me lu ju'i lobypli li'u 9 moi
Copyright, 1989, 1991, by the Logical Language Group, Inc. 2904 Beau Lane, Fairfax VA 22031-1303 USA Phone (703) 385-0273 firstname.lastname@example.org
All rights reserved. Permission to copy granted subject to your verification that this is the latest version of this document, that your distribution be for the promotion of Lojban, that there is no charge for the product, and that this copyright notice is included intact in the copy.
Number 9 - June-July 1989 Published by: The Logical Language Group, Inc. 2904 Beau Lane, Fairfax VA 22031 USA (703)385-0273
Ju'i Lobypli is the quarterly journal of The Logical Language Group, Inc., known in these pages as la lojbangirz. la lojbangirz. is a non-profit organization formed for the purpose of completing and spreading the logical human language "Lojban". The newsletter section of Ju'i Lobypli is separately published under the name le lojbo karni. Issue #9 of le lojbo karni (LK9) was mailed in early May, as part of our effort to move our publication schedule up by one month. You should already have received that issue by the time you receive this. Issue #10 of both publications should be distributed in August or early September. Some 280 of you will receive this issue of Ju'i Lobypli; we are approaching 500 subscribers to le lojbo karni.
Our thanks to those of you who responded to our plea for contributions and donations. We aren't out of the woods yet, but neither are we afraid of not being able to pay for this issue - a real fear last time. We received several donations of $100 or more, making up the deficit in the bank account, and for the first time, we have over 100 people with positive balances. We still have a negative net worth, though. So, we would still like to have those of you with negative balances either to bring them positive or to let us know that you are interested even though you cannot afford to pay.
Potential donors please note: we still have not received IRS approval for Section 501(c)(3) status, which will officially allow your donations (not contributions to your voluntary balance) to be tax-deductible. We hope to have such approval by the end of the year. We are operating in accordance with that section, and your contributions now should be deductible if approval is obtained later, although there is always the possibility of disapproval. You may,if you wish, make your donation contingent on our getting such approval. We will inform all donors at the end of the year as to the status of deductibility of their gifts. We also note for all potential donors that our bylaws require us to spend no more than 30% of our receipts on administrative and overhead expenses, and that you are welcome to make yougifts conditional upon our meeting this requirement.
Your Mailing Label
Your mailing label reports your current mailing status, and your current voluntary balance including this issue. Please notify us if you wish to be in a different mailing code category. If there is an x or y code after your mailing status,you will be reduced to the code after the letter if we do not hear from you before next issue. Balances reflect contributions received thru 11 July. Mailing codes (and approximate annual balance needs) are defined as follows:
Level B - Product Announcements Only Level R - Review Copy for Publications Level 0 - le lojbo karni only - $5 balance requested Level 1 - le lojbo karni and Ju'i Lobypli - $15 balance requested Level 2 - Level 1 materials and baselined/final products - $20 balance requested Level 3 - Level 2 materials and lesson materials as developed - $50 balance or more
Contents of This Issue
This issue contains an attachment with the rafsi sorted alphabetically. There are three lists, in CVC, CVV, and CCV forms. We have a discussion of the Lojban design of mathematical expressions (MEX). Sheldon Linker provides some latebut significant comment on the algorithm used by us (and in previous versions of the language by Dr. James Cooke Brown)to make gismu. We have Lojban text from class members and two self-teaching students, the text of the skit (written byNora) presented at LogFest by the DC area class. We have a short report on some recent major events occurring since LK9, including LogFest 89; a more complete report will be in LK10.
News Update -----2 Commentary on the Loglan/Lojban gismu Lexicon -----4 On Constructing a Lexicon for an Artificial Language - by Sheldon Linker -----4 Bob's Summary of the gismu Algorithm -----6 Bob's Response to Sheldon's Critique -----7 Special Section - Mathematics and Lojban ----11 'An International Language for Mathematics' by Robert Strichartz - reprinted with permission from "The Mathematical Intelligencer" ----11 Our Submitted Response to Dr. Strichartz's Article ----13 On the Lojban Design for Mathematical Expression (MEX) ----15 le lojbo se ciska ----19 Self Descriptions by Lojban Students ----19 With Apologies to Spock by Bob LeChevalier ----20 A Story from Another Culture by Preston Maxwell (with corrections by Bob and Nora) ----21 la sinderelwud. by Nora LeChevalier ----21 lei lojbo by Nora LeChevalier ----28 Translations of le lojbo se ciska ----28 SCRABBLEtm for Lojban ----33 Ju'i Lobypli Editorial Conventions ----36 Attachment - Lojban rafsi Sorted Alphabetically
This issue was delayed a bit due to preparations for LogFest, coupled with an effort to get copies of the updated Synopsis to all people at level 1 or higher. We also had our normal class preparations, and a pleasurably higher level of new responses, contributions, and orders, all of which took away from my time to prepare this issue.
Some significant things have happened since LK9 was written, and we don't want to make people wait until August to hear about them.
- Some 40 of you are now receiving the lessons as they are prepared, as opposed to waiting for the textbook completion, currently scheduled for late this year.
- I have short update notes on the draft cmavo list that will be sent to all of you who I have down as having received the draft list. This is being sent on a fairly low priority, though.
- Athelstan was forced to cancel his trip to the West Coast due to a severe case of poison ivy, compounded by (Murphy's Law!) problems with his car. We apologize to all of you who were expecting him. We also want to thank those of you who offered sleeping space or who were attempting to organize meetings with him. Hopefully, at the very least the effort will stir some of you to get together with others who are in your area.
- The day I took LK9 to the printer, I was told that we would now be charged only 3 cents/page for orders of 25 or more copies instead of the previous 100 copy minimum. This allows us to significantly reduce our prices for draft text- book lessons, draft cmavo lists, and the draft machine grammar from the prices included in the last issue. We are going to a 2-page order form that contains more description of the materials and explains our current prices. The prices indicated have been in effect since approximately 1 June.
- Athelstan was able to attend LogFest 89 (albeit in considerable discomfort), along with 20 others, half of them from out-of-state. Most people stayed for the whole weekend. We had hoped for a larger attendance, but those who came were among the most serious students of the language - thus LogFest activities were more advanced than in previous years. The local class put on the skit whose text is included in le lojbo se ciska below.
Brad Lowry and new Lojbanist Elliott Deal videotaped significant portions of LogFest, and interviewed most attendees. The results will be used to make one or more Lojban videos. The first, a short loop tape 'commercial', should be completed within a few weeks, and will be used at Worldcon and other conventions and meetings.
Next year's LogFest will again be on the third weekend of June. Unfortunately, there is no good alternative that allows John Parks-Clifford (pc) to attend and which also avoids the hottest part of the summer here in DC. June is hot enough.
Ralph Dumain attended LogFest, and led a discussion of his and others' ideas on the Sapir-Whorf hypothesis and how Lojban/Loglan might be used to test it. There was excellent cross-fertilization of ideas, and we hope to have three or four viewpoints summarized in the next issue of JL. As noted in last issue, Ralph wrote a response to previous comments on his paper in JL6, and added a bibliography of some books he found interesting in regards to Sapir-Whorf and theories of 'linguistic relativism'. We have copies of Ralph's essays (about 10 pages) available on request, along with a similar length collection of material from JL6 and JL7 on the subject.
- The Loglan Institute, Inc. has announced publication of a new edition of Dr. James Cooke Brown's (JCB) Loglan 1. This book describes an update to what we believe to be a flawed version of the language. The Institute is charging $24.00 for the paperback book of 600 pages. It was supposed to be shipped as of 1 June, but, as of publication, neither we, nor any of our contacts, have received copies that were ordered. The announcement includes a copy of the table of contents, which reveals that the basic structure and content of the book is unchanged from the 1975 edition. There is a significant portion of the book devoted to various sorted gismu lists, rafsi lists, and cmavo lists (these are of course incompatible with, and superseded by, the Lojban lists). The notes on each chapter have been expanded; in the previous edition, these usually contained technical commentary and defenses against presumed criticisms of Dr. Brown's design. The book contains a new chapter (about 30 pages) on testing the Sapir-Whorf hypothesis.
As can be determined from the table of contents, the book's structure remains flawed; it is an overview and not a teaching tool. In addition, it is known to be seriously incomplete. JCB indicated in a separate letter to some people that he was intentionally omitting a complete specification of the grammar from the book in order to keep that grammar secret from la lojbangirz. for a while longer. He does not intend to make this grammar public for at least a year, and his version of a parser will intentionally not contain the grammar in a human usable form. This doesn't hurt us - we have a satisfactory grammar that is probably more sophisticated than JCB's; it does make his whole effort rather worthless, since a language with a secret grammar is uninteresting and unusable.
The most serious criticism of JCB's scientific work is that he has failed to submit it to serious critical review by qualified reviewers who have knowledge of the language, most of whom are associated with us. He states that the language has changed significantly since the publication of Notebook 3 two years ago, a publication that was received by only a few dozen people; presumably only a couple of these reviewed the new book. JCB has worked in a near-vacuum of outside review and support for the last 5 years, and we expect that his results will show this. We doubt that anyone other than JCB, members of his family, and possibly Bob McIvor, who has been doing the computer work for the Institute, have any real knowledge of the current state of his version of the language.
We hopefully will have received a copy by next issue, and will offer a more thorough review at that time. We are also soliciting reviews and comments from any of you who receive the book. All comments, both positive and negative, will be collected together and made available to anyone who wants them (including JCB if he asks). We are recommending that those of you who still have balances with the Institute consider using that balance productively to order the book. Until there is evidence of lasting value, we cannot recommend sending additional money for the book.
- la lojbangirz. cannot be considered an unbiased reviewer of JCB's work, however we may try. We were formed in opposition to his policies and his language design decisions. In addition, we are now involved in a legal dispute with The Loglan Institute, Inc. On May 23, after a unanimous vote of the Board of Directors, we filed a Petition for Cancellation of the Institute's registered trademark on 'Loglan' for 'dictionaries and grammars' on the grounds of fraud, and invalidity of the name as a trademark. Bob, Nora, and Jeff Prothero had retained a legal firm specializing in trademark after the Institute threatened us with legal action last year. With our incorporation, it was recommended that the organization file the petition rather than individuals, as the trademark has a more serious effect on the organization as an entity, than on us as individuals. We of course believe that the trademark damages our ability to reach the community of people who know the language and language project under the name 'Loglan'.
Because of our commitment to you not to use your money to fight this legal battle, we are maintaining a separate legal account. We will use only money donated specifically for this legal fund for expenses related to the petition. Bob, Nora, and Jeff have agreed to donate such money as is needed to pay the legal bills. We will accept donations from others, but ask that your priority on donations be for general support of the organization.
Our expenses on this case will be included within the constraint of administrative and legal costs not exceeding 30% of total expenses. This is not expected to be a problem.
We continue to seek an amicable settlement of the issue with the Institute. However, after delaying for 3 months, JCB rejected our proposed settlement without comment or counter-proposal, then went sailing in the Caribbean. His negative response coincided with his announcement of the new edition of Loglan 1, leading us to believe that the delay was an attempt to stall until the book was published. In any case, you may wish to consider that payments to the Institute for Loglan 1 and other materials might be used in a legal battle against us.
Since the Institute has specifically indicated that it will not sell its grammar in the near future, and the obsolete 1976 dictionary has been out-of-print for over 10 years, its claim that it is using the supposed trademark for these goods seems ludicrous, but that is for the U.S. Patent and Trademark Office to decide.
We will provide an update and more complete information in the next newsletter. Your comments and questions are welcome.
Commentary on the Loglan/Lojban gismu Lexicon
The following paper discusses the methodology used to build Loglan/Lojban gismu, first devised by Dr. James Cooke Brown (JCB) and described in the 1960 Scientific American article. We will summarize the algorithm after the paper for those unfamiliar with it.
Notwithstanding our baseline policy, it is still important to examine any and all assumptions that were a basis for the language design. Lojban will have to withstand questions and criticisms on several fronts in order to be accepted for any of its purposes. Sheldon's paper, written by a friend of the language, questions these assumptions and allows us (and you) to evaluate the quality of our response in a non-hostile environment. We thus welcome Sheldon's paper and commend his points to your con- sideration. We will respond to some of them afterwards.
(Note - the sample gismu used by Sheldon are from an earlier version of the language, although his points apply to Lojban and other more recent versions.)
On Constructing a Lexicon for an Artificial Language
by Sheldon Linker
13612 Onkayha Circle, Irvine, California 92728
714-552-1904 voice 714-552-6985 fax
Abstract: This document presents a discussion of the Loglan/Lojban lexical axioms, and takes a second look.
Statement of Alignment
Before I begin the technical discussion, I must take the time to include a note of a personal nature. I have been asked (regarding the great Loglan/Lojban debate) "whose side" I am on. I have always been interested in the type of technology which Loglan and Lojban represent. Dr.Brown, in allowing me the incursions into Loglan which he did, forwarded my education immensely, and more than he realizes. For this, I will always be grateful. Why, then is this article in Lojban, and not in 'The Loglanist'? I believe that here I will find a more receptive audience for this particular article. In short, I am on the side of advancement of this new art.
Both Loglan and Lojban have a lexicon based on a few simple tenets:
- The lexicon must be automatically separable:Some formal process must be able to separate words.
- Certain word forms belong to certain grammatical classes.
- There exists a formula (well known to the readers of this publication) which defines the recognition factor (the "score") of the proposed word.
Unfortunately, this formula has never been tested. It seems strange to me that we have spent 34 years working on a set of languages without ever attempting to test one of our basic assumptions.
Purposes of the word-building rules
There are several major purposes of the word-building rules:
- Understandability: The word-building rules are claimed to lead to the highest possible recognition factor, meaning that for the world's population, the greatest number of people will be able to recognize and understand the formulated word.
- The Sapir-Whorf hypothesis: The word-building rules take into account words naturally built by a variety of languages and cultures, and intermix them,supposedly building a culture-free lexicon.
Problems with the word-building rules
First, let me take on problems with the Sapir-Whorf idea behind the word-building rules. I claim that there is no such thing as a cultural connection to a language's lexiconconstruction. A word is a collection of symbols, elicitingan ordered set of sounds, which in turn will or will not be recognized. Once the word is recognized, cultural biases of the word or predication may be introduced by the listener, but not until then. Thus, any purpose to word construction other than a means to recognize the given word down not apply.
Second, if the Sapir-Whorf connection did apply, the words should be built evenly across all languages, without regard to the populations speaking those languages. Perhaps, if this were actually a factor, the language should be built evenly from Hopi, English, Chinese, and Swahili. I'll not belabor this point.
If, on the other hand, we are not interested in culture- free juxtapositions of letters taken from the English alphabet, then we must only consider the recognizability of words. This leads me to the number of languages which are used to build the lexicon. It seems to me that there are only three ways to go when deciding on the number of languages to be included in the word-building formula:
- Use every language spoken in the world: This would include little-spoken languages, such as Cornish, and languages spoken only by people who know other languages, such as Ladino. The logic here is that if we are truly to have an accurate mixture of languages, let's not stop at an arbitrary limit. If we should include all of the languages, we should include all of them.
- Use the top seven languages in the world: Fine, some say that a limit must be set somewhere. But why not six? And why not eight?
- Use only one language: While this may seem a radical proposal, this is actually a pair of radical proposals: One method is to use only one language and to make everyone else learn that language. Since virtually every Loglanist and Lojbanist in the world speaks English, this would give a good place to start. The other option is to recognize that since words are just tokens, they may be mindlessly translated by a machine. Thus, it is practical to have an English-based Loglan/Lojban vocabulary, and a Spanish one, and a Chinese one, and even a Latvian and a Sing Language version.
Of all these methods, I personally favor the idea of having a number of single-language-based lexicons. It would make the language much easier for all of us to learn. It would also make the language learnable by someone who is not an aficionado; the common folk might even see a properly constructed lexicon as a dialect, with a "funny grammar". If I write a letter using such a lexicon, and the intended recipient of the letter speaks only Japanese-based Loglan/Lojban, I could push a button on my word processor, and in a fraction of a second, a simple dictionary look-up gives us a translation, fully correct, as the grammars are identical.
There are more completely untested axioms: Do we use, as a weighting factor, the percentage of the world's population which speaks each language as their main language? Do we use the percentage which speak the language, multiplied by the grade-level of use and under- standing, divided by some constant? Do we use the percent- age of the world's population who speak the language as their first of second language? Do we use the percentage of the population which speak the language to any extent at all?
There are questions which I am not prepared to answer. My point in bringing them up is to point out that nobody knows the answers to there questions. Nobody has ever tested the recognition factors or anything like it to see what the weighting factors should be.
There are some factors not previously considered in determining the recognition factor of a proposed word:
- I believe, but have not tested this, that only contiguous groups of letters should count in building the recognition factor. For instance, in the Loglan word MRENU, we are told that we have a 100% match (for English) on MEN. However, I do not believe that anyone's mind joins the M to the EN in MRENU. I believe that the R forces us to stop, and that only the EN component of the word can be used in the calculation. Thus, using the rule that only contiguous letters count, the English factor for MRENU is limited to 67%.
- I believe, but have not tested this, that we should not be counting or comparing the words letter-by-letter using Loglan or Lojban letters. Rather, we should use the object language's phoneme set to count letters. Thus, Loglan's MADJI should not be compared to English's MAGI as MADGAI (yielding 83%). Rather, Loglan's MADJI should be rendered to object language to be sued, a comparison alphabet must first be devised. There exists an alphabet, called the International Phonemic Alphabet, commonly known as IPA, ideally suited to this purpose, for most languages.
- If there exists a bigger word in the object language, then a smaller word cannot be used, as the larger word provides a bigger "handle" for cognition. For example, in the Loglan word GOTSO, we get a 100% match on GO. We also get a 100% match on SEW, GOAT, and GOATS. Thus, we can keep the 100% match score for GOATS, but any attempt to use this word for GO or SOWs SHOULD RESULT IN AN English component score of 0%.
- If there exist more words in the object language the same size as the subject word, then the subject word score should be divided by the number of such words, as there is equal likelihood that the wrong word will be recognized as there is the likelihood that the correct word will be recognized. For instance, in the Loglan word MRENU, we consider the letters EN (other letters could be considered, but this is a limited example). EN appears as 67% of DEN, HEN, KEN (the verb), MEN, PEN, TEN, WHEN, YEN, ZEN, and END. Thus, there are ten English matching words, not counting proper names, so the 67% score is divided by this ten, yielding a 6.7% score.
- If, in the object language, there exists a word in the language with a higher score than the subject word, then that subject word should receive a zero score, as it is not the most likely word to be recognized. Thus, in the Loglan word MRENU, EN has a raw (undivided) score of 67% from MEN, but REN gets an undivided score of 75% from RENT. Once we start dividing, we get 25% (75% by RENT, WREN, and REND), as opposed to 6.7%. Thus the score for MEN in MRENU is zero, as it will be overpowered by other words.
- There should possibly be a factor for commonness of use. For instance, a word based on UNDERSTANDING might be more likely recognized than a word based on KEN, and a word based on SAID might be more likely recognized than a word based on QUOTH.
I may be correct in my statements and suggestions listed above, and I may be wrong in my suggestions for additional rules. But when anyone says that I am wrong, I will politely ask them to prove it. But if anyone tells me that I am right, I will shout "Prove it!"
Bob's Summary of the gismu Algorithm
Contrary to Sheldon's note, I suspect most readers have joined the project without having been told of the details of the word-making algorithm. For those unfamiliar with the algorithm, here is a brief description:
JCB used 1950 language population estimates to select the 8 most commonly spoken languages. (He claims to have investigated possible use of Arabic, then in 9th place, and found it to have no positive effect; he has never published supporting data, nor does he indicate what standards he used to decide on excluding it.) For each proposed gismu to be constructed (which he called C-Prims), he looked up one or more word for the concept in each language.
Possible constructed words were then evaluated using a scoring algorithm for recognizability weighted among the source languages. (Language population weights were based on first language speakers plus 1/2 of second language speakers. The number of speakers of various languages is too uncertain to use any more refined algorithm.) The words were expressed in 'Loglan phonemes' similar to the process described for Lojbanization of names in lesson 1 of our textbook. A score of 2 was given for two adjacent letters in a possible constructed word matching two adjacent letter phonemes in the Loglanized source word, or for the two same letters separated by a letter (possibly different) in both source and possible word. A score of 3, 4, or 5, was given for the same letters appearing in both words in the same order, regardless of whether they appeared together or in matching patterns.
Scores were then divided by the length of the source word, and weighted by the fractional population for the source language. Thus blanu (the word for 'blue' in all versions of the language, including Lojban) gets 3/3 x 28% for 'Loglanized' English "blu" and 3/3 x 25% for 'Loglanized' Chinese "lan". Chinese tones were ignored in scoring.
JCB claims to have investigated the words in depth, and to have used speakers of the source languages for reference. The results recorded in the 1976 dictionary reveal weak language scholarship often involving use of the wrong word, the use of several different Romanization techniques for Chinese that are contradictory, and the use of questionable synonyms which heightened the scores for English and Spanish. (It is natural to be able to think of more cue words for a concept when you know the language better; this happened to us with Lojban, as well.) The words selected were often not the highest possible scores, as he found when he 'remade' several of them around 1980 in order to assign unique rafsi.
The Lojban gismu were made using the same algorithm, but using more recent linguistic population data. We used recent dictionaries, but make no claims for our scholarship in Hindi, Arabic, and Chinese, although we did spot checks with speakers of these languages where possible. We used only 6 languages for four reasons:
- there was ambiguity in the order for the next few languages;
- we could not get a good dictionary for Bengali, which was a plausible 7th or 8th language;
- each language added increased research and computer processing time;
- we tried experiments in which we found that using a language from another language family (like Malay-Indonesian, another candidate language) tended to muddy the results, giving many candidate words with nearly identical scores even though there was little similarity among them.
We also experimented with variations in the scoring algorithm. Including Chinese tones merely reduced the Chinese weights by a predictable fraction, since every Chinese word has tones that are lost in Lojban. This also resulted in many tie scores, since that reduced weight nearly equalized Chinese and English scores. We have retained all of our input data and results for later analysis and reference.
It was noted that the computer usually found at least one possible word form higher than the typical human guesses for the best choice, thus causing greater suspicion of the results of JCB's manual techniques.
In our final word choices, we often took choices that were not the highest scores in order to increase coverage of rafsi assignments.
In spite of the frequent ties between unrelated words, the concept of randomly assigned words, chosen solely for rafsi assignment purposes, was discussed by the community (at a LogFest) and rejected by consensus. As part of the gismu baseline decision last year, however, any future words will have a lower priority placed on recognition score than on rafsi assignment needs. It was also decided that words would not be remade based on questions of the suitability of the algorithm that we used.
The description of the algorithm answers some of Sheldon's questions, although perhaps not to his satisfaction. I'll look now at his reasoning and proposals.
Bob's Response to Sheldon's Critique
First of all, I agree with the concept behind Sheldon's basic argument: it is true that neither JCB nor the Lojban gismu list workers did sufficient scientific research as to the validity of the methodology used to build gismu.
An example: JCB has never given a real rationale for the claim that his algorithm gives any meaningful statistic about recognizability. I suspect that he made an educated guess based on his own experiences. Sheldon's counterexamples (for example with 'GOTSO') show that that guess may not have been perfect.
I have problems with most of the detailed ideas Sheldon presents. He leaves out a lot of important information about his counterexamples, information that I believe renders them invalid. I must start by questioning his assumptions.
Assumptions About Word-Making
The first two assumptions given in Sheldon's background section are fine, if somewhat oversimplified. In addition to what he has stated, the lexicon must be audiovisually isomorphic - the spoken language must uniquely reflect the written language and vice-versa.
The third assumption, which Sheldon questions, is not nearly as basic to the language as the others. Both JCB and la lojbangirz. have made some gismu without using the 'recognition factor' algorithm where we felt it appropriate - the formula was a means to an end, not an end in itself.
The desired goal is a lexicon which is reasonably independent of any single natural language, while still easy to learn. If the lexicon were tied to a single language, the semantics of that source language would heavily color Loglan/Lojban semantics. If Sapir-Whorf is true to any extent, linguistics experiments using the language would be expected to be affected by whether an experimental subject came from the same culture or from a different culture as the single source language. This is one of the principal arguments against an 'Anglan', as proposed and discussed in JL6 and JL7. See those issues for more on the several arguments against a single language source for the lexicon.
Sheldon claims that there is no connection between culture and the construction of a language lexicon. There is counter-evidence to this claim. Some languages like Finnish and Turkish are agglutinative, they build words out of root pieces, just as Lojban does. English is less so, but uses a few suffixes and prefixes. Other languages are highly inflected - the form of the word varies depending on its grammatical position. It can safely be argued that, if there is a tie between the structure of a language and culture (per Sapir-Whorf), that these varying approaches to a lexicon could affect and be affected by culture. If we do not assume that Sapir-Whorf might be true, we cannot use Lojban to test it.
Arabic is an extreme case. The roots of Arabic are reflected entirely in consonants - vowels are inserted and varied to cause small changes in grammatical usage or semantic meaning. In addition, the consonants can vary some as well, especially in whether they are voiced or unvoiced. This caused a lot of problem with the Lojban algorithm, which does not recognize a difference between vowels and consonants for scoring purposes, and considers voiced/unvoiced consonant pairs to be as different from each other as from other consonants.
The word-making algorithm ignores all of historical linguistics, which documents certain sound shifts as being especially common or likely in the development of a language's lexicon. These sound shifts tell us that some consonants are more alike than others, and that vowel differences in words are relatively unimportant. Learning these Indo-European sound shifts is especially useful to an English speaker studying other European languages, since cognates can be recognized that otherwise appear or sound very dis-similar to the native language word. Sound shifts have other important roles. In English, changing a word from a verb to a noun or an adjective involves adding a suffix and often performing a sound shift. Compare 'act' (/ekt/ in Lojbanized form) and 'action' (/ek-can/); the 't' becomes silent, and may not be as important to recognition as the /ek/ root. Russian, with its extensive set of declensions, changes stress and final consonants in many of them. The stress changes usually cause a Lojbanically significant vowel shift. We had no way to take this into account in the algorithm, and used the root form, trying to drop the shifting trailer wherever possible. It seems likely that there would have been better ways to reflect the sound shifts in estimating recognizability.
Other factors not accounted for by the algorithm include the appearance of non-Lojban phonemes in the source languages, such as tones in Chinese. A related factor stems from the fact that many Russian root words are longer than Lojban's 5-letter gismu, while Chinese roots are shorter. Chinese roots that end in a consonant almost always ends in a sound that Lojbanizes as 'n'. These factors all skew lexicons, and perhaps cultures. They definitely lead to those languages not being equally represented in Lojban according to their population weights or their recognition scores.
Sheldon argues that cultural neutrality would dictate that words be built evenly across all languages regardless of populations of speakers. Possibly true, but this leads to an unworkable algorithm, since for any given word we have only 3 consonants and 2 vowels to play with. Equal weights on languages would lead to tie scores on several possible combinations, giving no basis to choose words. (We tried it - in fact equal weights on Chinese and English alone caused this problem.) A 'better' solution to cultural neutrality would be random words, chosen to optimize the rafsi selections, but this leads to no apparent benefits in learning the words.
I think the algorithm we used, or something similar to that algorithm, maximizes cognate value, yet leads to most words being different from comparable words in each source language, minimizing semantic transference. This is a delicate trade-off, that give more learnability than random words, while promoting cultural neutrality and a separate identity as a language.
The weighting based on language population is only one possible way to get needed non-equal weights for non-tie scores. It gives at least the appearance of an attempt to not make the language too relatively easy for people from one particular culture (especially the English-speaking one), offering a little bit of a political advantage in winning interest from other cultures. The emphasis is placed on getting broad input from multiple source languages, giving speakers of each of them a few hooks to learn by.
Beyond the question of trading off cultural neutrality vs. learnability, there is very little reason to choose one algorithm over another. Lojban used the same algorithm as JCB did because our goal at the time was to remake the words with a minimum time and effort without introducing new questionable procedures, especially ones that would make it impossible to obtain reconciliation with JCB at some later time.
It is vital to see that the actual words used are relatively unimportant to Lojban's acceptability; this is fundamental to our argument that Lojban is essentially the same language as JCB's earlier versions. The major requirements are that they be a reasonably complete and ra- tional set of words, that they be compatible with the morphology, and that they have a viable set of rafsi. If they are easy to learn, that is a benefit. JCB's emphasis on the word-making algorithm in the 1960 Scientific American article exaggerates its true importance; there was really very little else that JCB could have talked about then, since the language was little more than a vocabulary list and a grammar concept when the article was written.
Problems With Sheldon's Counterexamples
Sheldon uses some 1976 gismu in his discussion of possible alternative rules. In a telephone conversation, he indicated that he had done an informal test of some of these rules. He doesn't clearly report this in his article, but I suspect that the test was invalid, anyway.
The reason I feel this way is the misconception embedded in his new rule #3. He claims that 'gotso' is more recognized as 'goats' than as 'go', and that to use it for any shorter word than 'goats' should rate a 0 score. This would be true only for someone who didn't know the basis for word-making. First of all, Loglan/Lojban would never have a gismu for 'goats', since we do not have separate roots for plurals. Second, we know that short English words are most likely to be found among the gismu concepts (by Zipf's law, which says that the length of a word is inversely proportional to its frequency of occurrence). Therefore, someone seriously learning the language and attempting to guess the word from pure cognate recognition would try the shortest common word that is recognizable from the gismu, not the longest.
I challenge Sheldon to come up with any valid gismu-form 5-letter word based on 'understanding' that is recognizable, much less more recognizable than one based on 'ken' (as he suggests in rule 6. It turns out that the Lojban for that concept, "jimpe", derives from 'comprehend' (Lojbanized as /kamprixend/ getting a score of 3/10). I will be the first to admit that I would never see this word in "jimpe" without being told it. But given that informa- tion, I may remember the 'mp' cluster in the middle of the word - letters I might otherwise forget. To give "jimpe" a score of 0 is underweighting; 3/10 of 18% (5.4) is a quite low score. It is true that "jimpe" more strongly suggests "jump" and "imp", but the tie to "comprehend" is real and worth a little bit, if not much.
This brings out my major point; the word recognition scores are supposed to reflect an ease in devising cognate aids to both your recognition and recall. Both directions are important; Sheldon concentrates on recognition problems, even though recall of the Lojban given the English is the harder skill to learn. While he may not see 'go' in 'gotso', surely given the word 'go', he will more likely think of 'gotso' than 'mrenu' as being the word to remember.
In practice, people develop their own memory hooks for each word that they try to learn. If the word has a cognate value, they have some clue to recall it by. If there is none, they have to devise some non-English cognate basis. So if they know one or more of the others of our source languages, they may get some cognate value from the words for the concept from those languages. But even if a word has a score of 0 (many Lojban words have a 0 English score), this doesn't mean that the word is difficult to learn. The Lojban word for "child" is "verba", obviously with a 0 score. After having had a seven-year old around during most of June, I have NO TROUBLE AT ALL remembering that one, either by the cognate "verb" for the activity level, or "verbal" for the noise level.
So 'recognition scores' are NOT expected to be an accurate measure of learnability for the word. We would expect there to be some correlation between 'recognition scores' and the rate of learning. Sheldon is right in saying that this particular algorithm has never been verified to correlate with learnability; he is almost certainly correct in saying that there are probably algorithms that would correlate better.
Sheldon proposes 6 rules that he would like to see incorporated in the word-making algorithm. He doesn't actually write a resulting algorithm, so it is hard to see how these rules would be incorporated. His rule #4 would be unworkable in Chinese - almost every word has multiple meanings determined by tone, and every partial word can be found in an even larger number of words. However, I can't say that his approach is wrong, because I haven't tested it.
Since the words are made, we could say that the point is moot and who cares whether recognition scores mean anything. We wouldn't redo the words if a negative correlation was found; this would be unfair to the people who have started to learn the language. Stability is more important than an idealized algorithm.
However, since Lojban IS first and foremost a scientific experiment, I agree with Sheldon that the algorithm should be tested for validity, if only to find out the truth about our assumptions on learnability as an exercise in scientific integrity. In any case, Lojban will be criticized and questioned for years to come by newcomers, and it would be useful to have an answer. We also might learn some meaningful truths about how people learn foreign languages, truths that would not be limited to Lojban learning. In this way we might make a contribution to education independent of any value Lojban has in itself as a language tool.
It turns out that a basic version of such a test is not at all difficult to conduct, and can be performed at any time. We simply instrument LogFlash to gather statistics on individual words, recording how long is spent learning both to recognize and to recall them. For any individual user, this data should correlate statistically with English (or whatever the person's native language is) recognition scores. So we could get meaningful data from a single user, and reduce dependency on other variables by taking data from several users.
Instrumenting LogFlash is trivial; we could do it in a day or two. We would need volunteers who could commit to consistently using the program. If the person skips a couple of days now and then, the scores for the words being studied will be skewed; we know from user experience that irregular habits are far more significant than other factors in learning rate. But we can even record lesson dates and times to allow filtering of invalid data, though I'm not sure what threshold for filtering would be appropriate.
We have our words, and their recognition scores using our current algorithm. Sheldon, or any one else, could generate recognition scores for our words using his or any other algorithm. Some scores using these algorithms on our words might be low, and some might be high; it doesn't matter. We simply look for a correlation between scores and actual learning rates. If an algorithm has a greater correlation with learning than ours, it suggests that that algorithm would have been better used instead of ours to make the words.
The experiment could be extended to make words using a new algorithm, but I think it would be invalid to do so. Lojban learners have motivation to learn Lojban words, so that they can use the language. There is no comparable motivation to learn another set of words that isn't being used; it would also be unfair to ask someone to spend time to learn a bunch of gismu that aren't going to be used.
But even the simpler test will meet Sheldon's standard, and prove or disprove the usefulness of our algorithm.
If people want to seriously volunteer for this experiment, and are willing to make the commitment to science to regularly use LogFlash, Nora and I will make an instrumented program. We'll even give people the program for free conditional on their following through and returning a data disk. We do need to know what your education level is, how many years of what languages you have studied, and at what grade level; this type of thing is needed to provide the scientists with information on other variables; scientists also may want other basic statistical data like age, sex, field of employment, etc. if we get data from enough people to make this useful.
We'll provide raw data (and possibly reduced data as well) to anyone who wants to research and test their own algorithm, including Sheldon and JCB (if he decides he wants to conduct his own tests).
Hopefully, Sheldon, you will be first on the list of LogFlash volunteers! Then you can prove it to yourself!
A Short Note on Another Experiment
By the way, there is already one other experiment that has been proposed and agreed upon, though unrelated to Sheldon's paper. pc has identified a standard test which measures 'logical thinking'. He has proposed it be given to a set of Lojban students before they start studying the language, and then again after they have completed a class (or become fluent). This could be a first data point in identifying a means of measuring possible 'Whorfian effects' of Lojban. pc expects to provide me with copies of the test when he gets back to St. Louis around Labor Day, and we will definitely be using it on any following local classes, while encouraging any people starting to learn the language elsewhere to help pc's research effort. We will announce details and criteria later, but it seems both interesting and relevant to note that the application of Lojban to scientific questions has begun.
Special Section - Mathematics and Lojban
The possible usage of Lojban for machine translation discussed in the last issue has particular application in the area of scientific and technical writing. Dr. Robert Strichartz of Cornell University wrote the following article describing what he perceived as the requirements for a 'logical' language oriented towards mathematical communication. Several readers pointed the article out to us, and we received permission from the editor of "The Mathematical Intelligencer" to reprint it. We also include our submitted response to "The Mathematical Intelligencer". We have been told that a shortened form of this response will be printed later this year.
Since the subject is mathematical expression, an area of Lojban's design that has been discussed very little, I am taking the opportunity to give an overview of our perceived requirements and the resulting design that is currently in the language.
JCB has made several attempts to deal with the problem he labelled as 'MEX' (for mathematical expressions). He has indicated that he feels the problem to be separate from that of completing the language, and something that he does not plan to address further (and does not address in Loglan 1, according to his Table of Contents). I and others within la lojbangirz. feel that the language is incomplete without some such design, even if it is less than perfect. I believe that the design described here addresses the requirements of international mathematical expression, while staying within the framework of the rest of the language.
I've had a thought-out concept for MEX ever since the LogFest in June 1987, but have never before tried to set it down in words. (The resulting design, of course, has been embedded in cmavo and an unambiguous grammar for several months.) This issue seems like a good place to open this concept to the review and scrutiny of others.
The presentation given is only an overview. Some of the design exists only as rules in the grammar and pre-assigned cmavo; these will be fleshed out when I write the textbook sections on the subject.
An International Language for Mathematics
An International Language for Mathematics
by Robert Strichartz
reprinted with permission from "The Mathematical Intelligencer"
An International Language for Mathematics
Robert S. Strichartz
Reprinted with permission from The Mathematical Intelligencer, Volume 11, No. 1. Copyright 1989 Springer-Verlag New York
Mathematics has long been an international discipline. Euler, a German-speaking Swiss, spent much of his career in Russia, but by writing primarily in Latin he could be assured that all his contemporaries could read his works. It has been a long time since Latin was the universal language of scholarly writing. Some would say - and many more would hope - that English has taken its place. If so, this has been a short-term phenomenon, a result of successful American cultural imperialism; we certainly cannot rely on it continuing indefinitely. And there are still many exceptions to the rule that all important mathematical writing is in English (or rapidly translated into English). The French persist in writing in their language, and the Russians produce a body of Russian language mathematics of large volume and high quality. The Germans, Italians, Hispanics, and Japanese have by and large capitulated; we do not know what the Chinese will do as they become a mathematical superpower. Therefore, to accept English as the international language of mathematics strikes me as short-sighted, presumptuous, and dangerous.
Given that mathematicians will be writing in a number of different languages, what are the prospects for rapid and intelligible translations? Recently, the mathematics library at Cornell suggested the possibility of dropping its subscriptions to some translated Russian journals in order to save money. The reasoning is that these journals are not frequently used, and we already receive the Russian-language originals, so our collection would remain complete. Alas, only the Russian-born among us can read them. The reason we do not drop the subscriptions to the originals is that their cost is negligible in comparison to the cost of the translations. The translation process is a time-consuming and difficult task, and the results are sometimes unsatisfactory when the translators do not understand the texts as mathematics. To translate an entire journal cover to cover certainly seems inefficient, when only a few of the translations will actually be read at a particular library. On the other hand, the cost of translating articles on demand may also be prohibitively expensive.
The computer would seem the natural solution to our translating problem. Of course, attempts so far to have automatic computer translation of general texts have not been successful. I would guess that some modification of G”del's Incompleteness Theorem would show that automatic translation, if not impossible, is at least an extremely difficult task. However, such problems would not necessarily arise for translation of texts that were designed in advance for easy translation. Mathematical texts use a limited vocabulary in a very stylized way, and thus suitable prepared mathematical text would seem the first plausible candidates for an automatic translation system based on an international language that would serve as an intermediary between different natural languages. Say we call this language Intermath; a mathematician writing in English would have her text automatically translated into Intermath. No one would read the text in Intermath, but a Chinese mathematician could have the Intermath text automatically translated into Chinese. To ensure the correctness of the translation, the author would simply have the Intermath translation of the text automatically translated back into English (presumably the computer would not be allowed to cheat by peeking at the original). If the sense of the text is preserved, then the translation into Intermath is correct; if not, the author will have to correct the errors.
What would be involved in setting up such a system? We already have a simple model of what a mathematical language might be in the formal languages of mathematical logic. In principle these languages are capable of expressing any mathematical argument, although in practice such expressions would be overly complicated. This is not surprising if we remember that these formal languages were created for the purpose of studying the algebra of mathematical proof, so a premium was put on limiting the number of symbols and the different ways they can be combined. Intermath, as a formal language, would be created to embody all the different ways of expression used by mathematicians in their normal modes of writing. The inventors of Intermath will have to examine actual mathematical texts to see how language is used, and then capture these uses in Intermath. A medium-sized vocabulary of technical terms will be needed; they will have to be updated from time to time to include newly coined terms.
An automatic translation program back and forth between all the natural languages and Intermath will have to be written, along with a program for modifying Intermath text by a user who does not know Intermath. The underlying as- sumption is that every sentence of a mathematical text has an unambiguous meaning, and so can be translated into an Intermath sentence. Undoubtedly there are many ways of realizing the meaning of a mathematical sentence in a natural language, so an Intermath translated text may lack the spontaneity and stylistic elegance of a well-written text. Also, a portion of the text, especially in the introduction and in motivating remarks, may be beyond the reach of the Intermath language. But on balance, aren't we better off having a translation that is accurate and intelligible, even if it is stylistically tedious and missing a few paragraphs?
How much would mathematical writers have to change their practices in order to produce a text that would be suitable for automatic translation into Intermath? This is probably the key issue that will determine whether an Intermath system can be used. If we cannot write, "there are an infinite number of primes," but are forced to substitute, "the cardinality of the set of prime numbers in infinite," I am afraid we will all curse the system and do what we can to sabotage its adoption. On the other hand, if all we are required to do is answer a few queries about ambiguous sentences, I don't think very many of us will object. Probably we cannot know how such a system would work until one is constructed and tested. But even in the worst case, if such a system proved impractical, I still think we would learn a lot about the way mathematics is written.
Who would construct the system, and who would finance it? I think the project is challenging enough that talented mathematicians, computer scientists, and linguists would work on it, provided it was funded. Therefore I appeal to the professional mathematical societies to consider organizing and funding such a project.
The benefits of such a project would extend beyond the mathematical community. No doubt other sciences and technical disciplines might imitate the idea successfully. Also, by strengthening the tradition of international co- operation in mathematics (and other cultural disciplines), the creation of an international mathematical language may help to build some bridges between the peoples of this planet. Our governments may never be able, on their own, to break out of the quagmire of hostility and warfare that has characterized international relations since time immemorial. But we, the mathematicians of the world, can maintain a system of international relations based on cooperation, friendship and mutual respect that can serve as a model for others to follow.
Our Submitted Response to Dr. Strichartz's Article
9 May 1989
Dr. Sheldon Axler
Editor, The Mathematical Intelligencer
Dept. of Mathematics
Michigan State University, MI 48824
This letter is submitted in response to Robert Strichartz's essay (Vol. 11, No. 1) on "An International Language for Mathematics". In that essay, Dr. Strichartz proposes the development of an artificial language, which he calls 'Intermath', as an interlingua that could be used in computer-aided translation of mathematical text.
Dr. Strichartz identifies some properties that such a language might possess:
- It should be modelled on the formal language of mathematical logic;
- It should extend beyond that formal language "to embody all the different ways of expression used by mathematicians in their normal modes of writing";
- It should have a vocabulary including technical terms which can be readily expanded to include newly coined words;
- It should embody the 'unambiguous meaning' underlying mathematical text, without necessarily preserving stylistic features;
- It should be amenable to machine translation techniques.
I am pleased to report that a language has already been developed which meets the criteria described by Dr. Strichartz.
Lojban (pronounced /LOZH,bahn/) is the latest version of the language first described by Dr. James Cooke Brown in "Loglan", Scientific American, June 1960. Dr. Brown started the Loglan project in 1955 as an attempt to develop a language as a test tool for experiments relating to the Sapir-Whorf hypothesis of linguistics. Dr. Brown's formulation of that hypothesis was that the structure of a language constrains the thought of a culture that uses that language. Thus, as a corollary, a language which embodied a structure that was drastically different from any natural language, and which minimized metaphysical assumptions that might be considered 'cultural biases', might detectably alter the hypothetical constraints.
Dr. Brown modelled his grammar on predicate logic, and specifically included grammatical techniques of clearly and unambiguously expressing logical connectives and other concepts that are not easily expressed in English and other languages. The language was also designed to be audiovisually isomorphic: spelling is completely phonetic, each sound corresponds to exactly one letter, and a stream of speech can be uniquely resolved into individual words. Dr. Brown also attempted to devise a mathematically- provable unambiguous grammar.
The project received minimal funding in the early 1960's and then stagnated. Dr. Brown published a microform description of the language in the mid-60s, then revised it into a printed book that was published in 1975. The early versions did not achieve the desired unambiguity, and did not live up to goals in vocabulary development. Only a couple of people ever learned any of these versions, and the language evolved so rapidly that no one ever mastered the language well enough to test the original goals. The handling of mathematical expressions turned out to be extremely deficient, a major drawback to the intended language audience.
The rise of the computer, and especially the personal computer, changed this. Computational techniques of proving computer languages unambiguous were first applied to Loglan in the mid-70s. By 1982, a core subset of the language had been verified unambiguous using the UNIXtm tool 'YACC' (Yet Another Compiler Compiler). Meanwhile, the word-formation rules were redesigned to expand the capability to add new vocabulary with easily identifiable etymologies.
Lojban is the latest version of this language, developed by The Logical Language Group from Dr. Brown's seminal research. It is the first version of the language to have a completely specified unambiguous grammar, verified by the YACC algorithm. Lojban is still based on predicate logic. The language definition phase is being completed in 1989; systems engineering techniques such as configuration management have arrested the uncontrolled evolution of the language, while still permitting carefully controlled growth into new areas as we learn more about specialized language needs.
Lojban's grammar includes a grammar for mathematical expressions that goes well beyond simple mathematical operations. This grammar includes a metalanguage capability for defining new operators, redefining the precedence grouping of operators (unparenthesized Lojban defaults as left-grouping), and intermingling of 'normal language expression' with mathematical expressions.
The Lojban design allows for standard notations as well as Polish, reverse Polish, and an indefinite variety of other systems - a feature vital for supporting the many varieties of mathematical expression used in specialized disciplines. Lojban also incorporates a scheme for borrowing words from the international scientific vocabulary, as well as an ex- tremely rich combinatorial technique of building words from Lojban roots.
In addition to its capability for mathematical expression, Lojban is designed to be a human language. Studies of language universals have led to the inclusion of stylistic features ranging from the observer-based indicators found in native American languages, a tense-location system that allows for optional treatment of space-time as a single concept, the isolation of concrete concepts from abstractions, and a variety of subjective emotional markers that distinguish the attitude of the speaker from the semantic content of the expression.
Lojban retains audiovisual isomorphism. The grammar is comparable in complexity with the more sophisticated computer languages of today. A core of 1300 root words and about 600 structure words (many of them mathematically-oriented) can be simply combined into an effectively infinite vocabulary. The small vocabulary and simple grammar make Lojban easy to learn.
What does this mean in terms of Dr. Strichartz's proposal?
Lojban is certainly capable of serving as 'Intermath'. A Lojban parser is under development, the first step in a machine translation system. Artificial intelligence experts have started looking at techniques that are especially suited for AI-based translation from natural languages into Lojban. Lojban's syntax resembles that of LISP and PROLOG, which are often used in storing knowledge- base information for AI processing. It is conceivable that the Lojban machine translation system might actually be written in a Lojban computer language. Lojban is probably capable of expressing even some of the stylistic features of a source natural language.
More important, a scientist or mathematician can write originally in Lojban, using its unique stylistic capabilities; this writing would be especially suited to mathematical texts due to the predicate structuring of sentences. Writing directly in Lojban eliminates the most difficult step in machine translation: going from an ambiguous natural language to unambiguous Lojban; the other direction should be much easier.
A mathematician should eventually be able to write text either in his/her native language or in Lojban. Journals would be distributed internationally in Lojban. Cornell, or any other university, would need only to translate from Lojban into English, and then only if the mathematician couldn't read the Lojban; this is much easier (and cheaper) than translating from a large variety of languages.
The ease of learning Lojban should not be understated. The first class is being taught In the Washington DC area during the first half of 1989; a textbook is being written simultaneously and is expected to be published by the time this appears in print. It is believed that the course being developed could allow reading fluency in a semester, and conversational fluency or the ability to write sophisticated texts (such as mathematics) within a year. The actual time to achieve fluency is mostly dependent on the rate of learning the root and structure word vocabulary. Students are already reading and writing fairly natural and comprehensible non-technical text, limited primarily by their mastery of vocabulary, after less than 30 classroom hours. Computer-aided vocabulary instruction tools have been in use for a couple of years and have proven extremely effective. When fully integrated with classroom instruction and standard techniques such as language labs, the learning process may even be accelerated.
Lojban's cultural neutrality and decidedly non-English syntax eliminates any dependence on 'American cultural imperialism'. There is interest in Lojban among the community favoring an 'international language', although this is not the primary aim of Lojban. Lojban certainly has some claim to viability as an international auxiliary language; its ease of learning and cultural neutrality help give Lojban standing in the idealistic internationalist community. On the other hand, Lojban's applicability to machine translation, AI applications, and scientific and mathematical expression, as well as its flexibility for new uses, combine to give Lojban a practical (and possibly commercial) short-term future.
Most surprisingly, almost all of the language development has been performed without normal academic funding. Dr. Brown funded much of the early work out of his own pocket, especially after he left the University of Florida in the early 1960's. Almost all of the work in the last 10 years has been performed by unpaid volunteers, and funded by small contributions and subscriptions to publications. Unfortunately, independence from academic funding has also avoided the necessary academic review process. While several linguists, logicians, and computer scientists have worked together to develop Lojban, no mathematician or scientist has actually put it to the test of usage. For too long, we have been without the critical inputs of the mathematical community which would ensure that our design meets their requirements.
With the language definition complete, the review and testing process by scientists, the development of acceptable application software for translation, and even the original linguistic research that inspired the language all will require funding. To earn that funding, and to earn the respect this long-term effort deserves, it is time for the Loglan/Lojban project to assume a proper role in the academic process and prove its value to the community that we expect will find it most useful.
The Logical Language Group, Inc. is a non-profit organization dedicating to developing Lojban, teaching it, and encouraging its practical use. We have teaching materials available in addition to computer-aided tools. A quarterly journal, "Ju'i Lobypli", discusses the project goals and technical issues in the language, and provides some amount of Lojban text in a variety of styles. Recent issues have dealt with both mathematical expression and ma- chine translation, and reprint copies are available. A short newsletter, "lojbo karni", is also published to inform interested but less active participants.
We call for mathematicians to support our efforts, and if possible to learn Lojban and to test its applicability to communication within their specialties. We would like to see classes taught in the university environment, and perhaps some graduate research projects that depend on Lojban could be undertaken. (Research activities that involve communicating with native speakers of other languages might find Lojban an especially valuable tool.)
Your support will help determine the ultimate value of Lojban, whether for scientific communication, linguistics research, or any of the other goals that have been proposed. We are especially interested in ideas for research and application approaches that will attract funding and participation from the scientific community. We look forward to your interest, comments, and questions on our efforts.
Bob LeChevalier, President
The Logical Language Group, Inc.
On the Lojban Design for Mathematical Expression (MEX)
Let us start by reiterating Dr. Strichartz's guidelines for an international mathematical language:
- It should be modelled on the formal language of mathematical logic;
- It should extend beyond that formal language "to embody all the different ways of expression used by mathematicians in their normal modes of writing";
- It should have a vocabulary including technical terms which can be readily expanded to include newly coined words;
- It should embody the 'unambiguous meaning' underlying mathematical text, without necessarily preserving stylistic features;
- It should be amenable to machine translation techniques.
These are merely a starting point, but clearly Lojban's handling of MEX must address all of these. Some of these have been discussed previously; we addressed machine translation issues in JL8.
Lojban is, of course, designed as a predicate language and is thus modelled on mathematical logic. We have, incidentally, tried to allow for non-traditional logics, such as 'fuzzy logic'; in general, however, these new logics must be covered under the requirement to support new technical terms and forms, and need not be in the underpinning of the language. There are a myriad of possible new approaches to mathematics and logic; we cannot hypothesize that any particular one of them is more important or more lasting than the others, or this might skew Lojban's long-term usefulness, as well as its claim to being unbiased.
This leads me to emphasize the second guideline listed above, "to embody all the different ways of expression used by mathematicians in their normal modes of writing". Lojban is not, and must not be, an attempt to impose a new system of expression on mathematics, nor to constrain mathematicians to use one system on the grounds that it is 'better' than any other.
Much of the debate about the design of MEX during the history of Lojban and predecessor versions of the language has revolved around this last issue. Proponents of Polish notation, reverse Polish notation, and other varieties of expression have argued for their preferred notation as being superior to others, inferring therefore that Lojban MEX should require mathematics to be expressed in that form.
There are several flaws with this thinking.
First, a notation that is good for one form of mathematics may be difficult to use with other forms. For example, logicians often find advantage in using Polish notation for certain types of logical analysis. On the other hand, computer mathematics is often organized around reverse Polish notation.
In a similar vein, the existence of two forms of calculator mathematics (standard and reverse Polish), each with its own proponents, is a simple example demonstrating that the superiority of one form of expression over another is arguable even within one area of mathematics.
Lojban is an attempt to structure language to allow for unambiguous logical expression. Regardless of which of the forms of expression is used, the international language of mathematical expression is already both unambiguous and logical. It is unambiguous in that any mathematician, anywhere in the world, can read any written mathematical expression and will 'parse' that expression the same way. Mathematical language is logical in that it can be used to make inferences by simple, regular, manipulations.
Note that I qualified the previous point by referring to written mathematical expression. Even in non-Romanized languages, mathematics is written using the standard Romanized nomenclature, and is read and understood in exactly the same way as in other languages. However, where the system breaks down is in the spoken form of these expressions. Because natural languages are neither unambiguous nor designed to express mathematical thought, the spoken forms of mathematics are often both clumsy and ambiguous.
We are attempting to make the Lojban design both unambiguous and able to express mathematics. Furthermore, we desire Lojban expression to be audio-visually isomorphic - we want any written form to be read in one unique way, and any spoken form to have a unique written corresponding form.
It turns out that we can do this with minor allowances. The proposed MEX grammar allows for a unique spoken form for a given written form. It also defines a written form that complies with international mathematical conventions.
Unfortunately, that form is not a unique form because international mathematics does not prescribe unique forms. The best example is the elision of the multiplication operators in '3x + 2y'. Because we must preserve unambiguity in speech, and there are other possible interpretations for the strings '3x' and '2y' than as mathematical multiplication, the spoken form of this expression must not elide the multiplication operators in its spoken form.
In addition, Lojban cannot make assumptions about the precedence of mathematical operations. First of all, Lojban must allow for the addition of new operators as new mathematical forms are developed. Lojban must also permit the specification of nonce operators, such as those used in abstract algebra in discussing the properties of rings, fields, and other systems that are defined by their permitted operations. Obviously, any new operation may have a precedence which is higher, lower, the same as, or in between the precedences of other existing operators. Furthermore, it is permissible in some mathematics to define systems where precedences differ from those commonly attributed to mathematical operations.
For examples, I reference Dr. M. Ecker in The REC Newsletter, Vol. 3, No. 8, Dec 1988. Dr. Ecker reports that of many spreadsheets tested, one processes '- 5 ^ 2' to be '-25', the rest put negation as higher precedence than exponentiation. Many computer-based calculators even ignore precedence between addition and multiplication, evaluating '1 + 2 x 3' as '9'. Regardless as to whether these are 'right' or 'wrong', there exist mathematical processes that differ in regard to precedences. To discuss those differences, you must be able to express them. So Lojban must be able to specify non-standard precedences and to specify the relative precedence of new operations.
Before I proceed with a description of how we do this, I want to summarize some other requirements for Lojban MEX. The subject of MEX requirements was discussed in some detail by Lojbanist Bill Mengarini in The Loglanist, Vol 5, No. 1. I will summarize the list of requirements presented there.
- Lojban must be 'continuously and infinitely extendible, capable of having new constructions added to it as easily as new metaphors can be added to the [Lojban] vocabulary'.
- Lojban must be able to handle data aggregates such as matrices, arrays, records, heterogeneous lists, stacks, queues, trees, etc.
- Lojban must be able to express values in a variety of bases, including octal, hexadecimal, and binary.
- Lojban MEX must be extendible to handle the widely varying formulaic structures of the hundreds of computer languages in wide use, as well as semi-mathematical expressions such as chemical formulae and equations.
As partial solutions to these requirements, the following derived requirements are proposed:
- Lojban must be able to include text from other languages inside quotes, including text that does not fit Lojban phonology, morphology, or orthography.
- Lojban must have an extendible set of special characters and symbols for reading off non-Lojbanic and mathematical text as strings of symbols. We must be able to make up words for symbols which are easy to remember.
- There must be a way to incorporate Lojban predicates into MEX functions and operators. This allows the power of metaphor and borrowing to be applied to MEX.
(I extended this requirement to include the ability to incorporate predicates as numbers or quantifiers - as we do in English when we talk about a 'brace of oxen', 'a herd of sheep', or 'an exultation of doves'. These are folk quantifiers that do not have general mathematical applicability - I don't know what you get when you add two 'exultations' together, or even worse, multiply them. But people have and do use them either as numbers or collectives in expressions, and Lojban should not presume to say that they cannot be used.)
Finally, Bill describes a 'wish list' of features he sees as being needed in the language. Starting with his list, I've expanded and generalized some areas. The following are major elements:
- scalar radices (bases) such as binary, octal, decimal, and hexadecimal, including the capability to specify a default;
- vector radices, such as time, where the base changes as you move from hours to minutes to seconds;
- data-dependent radices, where the base is determined by the data itself, such as dates. Americans can tell just by looking how to interpret the dates 6/89 and 6/18/89. (Other countries sometimes use a different order.);
- Roman numerals and other non-radix number systems;
- scientific notation;
- repeating decimals;
- irrational numbers;
- numbers typically represented by symbols, such as 'e', 'pi', and 'i';
- complex numbers of arbitrary degree;
- rational and mixed numbers;
- a means of specifying the precision of numbers;
- intervals, both discrete and continuous, open and closed;
- an open ended set of characters, capable of dealing with ASCII, and other character sets;
- a means of indicating spacing in strings;
- quantifiers must be able to include strings, and we must be able to do operations on strings;
- Data Types
- a means to indicate restricted data types such as those use in computer languages (Boolean, Integer, Floating Point), including user-defined data types;
- a means to discuss the attributes of such data (data declarations);
- Data Aggregates
- sets, including multisets that allow duplicates, fuzzy sets with uncertain membership, and ordered sets;
- multi-dimensional sets, such as arrays and matrices;
- graphs and directional graphs (digraphs), including trees;
- Units of Measurement
- being able to talk about dimensioned numbers and dimensionless numbers;
- simple dimensions (meters) and compound dimensions (furlongs/fortnight);
- a means of talking about the dimensionality of numbers;
- a set of 'primitive' functions or operations (I am not using the mathematical definition of function here; some operations may have more than one answer, such as the 'zeroes' of a polynomial);
- a means of defining an extendible set of new functions/operations;
- functions/operations which can operate on non-numeric operands, including equations;
- functions and operations which return more than one type of result (such as multiple parameter subroutines);
- greater than, less than, equation, similarity, set membership, vector componency are but a few of these;
- we must be able to distinguish between assertions and assignments ('x = 5' and 'set x = 5';
- we must be able to express predications (bridi) about quantifiers, strings, functions, and relations;
- Formulae and Algorithms
- means of expressing sequences of operations, including precedences;
- means of asserting a default precedence;
- means of expressing data dependent variations in the sequence of operations;
- Interfaces to Non-MEX Lojban
- each component of MEX must be discussable as a sumti in a non-MEX bridi;
- each component of MEX must be convertible to one or more appropriate predicates (kunbri or bridi);
- non-MEX constructions must be convertible to appropriate MEX equivalents (such as 'exultations' of doves as quantifiers).
- Experimental forms
- For example, Jerome Frazee has proposed a set of systematical mathematical operators that clearly indicates the relationship between addition, multiplication, and exponentiation, and higher degrees in the same vein. While it is not clear that these operators are useful in most mathematical applications, they must be able to be experimented with either within or without the rest of the MEX system.
Pinning down the bounds of international mathematical language is difficult, and probably impossible; no one knows all fields of mathematics. Perhaps the editors of the best mathematical journals are the best equipped in this area, or perhaps mathematical linguists. No one will ever be certain that we have covered all possibilities. The above list is probably a good starting point, however.
All of these bits and pieces come together as a large order, especially for an unambiguous language. However, the international language of mathematics manages it, and Lojban should be able to.
Let us now describe how the current Lojban MEX design meets these multitudinous requirements. It turns out that there is a fairly simple solution - the same one used by international mathematics. That is to assume a certain default set of types and parameters, and to have a means of specifying deviations from that default, both for the short term, and for the long term.
By long term deviations, I mean that if we are talking about chemical equations, we must label that we are doing so, and then use the defaults of chemical expressions until the default is reset. By short term deviations, I mean such things as marking a single number as being in a non- standard base and assuming all others to be in that base. When you read a mathematical expression, a chemical formula, a computer program, or a graph, you know the conventions that apply. Most important, you handle these conventions more as a semantics problem than one of grammar.
Lojban can handle a variety of systems of notation because of its predicate structure. There is no requirement in Lojban that particular sumti of a predicate be always on the left or the right as long as they are in the right order, or if not, some type of marker indicates that they are not. Lojban does vary slightly from this, in that it requires a marker if all of the sumti are to the right of the kunbri, in order to distinguish this form from an observative. Lojban MEX need not include observatives, and hence is simpler than non-Mex grammar.
In fact, it turns out that a large number of features of non-MEX Lojban grammar have little or no value in MEX, and the resulting grammar for MEX is simpler than the grammar for non-MEX. If you learn the regular Lojban grammar, you already know the basics of MEX.
The major 'problem' with Lojban MEX as a representation of the international mathematical language is the default precedence. Because of the requirement for flexibility and extensibility, Lojban cannot assume a fixed set of prece- dences. This is a good thing, since each precedence level assigned in grammar rules doubles the number of rules required.
Instead, Lojban uses its normal default system of left grouping. It allows parenthesis of a couple of types to attach operands unambiguously to their correct operations, and to specify the order of operations. In spoken Lojban, you would express these parenthesis in order to be unambiguous. In reading off an international mathematics formula in any notation, you must therefore be able to add them in easily while reading, and it turns out that this almost trivial. You also must insert elided operations, such as the missing multiplications in '2x + 3y'.
The Lojban grammar will left-group all constructs without parenthesis. You can specify non-left grouping precedence in three ways. Most clearly, you can add in parentheses to force the verbal grouping to represent your precedence structure. When you do so, of course, your verbal expression will no longer be identical to the formulaic expression.
As a second choice, you can adopt the conventions of international mathematics for whatever notation system you are using, and then use those conventions to override the default grouping as part of the semantic interpretation of the expression. In other words, the Lojban text will have parentheses that indicate how operators and operations are attached. It will also have parentheses relating to operation precedence that can be ignored when analyzing what the expression really means.
The third and most complicated choice is to specify the precedence of each operation with the operation. Lojban provides the capability to attach any MEX quantifier to an operation as a precedence marker. You can then use the procedure discussed for the second choice to analyze the expression semantically.
There are arguments that this approach violates Lojban's unambiguity. I believe those arguments to be invalid unless you are willing to label international mathematical language to be similarly ambiguous.
If you make that concession, then I concede the argument. But Lojban, and Lojbanists, are NOT going to change international mathematics now or in the near future; certainly not to impose a more rigid standard of unambiguity than has served for centuries. Lojban is merely trying to represent that mathematical expression at the level of sophistication that serves adequately at present. We go beyond that by providing explicit means to specify conventions and protocols, and a unarguably unambiguous, if unorthodox, expression form which can use either left-grouping or parenthesization.
So Lojban MEX is sufficiently unambiguous to encompass existing international mathematical expression, while providing extra capability to go beyond that form if necessary or desirable.
I'm not going to go into examples at this point. For one thing, the semantic conventions and precedences haven't been defined. I've merely made sure that there is room to do so. For another, even I don't know the system well enough to demonstrate it fluently, nor do I really consider myself a master of the linguistics of mathematics. Even the textbook will only skim the surface, since many of the more interesting examples of mathematical expression will be unintelligible except to those who are familiar with the particular branch of mathematics involved.
Some day, perhaps, someone will write a general mathematical linguistics text for Lojban, probably as part of implementing the machine translation project proposed by Dr. Strichartz. I may help, but it is out of my area of expertise.
Until then, we will have a system that we know to be unambiguous at the parser level, is known to be flexible and extendible in numerous ways, and which can easily handle the basic mathematical expressions used in everyday speech, while doing so in a manner that is compatible with the international mathematical language.
le lojbo se ciska
Self Descriptions by Lojban Students
le ve skicu be mi
Lesson 5 is the first lesson that asks a student to write more than single sentences; we ask people to write a short self-description paragraph. Some students know more grammar and vocabulary than is covered in the first five lessons, but in general, the following reflect the grammar in the first five paragraphs. Darren Stalder, Sylvia Rutiser, and Paul Francis O'Sullivan are in the DC area class. Jamie Bechtel is self-teaching in Omaha, but wrote his self-description at LogFest. Each lesson thereafter asks the students to do a little writing that exercises the points covered thus far. I'm asking all students receiving lessons to send me their self-descriptions and later writings. We will check and comment on them, and if the students permit, will include some of them in later issues as additional examples or inspiration for the others. Jamie has already send us his Lesson 6 writing, which I believe is the first Lojban science fiction. Maybe next issue.
As usual, translations will appear after the end of this section, so that people who are learning the language can try to read the text without help first. I'm going to be sparing on commentary on these writings; each reflects the author's personal style, which I want to encourage, and each has been checked by several people, and/or by Jeff Taylor's parser, for correctness.
la deron. stalder.
coi .i mi du la deron. stalder. .i mi ba speni la rebekas caivlis. goi ko'a .i .iu ko'a traji leka melbi be mi kei le zmadu roda .i ko'a galfi leko'a cmene lemi cmene lenu speni poi ka lijda cfari ca le la xamast. detri .i la'edi'u ka clani lisri kei gi'e canai se skicu .i mi barda nanmu gi'e ckaji lo barda ke xance je jamfu .i lemi kerfa cu bunre .i lemi kanla noi dasni le lenjo cu crino .i mi xebni le tagji ke jibri cutci .i mi nelci lenu lemi jamfu cu lunbe .i le girzu skami preti danfu cu jibri mi
.i mi nelci la'edi'u .e lei jibri jdini .i lemi karce caxe klama ri'anai leka slabu .i .uu lemi karce cu mutcepilno le karce muvdu jisra .i mi ca cliva do .i .u'i micu pamoi troci le mu'i clani ke lojbo nu cusku .i co'o.
mi du la silvian. .i mi se jibri loinu minji platu finti .i le na jibri nu gasnu po mi du la lojban
ni'o mi cu litru fi le cmalu ke blanu karce noi se jadni loi dertu ri'a lenu mi na lumci
ni'o lo so'imei zdani dinju poi nenri la .aleksendrias. ca zdani mi .i lemi zdani goi ko'a cu se canko le so'imeiki'unai lenu ko'a cu berti .i la'edi'u mukti lenu mi ponse la .afriikan. vailet. poi spati
ni'o mi cu nelci loi cizra gerna
la polpronCIIS. .osuliaVAN.
coi. .i mi du la polpronCIIS. .osuliaVAN. .i mi cmaci prenu .i mi zasni gunka .i le citri citri logji cu pluka mi.i lo su'ore drata ke logji ciste cu pluka mi .i mi pacna lenu fanva le gerna la lojban. la esperanton. .i mi pacnalenu fanva le pagbu be le jegvo cukta le su'ore slabu bangu le lojban. .i mi nanmu gi'e na denmi ke grusi xunre sekerfa
mi du la djeimis. .i lemi kerfa cu kalsa je ke bunre narju .i lemi kanla cu blanu .i mi carmi nelci le nu cinki tadni.i mi rirni le jdaselsku cinki pe zoi le. religioso senonsis .le .i mi xebni leka mintu gasnu .i mi tadni lenu misanji senva .i mi pujeca na snada ri .i mi ckule tadni le prije larcu.
With Apologies to Spock
Row, Row, Row Your Boat in Lojban by Bob LeChevalier
I haven't seen it yet, by I've been told that the latest Star Trek movie has the stars sitting around a campfire singing'Row, Row, Row Your Boat', while Spock attempts to analyze the song logically. I was challenged at LogFest to attempt atranslation, which I did in 10 minutes while the others were arguing about what the song really meant.
.a'i .a'i ga'acpu ledo bloti bainaiku le cmari'e terflecu ci'o leka gleki leka gleki leka gleki .i ki'ubo loi nunji'e cu nunsne
.a'i .a'i ga'acpu ledo bloti /ah,hee(pause)ah,hee gah,HAH,shpu leh,do BLO,tee/ / - / - / - - / - bainaiku le cmari'e terflecu /bai,nai,ku,leh shmah,REE,heh ter,FLEH,shu/ / - - / - - / - ci'o leka gleki leka gleki leka gleki /shee,ho leh,kah GLEH,kee leh,kah GLEH,kee leh,kah GLEH,kee / - / - / - / .i ki'ubo loi nunji'e cu nunsne /(pause)ee kee,hu,bo loi,noon,ZHEE,heh,shu NOON,sneh/ / - / - / -
A Story from Another Culture
by Preston Maxwell (with corrections by Bob and Nora)
Preston has been working on a series of simple stories taken from non-Indo-European cultures, attempting to translate or re-express them in Lojban, partly as teaching examples, partly as inter-cultural education. His translations are stillat an early stage, since he is only up to Lesson 6. But he has done a lot of work, and I want to see it recognized as an example for others. Nora and I took one of his stories and polished up the translation to make it minimally grammatical, and here it is. We hope to have more of Preston's stories in later issues.
so'o pendo puca pinxe lo vanju .i ko'a goi pa le pendo cu djica lenu jarco leko'a pixra cupra ka certu .i semu'ibo ko'a cupra co pixra lo since .i leko'a pendo cu bacru zo .ii .i le pixra cu selcertu .ia .i ko'a pensi cusku lu mi ba xagzadri'a levi pixra gi'e semu'i cupra co pixra lo tuple be le since .i leko'a pendo cu bacru lu .iinai lo'e since cu na ponse lo tuple .i seni'ibo le pixra pujecanai xamgu li'u
la sinderelwud. by Nora LeChevalier
****************************************************** SCENE 1 - At the Home of Cinderelwood ****************************************************** (Father): ju'i doi sindereluyd. ko cazi brulu'i le loldi Hey, Cinderelwood, right-now sweep the floor. (Cinderelwood): .ai doi mamtyspen. Yes O MotherSpouse. (1st son): doi paf. la relber. pu porpi lemi selkei O Father, 2nd-Son broke my toy. (2nd son): mi pu na porpi ri .i la pavber. pu kalsyri'a lemi melbi kerfa I didn't break it. 1st-Son messed-up my beautiful hair. (1st son): .o'o naku mi go'i (anger) I did not! (** 1st son hits 2nd son **) (2nd son): .o'e la pavber. pu darxi mi doi paf. (OUCH) 1st-Son hit me, O Father. (** 2nd son kicks 1st son **) (1st son): .o'e la relber. pu tikpa mi doi paf. (OUCH) 2nd-Son kicked me, O Father. (Father): ju'i doi sindereluyd. .o'o do naku xu mulri'a lenu brulu'i le loldi Hey, Cinderelwood, (anger) aren't you done with the sweeping of the floor? .i ko jbikla Approach! .i ko komnicri'a le kerfa be la relber. Comb the hair of 2nd-Son. .i baku ko cikre le selkei po la pavber. Afterwards, fix the toy of 1st-Son. .i baku ko mulri'a cai lenu brulu'i le loldi Afterwards, finish the sweeping of the floor. (Cinderelwood): .ai doi mamtyspen. Yes O MotherSpouse. (** Aside **) .i .uu lomi ni se bilga cu zenba (sorrow) My duties increase. ****************************************************** INTERLUDE ****************************************************** (** Act out each line in background as it is spoken by narrator. **) (Narrator): ko'a goi la sindereluyd. komnicri'a le kerfa be la relber. 20 He, Cinderelwood combs the hair of 2nd-Son. .i baku ko'a cikre le selkei po la pavber. Then he fixes the toy of 1st-Son. .i baku ko'a [kefcfa|ke'ucfa] lenu brulu'i le loldi Then he again-starts to sweep the floor. ****************************************************** SCENE 2 ****************************************************** (** Fanfare sounds; knock on door heard; Father opens door; Herald enters **). (Herald): coi doi se zdani Hello O occupant(s). .i ca'i le banli nolrai pe la noltaj. mi cpedu lenu ro nanmu po'u vi le gugde ba klama le la noltaj. nunsaldansu By-the-authority-of the great king King, I request that all men in the country come to the King's ball. .i la nolraitix. poi tixnu be fi la noltaj. cu cuxna lera ba'ispe leva zvati KingDaughter, who is daughter of the great king, chooses her fiance from the attendants. .i ti pikta pi'o ro do Here are tickets for all of you. (** Herald hands tickets to Father and leaves **) (1st son): .ui mi ba zvati le banli nunsaldansu (happy) I will attend the great ball. (2nd son): .ui mi ba zvati le nunsalci (happy) I will attend the celebration. (** Preens his "beautiful" hair **) .i mi ba mebrai leva zvati I will be the most beautiful of the-there attendees. (1st son): .o'o naku do go'i .i mi go'i (anger) You will not! I will! (2nd son): .o'o naku do go'i .i mi go'i (anger) You will not! I will! (** 1st son hits 2nd son. 2nd son kicks 1st son **) (2nd son): .o'e la pavber. pu darxi mi doi paf. (OUCH) 1st-Son hit me, O Father. (1st son): .o'e la relber. pu tikpa mi doi paf. (OUCH) 2nd-Son kicked me, O Father. (Father): .e'o ko smaji doi zanbre be fi mi (please) Be quiet O favorable-context sons of mine. .i le remei pe do ba mebrai The two-some of you will be the most beautiful. .i .iacai le nolraiti'u ba cuxna pa do (intense-certainty) The princess will choose one of you. (Cinderelwood): .ue mi mo'u pa'a do pu se cpedu (surprise) I, in addition to you, was requested. .i mi ba klama le nunsalci I will go to the celebration. (Father): .o'o na go'i .i do ba stali ti gi'e gunka (anger) No you won't. You will stay here and work. .i mi mipypu'i ledo pikta I hide your ticket. .i ko mulri'a lenu brulu'i Finish sweeping. (** Hides one ticket in a lamp **) ****************************************************** INTERLUDE ****************************************************** 21 (Narrator): ni'o ca le nunsalci djedi le lanzu cu redri'a pida'ari On the celebration day, the family readies all-but-one-of themselves. ****************************************************** SCENE 3 ****************************************************** (1st son): la relber. pu malcpa lemi degjai 2nd-Son stole my ring. (2nd son): mi na go'i .i la pavber. pu malcpa lemi kerfa dasri I did not. 1st-Son stole my hair ribbon. (1st son): naku mi go'i I did not. (** 1st son hits 2nd son. 2nd son kicks 1st son **) (2nd son): .o'e la pavber. pu darxi mi doi paf. (OUCH) 1st-Son hit me, O Father. (1st son): .o'e la relber. pu tikpa mi doi paf. (OUCH) 2nd-Son kicked me, O Father. (Father): ju'i doi sindereluyd. ko cpacu le drata degjai .e le drata ke kerfa dasri Hey Cinderelwood, get the other ring and the other ribbon. (Cinderelwood): .ai doi mamtyspen. Yes, O MotherSpouse. (** Cinderelwood gets the ring & ribbon. Others finish dressing **) (1st son): mi bredi le nunsalci I'm ready for the celebration. (2nd son): ba'e mi bredi lenu se cuxna la nolraitix. I'm ready to be chosen by KingDaughter. (1st son): naku do ba se cuxna la nolraitix. .i mi ba go'i You won't be chosen by KingDaughter. I will. (2nd son): naku do go'i .i mi go'i You won't be. I will. (** As the three leave the house, 1st son hits 2nd son. 2nd son kicks 1st son **) (2nd son): .o'e la pavber. pu darxi mi doi paf. (OUCH) 1st-Son hit me, O Father. (1st son): .o'e la relber. pu tikpa mi doi paf. (OUCH) 2nd-Son kicked me, O Father. ****************************************************** INTERLUDE ****************************************************** (Narrator): ni'o la sindereluyd. cu na selkansa Cinderelwood is alone. (** Fairy Godfather enters, hidden **) .i .ue .ianairu'e go'i (surprise) (maybe not) That's so! ****************************************************** SCENE 4 ****************************************************** (Cinderelwood): mi na selkansa .i naku .uu mi klama le nunsalci I am alone. It is not the case (sorrow) that I go to the celebration. .i mi bilga lenu gunka I am obligated to work. (Godfather): coi. .elwyd. .i mi ridrirni do Hello, Elwood. I am fairy-godfather to you. (Cinderelwood): .ue 22 (surprise) (Godfather): .ai do ba klama le nunsalci .i mi sidju do (intention) You will go to the celebration. I help you. .i ko zgana Observe. (** Magic motions cause ticket hiding place (lamp) to light up. He gets ticket, and hands it to Cinderelwood **) .i do ca ponse le pikta You now have the ticket. (** More magic motions **) .ije le selgunka ca mulno And, the work is done. (Cinderelwood): .ui .i mi ca kakne lenu klama le nunsalci (happy) I can go to the celebration. .i .y naku .uu mi ponse lo drani taxfu (hesitation) I don't have any appropriate clothes. (Godfather): ko denpa Wait! (** More magic motions. Opens drawer showing fine clothes & holds them up to Cinderelwood. **) .i ko zgana do le minra Observe yourself in the mirror. (Cinderelwood): .ue le taxfu cu mutce melbi (surprise) The clothes are very beautiful. (Godfather): do mo'u pa'a cu mutce melbi .i .o'u ko se kajde You, also are very beautiful. (Warning) Be forewarned! .i do bilga lenu cliva le nunsalci pu la parecac. You must leave the celebration before 12 o'clock. .i ca la parecac. ro ledo melbi taxfu ba binxo ledo mabla ke slabu taxfu At 12, all your beautiful clothes will change into your old and broken clothes. (Cinderelwood): mi ba morji doi ridrirni .i mi mutce ckire do I will remember, O fairy-godfather. I am very grateful to you. (Godfather): .e'o ko klama gi'e lifri le pluka (please) go and experience the pleasant. (Cinderelwood): .aicai (intense-intention). ****************************************************** INTERLUDE ****************************************************** (Narrator): ni'o mi'o bazi zgana le nunsalci vi le nolraizda po la noltaj. We will now observe the celebration at the king-house of King. ****************************************************** SCENE 5 - At the Castle ****************************************************** (Princess): mi xanka denpa lemi ba speni I nervously await my future spouse. (** Father & 2 sons enter **) (Herald): la paf. mamtyspen. noi selkansa la pavber. .e la relber. Father MotherSpouse, who is accompanied by 1st-Son and 2nd-Son. (Princess): (** Points to sons **) .ue ta xajmi dasni (surprise) They are comically dressed. (1st son): (** Flexes muscles to impress princess **) coi nolraiti'u .i mi tsarai Hello Princess. I am the strongest. (2nd son): (** Preens **) coi nolraiti'u .i mi mebrai 23 Hello Princess. I am the most beautiful. (1st son): .o'o naku do go'i .i mi go'i (anger) No you're not. I am. (2nd son): .o'o naku do go'i .i mi go'i (anger) No you're not. I am. (** 1st son hits 2nd son. 2nd son kicks 1st son **) (2nd son): .o'e la pavber. pu darxi mi doi paf. (OUCH) 1st-Son hit me, O Father. (1st son): .o'e la relber. pu tikpa mi doi paf. (OUCH) 2nd-Son kicked me, O Father. (Princess): coi vitke Hello, visitors. (** To Herself **) .i'u ti mabla seicni (disgust) They are self-interested. (1st son): (** grabs princess **) mi dansu co kansa do I dance with you. (Princess): (** To Herself **) .o'e ti dansu co cpana lemi jamfu (OUCH) He dances on my feet. (2nd son): (** grabs princess **) mi dansu co kansa do I dance with you. (Princess): (** To Herself **) .o'i ti dansu co dukse le kazyjbi (annoyance) He dances too close. (** Elwood enters **) (Herald): la miprycmen. SecretName. (Princess): .ue .ui ti drani dasni (surprise) (happy) This-one is appropriately dressed. (Elwood): coi doi nolraiti'u. .i mi manci lesi'o rinsa do Hello, O princess. I am awed by the thought of greeting you. (Princess): coi. miprycmen. i do melbi cusku Hello, SecretName. You beautifully express. (Elwood): .i do melbi .i .e'o xu do curmi lenu do dansu co kansa mi And you are beautiful. (please) Will you allow that you dance with me? (Princess): .ai .ui Certainly. Happily. (** Elwood and the princess dance beautifully **) (Princess): (** To Herself **) .ui ti certu dansu (happy) This-one skillfully dances. (** Clock begins to strike 12 **) (Elwood): .ue ma tcika (surprise) What is-the-time? (Princess): la parecac. Twelve. (Elwood): .o'i .i .uu .ei mi cazi cliva .i co'o doi se manci (annoyance). (sorrow) (obligation) I right-now leave. Good-bye O awesome-one. (** Elwood runs out, catching his belt on the door. Elwood leaves; his belt doesn't. Optionally, his pants fall and he picks them up before he gets out **) (Princess): ju'i. miprycmen. .i do puzi cirko ledo befydai Hey SecretName. You just lost your belt. ko denpa Wait! (** To the Herald **) ko sisku lepu dasni be ti .i mi minde la'edi'u 24 Seek the former wearer of this. I command it. (Herald): .io doi nolraitix. Yes O KingDaughter. (1st son): (2nd son): ju'i. nolraitix. ko catlu mi Hey KingDaughter, look at me! (Princess): .oi (complaint) ****************************************************** INTERLUDE ****************************************************** (Narrator): ni'o caki le bavlamdei vi le zdani be la sindereluyd. The next day, at the home of Cinderelwood. ****************************************************** SCENE 6 - At the Home of Cinderelwood ****************************************************** (** Cinderelwood is sweeping again. Fanfare; a knock; Herald enters **) (Herald): coi doi se zdani Hello, O occupant(s). .i ca'i le banli nolraiti'u pe la nolraitix. mi cpedu lenu ro nanmu po'u vi le gugde ba troci co mapti levi befydai By-the-authority-of the great princess KingDaughter, I request that all men in the country try fitting this belt. .i la nolraitix. sisku lepu dasni be ti KingDaughter seeks the former wearer of this. .i la nolraitix. cu jdice lenu spebi'o rixici KingDaughter decides on marrying the third-forementioned. (1st son): .ai mi troci pamoi .i mi kakne lenu dasni ta I try first. I can wear that. (2nd son): na go'i .i mi go'i No you can't. I can. (** 1st son hits 2nd son. 2nd son kicks 1st son **) (2nd son): .o'e la pavber. pu darxi mi doi paf. (OUCH) 1st-Son hit me, O Father. (1st son): .o'e la relber. pu tikpa mi doi paf. (OUCH) 2nd-Son kicked me, O Father. (Herald): mi ba troci co mapti ro do .i ko jbikla I will try fitting all of you. Approach. (** Have fun with Herald trying to fit belt on too-big 1st son **) (1st son): xu mi mapti le befydai Do I fit the belt? (Herald): na go'i .iacai No (intense-certainty). (** Have fun with Herald trying to fit belt on too-big 2nd son **) (2nd son): .i'acai mi mapti le befydai (intense-belief) I fit the belt. (Herald): na go'i .iacai Not so (intense-certainty). .i ju'i brulu'i ko jbikla ki'u lenu troci co mapti le befydai Hey sweeper, approach to try fitting the belt. (Father): la sindereluyd. na mapti Cinderelwood does not fit. .i ri na pu zvati le nunsalci He did not attend the celebration. .i ko ranji lenu brulu'i doi sindereluyd. 25 Continue to sweep O Cinderelwood. (Herald): la nolraitix. pu minde lenu ba'e ro nanmu cu troci KingDaughter commanded that ALL men try. .i ko jbikla doi brulu'i Approach, O sweeper. (** The Herald tries the belt on Cinderelwood, and it fits **) (Herald): .ue .ui do mapti (surprise) (happy) You fit. (Cinderelwood): .ia .i mi pu dasni le befydai ca le prulamdei Certainly. I wore the belt yesterday. (Herald): ko klama la nolraitix. mu'i lenu spebi'o ri Go to KingDaughter to wed her. (Cinderelwood): .ai .uicai I will (intense-happiness). ****************************************************** INTERLUDE ****************************************************** (** In the background are Elwood and the Princess, obviously happily reunited **) (Narrator): .i la .eluyd. spebi'o la nolraitix. Elwood marries KingDaughter. .i le remei bazize'o gleki ke kansa lifri The two-some very-soon-and-indefinitely-beyond happily together experience. .i le drata mo And the others are doing what? (** Father, 1st son and 2nd son are apart from Elwood and the Princess, looking unhappy **) (** 1st son hits 2nd son. 2nd son kicks 1st son **) (2nd son): .o'e la pavber. pu darxi mi doi paf. (OUCH) 1st-Son hit me, O Father. (1st son): .o'e la relber. pu tikpa mi doi paf. (OUCH) 2nd-Son kicked me, O Father. (Father): .oi (Complaint!) ****************************************************** THE END ******************************************************
lei lojbo by Nora LeChevalier
Translations from le lojbo se ciska
le ve skicu be mi
The Description of Me
la deron. stalder.
coi .i mi du la deron. stalder. .i mi ba speni la rebekas caivlis. goi ko'a .i .iu ko'a traji le ka melbi be mi keile zmadu roda
Greetings! I am the same as the one called 'Darren Stalder'. I will be married to Rebecca Shively (ko'a).(Affection!) She is superlative in the property of beautifulness-to-me in direction the greater than all-somethings.
.i ko'a galfi leko'a cmene lemi cmene lenu speni poi ka lijda cfari ca le la xamast. detri .i la'edi'u ka clani lisrikei gi'e canai se skicu
She modifies her name into my name at the event-of-being-spouse which is (religious initiatedness during the 6th month'sdate). The-referent-of-the previous (the modifying) is a long-story-ness and a not-now described thing.
.i mi barda nanmu gi'e ckaji lo barda ke xance je jamfu .i lemi kerfa cu bunre .i lemi kanla noi dasni le lenjo cucrino
I am a large-man and am characterized by big hands-and-feet. My hair is brown. My eyes, which incidentally wear thelenses, are green.
.i mi xebni le tagji ke jibri cutci .i mi nelci lenu lemi jamfu cu lunbe .i le girzu skami preti danfu cu jibri mi .imi nelci la'edi'u .e lei jibri jdini
I hate the tight job-shoes. I am fond of the state of my feet being bare. The group-computer question answerer(computer networking consultant) be-jobs me. I am fond of the referent of the previous (the be-jobbing) and (as wellas) of the mass of job-money.
.i lemi karce ca xe klama ri'anai leka slabu .i .uu lemi karce cu mutce pilno le karce muvdu jisra .i mi ca cliva do.i .u'i mi cu pamoi troci le mu'i clani ke lojbo nu cusku .i co'o.
My car is presently a mode of transportation despite the oldness. (Regret!) My car is a much-user of the car-movejuice. I now leave you. (Pleasure!) I first try the purposefully (long type-of Lojbanic expressing.) Good-bye!
mi du la silvian. .i mi se jibri loinu minji platu finti .i le na jibri nu gasnu po mi du la lojban
I am the same as the one called 'Sylvia'. I am be-jobbed by Machine-design Invention. The not-job doing-without-purpose which pertains to me is the same as the thing called Lojban. (My hobby is Lojban.)
ni'o mi cu litru fi le cmalu ke blanu karce noi se jadni loi dertu ri'a lenu mi na lumci
(New topic.) I travel using the small blue-car, which incidentally is adorned with dirt because of the state of my not-washing.
ni'o lo so'imei zdani dinju poi nenri la .aleksendrias. ca zdani mi .i lemi zdani goi ko'a cu se canko le so'imeiki'unai lenu ko'a cu berti .i la'edi'u mukti lenu mi ponse la .afriikan. vailet. poi spati
(New topic.) A manysome-nest building which is inside the thing called Alexandria is a nest for me. My nest (ko'a) isbe-window-ed by the many-some despite (justification) the state of them being-to-the-north. The referent of the previous (the be-window-ing) motivates the state of my owning that called 'African violet', which is a plant.
ni'o mi cu nelci loi cizra gerna
(New topic.) I am fond of Bizarre Grammar.
la polpronCIIS. .osuliaVAN.
coi. .i mi du la polpronCIIS. .osuliaVAN. .i mi cmaci prenu .i mi zasni gunka .i le citri citri logji cu pluka mi.i lo su'ore drata ke logji ciste cu pluka mi .i mi pacna lenu fanva le gerna la lojban. la esperanton. .i mi pacnalenu fanva le pagbu be le jegvo cukta le su'ore slabu bangu la lojban. .i mi nanmu gi'e na denmi ke grusi xunre sekerfa
Greetings! I am the same as the one called 'Paul-Francis O'Sullivan' (Irish pronunciation). I am a mathematicalperson. I temporary work. The history-history type of logic pleases me. Some (numbering at least two) other type-oflogic systems please me. I hope for the state of translating the grammar from Lojban to Esperanto. I hope for thestate of translating the page(s) of the Jehovah-book (the Bible) from the (at-least two) old languages into Lojban. Iam a man and not-dense grayish-red be-haired.
mi du la djeimis. .i lemi kerfa cu kalsa je ke bunre narju .i lemi kanla cu blanu .i mi carmi nelci lenu cinki tadni.i mi rirni le jdaselsku cinki pe zoi le. religioso senonsis .le .i mi xebni leka mintu gasnu .i mi tadni lenu misanji senva .i mi pujeca na snada ri .i mi ckule tadni le prije larcu.
I am the same as the one called 'Jamie'. My hair is chaotic-and-brownish-orange. My eyes are blue. I am intensely fond of the event(s) of being an insect-student. I rear the (religious-expressor) praying insect who is "religios osenonsis". I hate the property of same-doing-ness without purpose (conforming). I study the state of my consciously dreaming (being aware and able to control my dream). I was-and-now not-succeeding at this (the state of conscious dreaming). I school-study the wise-art (philosophy).
With Apologies to Spock
by Bob LeChevalier
.a'i .a'i ga'acpu ledo bloti (Effort!) (Effort!) Rod-pusher of your boat
bainaiku le cmari'e terflecu not-forcedly to the small-river place-flowed-to
ci'o leka gleki leka gleki leka gleki with emotion of happiness about happiness about happiness.
.i ki'ubo loi nunji'e cu nunsne Because (justificational) Life is a dreaming.
You will note that the translation is not literal. I tried to capture my sense of the song. I'm proud of my means of acquiring the song's necessary repetition both grammatically and productively. The first two attitudinals emphasize the strokes of rowing, as in the English version. The nested 'happiness's put some meaning to the repeated 'merrily's ofthe English beyond more repetition emphasizing oar strokes. It also ties in with my interpretation of the song.
There has been considerable debate about the last word. I have chosen the interpretation that the song compares life toa (perpetual) state of dreaming, easy and gentle, relaxing and merry, like rowing a boat downstream. If the last wordis "selsne", my original choice, the song is harder to pronounce. It however suggests the interpretation that life is a dream (and presumably not real). The more I thought about it, the less I saw the first three lines as agreeing with this interpretation. But poetry is in the mind of the beholder. (Hmmm! If I want to be consistent, perhaps the first two words should be ".a'inai .a'inai" since I want to stress the ease of rowing downstream. This suits the rhythm better, too. What do you think?)
Now I can go see what Spock thinks it means. I welcome other viewpoints as to what the song means, especially if you translate it. (I won't print them unless they are in Lojban, though.)
A Story from Another Culture
by Preston Maxwell (with corrections by Bob and Nora)
so'o pendo puca pinxe lo vanju
Some friends were-then drinking some wine.
.i ko'a goi pa le pendo cu djica lenu jarco leko'a pixra cupra ka certu
He (one of the friends) desires the state of showing his picture-producing expertness.
.i semu'ibo ko'a cupra co pixra lo since
Therefore, he picture-of-a-snake produces.
.i leko'a pendo cu bacru zo .ii
His friends uttered ".ii (Approval!)"
.i le pixra cu selcertu .ia
The is be-experted-at (expertly done) (Certainly!).
.i ko'a pensi cusku lu mi ba xagzadri'a levi pixragi'e semu'i cupra co pixra lo tuple be le since
He thinkingly-expresses "I will better-cause (improve) this picture and therefore (motivationally) picture-of-some-legs-of-the-snake produces.
.i leko'a pendo cu bacru lu .iinai lo'e since cu na ponse lo tuple .i seni'ibo le pixra pujecanai xamgu li'u
His friends utter ".iinai (Non-approval!) The typical snake not-possesses some legs. Therefore (logical entailment) thepicture was-but-is-not-now good."
lei lojbo by Nora LeChevalier
This issue's strip introduces a new character, Bob and Nora's daughter la katrin. I can't say what her English name is; there are several related forms of this name in English ranging from Catherine to Katrina. I believe the latter is the intent, though, since it is a name Nora and I have considered if we ever have a daughter. It has the unusual advantage of being an acceptable English name for a girl and being both positive and meaningful in Lojban aswell: ka trina translates as "attractiveness". la katrin. is a valid way to turn this kunbri into a name.
The humor of the comic is based on the ambiguity of English when translated into Lojban. It shows just how hard the translation problem is if you are completely open-minded and have no context clues. ALL of the interpretations have Sam the computer assuming that English "after" refers to time sequence, when a correct interpretation requires the semantics of physical sequencing. (Either no one has told Sam about the multiple meanings of "after", or he hasn't gotten there yet by the end of the comic.)
Note that all of the interpretations given differ from each other in Lojban grammar.
katrin.: doi sam. mi na jimpe levi jufra be fi le glibau "O Sam, I not-understand the-here sentence in-language the English-language." "Sam, I don't understand this English sentence." sam.: mi ba sidju pe'u. katrin. "I will help you, I request you, O Katrina." "Can I help you, Katrina?" katrin.: zoi .by. Spot ran after the ball .by. "Spot ran after the ball" (This is presumably read aloud slowly with no particular emphasis, as would happen when a child reads words without understanding.) The "zoi" marks a delimited quote of non-Lojban text, with the delimiter being a single non-compound Lojban word, in this case "by", the letter "b". sam.: ko viska la ran. mo'u ba le bolci "(Imperative You) See Ran after (you see) the ball." "Spot Ran after the ball!" This panel shows Katrina seeing a ball, and then seeing la ran. ("Ron"?, or "Ran" from "Randall") at a later time. "mo'u" attaches "ba le bolci" as a modal modifier to "la ran.", so that you know that the parallelism is between "la ran." and "le bolci". Compare "ko mo'u ba le bolci cu la ran." meaning "See, after the ball does, Ran." sam.: .y sa. "...Ummm... Take that last sentence back." la spat. mo'u ba le bolci pu bajra "That named 'Spat', after the ball, past-run." "Spat ran after the ball." This panel shows the ball running, and then a plant ("spati") named "Spat" from the Lojban word for plant. The grammar uses the alternative parallelism in the last panel, but also plays on possible meanings of the word Lojbanized as "spat". sam.: .y sa. "...Ummm... Take that last sentence back." la spat. bajra ba le nunsaldansu "Spat is-a-runner after the ball (event-of-celebration-dancing)." "Spat ran after the ball." This panel shows a ball, a celebration dance, and then the plant named Spat running at a later time. sam.: .y sa. "...Ummm... Take that last sentence back." le barna mo'u ba le bolci cu rumpe'a "The mark, after the ball does, runs (meltingly-spreads)." "Spot ran after the ball." Playing off both "run" and "spot" this time, with a pictorial reference to modern surrealistic art. sam.: .y sa. "...Ummm... Take that last sentence back." la spat. ran. po'u ba le nunsaldansu "The one named Spat Ran, after the ball." "Spat Ran after the ball!" A person named "Spat Ran", presumably because he is wearing a plant on his suit (Yes, it's hard to see this.) has sore feet (o'e = "Ouch" is the indicator for pain or suffering), after his celebration-dancing. In afterthought, there is another plausible meaning of "nunsaldansu" besides a celebration party with dancing. Specifically, it could refer to the informal, more individual-oriented dance of happiness one might perform in celebration of something. In spite of the usage in this strip as well as in Cinderelwood of "nunsaldansu" for 'ball', note that all such lujvo at this point are still proposals and are not 'set in concrete'. This new interpretation may be better, since "ball" might require some sense of 'party', or a mass of people: "nungumsaldansu". Other tanru are possible. These are the type of questions to be resolved through usage before we write the dictionary. We do not want to enshrine lujvo with 'official' meanings until usage backs them up as the 'best' or 'most useful' or 'clearest' interpretation. None of us can claim that they have the Lojbanic skill or the cultural neutrality to make these decisions alone. At this point, when we are trying to convey stability of the basic language elements like the grammar and gismu, we don't need people thinking that Bob's and Nora's off-the-wall lujvo inventions are to be enshrined. Lojban is YOUR language, not ours. sam.: .y sa. ... "...Ummm... Take that last sentence back. Presumably Sam will follow with several more possible interpretations, but Katrina runs out of the room and doesn't hear them. The panels given have played upon every word except "the", which has at least half-dozen interpretations in Lojban (most of which are difficult to differentiate in a comic strip, but Nora might try that for next time). And eventually, Sam might consider the other meanings of 'after'. katrin.: mi bajra ba lenu se cfipu "I run, after the state of being-confused-by." "I'm leaving, since you've confused me." Katrina runs away, her mind blown by Sam's interpretations, incidentally mirroring another grammatical form that could be used to interpret the original "Spot ran ..." sentence.
SCRABBLEtm for Lojban
After several issues of promising it, I'm finally managing to get this into this one.
JCB first wrote up some rules for playing SCRABBLEtm back in The Loglanist, Volume 2, in the late 70's. No one ever reported playing the game other than JCB. his rules became obsolete around 1980 with GMR, when he remade many words as well as the rules for making lujvo. Certainly, letter frequencies did not stay constant through this change.When we again rebuilt the gismu list for the Lojban version of the language, we also changed these frequencies.
Since we've now baselined the list of gismu, that portion of the frequencies should not change. The cmavo list isn't baselined, but we know that 90% of all possible V, CV, VV, and CVV combinations have a meaning; the holes are nearly random, except for the xVV cmavo, reserved for experimental use. So we can estimate the final letter frequencies with some expectation of validity.
lujvo are harder to deal with, since there isn't even a list of them yet. However, in developing and tuning the rafsi assignments, I had gathered statistical data from several thousand tanru proposed over several years by JCB, the old Word Makers' Council, Eaton project volunteers, contributors to TL, and our own gismu list workers. This data reflects no actual set of words, only word proposals. However, its statistical size suggests that it can't be too far off in representing the eventual letter-frequency distribution for Lojban.
I did these letter frequency estimates right after the gismu list was baselined, and in November, converted them to SCRABBLEtm frequencies. Nora and I also talked about some rule modifications we think are needed to practically play the game with Lojban. I'll report my conclusions on these in a moment. First the letter frequencies, which I give both including and excluding the lujvo data:
Letter non-lujvo frequencies with-lujvo frequencies occurs Lojban/points English/points occurs Lojban/points a 991 12/1 9/1 2949 10/1 e 496 6/1 12/1 1560 5/1 i 1045 12/1 9/1 2678 10/1 o 395 5/2 8/1 1046 4/2 u 642 8/1 4/1 1755 6/1 y 19 0 2/4 5553 8/1 ' 316 4/1 - 1012 4/2 b 212 2/5 2/3 865 2/4 c 360 4/2 2/3 1040 3/3 d 219 3/3 4/2 862 2/4 f 149 2/5 2/4 616 2/4 g 146 2/5 3/2 589 2/4 j 249 3/3 1/8 1008 3/3 k 285 3/3 1/5 1107 3/3 l 348 4/2 4/1 1395 4/2 m 254 3/3 2/3 1048 3/3 n 563 7/1 6/1 2047 7/1 p 203 2/5 2/3 872 2/4 r 460 5/1 6/1 2979 7/1 s 339 4/2 4/1 1363 4/2 t 361 4/2 6/1 1359 4/2 v 119 1/8 2/4 490 1/6 x 108 1/9 1/8 532 1/6 z 87 1/10 1/10 359 1/9 blank 2/- 2/- 2/- h 2/4 q 1/10 w 2/4 Total 100/184 100/187 100/190
I've given separate statistics for non-lujvo words so that you can play without lujvo. The significant occurrence of "y" only in lujvo requires different statistics. Read the rules notes following to decide whether you want to include lujvo or not.
Thus, to turn a standard set of pieces into a Lojban set you need to make the following changes (presumably by marking them in pencil or pen on the English pieces):
Without lujvo: 3 e's changed to i's 3 e's changed to a's 3 o's changed to u's with remaining o's changed to 2 points 1 r changed to n 2 w's changed to ' with score changed to 1 point 2 t's changed to ' with remaining t's changed to 2 points 1 g changed to m with remaining g's changed to 2 points 2 y's changed to c's with 2 points, other c's also changed to 2 points 2 h's changed to j's with 3 points, other j also changed to 3 points 1 q changed to u with 1 point 1 d and 1 v changed to k's with 3 points, other k also changed to 3 points remaining d's changed to 3 points, remaining v changed to 8 points Change scores on all p's and f's to 5, on all l's and s's to 2, all b's to 3, and the x to 9.
With lujvo: 1 e changed to a, 2 e's changed to u, 1 e changed to i, 1 e changed to n, 1 e changed to r 1 e changed to c with value changed to 3 points 2 existing y's changed to 1 point 4 o's and 2 t's changed to y, with remaining o's and t's changed to 2 points 1 q changed to ' with value changed to 2 points 2 d's and 1 g changed to ', remaining d's and g's changed to 4 points 1 v changed to m with value changed to 3 points, remaining v changed to 6 points 2 h's changed to k with value changed to 3 points; other k also changed to 3 points 2 w's changed to j with value changed to 3 points; other j also changed to 3 points Change score on all b's and p's to 4 points, on all l's and s's to 2 points; change x to 6 points and z to 9 points.
Now for rules changes:
Since there is no standard dictionary that can be used as a reference, as in the English version, we have to have away to verify that words that are made are valid. Since this is presumably also a teaching tool, we want to eliminate wild guesses, which is a problem for cmavo. Let's handle each word type separately.
gismu may be verified on the standard list, although optionally the person may be required to give the keyword,place structure and/or to make a sentence that reasonably uses the word as a sumti or as part or all of a kunbri; these are thus relatively easy. Challenges can be easily resolved; mis-spellings are the most common mistakes to watch for.
cmavo should be based either on the latest draft cmavo list or on the textbook - in the latter case, only those cmavo that have been covered up to an agreed upon point in the textbook, as well as all cmavo in the flash card list, should be allowed. In any case, the person using a cmavo MUST express a sentence using that cmavo in a rational way, and if challenged, must explain what the sentence means and how the meaning is affected by that particular cmavo. rafsi are NOT words, and should not be accepted as valid.
Because nearly every possible V, VV, CV, and CVV is a valid cmavo, someone who knows the cmavo can fairly easily go out, possibly without ever drawing. As counter- motivation, all letters in cmavo count with a value of 1, ignoring both the face value AND multiplier squares. You can get rid of any piece, but it isn't worth much. On the other hand, if you make one or more gismu (and/or lujvo) at the same time as the cmavo, then the points count normally when scoring those words.
I recommend that cmavo compounds either be disallowed, or that they be limited to those compounds recognized as valid by the machine grammar lexer (for those having a copy, these are compounds that are flagged with a lexer lexeme). Note that this is NOT the same as the set used in Ju'i Lobypli; in our text, we write some words as compounds that are broken into two words and kept apart by the machine grammar (such as "lenu" and "lemi"). See the note following this article for more on this subject. In any case cmavo compounds should be judged under the same rules as cmavo, except for adding the machine grammar as a reference for which words can be compounded. Scoring for compounds should be pre-decided - either 1 point per token as for cmavo, or normal scoring; it seems that the restrictions on valid compounds may reduce the valid set so that special scoring is not needed to make the game work properly.
lujvo can be evaluated either by requiring that the person make a valid sentence using the word, or that the person give both the tanru that the word is based on AND a reasonable place structure and/or explanation that shows that the word has a potentially usable meaning. Be sure to check for valid morphology, and that the rafsi used correctly represent the tanru. 'Reasonable' meanings, and 'valid' sentences are going to be arguable, especially among new Lojbanists. I suggest that, if possible, it should be left to the other players to vote on acceptability, with an urging that open-mindedness be emphasized - let Lojban stretch your mind.
le'avla borrowings were not included in the statistics. If permitted, it is recommended that the person be required to show that the word is of proper morphology (it cannot break down as a lujvo; if preceded by a cmavo, it cannot suck up the cmavo and then break down into a lujvo or gismu; and it must have a a consonant cluster in the first 5 characters, use no 'y's, and have no unpronounceable clusters caused by adjacent impermissible medials in the same syllable), and that it is reasonably Lojbanized from an identifiable root in another language. I also recommend that only le'avla with rafsi categorizing on the front end be permitted, this is not an official restriction on le'avla, but makes evaluating validity a lot easier. le'avla are the most arguable of word forms, and should not be allowed unless all players are reasonably familiar with the rules for making and evaluating them. While I can make le'avla in my head, I would still choose to leave them out of the game at this point.
House rules should be decided in advance as to whether sentences demonstrating usage must be fully grammatical, or whether a sentence with a grammatical error unrelated to the word being demonstrated is acceptable. The latter rule seems appropriate to me for evaluating gismu and lujvo and le'avla, but I think cmavo sentences must be completely grammatical.
Some hints that are applicable if you use lujvo:
Often, you can turn a gismu into a lujvo by adding a trailing vowel. Make sure you can identify and define it, though.
Similarly, cmavo can easily be turned into brivla. Most CVV cmavo are also rafsi that can be included in a lujvo.
Prefixing a CVVr/n, CCV, or CVC/CVCy rafsi on any brivla makes a longer lujvo, if you can determine an acceptable meaning.
Except for scoring cmavo, and adapting the rules for determining word validity to fit Lojban's lack of a dictionary, the game should proceed along familiar lines. We haven't actually play-tested these rules, but we welcome you all to try them and to report on your results. We may print a sample game in a later issue if a good one turns up.
Last minute note: John Hodges has reported to me that he had devised a variation on Scrabbletm that is useful for teaching. He is planning on using it in the Blacksburg class, which is studying Lesson 3. Fleshing out his brief description to me gives the following:
You can use the flash cards (up through an arbitrary lesson) and additional cards cut to the same size, or you can use a few hundred regular index cards with gismu (again, up through an arbitrary lesson) written on them. Extra cards are left free (blanks) and put in a stack; a person can take one or more of these and write any needed cmavo (you may or may not want to limit the number of pro- sumti, but don't give too few). The cards in your hand are lesson vocabulary words, and they are the only ones that score (probably 1 point a piece). cmavo from the vocabulary list might or might not be worth anything. Players must build off of other people's sentences on the board. If it is grammatical, you score the number of vocabulary words that you used. Elision is optional; if you elide correctly, though, you leave fewer cmavo lying loose for someone to build off.
The sentence must be grammatical, and if challenged, the person should be able to at least paraphrase it correctly into English.
This is a brief outline. I may report next issue on how it worked out for John's group. Feel free to play with it and send me any ideas for improvements or details that I've omitted. The advantage of this game is that it encourages interactive learning, and it also gets people to build sentences in Lojban directly from the gismu, rather than making up sentences in English and translating them. People who don't write a lot often have trouble thinking of topics for sentences; the limited set of gismu on the board or in your hand helps you focus on specific topics. People who translate from English often try to express more com- plex ideas than they've studied the grammar for.
My writing style is not formally decided on, and may not be the best or most appropriate; but it at least is fairly regular. I'm open to suggestions and changes. We could, for example, restrict compounds to those treated as such by the machine lexer. However, some machine lexer compounds would be hard to read due to length, and we might not want to require these to be written as one word. Students of the old language, who are used to compounds like "lemi" and the old "lepo" that is now written as "lenu" may like the compounds as I currently write them.
Remember that by the rules for Lojban morphology, ANY cmavo that are adjacent in an expression may be compounded and written as a single word. NO cmavo may be compounded with a brivla or name. However, while there is no rule forbidding it, if you compound across an elidable terminator, you probably will make it difficult for a human (though not a computer) to figure out.
If compounding joins two words separated by a mandatory pause, the period representing the pause is mandatory. Otherwise, these periods are optional aids to the reader. In Ju'i Lobypli, we will try to always include the periods, or to add them in if the author did not do so.
Since we are a major source of 'good' Lojban, I can't print things with known grammatical errors or obvious mistakes in, for example, tanru. I will correct these if the change is obvious. If I make more extensive changes (as I did with Preston's story above), then I will indicate that it is corrected. If there are still mistakes then, you can blame it on me. If you want, I will optionally return such changes to you for verification and approval (or rewriting) before printing, in which case it will not be marked as 'corrected by'.
If you annotate your work to indicate that a given non-standard writing or grammar is intended, I will print it as you ask. Similarly, I will use your compounding practices instead of mine if specifically asked. For exam- ple, Nora chooses to often write "lenu" as two words, as shown in the last panel of lei lojbo.
Something known to be invalid by the current rules of Lojban (as opposed to my writing style policy) will be marked with an '*'. If of uncertain validity, it will be marked with a '?'. There may be instances where, for poetry or example, you want to have something intentionally invalid. As long as it is marked as such, fine.
Ju'i Lobypli Editorial Conventions
I'm starting to receive a variety of Lojban text for printing in Ju'i Lobypli. There is enormous variety among these writings as to which cmavo are written as compounds. With handwriting, I can't always tell if compounds are intentional. I am following the practice of regularizing which words I compound to match my own writing style. This both proposes a standard, and makes it possible to use my word processor spelling checker in a useful way. Proofreading Lojban text is hard enough without trying to deal with a variety of writing forms.
Well, that's it for this issue. As I write this, it is starting to feel that we really have a language, now that several people are learning and using Lojban. Keep it up and help make next issue even better.