A couple of trends

[janKipo]

I've noticed a couple of trends in recent tp submission in various venue by people of various skills levels. I started by correcting these patterns. then moved on to just questioning them and commenting on them. I am now to the point of wondering if they are not on to something inherent in tp that gets lost in too strict application of the rules. Let's talk about them a bit.
The first new pattern (maybe not so new, as I look back) is iteration, repeating the same word immediately after itself (occasionally even three times). The most common case is 'mute mute' "very much", but usually attached to another adjective, so "very very", an intensive, a common use of epetition in other languages. That this seems to be the point in tp is confirmed by the fact that the other common words in this pattern are 'pona' and 'ike', used in context that call out for extremes, where 'pona mute' or 'ike mute' would seem to work. All the adjectives used in this way seem to be evaluative, in some sense, but there are not yet many other cases. There are, however, a couple of cases of verbs being repeated. I put those down to dittography at first, but now see that they might have been meant to indicate repeated or intense activity. Both 'tawa' and 'pali' obviously allow interpretations of this sort "running" or "wandering around", say, and "overworking" or "slaving away". So far, I don't think I have seen a repeated noun to suggest plurality. I do note, however, that all of these cases can be covered by 'mute' as a modifier.
The other trend started with these same cases but now with the notice that these expressions as modifiers were not set off by 'pi', as the rule requires. So there was 'pona mute mute' (even with a third 'mute') rather than 'pona pi mute mute' and similarly with 'jan ike ike', even in writers who were generally very careful about their 'pi's. It is as though the repeated words were not taken as two words but just as a strong one. And that set off a chain of thoughts: if 'ike ike' is just one word, "strong 'ike'", then why is 'ike mute' two word, or 'pona lili', for that matter. So, maybe the 'pi' rule needs fine-tuning to exempt the intensive forms even when they take up more than one word (though 'pi's might be needed to specify the scope of the intensification). But why stop at intensifiers? Well, because it is hard to know where to stop once you get started. But some cases seem simple. We already have that any string of numerals meeting certain conditions is a unit, without a 'pi'. And that extends to ordinals as well. What about the case of colors, where most of our familiar colors are actually two or more words in tp, must we say 'kili pi jelo loje' rather than just 'kili jelo loje'? And what about other cases of compounds of words in the same semantic space, body parts, say? I don't know where this might end, except either in returning to the simplicity of the current rule, or advancing to a welter of complexity and dispute, but it seems a good and timely idea to consider it.

[janMato]

re: reduplication as intensification
When toki pona, in its small toolbox has a toki pona way of doing something, normatively, we should do that. So pona kin is better than pona pona or pona mute mute. But the corpus so far disagrees a good percent of the time-- maybe it is a creole universal.

The corpus has lots of reduplication as intensification, I agree, it does fit nicely into the head+modifiers string, nor a head/modifers/pi/head/modifiers string. So I modeled it as "tags", words that modify a head without a pi.

I did the same thing with anu seme because while anu seme arguably has some internal structure, if you think about it as anuseme, which is kind of how it gets used, then it is just a tag. Ditto for X ala X.

re: reduplicated verbs
I just looked up Tagalog verbs, they reduplicate and the tense, aspect and the like changes. In my own household, my wife started saying to the toddler things like "butt butt" and other reduplications which are diminutives. I have no idea where that came from, unless it was from the Filipino mother in law, from the land of halo halo (mix-mix, a ice cream like desert)

[janKipo]

I forget about 'pona kin': 'ni li pona kin' "This is GOOOD!" or, boringly "This is good indeed" Probably not much used in speech, where you can actually emphasize the 'pona'. Now that you mention it, a lot of the pidgins and creoles I have seen do use iteration for something or other: emphasis, repeated action, or repetitive, diminutive/affecctionate; tense/aspect changes are a new one (except as above).
In any case, they do seem to be here to stay and maybe the notion of a tag is a useful one" word sequence that functions essentially as a single word, including not requiring 'pi', etc. We already have the numbers, of course. and it would save trying to give a grammatical explanation for 'x ala x' verbs (and elsewhere?). 'anu seme' doesn't seem to me to be so much a problem, but, then, we haven't done a lot with 'anu' really. We do need to firm up the notion of a tag a bit, though, else anything will start to work as such and we lose what little precision tp has.

[janKipo]

About tags, consider some possible cases and the consequences for various systems.
'x ala x', at least for main verbs, has a nice transformational history but nothing that makes sense to a linear parser (nor does introducing the modal Y/NQ help much) and it doesn't seem to impinge on other structures or functions. So it seems to be a paradigm case.

'anu seme' is so ambiguous from a technical standpoint and the ambiguities are so irrelevant generally that there seems to be little point in sorting it out. At worst, you get (very) occasional odd (maybe even funny) answers to 'sina wile tawa sitelen tawa anu seme': beyond 'wile' and 'ala'. you might get 'tomo' or 'moku e kala' or (in your dreams) 'unpa' or 'li weka e sama'. If the parser is primed for all of these, then it has to produce a lot of strings, but better to just take the 'anu seme' as a floating chunk of some sort, and explain the odd ones if they arise. Not a paradigm case, but a clear case of a practical solution. Occasionally someone will challenge the Y/N nature of the question, but, if they give an answer, we can come up with an alternate analysis.

Number strings, at least those with the number words in non-rising order (I don't know what we do with 'wan tu luka', which still seems to add up to 8) seem another paradigm case. But there are possible conflicts as in "The general needs two centuries" (a Roman general, of course) 'jan lawa li wile e kulupu pi jan utala 100 tu' . Of course, we would intuitively solve this by moving the 'tu' to attach directly to 'kulupu', but it is not clear this is always possible or feasible. Still, for now, this looks like a safe case.

Compound colors, like 'laso jelo' or even 'loje jelo walo' work on the notion of many words just meaning one thing. That it is one thing is, of course, L1 chauvinism for the most part. And it does conflict with reasonable alternatives. I had no sooner thought that 'laso jelo' was not a problem than I remembered my wife complaining about how yellow the broccoli had become 'kasi laso jelo li jaki. taso kasi pi laso jelo li pona mute'. Now, in the preparser, you may want to stick in a hyphen or whatever in the latter case (but you don't need it) but I don't think this will work as a tag in the language generally.

Repetitions. There seem to be several slightly different cases here. Even the main cases, 'mute mute (mute)', 'pona pona' and 'ike ike', of intensification, seem to work differently. The non-count words (those except for 'mute' and sometime 'lili') seem to rarely give problems in meaning:
'jan pona pona' does differ from (official) 'jan pi pona pona') but only because of the idiomatic nature of 'jan pona'; 'soweli pona pona' is not a problem. But problems might arise in more complex cases, just as with numbers: 'kulupu pi mani pona pona': I assume that there is occasionally a difference between a herd of very good sheep and and good herd of good sheep. But, as in the case of numbers, shifts seem to come to the rescue.

The problems with 'mute' and 'lili' (and some other considerations) also doesn't arise at the first level: 'jan mute mute' is going to mean quite a few people however it parses, though you may have different ideas about how they are grouped. But 'soweli pona mute mute' goes off in three ways (at least) if we allow reduplicating tags: "very many good critters", "many very good critters" or just "very good critters". But, this goes too far in ignoring 'pi'; keeping serious use of 'pi', the last two are properly 'soweli pi pona mute mute' and the second would, following the now familiar pattern, be changed to 'soweli mute pi pona mute' (with a little philosophical violence, but none practical). So, the final enumeration position can stand repetition (and presumably also 'mute'-'lili' mixes) without needing 'pi'. And the same would be true for the "very" (and "barely"?) uses at the end of string, provided we always move enumerations forward to head (where the repetition can also be used). [5/20 But I just had occasion to suggest 'jo e linja lili mute' for "fuzzy" and that clearly has to be distinguished from 'linja pi lili mute' "nearly bald" and no shifting is available that doesn't make thing worse. I am now of the mind that the 'pi' rule had better be rigidly adhered to.]

Repetition for verbs (intense or repeated or repetitive actions) don't see to make much of a problem, since the modification structure around verbs is markedly less complex (or less well studied or used). The case of modals is possibly tricky semantically, and could even introduce extra levels structure if tags were not used, so tags seem appropriate here (but there aren't many examples yet . but i had a vision of a guy dancing around saying 'mi wile wile pana e telo jelo' "Man, i really gotta go!")

I can't think what reduplicated prepositions would do.

If 'mute lili' and the other combinations are allowed as tags in some places, one wants to ask about structures with 'ala', but the history of philosophy tends to say that the scope of "not" has to be kept clear everywhere or all Hell breaks lose. And similarly for 'ali'.

[janKulisa]

I know the last post on this thread was about a year ago, but it seems to still be relevant.

Now, I have to give my disclaimer that my use of toki pona is primarily based on my gut feeling, which may or may not be English-influenced and may or may not be different from someone else's gut feeling. But anyhow...

I see this kind of reduplication as emphatic, but also parse it linearly; that is, a "jan lili lili" is emphatically small, but is also a [[jan lili] lili], as small sort of small person. Likewise "pona mute mute mute" is good, and very good, and very very-good, and very much so. Now, one could say "pona mute pi mute mute" but while that is grammatical, it just doesn't have the same ring to it.

So back to "jan lili lili"; we understand that people come in different sizes, and a "jan lili" is in the smaller end of the range. But small people also come in different sizes, so a "jan lili lili" is small, for a small person. Of course this is complicated by the idiomatic use of "jan lili" for any human (or other person-like being) offspring, regardless of size; if I have a kid and he or she grows up to be six feet tall, I'd still say "ona li jan lili mi"; even though I'd be the physically smaller of the two. So "jan lili lili" might specify a child who is still small, as opposed to one who has grown up. Meanwhile "jan pi lili lili" makes the second "lili" modify the first, instead of modifying "jan lili", which leaves us with a slightly small person — a rather different concept.

[janKipo]

Here is where guts meet grammar. From a strict traditional point of view, 'jan lili lili' means "a few small people" [jl]l. The only variant available to that is 'jan pi lii lili' which means "a slightly small person", I suppose just below the mean or so (how 'lili' works as an adverb is not well worked out). For "a very small person" we need 'jan pi lili mute'. There is no place in the grammar as now constituted for anything using 'lili lili' to mean "very small" (and no room at all for 'lili lili lili' except 'jan pi lili lili, lili' [j[ll]]l "a few slightly small people" or 'jan lili pi lili lili' [jl][ll] "not so few little people" (very roughly). Life is a little better with 'mute' , since 'jan pi mute mute' does mean "very many people" and 'jan mute mute' does not have a good standard meaning -- and adding more 'mute's on doesn't help or hurt. I suspect that this is the source of all this (together with some knowledge of other languages that use this format productively), but it just seems a bad fit for tp. On the other hand, if we leave 'lili', with its two semantic areas, and 'mute' with its special uses, there seems to be less of a problem: 'jan pona pona pona' resolves grammatically without too much outside interference ('pona' sometimes means "enough" or "the right amount") and the standard interpretation comes out to about what was intended: [[jl]l]l "the good among the good among the good people" I still think that 'lili mue' or 'lili kin' is safer.

[janKulisa]

So, would you say that "jan pona lili" necessarily means "a few friends/good-people"? I don't see why "lili" has to mean "a few" in one place, and not in another. Actually, I never use it to mean "a few"; that's always "pi mute lili"; but that's my personal preference and not necessarily relevant. Supposing it can mean that, why would "jan lili lili" necessarily mean "a few little people", when "jan lili" doesn't necessarily or even usually mean "a few people"?

[janKipo]

At a guess, 'jan lili' doesn't usually mean "a few people" because of the commonness of the meaning "child". Ordinarily a measure word in last position takes on its measure role, the more so the longer the modifier string (with the variation that it does it early on if the string contains 'pi' phrases). In the case of doublets like 'jan lili lili', the role is forced on it, since the other role would require a 'pi' (we're talking current standard grammar here, of course -- and the usual assumptions about modifier place). Actually, as I write this I realize that the "few" meaning of 'lili' just doesn't get used much at all -- unless it is forced, as above, or by context. Most people probably say 'pi mute ala' or 'pi mute lili' for the numeration place. So one useful move might be to take "few" out of the dictionary for 'lili'. That then makes 'jan lili lili' just another case of duplicated modifier and maybe opens the way for general duplication for emphasis. 'mute' remains a numerator, however and 'jan mute mute' an enigma, grammatically.

[janKipo]

One of the neat things in Loglan was that the primal adjectives were all comparatives: {blanu} meant not "1 is blue" but "1 is bluer than 2". Loglanists hated this since, they said, using common practice, rather than the actual rules, that, if you wanted to say "x is blue", you had to leave open the second place and that became just a particular quantifier. That is "x is blue" means "x is bluer than something" and so just about everything is blue. Loglanists used it anyhow. Lojban dropped the whole notion, despite its good logical and linguistic credentials.

The actual Loglan rules called for the gap to be filled with "the average X" where X is the tightest class that intensionally includes x (typically the one in the description) and "average" is left vague among various statistical (mean, median, mode) and conceptual(typical, paradigm, usual, ...) possibilities. The important thing is that the meaning of an adjective is always tied intimately to the noun to which it is attached. We know this in practical cases: an enormous ant is nowhere near as big as even a tiny galaxy and a good steak is not to be preferred morally to a man. And so on.

In tp then, an adjective is relativized to the noun phrase it modifies. A big red dog is, more or less, bigger than the average red dog, which, in turn, is redder than the average dog (but the color doesn't seem to affect size, so it is probably bigger than the average dog regardless of color). Now, apply this line of reasoning to the case of 'ijo pona pona', "a very good thing". So, ijo pona is a thing better than the average thing, in the upper 49% at least. But, by the same reasoning, 'ijo pona pona' refers to something that is better than the average ijo pona, in the upper 49% of that upper 49%, say, or in the upper 25% of ijo altogether with respect to pona. This seems to be in the range of 'ijo pi pona mute' "very much better than average", as the usual understanding of repetitive adjectives suggests.

This is not an argument that adjective repetition as intensifiers should be welcomed or that, if allowed, they should be used without 'pi'. It is, however, a way to make a case for their legitimacy -- and a guide to understanding them -- should they be accepted.

[janKulisa]

I still don't get the "role is forced on it" thing... Where does this rule come from? As far as I'm concerned, something that means "little" in one modifying position can mean it in another. But then, I don't speak Loglan, and as far as I know Sonja doesn't either.