From cbmvax!uunet!cuvma.bitnet!LOJBAN Thu Feb 13 11:29:57 1992 Return-Path: Received: by snark.thyrsus.com (/\==/\ Smail3.1.21.1 #21.19) id ; Thu, 13 Feb 92 11:29 EST Received: by cbmvax.cbm.commodore.com (5.57/UUCP-Project/Commodore 2/8/91) id AA13613; Thu, 13 Feb 92 11:23:45 EST Received: from rutgers.edu by relay2.UU.NET with SMTP (5.61/UUNET-internet-primary) id AA28628; Thu, 13 Feb 92 11:14:05 -0500 Received: from cbmvax.UUCP by rutgers.edu (5.59/SMI4.0/RU1.4/3.08) with UUCP id AA22138; Thu, 13 Feb 92 08:48:34 EST Received: by cbmvax.cbm.commodore.com (5.57/UUCP-Project/Commodore 2/8/91) id AA24493; Thu, 13 Feb 92 07:36:18 EST Received: from CUVMB.COLUMBIA.EDU (via uunet.UU.NET) by relay2.UU.NET with SMTP (5.61/UUNET-internet-primary) id AA24735; Thu, 13 Feb 92 03:31:36 -0500 Message-Id: <9202130831.AA24735@relay2.UU.NET> Received: from CUVMB.COLUMBIA.EDU by CUVMB.COLUMBIA.EDU (IBM VM SMTP R1.2.1) with BSMTP id 6229; Thu, 13 Feb 92 03:30:16 EST Received: by CUVMB (Mailer R2.07) id 0047; Thu, 13 Feb 92 03:29:05 EST Date: Thu, 13 Feb 1992 02:44:49 -0500 Reply-To: Logical Language Group Sender: Lojban list From: Logical Language Group Subject: more detailed explanation of generation 1 and textbook plans X-To: lojban@cuvmb.cc.columbia.edu To: John Cowan Status: RO I've been asked a couple of pointed questions about the textbook and grammar status - why they haven't been done yet, and to what extent the current plans fulfill our promises of long ago. I will lead up to a specific set of questions, published in a separate message for easier response, to give me a better sense of what aspects of our teaching materials are most important to you. The result will largely determine the emphasis I will use on assembling and updating the teaching materials for the first generation learning-materials-book/ textbook. Many responses are sought, from people of all levels of activity. Lurkers, speak up! Let me give an outline history for the discussion that follows. Relative newcomers will be able to make more sense of the discussion with this common basis. Dec 87 was when the original gismu word-making was completed. At that point we had not even fully accepted that we would have to recreate the grammar from scratch to avoid JCB's copyright claims. That grammar work started in earnest the following June. Oct-Nov 88 we had the first complete 'Lojban grammar' done. In Oct 88, I also put out for review the first 30 pages of a detailed grammar description explaining the whole grammar in depth from a 'big picture' point-of-view. Review comments from Dave Cortesi, among 8 others indicated that while the explanation was good, it just wasn't tutorial enough. Dave suggested that we write a textbook first, then write a 'reference manual' later. Jan 89 - we started the first Lojban class, intending to cover lessons at one per week and finish the whole language by June. I was attempting to write lessons at that rate, and came close to succeeding at first. But the pace was too ambitious for the class, and for me. Still, the 300,000 words or so of the draft lessons that I completed by May stands as my high point of productivity - one that I have suffered a lot of stress trying to live up to ever sense. The draft textbook outline, some 18 pages, is still a reasonable summary of what a systematic study of the entire grammar should include, even though the last 8 lessons outlined consist mostly of miscellanies that disjointedly fill in gaps in the first 10 lessons designed to achieve conversational ability. Jun 89 - LogFest that year interrupted my textbook writing, and I never really got back to it. The class continued, but the students were lagging in vocabulary. Questions arose in that we couldn't answer quickly, notably the 'negation' question, that led to 8 months of thorough rethinking, which was followed by detailed rethinking of virtually every other point in the grammar in successive examinations of similar depth, ever since. Put succinctly, while we could teach the language we knew with little problem, it was difficult to answer questions that used features of the language that we and others hadn't used, and that no one had thought about much, either. But as the language community grew, the odds were that every new active Lojbanist would stumble onto a new item that we hadn't thought about, and ask an unanswerable question. Nov 89 - after a month-long interruption for Worldcon, we declared the first class 'over' after having achieved the original aim - sustained conversation totally in-language for several hours by people who hadn't participated in the design. But the bodies left along the way showed that the teaching philosophy had to be redesigned. The 2nd class, in Blacksburg VA, had started in April but folded after some 4 months after completing lesson 3 - the draft lessons were OK for self-paced self-study but insufficient for a group study trying to work together at a single pace, led by someone who didn't know much more than the rest. Another problem was the varying groups of students and their approaches to learning, as discussed below. Nora wrote the grammar outline, now the diagrammed examples discussed in my just-posted response to Frank Schulz, intending it for people who had finished the 6 draft lessons and needed a quick reference to 300 pages of detail. We ended up also giving it out as part of introductory materials after being plagued by questions/complaints from people who had hardly seen any examples of the Lojban grammar features in either the brochure or the Overview of Lojban. Alas - that outline without any explanation never served the introductory function well, and the new version I started circulating last fall is padded with some explanation aimed at the introductory audience. Jul 89-Jan 90 - Athelstan paved the way to a new approach at teaching with his first minilessons, finally realized in print in the draft circulated last fall. In a 1 hour session, using a vocabulary limited to some 40 words that referred to concepts that could be pointed out and used in a typical lecture room, Athelstan covered most of the material of the first 3 draft lessons. He regularly got everyone in a group of novices, able to say something in Lojban and know what and why they said it within that 1 hour. This was astoundingly successful with the majority of those who saw these mini-lessons because it gave a good sense of the 'big picture'. It hasn't worked so well in print, because what was successful as class exercises, where novices learn from each other, fails in print, where it merely appears that there are a lot of exercises with too little explanation. We know how to fix this, but Athelstan has been distracted by numerous personal crises for over a year now and hasn't yet finished even the answer key (alas, for those still waiting). The major problem with the minilesson is the opposite of the one found in the draft textbook. Concentrating on minimal vocabulary, and getting a couple of big ideas across in a hurry, the minilesson gives no sense of the depth of the language design, with all of the flavor of its new ways of looking at the world. None of the cute entertaining, aspects of the language are included (nor are they in the draft lessons). The problem is that these features, while interesting and mind-expanding, tend to be rarely applied in real usage. (As an example, I refer to the classic use of 'enough' as a number by JCB, leading to what in Lojban is "mi raumoi" - I am enoughth (in a line of attendees for a sold-out movie). So we don't teach them early - they are bells and whistles. But they are also what makes the language worth studying for some. A related problem is that, while a couple of basic concepts are learned well using the minilesson, we've had very little sign that anything more than those basic concepts have stuck, and the latter only minimally and at the conceptual level. Athelstan has several times tried to do a second minilesson to follow on for people who have had the first one - a mini lesson that might be taken the next day, or maybe several months later at a different SF convention. They've never worked except for people who have just had the first lesson - too much was forgotten. The sense I (possibly different from Athelstan) have, is that the minilesson effectively teaches the 'what' of the language, but not the 'why', and that lack of 'why' means that people don't have the wherewithal to continue on their own. The draft lessons, by comparison, are heavy with 'why', but inefficient at getting you to a point of being able to express yourself in the language, and even more important, with being comfortable doing so. This can only be done with LOTS more examples and exercises. The discussion of the diagrammed examples is a constant problem I've had in writing teaching materials, from intro stuff to the textbook. I've been told by people who are skilled at teaching that at the beginning of a teaching sequence, you must simplify, and if necessary oversimplify, to avoid confusing the learner. A senior Lojbanist who teaches martial arts says that a good instructor will positively lie in the first lessons, because the untrained person will not only not understand, but because of the differing cultures, etc., will jump to wrong conclusions. The draft mini-lesson, and the older draft textbook lessons, are patterned after this idea of starting with slow, somewhat oversimplified situations, and working up to complications later. The ultimate of this, of course, is JCB's Loglan 1, which teaches a bunch of oversimplified pieces of the language but gives no idea how to put the pieces together coherently. I think the draft textbook did much better than that, being that by lesson 6 it had worked up to a full-page narrative. Other people, especially those who do not intend to spend a lot of time learning, but who want to get the big picture quickly, find the diagrammed examples, and the EBNF to be just what the doctor ordered. They find the elaborate explanations of the longer texts distracting. Indeed, this is reflected in the two types of complaints most often received about the draft lessons. GROUP I - One group basically complains that after lesson 3, 4, 5, or even 6, they don't have enough grasp of the features of the language to say what they want to say in print OR in conversation, and hence are not motivated to try (things like lujvo making, logical connectives, and non-trivial tenses). When I ask people in lesson 3 to make up a sentence expressing a simple bridi, a common response was, "how do you say this?" followed by something that is considered simple colloquial English but which is not within the power of lesson 3 grammar (lack of knowing syntax for relative clauses was the typical key to these problems, for those who are interested - these features are covered in lessons 5 and 6 - certainly of interest to you, Frank). One of the first class students, Albion Zeglin, asked a question about negation (interactions with quantifiers) about the time we finished lesson 6 - one that we hadn't yet considered - it was so abstruse by the standards of even lesson 6 that I hadn't worried about the answer, much less teaching it. The result 8 months later was the negation paper, now an unofficial 'lesson 7' without exercises that we include with the draft lesson set because it presumes a person knows the material of those lessons. The type of 'big picture' overview in that paper represents what the Group I student usually wants. The upcoming tense and MEX papers by John Cowan, and a whole slew of other papers still on the drawing board address this approach to the language, which is that I originally intended back in 1988 - a thoroughly documented summary of the grammar features explained in a moderately tutorial fashion. GROUP II - The other group found the pace of the textbook to be too fast, not in terms of number of pages spent, but simply the rapid pace at which new ideas were presented. When I completed a topic, I moved onto the next, presuming that the previous topic was covered and therefore understood. But if a person was uncertain about a point in lesson 2, then getting a bunch of new information in lesson 3 only increased the confusion. For all the examples and exercises that I had put in the draft lessons, it turned out that there weren't nearly enough for these people. The comes a third situation, closely matched with the second situation. I designed the draft lessons for a serious student who would be doing most or all the exercises, studying the examples, and using LogFlash on the side for rapid vocabulary growth. The draft lessons assume roughly 100 new gismu words per lesson are being studied - assuming a LogFlash student working daily with 20 words a lesson, this would mean a pace of at least a lesson a week. The goal was to get people as quickly as possible up to a threshold of about 900 gismu (total vocabulary around 1200 words), which was then and is still seen as about the level at which people are comfortably willing to try to converse. But even the best students in the first class didn't stick with the program, and only one got through the 900 words - and then after EIGHT MONTHS, while those of us who had designed Lojban had done the full 1300 words in 2 months, then gone through them the critical second and third passes in another month. The rapid vocabulary pace adds to the whirlwind feel of the draft lessons. This feature coupled with the fast buildup of grammar discussed for the second group, made the language especially hard for those people. The first group didn't appreciate the faster growth, because they even more than the second group, were attempting to bypass LogFlash and other systematic vocabulary building approaches and get to the meat of the matter. Now - Thus at the current state, we have a variety of projects in the works to improve our teaching materials. The minilesson will be rewritten to provide that brief push for those who want to get started - it will become part of our intro materials. Among other things, completing it will be a sign of good faith and commitment for the limited number of people we can give support to without payment. The diagrammed examples will also help provide the big picture at the introductory level. The draft textbook will offer the other extreme, the slow detailed approach that will get people understandingly to the point of trying to express themselves in Lojban. We now feel that the 6 lessons contain some 75-80% of the grammar needed for beginning conversation expression on most topics, and the remaining portions tend to be picked up rather easily by simply studying what other people have written and trying to write on your own. Nick, Mark, John Cowan, and many others have proven that this works. Enough for expression is NOT enough for translation - the stuff Nick and Mark are doing now come some 6-9 months after finishing those draft lessons and continuing to work with the language at a moderate level. But Frank Schulz should feel much more comfortable with the type of thing he is writing now for the list. The generation 1 book will also contain the introduction and 1st lesson of the revised textbook, a much more solidly written text that shows the lessons we've learned from the minilesson, while maintaining the depth of a textbook, and adding LOTS of examples and exercises. John Cowan is working on what will effectively complete the Oct 88 grammar summary, but in a bit more tutorial style rather than straight reference, a coverage and style similar to that in the negation paper. Many known aspects of how Lojban semantics reflects its grammar will be presented. There are still holes (and always will be), but they will be small. His initial attempt at a reference, the dictionary-like "selma'o catalogue" suffered from excessive need to cross-reference to other sections; it tended to read better from front-to-back, rather than as a reference where each item stands on its own, the way people use a dictionary or encyclopedia. There are about a dozen papers on his list, of which he will write all but a couple. Those which are done will be included somewhere within the generation 1 books, but because they are a cross between reference and tutorial, we have wavered between putting them in a separate book, or combining them with one of the other books. What happens will probably depend on how much he's gotten done at the time the rest of the generation 1 stuff is ready, within a couple of months. The leaning now is to three books, with such semi-reference materials kept separate from both student text and pure reference. This allows us to include some of the better explanatory materials from JL. The last generation 1 book, a pure reference if John's stuff and similar materials are published separately, is looking more and more like what people want to see as a dictionary all the time. It won't be what >I< have in mind for a dictionary, and in many ways it won't be what some people are looking for, but I think I'm going to be able to include many lujvo without too much work, and perhaps twice as many English entries as Lojban entries, possibly putting the word-coverage almost comparable to that of JCB's 1975 dictionary. The definitions will be sparse, generally limited to the 100 characters of the LogFlash gismu list, but that list is being tightly packed with information in its new revision. This will be the first book completed. We expect that people already working with the language can leap on ahead without a lot of help if they have this book, and the textbook and other materials can then piggyback on their added person-power. It also forms a basis of commitment to stability in our baselines that has thus far been sufficient to motivate some to learn the language, but not all, but isn't quite solid enough to give me confidence when textbook writing. (Imagine what it would be like to repeatedly go through textbook drafts trying to catch errors cause by place structure and cmavo changes of the last year. It is much easier to relearn the minor changes than to go back and bring the past work up-to-date. This nitty-gritty detail work is what makes the job take so many months. But when we don't do it, we get sloppy errors like the one R. Miller and others caught in the Diagrammed Examples dealing with the pronuciation of 'ai'. lojbab