KOhA 7 8

From Lojban
Jump to: navigation, search

KOhA 7 & 8

Many of these items are fairly unfamiliar, so any proposal about their future use has to begin with a survey of their apparently intended uses.

KOhA8

The only member here is ce'u, defined in the wordlist as a lambda, that is a devise for creating the name of a property (etc) out of the predicate for it by putting variables into the sumti slots. The lambdas in logic have the grammar of descriptors – that is they bind a free variable to make a term: ‘lxFx' for “the property F” like ‘ixFx' for “the thing that has F” ce'u however is just the variable in situ, the prefixes lambda x is omitted, thus F ce'u F, ‘c F' for the first case above. As a result, successive ce'u in the same abstraction are each a new bound variable; to repeat the same variable requires using meaning-anaphoric pronouns (not Lojban’s most reliable device). Sentences are created in this system by the application of a lambdaed predicate to an argument or set of arguments, which results in the arguments in left to right order replacing the lambdaed variables in left to right order. Thus ‘(cFc)a b’ => ‘aFb’. Too few replacing terms creates a new predicate with fewer open places, to many leaves that last few ignored in the process.

The questions about ce'u are mainly about when it is necessary: is broda somehow different from ce'u? Can ce'u be omitted in places other than the first, or will unmarked su'oremoi places be taken as zo'e or zi'o or even zu'i (see below)? The standard advice (whether or not practice the data is too small to be sure) is to put in explicitly any deviation from the lambdaed first place and zo'ed the rest. Thus, a two-place relation requires at least one ce'u (in the su'oremoi place) or two if the first lambdaed place is not the first place in order. Further, if a place is to be eliminated altogether rather than just given an indifferent or expected value (current zo'e), it has to be marked by zi'o (or the predicate modified as discussed below). What practice there is has mainly been in technical discussions, where the extra precision expected has led to all ce'u, even those in the first place, being displayed.

The only suggestion about changes for ce'u has been the radical one of changing it from standing for a lambdaed variable to having it play the role of lambda itself, with regular variables as variables: da F} for the running example (whether the forms need to prefixed and the prefix marked are open questions as is whether each variable in such a prefix needs a separate ce'u or whether an initial one would cover the whole list). The two advantages of this suggestion are that it eliminates the need for relying on pronouns when some value recurs (‘cxFx' for a reflexive relation rather than ‘c F it') and the permission to rearrange the order of arguments without having to use SE or FA: ‘cxcy(yFx)' rather than ‘cy seF cx,' for example. The disadvantage is that it makes almost all uses of ce'u longer and eliminates the choice to drop explicit reference in the first place. Much of this objection could be overcome by the conventions that 1) the variable need not be explicit if there is only one and it is not repeated (reducing da to ce'u in that case) and 2) as now, a ce'u by itself can be dropped from the first position. This probably means that most actual uses are unchanged; the expanded ones will mainly be in precision work where the detail might be welcomed.

KOhA7

This class consists of five cmavo that play a variety of roles in sentences, mainly as sumti placeholders, but usually with some significant role to play.

ma is the most obviously significant member of this class. When it fills a sumti gap, the sentence in which it occurs is converted into a question, a request to fill the matrix around the gap supplying a sumti that makes for true sentence (the technical description of what happens need not concern us here). This is radically different from any other gap filler in this class since it does not affect truth-value but changes the nature of the sentence from assertion to question.

So far as I can find, no one has suggested a change for ma or pointed to any problems with it. To be sure, if the questioned referent is repeated, we are thrown back on referential-anaphoric pronouns since each occurrence of ma introduces a new question (gasnu is “Who does what?,” not “Who does himself”). But no one has suggested that ma become a term-making operator as has been suggested for ce'u.

ke'a is used in relative clauses to refer back to the sumti to which the clause is attached. It is a referential-anaphoric pronoun of restricted use, then. The standard seems to be that ke'a can be dropped when used for the first argument of predicate in the relative clause but otherwise is required (and it is permitted – and often used, especially in technical discussions – even in first place). All occurrences of ke'a in a clause have the same referent, so pronominalization is not needed. No one has suggested a change for ke'a in this use. There have been suggestions for using ke'a in other environments – in abstracts, for example – to pick up “obvious” referents from the context. No good case has been made for these suggestions so far.

zi'o, zo'e, zu'i (and [1]) Usually, when we use a polyadic predicate, we do not fill all the slots: djan without mentioning whence or by what path or in what vehicle. It may be that we don't care at the moment or it may be that the answer is too obvious to be worth mentioning. What we usually do in that case is simply say nothing. If a fussy logician were to challenge us for leaving places unfilled, we might say “Oh, just put in ‘something'”, a particular quantifier of unspecified – so presumably appropriate – subject, in Lojban a bare da or da. But da may be viewed as saying either too much or too little: on the one hand it encourages the question “Which what?” which is too much if we don't give a damn; on the other hand it does not do justice to our claim if we have some obvious object in mind. So, when we need to fill the places – to soothe the fussy or to indicate that the important stuff goes not in the next place but in one after that – we use zo'e. If we want not only not to specify what object is there but to get rid of consideration of that place altogether, to convert the n-place predicate to an (n-1)-place predicate by dropping that place, we close the gap with the plug zi'o (or compound the original predicate with the prefix zil- and a place marker: zilmumrklama is a four-place predicate just like klama except that the question a vehicle does not enter into it at all). And finally, in general sentences, when we are just making general claims (I think this is how it goes but we don't have examples to test) we use zu'i to indicate that what foes in there is a reference to a typical fill, whatever that may be in the situation.

zo'e inherits from the blank space a double reading, both of which have pretty clearly occurred – but which are not well separated. In djan we may not care about whence and along what path or in what vehicle or we (speaker and hearer both) may know that John only goes to the store from his office along Route 7 in his Dodge Dart. We may not want to waste words on what is already known. Or we may know and still not give a damn. It might occasionally be useful to know which situation obtains, so that the hearer will not bother asking about the “missing information” or feel left out if he does not know what he takes it that others do. One suggestion about this has been to use zo'e for the "don't give a damn" case when something is said and zu'i for the “obvious” case (zu'i is not being used for “the typical value” and that isn't very clear anyhow). We can leave the blank ambiguous, to be relieved by the more explicit marks if need be – or eventually in the case of “obvious” by supplying the references which everyone was taken as knowing. Thus, the tugni in response to a proposal is still enough; we do not have to go over to zu'i tugni zu'i zu'i zu'i let alone to tugni.

There is a current usage of zo'e in yet a third sense, where it turns up as an independent pronoun, not just filling a place but capable of being further modified: noi being the main pattern. The idea here seems to be to get to a totally unspecified thing (well, it has to be of the sort that can broda) without using quantifiers, with the problems of scope and the like these bring with them. What this means in terms of any discipline: syntax, semantics or pragmatics, is unclear. zo'e is pretty clearly not meant as a mere placeholder in a formula, to be filled by a more explicit term to make a claim, like the xns in definitions, for example; it is taken to be referential as it stands but without in any way fixing on the referent. The aim is often put as talking about a “generic individual” and the purpose is to find a way of expressing general claims more clearly. But the notion of a generic individual (even of certain sort of thing) and of using one individual to determine what is true in general, seem logically misguided as well as metaphysically.

From all of this I extract the following proposals:

ce'u becomes something with pretty much the grammar of a quantifier: goes before a variable to form a term, can stand alone as a term in an abstraction that does not contain another ce'u, is used only in abstraction (and we throw in the convention about dropping from the first place in the abstraction).

zo'e is used to make explicit the speakers indifference to what the referent in a given place is.

zu'i is used to make explicit that the place is to be filled with the “obvious” referent in this context.

Otherwise, ma, ke'a and zi'o are unchanged.

The ce'u will add variables to some occurrences of ce'u and use those to replace some anaphoric pronouns. The cases will be few, I think. Most cases will fall under the conventions about single ce'us and dropping the first place ones.

A few zo'es will become zu'i - and it may be hard to tell which in some cases (which was why the change in the first place). Apparently, no cases of zu'i – outside CLL – occur, so no usage is affected with these. And zu'i gets a more useful meaning.