From lojbab Wed Sep 21 01:45:30 1994 Received: from access1.digex.net by nfs1.digex.net with SMTP id AA24960 (5.67b8/IDA-1.5 for ); Wed, 21 Sep 1994 01:45:20 -0400 Received: by access1.digex.net id AA05793 (5.67b8/IDA-1.5 for lojbab); Wed, 21 Sep 1994 01:46:44 -0400 Date: Wed, 21 Sep 1994 01:46:44 -0400 From: Logical Language Group Message-Id: <199409210546.AA05793@access1.digex.net> To: holmes@diamond.idbsu.edu Subject: Re: Chief logician? Cc: lojbab@access.digex.net, lojban@cuvmb.cc.columbia.edu Status: RO Yes, some of our analyses go against the grain of 'natural analogy', but so far, for the most part, people haven't had much trouble. For one thing, we have more explicitly defined negation as a predicate negation that has full sentence scope, whereas natlang negation tends to only have scope over 'the rest of the sentence' following the negation. We define our negation as being immediately exportable to the prenex by definition without change inquantifiers. In my own usage, therefore, when I rarely have to resort to quantified variables, I just leave the negation in the prenex and do not use the 'short forms' of natlangs where the variables are not explicitly quantified. I find this form easier to understand, almost error-free in manipulation, and the circumstances where I have been required to use it rare enough that Zipfean shortening doesn't seem to enter in. (We do have a natlang style negation form, but it isn't much used except in translation.) The easiest way to exemplify our variuos forms of negation is to refer you to John Cowan's negation paper on ftp.cs.yale.edu pub/lojban/draft/refgrammar At last report there was an error in one section (a DeMorgan expansion) that might not have been caught, but it has lots of deatails of the various forms of negation. If I recall the discussion, the two more important kinds of negation are defined by reference to a classic (I think Aristotle) 'box' diagram, where contradictory negation is expressed by moving diagonally across the box, and contrary negation by moving horizontally (ort is it vertically and horizontally or ...). I am not walking to the store could mean 1) simply that it is false that I am walking to the store 2) that I am driving or running to the store 1) is contradictory negation 2) is contrary negation Metalinguistic negation deals with things like Have you stopped beating you wife? when you haven't started - it says that there is some hidden assumption that is incorrect that makes the sentence not only false but meaningless from a truth-functional point of view. Contrary/scalar negation works very well on arguments: I don't have 3 children (I have 2 or 4) is expressed in Lojban better by nbenegating the argumetn "3 children". Metsalinguistic negation crops up in a variety of ways. Grammatically in Lojban it is an attitudinal (actually a discursive - but same grammar). Contrary negation can be applied to arguments, can be applied to predicate words within a metaphor, or can be applied to an entire metaphor or predicate 9and a few other things). Contradictory negation only applies to sentences, and all forms of contradictory negation of a sentence negate the entire sentence (Modulo the special natlang scope form which is used to exempt quantified variables in a 'natural way') ==== Abstract raising has reared its uglu head in so many ways in Loglan/Lojban that I hesitate to try to cover all of them. One is underlying the current discussion on nitcu/need. Do you need an object, or do you need a state/event involving that object? Causality, is it 'guns' that kill (dead-cause) people, 'people' that kill people, or 'people shooting guns' that kill people. The first two are probably raisings of the third. Most Lojban causality is now expressed as 'event causes event' rather than 'object causes event' or 'object causes object' or 'event causes object'. But we have a form, now commonly used that recognizes that we aren't always sure exactly what the event is, or how to express it (or simply that it doesn't matter), so we can explicitly raise the argument from abstract to object with a marked form. e.g. my work is done le nu mi gunka cu mulno (lojban) lepo mi turka ga kapli (TLI) and not *lemi gunka cu mulno ?lemi turka ga kapli but Lojban allows tu'a mi mulno raise I am-done "I'm done." which if the L1 place styruture is correct "mi kapli" should never be said, unless perhaps at one's point of death if you want to view yourself as an event. We had to go through all of the place structures and decide which of them took events as a norm, while which took objects (there are circumstances where cossover is permitted and works without marking though), and in some cases mosdify place structures to eliminate raising. TLI Loglan words that have hidden raising are most evident in predicates where there is an abstraction in one argument, and a concrete (often an agent) in another argument, such that the concrete is ALWAYS found as a place in the abstraction. This is not always 'bad', but we have found it worthwhile to know exactly where it occurs, because logical errors seem to explode out of such constructions. (For one thing, quantified variables inside an abstraction do not necessarily export to the sentence level. But if the two levels are muddled, you can get errors in scope.) ==== Oh, I failed to address one point under 'negation' - termionology. We much prefer to use 'real' definitions of terms rather than jargon definitiuons. Only to a logician does 'negation' apply only to propositional forms. English words have English meanings, and most people think of 'negation' as being tied to the meaning of "not" in a wide variety of usages. You can control discussions by insisting on narrow meanings for words, but then you find that the results of the discussion do not have wide applicability. WE chose instead to bite the bullet and tackle the fuzzier issues of negation along with the purely propositional ones, because then people know how to handle the concept "not" even when it is not purely a propositional usage. lojbab