Return-Path: Message-Id: Date: Tue, 11 Jun 91 02:39 EDT From: lojbab (Bob LeChevalier) To: lojban-list Status: RO X-From-Space-Date: Tue Jun 11 02:40:32 1991 X-From-Space-Address: lojbab Subject: Story of a Proposal - A Solution to Cleft Places & all the worlds' ills 1st of 6 related messages A couple of weeks ago, I hinted that Nora and I had come up with an idea about cleft place structures that we thought would solve that problem, but we wanted to hear what people said about the problem first. That proposal died while you-all were discussing. John Parks-Clifford and I then developed a new solution on Sunday a week ago (2 hours on the telephone beats a lot of letters - via email or otherwise). This solution had the additional benefits of tracking with linguistic analysis of related phenomena in the natural languages, and also of suggesting a solution to the nagging problems of indirect questions, the place structure of "djuno", and providing some unintended niceties as well. Expertise in the language is not always a boon. Last Tuesday, I presented the solution to John Cowan, who thought it sounded OK, and to the Lojban conversation group. It took Sylvia Rutiser (far less 'expert' in Lojban than the rest of us) 2 minutes to find a big hole and Nora 2 minutes longer to drive a truck through it and totally demolish the idea. Well, almost. I came up with a little change (what is now tu'a in the discussion below), which got around the problems and got the idea back on track. pc, Nora, John Cowan, and I agree this one might work. Do you agree? John Cowan wrote up what became the following discussion for that Tuesday review meeting. He gets the credit for doing so, and for generating the last debate on the topic, which occured when I went over the plan with Athelstan on Friday. Athelstan agreed in principle with the package, but did not like the implications of John's wording of one part of one line. The result was a 2-hour discussion that lead to 2 alternatives, which we will now place before you-all for your opinions. The discussion is broken into several messages due to length. Please read all of them before commenting, since comments in one may answer your question in another. Ready! Aim! Fire! (Anyone who doesn't know what the cleft structure problem is, and cannot figure it out from the proposal or the ensuing discussion should send me a message, and I can send you my original posting describing the problem. Warning - this is deep in the nitty-gritty of Lojban grammar and place structure theory, and I rely in my discussion on people knowing our common terminology used in Lojban grammar, and also that you have copies of the grammar and cmavo list.) Proposals: 1. Sumti Raising: Natural language has a feature whereby the subject, or less frequently the object, of a subordinate clause is brought out of that clause up to the main level of the sentence, with the actual subordinate clause deleted, and remaining only by implication from the remaining subject/object. This feature is called subject-raising or object-raising, depending on what is 'raised' to the main level. In Lojban, there is no difference between 'subject' and 'object', because free rearrangement of arguments (sumti) using conversion with selma'o SE can lead to any sumti being in the first 'subject' position. The proposal thus generalizes 'sumti-raising'. Initially, the proposal intended sumti-raising to be unmarked - relying on semantic knowledge of whether a given place in the place structure takes a concrete sumti or an abstract sumti based on selma'o NU. As an example of this earlier form: mi troci le vorme (1) I try the door. based on the place structure of troci x1 tries to do/attain x2 by x3 (1a) which I interpret as expanding to: x1 tries to bring about the event/state/process/activity x2 by method x3 (1b) (1b) clarifies that x2 is an abstract clause. I then view "le vorme" is a sumti-raising from one of two possible x2 abstract sentences: mi troci lenu mi karyri'a le vorme (2) I try the-event I open-cause the door. mi troci lenu le vorme cu kalri (3) I try the-event the door is-open. (There is further possible complication in (2) in that rinka (cause), the basis of "karyria" would normally take an event abstraction in its x1 place, thus making the "mi" in that position a sumti-raising as well. (2) thus can be further expanded to: mi troci lenu lenu mi lacpu/catke le vorme cu rinka leka le vorme cu kalri /lenu le vorme cu kalri mi troci (lenu I try the-event cu rinka (4) the event I pull /push the door causes ) (4a) the-property-of the door open(-ness) /) (4b) /the-event-of the door is-open and it can be seen in (4a) that (3) could be a kind of sumti-raising wherein a property abstraction turns into an event abstraction. It can also be seen that there is semantic information lost in sumti raising - we don't know in (2) whether pushing or pulling the door is necessary (or hitting the elevator door button, for that matter) There clearly is a lot of semantics hidden in "le vorme" in (1). The flaw in sumti-raising as presented is revealed when we put the 'concrete' sumti "le fasnu" in x2. "le fasnu" appears to be concrete, but the x1 place of fasnu being implied is the epitome of an event abstraction. You cannot assume that: mi troci le fasnu I try the event. is a sumti-raising. The x2 value of this sentence represents the first place of fasnu, which is itself an event abstraction. Thus what looks like a concrete sumti is really an abstract one. "mulno" is even worse - it means complete/whole, with 'complete' as the translation for x1 as an event, and 'whole' as a translation for a concrete x1. Sylvia Rutiser noted that one who knits might talk about a sweater as being whole (which is not sumti raising) or can say "the sweater is complete", which is sumti raising from "The knitting of the sweater is complete." The question of whether we are dealing with abstract or concrete sumti suddenly becomes ambiguous without context; this kind of ambiguity, while technically semantic in Lojban since there is no grammatical difference between the two interpretations, is properly seen to be an unacceptable GRAMMATICAL ambiguity, because we don't really know what the ROLE of the words is without context. On the other hand, sumti-raising is clearly valuable to Lojban speakers. It allows brevity and makes the language seem more 'natural'. sumti-raising also would answer the question as to how we interpret a variety of bridi sentences in which the value in a sumti place seems not to fit the formal place structure. Finally, as we observed in the expansion from (2) to (4), there are many levels at which sumti-raising can creep in subconsciously - and we who are raised to speak natural languages with implicit sumti-raising may be unable to eliminate these subconscious lower levels of sumti raising. In effect, then, sumti-raising will be a part of Lojban whether we want it or not. Better to recognize it and encourage it to be used properly, as in solving the cleft structures problem. The solution to the problem of determining that sumti-raising is intended is to explicitly mark it, and the cmavo "tu'a" has been proposed in selma'o LAhE to mark a sumti as indirectly indicating some full subordinate abstract bridi that would fill the space. (An unmarked concrete sumti in a place that normally takes an abstraction would also be taken as sumti raising per the above discussion - communication would likely take place, but this would be a potential semantic error of the type that can result with "mulno". Thus we will not teach unmarked sumti-raising as valid, and will discourage it when we notice it.) That sumti-raising solves the cleft place structures problem is shown in the following, from John Cowan: sumti-raising allows a concrete sumti (one not built on selma'o NU selbri) to appear where the place structure calls for an abstract sumti. This construct is taken as elliptical for "le NU broda" for some NU and some broda. Simple example: lenu mi cinta cu galfi le bitmu The event-of (I paint) modifies the wall tu'a mi galfi le bitmu [The abstraction-of] I [doing something] modifies the wall. This allows uncleft place structures to mimic cleft ones for user convenience. The consensus so far is that uncleft place structures are the preferred way to go in Lojban. sumti raising allows the full range of natural language expression in an uncleft place structure language by explicitly analytical means. ---- lojbab = Bob LeChevalier, President, The Logical Language Group, Inc. 2904 Beau Lane, Fairfax VA 22031-1303 USA 703-385-0273 lojbab@snark.thyrsus.com