in my underwear.

mi moli e soweli suli lon len anpa mi. mi sona ala e ni: soweli li len e len anpa mi kepeken seme?

[janKipo]

Why a duck? (in accent) To get to the other side. Thx, Julius. But the tp isn't ambiguous, officially (and wouldn't be anyhow if noun-modifying prep phrases needed pi -- the reason for the proposal).
Sure, precision is very difficult (Lojban/Loglan claims to have it in this area, but it is not clear that anyone speaks/writes those versions of the languages -- or could). The usual problems are those of adequate foresight -- or memory for hindsight, RPN. I'm not even sure I see how your infix system works, let alone see how to use it in running speech.
But that is not tp's problem, since it makes no claim to unambiguity. The question is more whether it can get most situations in a reasonable range of possibilities. Here we have two possibilities that are covered by a single form and that is probably not too bad a situation. We can ask -- if resolution seems required -- whether it is the girls or the school which is pretty. Of course, the problem gets worse as the string gets longer, but then we have the practical advice to break it down into shorter bits. Which latter is the moral of the whole exercise, I suspect.

[janMato]

the reason for the proposal).

Was the proposal to use "pi pi"? It will sound just like insect, pipi.

I'm not even sure I see how your infix system works, let alone see how to use it in running speech.

The first was just verbal parenthesis. I suppose it could be expanded to have a different verbal paren for the outermost, the 2nd outermost and innermost

foo/bar = { }
fizz/buzz = [ ]
heebee/jeebee = ( )

foo fizz heebee pretty little jeebee girls buzz school bar. = {[(pretty little) girls] school}

The second idea was to show order of operations, the same way we resolve 5 + 2 * 10 / 2 + 4 % 6 + 2 (modulo probably first, then *,/, then +,-)

But that is not tp's problem, since it makes no claim to unambiguity.

Agreed. Ambiguity reduction is good, ambiguity removal is going to be hard without resorting to a spoken mathematical language.

...we have the practical advice to break it down into shorter bits. Which latter is the moral of the whole exercise, I suspect.

I'm too lazy to write an good example now, but I suspect breaking things into sentence will sometimes cause ambiguity. If the following were spoken (and we didn't have the period to disambiguate)

mi wile moku telo li anpa

I want to eat. It is raining.
I want to drink and to fall.

[janKipo]

The proposal was to require pi before prep phrases modifying nouns.
Still don't get the infix notation -- why is the first pairing listed as two, and the second as one?
Yes, as paleographers know, punctuation is important -- and proper juncture in speech (but is eating wetly really the same as drinking? -- doesn't matter, it is a legitimate expression, whatever it means). Still, the advantages of stretching things out probably outweigh these occasional difficulties (not to mention the clarification one gets of the modification relation). Of course, it is prolix, but it is good advice, whenever you find yourself wondering how to say something (in tp -- though perhaps not bad in general).