'pi' still

Language learning: How to speak Toki Pona, translation problems, advice, memory aids, tools and methods to learn Toki Pona and other languages faster
Lingva lernado: Kiel paroli Tokiponon, tradukproblemoj, konsiloj, memoraj helpiloj, iloj kaj metodoj por pli rapide lerni Tokiponon kaj aliajn lingvojn
Post Reply
janKipo
Posts: 3064
Joined: Fri Oct 09, 2009 2:20 pm

'pi' still

Post by janKipo »

The short specification of 'pi' in modern tp dialects is that it introduces a modifier unit of more than one word in a head-modifier string. The 'pi' and what follows it then acts as a unit, the whole modifying what precedes it. Internally, the expression after the 'pi' is usually itself a head-modifier string, possibly also complex. but that internal structure does not carry over to the larger context, being treated as a unit, like a word. Thus, in 'jan suli mute', the rule of left grouping says that 'suli' modifies 'jan. and 'mute' modifies 'jan suli' "many big people" (which has the mirror structure under English right grouping). But in 'jan pi suli mute', the whole 'suli mute' (or 'pi suli mute' depending on how you do your analysis) modifies 'jan', while, within the modifier phrase, 'mute' modifies 'suli', giving "very big people" (English grouping shown by the adverb-adjective pattern missing in tp at the vocabulary level). It is mainly to mark these different groupings that 'pi' has come to be so important in tp. But it has a more complicated history than simply a device for this purpose (which it does not, alas, completely solve even now). And this history accounts for some the peculiarities in some presentations of 'pi' in textbooks.

It is not clear when 'pi' started to be used in this disambiguating way, since it is already there in the earliest records I can find (though not always consistently -- or the records contain uncorrected errors relatively to the likely meanings). But also present in the earliest records, and written up in the earliest textbooks is another use of 'pi', to mark possession, "belongs to". This was not confined to modifier string but was also used in predication. So, Pije's grammar contains examples like 'ni li pi mi' "This is mine," where 'pi' is not only a predicate (preposition? particle?) but is followed by a single word. While this predicative use clearly extended to other words than 'mi' and to longer phrases, especially names like 'jan Pije', it is not clear whether the single word usage also applied to modifiers, whether, that is, 'tomo pi mi' was legitimate at that time, or whether the 'pi' was obligatorily dropped when the predication shifted to modification. In any case, the 'pi' clearly carried over when the possessor was identified by an two or more word expression. The shadow of this usage persists in the textbook tradition with the definition of 'pi' as "of", a standard possessive marker, and the line that what follows 'pi' is a noun + adjective, a typical possessor marker.

Of course, in those same textbooks, within a page or two of that rule will be a case like 'jan pi suli mute', where what follows 'pi' is clearly an adjective + adverb. The usual dodge for this is to insist that whatever comes right after a 'pi' is a noun and its modifier therefore and adjective. So 'jan pi suli mute' is "people of great size" and so on. This is, given tp's fluid parts of speech, a possible reading, of course, but tends to add complications to understanding which are not necessary. It also tends to limit the scope of 'pi' in various ways, which simplify the language in one dimension but complicate it in others. In particular, this way of viewing things tends to (and in Lope's writings, does) limit what follows 'pi' to two words and does not allow that the modifier of the internal head (whether noun or not) can be itself a 'pi' phrase. Of course, every 'pi' phrase of more than two words opens a mass of possible ambiguities (only partly helped by judicious use of commas), so good style and consideration for readers or listeners suggest not nesting or grouping a lot of 'pi' phrases. On the other hand, going back from such phrases to the underlying sentences means lengthening your discourse and fragmenting it, with the danger of losing the threads. The present usage permits -- but surely does not encourage -- multiple 'pi' phrases and tries to temper the problems with them by using commas appropriately.

The 'pi + noun + adjective' rule also means that prepositional phrases as modifiers do not naturally get set off by 'pi', since they are often hard to read as 'noun + adjective', their "adjective" typically being an obvious noun phrase. Indeed, in much of the early text, PPs as modifiers were not set off by 'pi', the change being gradual, becoming more or less regular around 2009. The rationale for the change was simply that a PP as a modifier was obviously a modifier unit of more than one word and so should have 'pi'. That is, it assumed to general form of the 'pi' rule to be already in place, as it clearly was not. It does now seem to be, except for the retrograde influence of some textbooks, which is counteracted by corrections on the net.

From a constructivist point of view, 'pi' appears when an expression comes to be a modifier from some other function. The transition erases the particular value of the original function (agent, object, etc.) into a single "modifier" category.

The exception to this seems to be the case of "degrees of adjectives. We can say that something is good , "pona" but also that it is very good, "pona mute" and even that its very-goodness is very: 'pona pi mute mute'. And similarly for 'lili' (though not as common) and for 'ala' at any point and 'ali' at the beginning of the sequence. There is even the emphatic duplication, like 'pona pona pona', which is not official, but common in more colloquial styles and whose regularized use is not established (but is likely to require a 'pi' somewhere -- probably at least at the front -- when used as modifiers. In this way, we can generate a number of cases, to be discussed in another place, which often contain 'pi's from the start: e.g., 'pona pi mute pi mute ala' "very good but not very very good". In this way, predications that become modifiers may already contain 'pi', a fact that creates some problems later.

Another source of predications that already contain 'pi' is incorporation, whereby the periphery of a predicate come to be modifiers of its verb core. Thus, 'ona li alasa e waso telo, lon tempo lete' "He hunts ducks in winter" might become (as in English, etc.) 'ona li alasa pi waso telo, lon tempo lete' "He duck hunts in winter" or even 'ona li alasa pi waso telo, pi tenpo lete' "He winter duck hunts" and 'ona li alasa pi tenpo lete e waso telo' "He winter-hunts ducks" (the loss of the prepositions is another story, but very common for verbs and at least 'lon' and 'kepeken' and, to a lesser extent, 'tan'.) We mark predications that have undergone this process fully by enclosing them in slashes, /\. So '/alasa e waso telo ,lon tenpo lete\' is ' alasa pi waso telo, pi (lon) tempo lete'. Of course, it is assumed that any predication can be changed int any of its incorporations.

The oldest regular case is possession, where the fact that something possesses something else allow allows it to call it its own. The nature of this possession is deliberately left totally vague: consider what you may have: body, soul, arms, wife, children, income, ideas, future, wish horses, and so on. Older tp could sum all this up in the predication 'li pi ...' . Modern tp only has the rather inverted '... li jo e...' or the artificial 'li ijo ...'. More generally, one can claim as one's own just about anything one has put his mark on (even if only mentally) "my fish" even it got away, and so on. So, we have the very general pattern, x li A y B + C y D (where the caps are any old thing at all grammatical, and x and y are what we are interested in) => Cy{x}D , where the braces indicate that x comes out as x if it is one word or as 'pi x' otherwise. 'mi alasa e waso telo. taso waso telo mi li tawa weka. waso telo pi jan pona mi li moli.' (I admit I rather miss the old 'li pi' construction, since it would simply the rules a lot.)

With these three aside, however, we can account for all the regular cases of 'pi' as 'x li A y B +C x D => C x{/y\} D. In this pattern, the case of PPs as modifiers is not a special case, but completely regular, as are incorporated regular predications, like ''jan pi alasa pi waso telo, pi tempo lete', and degrees of adjective. 'jan pi pona pi mute pi mute ala'.
(I note in passing that the comma in the duck hunting example is problematic in the way that such commas often are. Here it is left over from the original predication, where it separated PP from DO. But in the present context, it could also separate 'tenpo lete' from 'alasa pi waso telo' and attach it directly to 'jan', "a winter person who hunts ducks", a possible -- though slightly odd -- description. I have no suggestions how to deal with this, since the whole notion of commas is in very early and tentative state.)

There remain some cases of 'pi' which do not yet fit neatly into constructive patterns. The patterns are set up for noun phrases, so cases of verb phrases are not covered (but are, happily, rare). There are also some cases where the best format is still unclear: what are the degrees of 'pona lukin', for example: ''pona mute lukin' or 'pona lukin mute', those these are not strictly about 'pi' yet.
Post Reply