me lu ju'i lobypli li'u 12 moi

From Lojban
Jump to navigation Jump to search

For a full list of issues, see zo'ei la'e "lu ju'i lobypli li'u".
Previous issue: me lu ju'i lobypli li'u 11 moi.
Next issue: me lu ju'i lobypli li'u 13 moi.


Copyright, 1990, 1991, by the Logical Language Group, Inc. 2904 Beau Lane,
Fairfax VA 22031-1303 USA Phone (703) 385-0273
lojbab@grebyn.com

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 12 - May 1990
Copyright 1990, The Logical Language Group, Inc.
2904 Beau Lane, Fairfax VA 22031 USA (703)385-0273

LogFest 90 - 15-18 Jun 1990

Details Inside. Also: Negation, Indicators, News, and more.

Ju'i Lobypli (JL) 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". la lojbangirz. is a non-profit organization under Section 501(c)(3) of the U.S. Internal Revenue Code. Your donations (not contributions to your voluntary balance) are tax-deductible on U.S. and most state income taxes. Donors are notified at the end of each year of their total deductible donations. We note for all po- tential donors that our bylaws require us to spend no more than 30% of our receipts on administrative expenses, and that you are welcome to make you gifts conditional upon our meeting this requirement.

Press run for this issue of Ju'i Lobypli: 350. We now have over 660 people on our active mailing list.

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. Balances reflect contributions received thru 21 May 1990. Mailing codes (and approximate annual balance needs) are defined as follows:

Level B - Product Announcements Only
Level R - This is a 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

If your level number has an '*' by it, we need to hear from you. This indicates that we have had no contact from you in over a year, and either have not paid for any materials, or have a negative balance in excess of $25.00. If we have not heard from you before late July (JL13/LK13), we will have to lower you to 'Level 0' at that time. We're perfectly willing to send JL to people who cannot pay, as long as we get feedback indicating that you are using the materials.

Contents of This Issue

This issue is coming out only a month after the last, as we try to catch up on our tardy publication from last year. Thus, no Lojban text this issue, except for incidental examples.

The issue is dominated by two long papers, describing the proposed modifications for negation and indicators/attitudinals. These are comprehensive, even more so than the textbook discussion will be. The negation paper may be tough reading, but has lots of examples. The attitudinal paper is lacking in usage examples primarily because of time, and the lack of a comparable system in English. The two papers are bound separately from the rest of the issue; many 'level 3' students will wish to bind these papers with their draft textbook materials.

Our regular news section includes information about LogFest 90, and we also present several lesser proposals to be considered and decided at LogFest. We want your opinion on these.

Please forgive any substandard editing, especially odd formats. We had a hard disk failure the day this was to go to the printer, and this represents an intricate but hurried reconstruction.

						   Table of Contents

News
  LogFest Plans						       --3
  In Memoriam -	Faith Rich and Preston Maxwell		       --5
  Finances						       --6
  Research and Development				       --6
  Growth and Publicity					       --9
  Education						      --10
  International	News - Athelstan's Travel Plans		      --10
  Products and Policy					      --11
  News About the Institute				      --14
A Brief	Introduction to	Predicates and Place Structures, by Athelstan --15
Proposals to be	Decided	at LogFest			      --16
Letters, Comments, and Responses			      --27
Enclosures - Proposed Revision of Attitudinals and Indicators, On Lojban Negation, Some	More Proposed Logos, Map to
  LogFest 90

Computer Net Information

If you have access to Usenet/UUCP/Internet, you can send messages and text files (including things for JL publication) to Bob at:

lojbab@snark.uu.net

You can join the Lojban news-group by sending your mailing address to:

lojban-list-request@snark.uu.net

and traffic to the news-group can be sent to: lojban-list@snark.uu.net

The ".uu.net" trailer on the above address is likely to change soon. We'll notify you by e-mail when we have firm information, and provide updated information here next issue. There will be some transition period after the change, during which this address will still work.

Please keep us informed if your network mailing address changes.

Compuserve subscribers can also participate. Precede any of the above addresses with INTERNET: and use your normal Compuserve mail facility (its possible that you can send only to addresses in the '@' format). Usenet/Internet people can send to Compuserve addresses by changing the comma in the Compuserve address to a period:

nnnnn.mmmm@compuserve.com

Whether you wish to participate in the news-group or not, it is useful for us to know your Compuserve address. For example, any decision for la lojbangirz. to obtain a Compuserve account will be based on a need to serve a goodly number of you that want to exchange information.

We've been requested to more explicitly identify people who are referred to by initials in JL, and will regularly do so in this spot, immediately before the news section. Note that some people choose to write under a pseudonym which is initial-like, as jyjym. did last issue - we prefer to print names of our correspondents, but will consider allowing a pseudonym if the author requests privacy and we feel the writing is important enough to publish. Note that 'Athelstan' is that person's real name, used in his public life, and is not a pseudonym.

'pc' - Dr. John Parks-Clifford, Professor of Logic and Philosophy at the University of Missouri - St. Louis and Vice-President of la lojbangirz.; he prefers to be addressed as 'pc'. As an incidental news item, pc is getting married on 18 May, and his new bride will be accompanying him when he comes to LogFest in June.
'Bob', 'lojbab' - Bob LeChevalier - President of la lojbangirz., and editor of Ju'i Lobypli and le lojbo karni.
'JCB', 'Jim Brown'- Dr. James Cooke Brown, inventor of the language, and founder of the Loglan project.
'The Institute' - The Loglan Institute, Inc., JCB's organization for spreading his version of Loglan.

News

LogFest Plans

LogFest 90 will be held from Friday thru Monday the weekend of 15-18 June 1990; most of the planned activities will be within the weekend proper. The annual meeting of the Logical Language Group, Inc. is set for Sunday morning, the 17th, by our bylaws, and is the one absolutely fixed item for the gathering. All other dates and times discussed below are subject to change depending on what attendees want to do and/or talk about. This meeting is for YOU, so make sure we know what is important to you. Especially let us know if you can only come for part of the weekend, and have a particular topic you want to talk about; we'll try to move things around to fit you in.

From the feedback we've been getting, we're expecting more people at LogFest this year than in previous years. For the benefit of our planning, please let us know if you are coming (or if you 'might be' coming - we don't need ironclad commitments). There's no penalty for not doing so - we simply have a smoother, better planned gathering if you do.

We have plenty of sleeping space, especially for those with sleeping bags and/or tents. Families are welcome, provided that they have sleeping bags too (there are motels within a couple of miles, but the DC area isn't cheap); we also need financial help if we're going to feed families.

Also, if your coming depends on finding transportation to Washington DC, let us know. There may be someone else coming from the same place or direction that you can rideshare with. This is especially likely for the Northeast corridor, from Boston down to Washington. Be sure to specify what dates you want to attend, the earliest you can leave home and the latest you can return home to make ride-sharing feasible. If you have access to the computer network, you might also consider posting a notice on 'lojban-list', so that others who are thinking "maybe" will more likely consider going, with the knowledge of reduced transportation costs.

What will be going on at LogFest? We will have two major themes this time:

  • How can we help more people learn Lojban easier?
  • What short term applications for Lojban are worth highest priority?

We also will have the goal, by the end of the weekend, of deciding on several last minute proposals regarding the language and declaring a conditional or unconditional baseline (freeze) on the grammar.

Now for details.

There will be several things going on simultaneously.

  • Some main track discussions, generally related to the themes above, which Bob will probably be leading or helping out with;
  • 1 or 2 computer rooms, with 2-3 MS-DOS machines, Art Wieners' UNIX machine, and whatever else people bring. Art and others will usually be in there demonstrating existing software, and working on new software including the parser. There will be a little bit of computer talk on the main track, but we will generally be trying to make it so non-computer types can understand what is going on.
  • Athelstan, and possibly others, will be helping with new Lojbanists. Athelstan will give mini-lessons at times scheduled below (if there are people interested then), but he will also be available impromptu for mini-lessons, explanations of grammar points, and demonstration of the flash card technique.
  • On occasion throughout the weekend we'll hold short sessions of all-Lojban conversation - typically a half hour or less so that non-speakers don't get bored, or frustrated. These sessions will probably occur in between main discussions.

Specific schedules as they stand:

Friday

7PM - Athelstan mini-lesson (about 1 hr.), if there are people present who haven't been to one

8PM - After the mini-lesson (or starting earlier if there is none), there will be a discussion of Jim Yorke and Celso Grebogi's proposal (later in this issue) to have Lojban image languages using gismu based on single languages to make them easier to learn. Side issues may include applications such as teaching Lojban to kids (including using such teaching to help them learn English better), and other aspects of making Lojban easier to learn for people than it currently is. Dr. Yorke will be at the session, but has schedule constraints that mean that he can only attend early in the weekend, so the Friday night schedule is relatively fixed.

Saturday

Saturday morning - additional discussion about teaching and learning Lojban, including any leftover discussions from the night before. Bob hopes to have new textbook lesson drafts for people to look at and comment on, so here's your chance to make a difference before the book is finished.

If there is interest, we will start side discussions at that point on how to organize and lead a class when you don't know the language yourself. Athelstan and John Hodges will both be there, and both have done this now. Eric Tiedemann will, we hope, be there to report on the initial work by the New York City class. We want to stimulate new class efforts, even if only 2-3 people are participating. Athelstan is hoping to launch 2 classes, one in Virginia and one in Maryland, at LogFest or immediately afterwards. We are also considering instituting a weekly Lojban conversation session to start with or shortly after LogFest (this may end up having to wait until fall - simply because Athelstan may be spread too thin.)

1PM - Another Lojban mini-lesson, if there is demand.

Saturday afternoon will include a discussion of voice recognition, the Lojban parser, AI work and various other computer processing projects. Art Wieners will tell about his fairly extensive efforts, including word recognition and analysis, as well as Artificial Intelligence applications for Lojban. Unfortunately, Jeff Taylor, who has been leading the parser efforts, is having to drop out of Lojban work for a while, and won't be coming. We will be looking for new parser helpers (C programming language, preferably in an MS-DOS environment), and there may be technical side-discussions. Eric Raymond, host of 'lojban-list', our computer network mail-server, has been working on porting LogFlash to Unix, and will talk about what he's done. Nora and I will talk about planned new software and enhancements, and we'll accept proposals for others. All software (and other products) will be available for demonstration, examination, and (of course) purchase. (Oh, yes; and everybody in the world will be trying to convince Bob and Nora to switch from MS-DOS to UNIX; Good luck!)

Saturday nite - Bob will lead discussions about textbook plans, dictionary plans, and take suggestions. We'll also talk about the recent change proposals to the language. Non-controversial ones will get a quick decision; the rest will be deferred for further discussion on Sunday or later - all attendees have a vote.

Sunday

Sunday morning is the la lojbangirz. business meeting. Expect business-y things. Most people tend to sleep in late on Sunday morning after two late nights, so we've put this at a low ebb time to keep from boring people. We hope to finish the meeting by noon. The agenda is not yet known. Topics typically include finances, plans and priorities. All attendees are welcome to participate informally.

Sunday afternoon, we will discuss gismu list change proposals, hopefully briefly, and make a group decision on whether we are ready to baseline the Lojban grammar. If so, the language is DONE!!!! Practically speaking, expect a conditional baseline, possibly excluding MEX, and possibly waiting either for the parser or the textbook to be completed or both, before it becomes a full freeze. (More on this below, under R & D).

around 3PM? - A final scheduled mini-lesson; this time an advanced lesson for people who have already had the basic mini-lesson. This may continue into a longer tutorial, if there is demand, but will otherwise last an hour.

Sunday nite and Monday will be left unscheduled for whatever comes up - invariably something runs overtime. Besides, many out-of-towners will have left by then, though pc and probably Art Wieners will still be here. pc and family plan to arrive on Wednesday the 13th. Art usually comes for a full week, but his dates are not yet known.

Other activities will be dependent on demand:

Some have suggested a group effort to write or translate something, perhaps a short essay reporting on LogFest. As a possibility, we have an ancient back burner project to update Jim Carter's short story in old Loglan to current Lojban, and a similar effort involving an old story by Bob Chassell. These are non-trivial and will not be finished in a weekend, but are easier than composing raw text; people can get a good start and those who want to can continue on their own. Carter's work, in particular, will take a fair amount of tanru and lujvo re-making; he used a lot of Institute Loglan words that are based on no longer acceptable tanru. In addition, we've found that the current gismu list has more expressive power than earlier lists, making some tanru simpler or even unnecessary.

There will be other side discussions as people desire. Speak up now with your favorite topic. Some possibilities include lujvo making, and good tanru for various words, etc. One topic we're not sure how many are interested in is the making of le'avla borrowings; any decisions on the algorithm have to be made before the textbook session on the topic is written.

We will be having Lojban conversation as mentioned above, possibly with on longer session towards the end (when more people feel comfortable with trying to say things) if there is demand for it. To keep costs down, we need people to bring their own word lists, etc., though we'll have inventory if needed.

If you are thinking about ordering stuff, it saves us postage to give it to you there and let you carry it home. But we need to have a good number of people pay for stuff there in order to do this. We will probably have no money in the bank after LogFest to restock inventory, except what we collect at the meeting.

Logistics:

We generally put up a good spread of food to keep people from going hungry We're asking for contributions of $5-15 per day towards costs (and preferably more, if you can help), but there is no mandatory admission charge. If we don't get enough volunteer cash flow, this will be the last year of voluntary giving. To pay for LogFest, we will probably need more than the $600 spent last year, depending on attendance, which we expect to be double or more. (That $600 doesn't include all costs, only those paid by la lojbangirz. - it averages out to about $30 per person, including those who only came for part of the weekend.)

As mentioned above, families are welcome if you bring sleeping bags and tent or trailer. We'll have at least one adult and one 13-year-old non-Lojbanist here for the weekend, so there will probably be some side activities such as sightseeing in DC or shopping at the malls, or whatever for disinterested spouses and kids. Let us know if you're bring extras along so we can plan for food, etc.

We are close to DC's Metro system and right off the freeway. There are detailed directions and a map enclosed with this issue. Of course, if you have any questions, just ask.

In Memoriam

We've recently heard of the deaths of two noteworthy supporters of the language, Faith Rich (Chicago, IL) and Preston Maxwell (Seattle, WA). Our sympathy goes to their families and friends, and we of the Lojban community will miss them.

Faith Rich has been for the last several years the most prolific inventor of tanru and new words for the language. While she did most of her work with The Loglan Institute, she also has assisted la lojbangirz., especially when we were attempting to baseline the gismu list. Faith may have devised over 3000 tanru and lujvo as part of the Eaton project, an attempt to provide comprehensive vocabulary coverage for the most frequently used concepts in 4 European languages. A long-time supporter of the international language movement, she was in her late 70's when doing this work, and had expressed no hope of living until the language reached fruition. Yet, she was a fervent believer that Loglan/Lojban was worthwhile and important work, and provided hard effort and constant support over the years.

The community knows less of Preston Maxwell, who joined us only a couple of years ago. However, Preston was one of the first people to try to learn Lojban on his own. Preston was a skilled translator with experience in several non-Indo-European languages. He was especially intrigued by Lojban's attempt to free itself of European biases and to support structures and concepts of non-Indo-European tongues. Some of his early writings, Lojban versions of non-Indo-European short tales, were published in earlier JL issues; these were the first attempts at writing 'natural Lojban', and the first writings by someone who had not previously learned earlier versions of Loglan. Preston also attempted to organize the Seattle community, though without success. Hopefully, others there will be inspired to proceed in his stead.

These two deaths remind us of our mortality. People are always saying that they'll learn Lojban 'later', when they have 'more time', or when the language is 'done', or has 'practical use'. Faith and Preston didn't wait till 'later', they made time for Lojban, and they found reasons why working on the language now is worthwhile. In their memory, we ask other Lojbanists to do the same.

Finances

Should we bore you by complaining about finances again? We have to. It wouldn't be JL, otherwise.

The best I can say is that we are paying our bills, or at least most of them. Our legal bills have to be paid off slowly because of our rule limiting legal and other expenses to 30% of all expenses. The side benefit of that rule is that the money reserved for those legal bills is the only thing keeping our bank balance positive.

To keep us financially afloat, we've figuratively 'put the brakes on' spending by la lojbangirz. We've spent about $3300 this year as of the publication of this issue, only 3/4 of what we had spent last year by this time. Unfortunately, our income is similarly lagging behind last year's. In spite of over 50% more people, we're getting less money in.

The major drop has been in donations. In spite of our 501(c)(3) status, only a couple of people have donated money to the organization above and beyond balance contributions. Thus we are hurt more by the large number of you who aren't contributing. Remembering that we don't make any money on our publications, every non-payee has to be supported from donations. The small amount of money we make on software doesn't cover the phone bills.

The price will eventually have to be paid. We are going to be mounting a major drive to add European Lojbanists during Athelstan's trip later this year. Overseas postage is enormously expensive, and we will not easily support these new Lojbanists.

The result is that we are likely to have to borrow money to publish the textbook and dictionary. We've worked hard to get a good credit rating, and can probably get loans of the size we need, but interest costs would seriously crimp our future budget. Loans also mean that we'll have to charge more for the books, perhaps as much as $5 to $10 more than we'd like.

To put it simply, we have to find between $4000 and $6000 to publish each book (more likely the latter figure). If every JL subscriber buys 1 copy, this probably means a $20 price tag each, which will probably hurt sales, especially among non-US Lojbanists.

This also means that we won't be able to publish the dictionary until we've sold a lot of textbooks.

Are you covering your balance? Could you help by $5 or $10 more? Could you consider donating to the textbook fund, whether $5 or $50 (or more - wishful thinking)?

The answer to these questions could make a BIG difference in Lojban's success.

Research and Development

Accomplishments - We've accomplished a lot in only a month. The two major papers in this issue are the product of several months work each, but were finally brought to fruition this month, thanks to pc's visit and Bob's sweating a lot of ink (Bob's aging printer is going through a ribbon a week these days).

Assuming the adoption of these two analyses, only three chunks of the language have significant open questions. These are the tense system, MEX, and the making of le'avla.

Tense - The most significant grammatically is the Lojban tense structure. When pc was in Washington, we discussed the subject thoroughly, and what remains is to incorporate the decisions into the grammar and write it up. The incorporation will take only a few days work - the grammar of tense is isolated from the rest of the grammar, and Bob has already redesigned it from scratch twice. Writing it up will take a little longer, but the write-up will not be nearly as difficult as that of negation.

MEX - MEX, the grammar of mathematical expressions, is a bigger problem, but is only a small and simple piece of the grammar. When pc was in Washington, we didn't need more than an hour or two to devise a new approach; it will take only days to see if that new approach works in the grammar.

MEX is a complex problem, but is easy to resolve. With negation and attitudinal indicators, half the battle is understanding the entire complexity of the problem. Once the goals were clear, the design was easy. With MEX, the goals have been understood for a long time; the problem is merely one of design.

Bob has had contact with Ivor Darreg, who proposed a MEX-like code called Numaudo about 30 years ago. Anyone familiar with this effort is encouraged to send comments to us.

le'avla - The other remaining design issue is the making of le'avla borrowings from other languages. Unlike the other design problems le'avla grammar and morphology are fully settled. The question is merely one of usage:

How can le'avla be made efficiently and intuitively so that language learners have the full power of these words at their disposal?

There are two conflicting goals expressed in this question. Any efficient and intuitive algorithm for le'avla inherently will cut down on the full potential use of le'avla.

Ideally, Lojban le'avla should be as much like borrowings found in other languages as possible. Thus, jargon words for foods, technical and scientific terms, etc. would be easily recognized for what they mean. By this rationale, the word for the food "spaghetti" should be "spageti", which is indeed a valid le'avla; the word for "integral" should be the valid "integra". These words would certainly be easy to learn.

There are three problems with this approach:

  • natural language borrowings and jargon fail Lojban's basic paradigm of Lojban is "one word - one concept". The word "integral" has several jargon meanings - which meaning will Lojban adopt, and how do you choose words for the other jargon meanings. Architectural 'integral' has no relation to mathematical 'integral'; each field based its word on a different etymological link to the English and earlier Latin concept embedded in the word 'integrate'.
  • recognition of the word chosen depends on whether the appearance or the sound in emulated. Someone inventing a le'avla 'on the fly' in conversation, will not be thinking as much about how it looks on paper, as whether it sounds like the borrowed word. Conversely, a writer will concentrate on visual recognition, often ignoring sound differences.
  • cultural bias. Most borrowing nowadays is from English to other languages, and such borrowing is often at the expense of perfectly good words in the native tongue. Barring adoption of the Yorke-Grebogi proposal below, there is no virtue in making Lojban look like an encoded English, and any massive dependence on borrowing from the one language that all current Lojbanists know will suggest uncontrolled cultural bias (whether it is true or not).

For any 'official list' of borrowed vocabulary, Lojban should rely on careful comparison of like words in several languages, and borrow from a single language only when a tanru cannot be made and when there is clear evidence that most other languages have similarly borrowed the word. Otherwise, we would be better to compose a word using a variation of Lojban's gismu-making algorithm.

Most le'avla are really names of a sort, applied to specialized concepts as labels to allow them to be easily referred to. There is no semantic breadth to these words and they are unlikely to be used in tanru other than in a most obvious manner, much as one would talk of a Broadway play or a Chrysler car.

Unfortunately, not all le'avla will remain 'name-like'. Borrowings have the habit of passing into common usage, ceasing to be jargon, acquiring the full range of connotations that other words have, and being used metaphorically to build other concepts.

To preserve its integrity, Lojban has to keep all le'avla distinct from each other, so that each has only one meaning. Thus, if two fields have claim to the word "integral", they both can't use "integra".

A few years ago, Bob proposed that all le'avla be labelled by some tag that associates the word with a particular jargon. Thus all foods would have a label saying that the word was a food. In short, all le'avla would be invented as a sort of lujvo, a compound of two words or parts of words, one a Lojban categorizer and one a base word having no defined meaning or place structure.

An advantage of Bob's approach is that, if you are unfamiliar with a jargon word, the label and morphology will at least tell you that it is a jargon word of a particular field. Thus these words are more learnable to a non-specialist.

The disadvantage is that if you ARE in the field, the word will no longer resemble the source language word, and will be more difficult to learn for people trained in the specialty that uses the word.

Ease of learning is important to using the full power of le'avla. However, ease in making new understandable le'avla is also important. In natural language, new words are created as borrowings all the time. Lojbanists need to be able to make new le'avla at will, with ease comparable to the making of lujvo.

Unlike lujvo and gismu, le'avla have a more amorphous morphology, and ease of making words requires some constraint on those words. A legal le'avla is defined merely as a Lojban word having a consonant cluster in the first 5 letters (which must be a legal initial if at the beginning of the word), only permissible consonant clusters within the word, no occurrences of 'y', a final vowel, and penultimate stress. It also cannot break down into a combination of gismu, lujvo and cmavo either directly, or if preceded by a cmavo. le'avla space is thus primarily defined by what it is not. Testing to see if a word fits the rules is difficult, subject to error, and virtually impossible to perform in your head.

It turns out, though, that if you always have something on the front of the word as a marker of 'jargon type', as proposed by Bob, there is an algorithm (proposed by John Cowan) that makes it as easy to make a le'avla in your head as it is to make a lujvo.

The words so formed are always good le'avla, and the front part can be any rafsi, which is glued on by a syllabic consonant like 'r' that acts like a vowel.

Unfortunately, a side effect is that the rest of the word often cannot stand on its own as a word. This means that all le'avla are forced to be at least one syllable longer (the attached rafsi) than a corresponding natural language source word, and that all jargon words in a field will be prefixed by a common and redundant syllable.

Bob favors this approach anyway - it encourages tanru and lujvo-making as a preference to borrowing from other languages, which is likely to be heavily biased towards English at least in the early years. Furthermore, the remaining le'avla word space need not remain unused permanently. Some review board can watch for le'avla that become 'more than jargon' by coming into common use and being using in tanru. Specialists in a field could also set up boards to approve 'unofficial' jargon le'avla that could be flagged as being special to the field without repeating the redundant field affix. (unofficial because someone has to resolve between two fields that try to use the same word).

'Official' short le'avla, on the other hand, would become a subclass that is as carefully made and evaluated as gismu. A new learner need not learn these short forms if she/he knows how to reborrow the word anytime it is needed.

Is Bob's approach adequate? Is the algorithm for building le'avla as easy to use as it needs to be? Are there other approaches to be considered? These are not really design issues, but usage issues that will be determined in future years by the people who learn and use the language.

Until then, we need to know what to teach in the textbook. This is not something that Bob should decide on his own. What do you think?

Grammar - With the expected completion of all substantive grammar changes within the next month or so, we can finally look to baselining the grammar. A decision on this will be (hopefully) made at LogFest 90. As with the gismu list, we expect a 'soft' baseline that will last several months, followed by a hard freeze on the grammar, with the intent of holding the grammar constant for 5 years, until fluent speakers can make decisions based on 'inside' knowledge of the language.

There are risks in baselining - we might freeze a fatal flaw into the language - but these risks are believed small. We've now taught 3 classes in Lojban, Bob has written draft lessons covering most basics of the language, and Bob and Athelstan have both done translations of complex and intricate works from other languages. We also have our record as a guideline. Over the last year, the grammar is sufficiently stable that we can still in good conscience sell the draft lesson materials written then. There have been changes, some drastic in overall effect on the language, but most do not affect the beginning speaker.

Writing the textbook may reveal more flaws. Bob believes that any such problems will be minor compared to the past history of the Loglan project, or even the recent history of Lojban. Of the two big redesigns described in this issue, only one even required minimal changes in the grammar.

The parser may reveal a few more problems, as may the random sentence generator. Most of the problems detected by the parser have been minor ones related to elidable terminators that turned out not to be elidable because of the restrictions of YACC grammars (thus an incomplete sentence consisting of bare sumti with no selbri must be followed by "vau" to be grammatical). Doug Landauer has proposed a more powerful parser-writing tool which might simplify the grammar and resolve these problems. Following a soft baseline, though, such improvement would have to be either invisible to the average user, or offer a benefit major enough to warrant changing the baseline (and the textbook that will be written based on that baseline).

A final source of changes may come from an analysis in progress by Carl Burke (and possibly to be assisted by David Bowen and others) to convert the YACC grammar into a simpler format called 'Extended BNF'. 'BNF' grammars are easier to read than YACC grammars, and the process of building them reveals irregularities and asymmetries in the grammar. We've already made two extremely minor changes as a result of Carl's work, both expanding the set of permissible Lojban expressions to include a symmetric equivalent of another structure. The negation analysis discovered a third asymmetry, which we repaired by removing several redundant rules from the language.

It is unlikely that anyone will notice any change to the grammar after the baseline, but our rules require that any such change be announced, explained to all Lojbanists, and approved only after a comment period.

The major risk to the grammar for some time to come will be the MEX component. The MEX grammar is impossible to test without trying to use it, and there is nobody competent enough in Lojban who could make use of the capabilities of MEX to the degree necessary to test it like we have tested the rest of the grammar. Only time will prove MEX. (Meanwhile, we'll try to be satisfied with one of our several attempts at grammatically expressing the opening phrase of the Gettysburg Address.)

Parser - Key to the final test of the grammar will be completing the parser. Unfortunately, Jeff Taylor, who has written the most comprehensive parser version so far, has indicated that he doesn't expect to be able to help during the critical phrase right after the preliminary baseline.

We are thus looking for people who know something about parsers to help out over the next several months.

We must identify volunteers who are willing to strongly commit to seeing the project through to completion. The parser project is too vital to suffer problems we've had in getting LogFlash ported to other computers. Committed people who know what they are doing, and who can and will stick to it, can implement the current changes in a relatively short time, and can minimize the amount of work needed to maintain the parser if testing shows that further changes are needed.

Some preferred qualifications:

  • ability to program in C, and a knowledge of how YACC-based parsers work, so you can understand what Jeff has done; rewriting or translating the existing, and working, code in a different language is not an effective use of time; porting the parser to other computers is a task for when the grammar has been baselined and tested with the first parser;
  • a demonstrated attempt to learn the Lojban grammar, which probably means that you are a level 3 student; most of the problems to be dealt with require a knowledge of why we are implementing a rule, not just what the rule is. The programmer needs to communicate with those who are not familiar with the inside of the parser, like Bob (some level 1 and level 2 people might be able to do this, but we'll be harder to convince);
  • an MS-DOS computer environment with a C compiler; this is for compatibility with Bob and with our community, which primarily uses PC-compatibles with MS-DOS. We eventually want to have a UNIX version of the parser, but nearly all of the people who will be involved with testing the parser cannot work with a UNIX version.
  • a willingness to work on the parser pretty much to the exclusion of other Lojbanic efforts for a while; this has to be voluntary, but only for a few months of part-time work, until a minimal working parser is produced;
  • there is a preference for a single person or a small co-located team, preferably near Washington DC - this is simply for communications and logistics purposes. Since Jeff lives in California, distance has proven to be a surmountable problem.

We have about 3000 lines of C code, which Jeff says is not especially well commented. The user interface is fairly solid. The primary work is inserting new YACC grammars into the parser, and modifying the hand-coded rules in the lexer to reflect similar changes in the YACC-verified grammar. The basic style is fairly well set, so we need someone who can read Jeff's code and complete it, not rewrite it. Time required is expected to be less than 100 total hours once you know Jeff's code, and make any conversions to run the code in your environment. Jeff tried to design the system so that inserting a new grammar would be very fast; and has demonstrated that his concept works several times.

What do you get for this? Only our thanks and credit in the final product. Since Jeff did the bulk of the work and has chosen not to retain ownership rights, it would be unfair to offer greater remuneration to anyone else.

I'm sure I've made this out to be worse than it really is. But there will be a lot of pressure to get the parser completed without constant support from Bob, who will be writing the textbook. Faint souls should not venture forth.

Any volunteers?

Growth and Publicity

Our growth rate continues at a phenomenal pace, although it has slowed some since the beginning of the year. Actually, growth may be continuing at the same rate, but a few people have responded to the questionnaire in JL10 by dropping out or indicating that they wish to be inactive until the textbook is done. We don't count these people in our active mailing list.

Our growth has been almost entirely in the more active levels. We've added another half-dozen level 3 students in the last month, and one of these, John Cowan, has already done his first Lojban writing, a fairly sophisticated translation of a single extremely complex (and weird) English sentence - but more on this next issue when we print it.

Much of our growth has been outside the US. There seems to be a budding class or study group in the Vancouver BC area, led by new Lojbanist Mischa Sandberg. He contacted us as a result of our interview on the CBC Canadian national radio program. Some French-speaking Lojbanists in Canada have started learning the language as well.

We'll have to leave details out until things are more definite, but one of these new French-speaking Lojbanists is a well-known science fiction writer, who wants to incorporate Lojban as a major element in an in progress French-language novel about a 'logical language'. We've offered to assist in incorporating 'good Lojban' and inquired whether an English translation might be considered by the publisher (How many of you would buy such a translated novel, merely knowing what we've told you?). If the book is published as expected, la lojbangirz. will try to make it available to the community.

Most significant has been the increase in membership in the Usenet/Internet computer mail-server, called 'lojban-list', which is approaching 100 people, including those cross-linked from Compuserve. Now all we have to do is get these people talking - the net has been relatively silent. Since Bob has to pay long distance rates to use the net, he can't lead activity. Thus people need to start talking about their own interests.

Our major publicity effort this year independent of the textbook publishing, and any advertising that is associated with that, will be Athelstan's trip to Europe, which is described under International news below. Bob and Nora are planning a trip to California in late August to early September, but haven't made specific plans yet.


Education

Athelstan finished his introductory class in Lojban in early May. Unfortunately, attendance at each session was inconsistent, so not as much was covered as he had intended; however, a good survey of the entire language was completed. Of 10 who signed up, 4 or 5 'completed' the course, in that they have moved past the basic material covered and can continue studying productively on their own using the draft textbook lessons. One of these graduates is moving to the Kansas City area, and we are hoping she will help the several others living in that area when the textbook is finished.

Eric Tiedemann has gotten the New York City people together, and two classes will be starting there. One will start 20 May, and run every Sunday afternoon (except Memorial Day and LogFest weekends) at his residence in Manhattan (near Columbia U.) for most of the summer; the other will be on Wednesdays in Brooklyn at Marc Glasser and Donna Camp's residence. We'll report on progress next issue.

We also have a potential class in the Vancouver BC area, as mentioned above, and Rory Hinnen has been trying for months to get people together for a Los Angeles area class (if you live in LA, and are interested, his phone number was in le lojbo karni #11, or write to Bob at the la lojbangirz. address and he'll try to get you in touch with Rory.)

International News

Athelstan's Travel Plans - Athelstan is doing detailed planning for his trip to Europe and Israel in August through October. He promises to stay out of the poison ivy this time. (For those not with us last summer, Athelstan contracted an awful case of poison ivy just as he was to leave on a cross country trip. After being bedridden for a full week, keeping his schedule was impossible, and the whole trip was cancelled.)

Here is his approximate route:

He will be in The Hague, Netherlands during the third week of August for the World Science Fiction Convention. He will meet with any Lojbanists who can make it there. After spending a week thereafter sightseeing in the Benelux countries, he plans to cross over into Germany, possibly travelling by bicycle. He will definitely be stopping in Goettingen. By now it is early September. He will then travel on a Eurorail pass south, stopping at several places in Germany, and passing into Austria and/or Switzerland (partly dependent on the Lojbanists in those countries).

Athelstan will pass from Switzerland going west; he intends to stop in Bessas in southern France to visit Lojbanist John Negus. He will then return eastward into Italy, stopping in the Turin area to give talks with the assistance of Lojbanist Silvia Romanelli. Next he will go south to the boot-heel of Italy and take the Brindisi ferry to Greece, and eventually will continue on to Israel. One of our level 3 (American) students is planning to move to Israel, and will help Athelstan organize and present talks there.

All this wandering will use up his month of rail travel, leaving him in early October returning by a much more direct route to the Netherlands around mid-October, from which he'll return to the U.S.

Another ambitious schedule, but more likely to be completed since he can sleep on the train between stops.

Brochures - By the time Athelstan goes overseas, we should have completed the translation of the brochure into French and Italian, and we are trying to arrange translations into Spanish and German (any volunteers to translate or to check someone else's translation?). We will have pre-shipped supplies to several places along his route so that he doesn't have to cart them all over Europe, but Athelstan should have copies of most of our publications to show off as he goes.

If you aren't on this itinerary, we'll have to try to catch you next trip, unless you can meet Athelstan somewhere on his route. To do any coordination of this sort, we need to hear from you as soon as possible.

Products and Policy

lujvo Analyzer - Nora has enhanced the lujvo-making program so that it also takes lujvo apart for you, giving you the source tanru, in addition to putting them together and drilling you on the latter skill. No price increase for the added capability. Nora is also adding a drill on taking lujvo apart, which will be available by LogFest. For those few who have made it to LogFlash 2, the lujvo drills in the program should significantly enhance your memorization rate.

List of Available Papers - My initial search (described last issue) turned up rather fewer items than I had hoped. The following are what I have readily available, with page counts (parenthetical notes indicate the subject of the paper - not always clear from the title).

There are a few other papers that I have, which I am less sure the authors intend for me to make public. If you have sent me something in the past that you think should be on this list and isn't, please let me know.

We are charging 15c/page for these to cover the extra work and Xeroxing charge for these essentially individual orders. Filling these orders will also be on a low priority basis, unless you are doing a volunteer activity requiring the materials for some reason, and these items require payment or existing positive balance.

Athelstan and Bob LeChevalier: Review of Loglan 1 - 12 pg.

James F. Carter: The Language Gua\ spi - 18 Nov 1988 (the latest paper we have describing Carter's tone-based language derived from Loglan - Carter occasionally revises this paper, and we'll send you the latest we have, though the page count may vary) - 23 pg.

Paul Doudna: Back to School - P*L*G*S Revisited ('Pretty Little Girls School' and the Ambiguity of tanru) - 14 pg.

Paul Doudna: Comments on Loglan 1 (1989) - 12 pg.

Paul Doudna: Comments on MEX as discussed in JL9 - 2 pg.

Paul Doudna: Consolidated Synopsis of the Categories from Roget's Thesaurus (a detailed analysis based on 4 different thesaurus versions) - 26 pg.

Paul Doudna: Lojban Predicate Categories (Semantic Categories of Lojban gismu) - 77 pg.

T. Peter Park: Lojban (two excellent introductory papers that will be the basis for a new Overview for new people when we get time) 23 pg.

T. Peter Park: Lojban Etymologies (sample gismu etymologies compared to roots in several languages, not just the basic 6 used to create them) - 6 pg.

T. Peter Park: On Reconstructing the Common Origin of Modern Languages (the work of Morris Swadesh and others postulating a common language origin, and a poem in Nostratic) - 21 pg.

T. Peter Park: Prehistoric gismu?: Lojban and the Human Proto-Language and The Morris Swadesh "Top 100" in Lojban (compares Lojban words and etymologies with hypothetical proto-language words; 2 papers) - 12 pg.

Eric Raymond: La Tenguar: A romantic Orthography for Lojban (Applying the Tenguar Alphabet of Tolkien to Lojban) - 6 pg.

I am also expecting some papers from Andy Hilgartner, relating General Semantics, Lojban, and novel theories of logic, sets, and behavior that he and his associates have developed over the last few decades.

The Planned Languages File Server - We've had a lot of requests for distribution of our materials 'on-line' via electronic mail. For reasons we'll go into in a minute, we have to constrain this somewhat. Meanwhile, we will attempt to make some of our most common items available via computer mail, via a new 'Planned Languages file server' service offered by Jerry Altzman and Mark Shoulson.

If you have network access and haven't heard of this setup, contact:

archive-management@ivory.cc.columbia.edu

(Compuserve people prefix with 'INTERNET:').

If you send a one-line message containing the word 'help' only to this address, it will send you help information - or you can ask more specific questions.

Items I plan to send to the file server include: - the gismu list

  • the rafsi list
  • the brochure in English and French
  • the Overview
  • the latest machine grammar (after the LogFest reviews)
  • the Saki story translated by Athelstan for JL10, with parser analysis and the JL10 English.

If I have time, these files will be on-line by the time you receive this.

I will listen to requests from people for any other items they wish to see distributed via this service. Decisions will be made on an item-by-item basis. However, we will not be putting up materials until we are reasonably sure of their long-term stability, and some materials like JL and the Synopsis, are too dependent on text formatting for us to readily make ASCII versions.

We have no problem with other people putting up their own interpretations and proposals for various aspects of the language (please send us a hard copy for the archive record please), but we would like people not to post copies of our materials and publications on electronic media, at least not without asking us first.

This is part of a large and complicated policy problem regarding materials distribution. Our policy that reflects numerous constraints, and it is not one we are necessarily happy with. Because this policy has been recently questioned, we will discuss it at LogFest. Meanwhile, we want to explain what that policy is, and why we feel bound to it.

Policy - Here is our distribution policy:

la lojbangirz. encourages the free exchange and distribution of information about Lojban. We thus have a copyright policy that encourages distribution 'without charge' of most of our materials. The actual design of the language, as opposed to the materials describing it, are considered to be fully in the public domain.

We can't, however, extend this policy to draft versions of our to-be-published materials. With this material, we do not want people freely distributing it, without checking with us first (at which point we'll probably OK it on an individual case basis).

While we would like to include everyone in whatever aspects of the project they choose, we cannot. For any given research and development effort, we have to limit distribution of the various draft proposals to a manageable number of people that we believe will contribute to that effort.

Finally, since the bulk of the decisions made on any issue require extensive interchange, usually both informal and verbal, most people get their chance to participate only at LogFest. Because it takes place only once a year, we tend to have too many issues 'on the table' at LogFest to discuss each of them in detail with everyone. Therefore we have to ask the community to trust its leaders to consider everyone's viewpoint. After all, we have ultimate accountability to you - if we don't do a good enough job on the language, you won't learn and use it.


Why the current policy? Several reasons, notably cost, time, schedule, and our image as a project.

Cost - It costs money to produce these materials, and this includes basic overhead costs and the R&D costs of producing the first copy of any document, as well as reproduction, postage, and everyday business expenses (for example, we pay state sales tax on Virginia 'sales' even when they aren't paid for, and on all stuff we throw out unsold). Every free copy we give out means that we either need to replace the recipient's share of our costs with donations, or charge everyone who does pay even more. This is unfair. With our voluntary balance system, we can at least let people know what their share is, and they can decide whether they can afford to contribute their share, and whether they will.

As a note on this justification, it does not save us the full cost of a copy to send you something via electronic mail. Even if we use a means of free distribution (as we plan to do for limited purposes as described below), the lower copy volume increases our per copy cost on the rest.

As anyone who has sold software under 'Shareware' can tell you, only a small fraction of the people who get an on-line Shareware product ever pay for it. Thus only a stable product requiring small support expenditures can be distributed, unless the product becomes extremely widespread, or the suppliers have independent financing. We don't.

Hopefully, this will change after the language is baselined and the books are written. During the time period before then, the survival of the language is dependent on la lojbangirz.'s financial solvency, and we need to ensure this continues.

Time - Time affects mostly the special orders policy, and the involvement of people in the decision-making process.

To put it simply, I (Bob) have only a limited amount of time and far too much to do if we want to have a textbook and dictionary soon; I already spend as much as half my time on overhead activities and filling orders. If I take extra time to fill special orders, the important work doesn't get done.

If I fill some special orders, I have to fill them all; it would be only fair. Thus, I must avoid filling any, unless the order is justified by a specific project goal.

Thus, if you want something that isn't on the order form, you can ask for it. If there is enough demand, we will consider adding it to our regular offerings.

Schedule - People have been waiting for 35 years for the language to be completed. The language is essentially done now. The research questions that occupy so much of our time and paper may seem major, but are really trivial corrections. The changes reflected in the 38-page negation paper, for example, can be summed up two lines of the YACC grammar - all the rest is 'teaching material' so that reviewers (and the textbook writer) understand the same interpretation of those two lines.

Most of the work on these minor issues require a lot of time, and a fairly deep knowledge of the language. We rely on people, like pc, with an immense historical knowledge of the discussions that have taken place undocumented over the last dozen years or more.

New people have a lot to contribute, too, and our pages are studded with names like Michael Helsem, John Cowan, and Albion Zeglin, and John Hodges, who have made contributions to the language by asking the right question at the right time, proposing a simple idea that offers major benefit without causing traumatic language change or relearning, or simply by showing us what the language looks like when someone uses it.

These new people are part of an 'inner circle' of their own creation, one of participation. There is, however, because of schedule, another 'inner circle' that is making the 'last minute' decisions about the details of the language. This circle is by no means fixed - it changes from issue to issue, and usually involves the proposer of the change, however new that person is. The core, however, has been pc, Nora, Bob, Athelstan, and for vocabulary issues, Tommy Whitlock. The primary qualification for this level of involvement has been a willingness to allow Lojban to disrupt and dominate the rest of your life activities (pc spent a couple of days reviewing the negation paper during the week of his wedding), but that is the price of getting the language done now.

There is no intent to exclude others, but we do, and we unfortunately must, as a side effect of trying to get the language done. For every new person we bring into a discussion of any issue, we have to provide that person with the history, and the language design affecting that decision. If the language is ever going to be finished, we have to draw the line fairly strictly. Any new person has to work hard, and gain an understanding of the issues based on their own study, and whatever occasional help we can offer.

Draft materials are the results of some one person's study of a particular research problem, and a given version circulates among the 'inner circle' that is associated with that problem. If the problem is critical, or if the results will take too long to complete in the face of high demand, as with the cmavo list, we release this draft to our entire subscriber list to look at, to use, and to comment on. For these drafts, the final quality isn't there, and stability isn't promised. Yet, because anything we 'publish' smacks of being 'official', our draft publications are treated more seriously than any informal writer's expressed proposals or opinions.

Image - Which brings us to the image problems of stability, veracity, and inclusiveness.

A lot of people have invested time in learning previous versions of the language, only to find the language they'd learned was changing like the desert landscape in a windstorm.

The Loglan Project has had a history of uncontrolled change, one that Lojban must live with until we have proved stability. As a result, Bob keeps fairly detailed records of who gets what version of what materials, so that updates can be provided on request. (This is the mysterious difference between 'level 1' and 'level 2'.)

Because of the aura of 'official-ness' that attaches to our publications, and because we are writing material intended to look authoritative when approved, you cannot always tell that a draft is incomplete or unapproved. Because so much of our work is paper hard-copy oriented, we don't have a system in place to reliably inform people of the current status of any particular document they get access to. It is thus likely that someone getting a copy of draft material other than from us will be working with outdated information. If someone invests time in learning from our materials, and then finds the stuff is obsolete, they get upset - with us - for misleading them. They also get the wrong impression that language change is not under control enough for them to bother learning.

Even if we had a good system for marking drafts, we still have to exercise some care. It can be argued that people who get well-marked drafts and study from them, do so at their own risk. But it is the language effort as a whole that is at risk. If these people ignore the warnings, but then feel misled anyway, they leave; hundreds of people have done so with prior versions of the language. The language was virtually dead in 1986-7 (less than a half-dozen people were still actively working on the language) when the effort that became la lojbangirz. got started.

The majority of Lojbanists have little interest in the language development project, other than to know that their concerns are listened to. They want a language, complete and ready-to-learn. That is la lojbangirz.'s primary priority.

After the language baseline, there will be a 5-or-more-year period of little or no change, rigidly controlled. By being slow and rigidly controlled, we can then involve everyone, though practically speaking, while a newcomer might find a problem, only someone who has learned the lan-guage will be able to devise a credible solution.

There will probably be as many or more proposals for change as there are now, but the people who learn the language will know that the proposals are just debate. After the five-year period, they will be the ones to select which if any changes to accept. The language will no longer be imposed on people from outside 'experts', but will be a living language responding to its own internal life. The freeze on the language will not end until the speakers of the language have enough self-confidence (and inertia) that they keep control over the rate of change they will permit (which will probably be as low as any other 'stable' language). This is the way it must be.

Finally, addressing one of the original issues that led to this discussion, we cannot orient the Lojban effort around the computer communications network. To do so would be to practice an exclusion of non-computer people that we cannot afford.

Too many of our most committed and productive workers have no access to the computer networks, and about half the community has no particular access to computers at all. Many of these people have complained about an apparent bias towards people with computers and computer knowledge. This is partly due to Bob's bad habits of accidentally using computer jargon throughout his writing.

We can encourage 'on-line interchange', and have done so. I'm spending far more time supporting 'lojban-list' then I intended to. But, we cannot afford to even appear to exclude people who don't have computers. In fact, we have to try especially hard to include people without computers, to make sure that they can participate to the same degree as everyone else.

News About the Institute

Readers have indicated that they want me to avoid commentary in this section, on the assumption that everyone who cares already knows about the political and intellectual property disputes between la lojbangirz. and the Loglan Institute. Anybody who doesn't know what's going on, and cares, can write to la lojbangirz. for explanation.

New Lognet Published - Jim Brown (JCB) put out another Lognet around the beginning of April. He reports about 300 copies of Loglan 1 sold, and about a dozen new members for the Institute. He also reports that the Institute is still losing money even at that sales level.

JCB also reported that he and his wife visited some Arizona universities and an anthropology oriented science fiction convention. JCB gave several talks on Loglan, and reported being encouraged by how easily people seem to learn pieces of the language. He also met with an unnamed colleague to work on a paper about Loglan which he plans to use as a delayed answer to Arnold Zwicky, who strongly criticized linguistic aspects of Loglan in the late 1960's.

Several technical proposals for changing the Institute version of Loglan were discussed. Most of the proposals either are not applicable to the Lojban version, or have been incorporated already in our design.

There are apparently several proposals under consideration for resolving two of the flaws in Institute Loglan that Athelstan and Bob specifically identified in their review of Loglan 1.

One reader argued against a proposal from the December issue (discussed in JL11), that would divide up Institute Loglan gismu into 'nouns', 'verbs' and 'adjectives' of several types, and given each group a peculiar series of inflections based on varying the final vowel. The argument against the proposal seems to have convinced JCB.

Interestingly, a new paper in the journal of the New York Academy of Sciences reports that linguists investigating proto-languages are hypothesizing that the earliest language had only one vowel, and that multiple vowels appeared in language to distinguish between related meanings - these differing meanings then became 'nouns', 'verbs', and 'adjectives', and then all the varying declensions of each. Thus the Institute may be repeating ancient history.

Rex May Resigns - Rex May has resigned as 'editor' of Lognet. He had held this position for the last two issues, although JCB did the production. Rex indicates that JCB made editorial decisions without consulting with Rex, and specifically refused to print an article Rex had written proposing a different morphology for Loglan words. Rex plans to submit it for publication in Ju'i Lobypli after rewriting it.

The resignation was due partly because of editorial disputes with JCB, and partly because of the dispute between la lojbangirz. and JCB. Rex printed a statement in the recent issue of Lognet that was critical of la lojbangirz. for not clearly demonstrating to him why we feel that JCB had made Loglan public domain. He has indicated that this statement was printed in order to stir a response from the community. Only two people responded, but the answers were sufficient to provoke Rex's resignation.

Rex has given us the following comment for publication: "I have decided that the 'ownership' of the Loglan language is at best equivocal. As editor I feared I was taking sides. I don't want to take sides on something I can't figure out."

Rex also wrote a review of Loglan 1 for Liberty magazine (March 1990), a libertarian publication. The review was very positive, although it concentrated more on the positive aspects of Loglan than those of the book. A rebuttal letter that severely criticizes Rex's review appeared in the May issue. The comments seem to be aimed at Loglan in general, although the comments are based in response to Rex's statements. The criticisms of the language are flawed, and appeared to be based on the writer's negative attitudes toward what Rex had written, rather than on any independent knowledge of the language. Thus, more than anything else, the exchange seems to have demonstrated the difficulty in writing very short articles about Loglan/Lojban for a broad audience; if specifics are given out of context, inevitably a few readers will incorrectly generalize or otherwise misinterpret what has been presented.

la lojbangirz. is considering a response to the Liberty exchange, especially since Lojban was not mentioned. Unfortunately though, the misinformed criticisms seem to have dampened Rex's enthusiasm for the language, if not others.


Athelstan has noticed that people seem intimidated by the concept of predicates and by the apparent amount of work required to memorize place structures. Athelstan here explains what a predicate is, and shows that learning place structures is familiar and natural to every English speaker.

A Brief Introduction to Predicates and Place Structures

by Athelstan

A predicate in Lojban - or in logic - is a relationship between one or more arguments. For example, if John is Sam's uncle, then we say:

la djan. cu rirbu'a la sem.

If John hits Sam, we say:

la djan. cu darxi la sem.

If John is taller than Sam, we say:

la djan. cu sraclamau la sem.

Although the English sentences vary in appearance and grammatical structure, the Lojban sentences are similar to each other. The type of relationship varied from state (English noun) to active (English verb) to characteristic (English adjective), but it remained that all three were relationships between John and Sam, so Lojban treated them all the same grammatically; this is the heart and soul of Lojban grammar.

Each relationship has characteristic arguments or roles associated with it. Uncle has the "uncle" and "nephew" or "niece". Giving has the "giver", the "recipient" and the "gift". Reading has the "reader", the "text", and the "document" or "medium of recording" (e.g. book, headstone, CRT screen).

In English most verbs have a rigid place for each of these arguments. For example (leaving off the last place):

John reads the sentence.
and
The sentence reads John.

mean two entirely different things.

Similarly:

Sam gives John the book.
John gives Sam the book.
John gives the book Sam.
and
The book gives John Sam.

are all different, and the change in meaning is directly related to the change in position.

Lojban predicates also have specific places for specific arguments. This place structure is denoted explicitly in the definition, unlike natural language place structures which are learned subtly as implicit usage of the word. In Lojban, for example, the place structure of tcidu (read) is:

A reads B from medium C.

The place structure includes the reader, the text, and the medium being read from. So:

la djan. cu tcidu le jufra
John reads the sentence.
and
le jufra cu tcidu la djan.
The sentence reads John.

are as different in Lojban as they are in English.

Lojban "dunda" has the place structure:

A gives B to C.

which has the giver, the gift and the recipient.

la sem. cu dunda le cukta la djan.
la djan. cu dunda le cukta la sem.
la djan. cu dunda la sem. le cukta
and
le cukta cu dunda la sem. la djan.

go through the same convolutions in meaning that their English counterparts do.

So place structures are not foreign to English speakers. We use them; we just don't think about them or talk about them. They also move in funny and subtle ways in natural languages, and we learn by example how to manipulate them. Lojban makes them regular and explicit and it's easy to manipulate them.

Proposals to be Decided at LogFest

The two papers separately attached are intended for comment, discussion, and (hopefully) approval at LogFest 90. They were written by Bob, and have passed a stiff review by Nora, Athelstan, and pc. If you cannot attend, feel free to send in your comments anyway.

The negation paper probably contains about the equivalent of 1 1/2 draft textbook lessons worth of technical information, and the attitudinal paper another lesson's worth, thus providing level 3 students with significantly more study material on the language. They are bound separately so that students can bind them together with textbook lessons in they wish. The attitudinal word list is bound separately from the paper, for the benefit of those who bind their various word lists together.

There are also several minor proposals which will be discussed. Some of these are included in the following pages.

It isn't clear how well detailed technical topics can be discussed at a meeting like LogFest, where attendees backgrounds and interests are widely disparate, so the amount of group discussion time given to each will be determined by how many are interested, and how much chance they have had to review the proposal. Objections to any proposal will be noted, of course, and small side discussions will take place as we have time, to resolve these. LogFest decisions are usually made by consensus, occasionally by direct vote; all proposals not relegated for specific further study will be decided one way or the other, though - in previous years this has meant wrap-up discussions on Monday, or even Tuesday.

Now for specific proposals.

gismu Baseline Changes - New gismu are required by our baseline rules to be discussed and approved. They are supposed to be analyzed in the 6 languages and 'made' before the proposal is decided, but we do not currently have an Arabic dictionary reader available (any volunteers?). This is relatively unimportant; rules also stipulate that words to be added will be 'coined' from unfilled space when the invented word causes a conflict, since we don't want to change the baseline word set unless absolutely necessary. Change proposals are numbered for clear reference.

  1. A major reason for adding or changing a word is the desire to give the concept a rafsi. This is the reason for last issue's proposal to change "ckamu" to "mleca". There are no other outstanding proposals to change words.
  2. Last year's LogFest decided that the concept "text" probably did not need to be a gismu, but no general solution was found expressing the concept. Recently, Athelstan recognized that nearly all meanings of the word "text" can be covered by "se tcidu" ("selci'u" = 'x1 is read-by/readable by ...', with the added tanru qualifier "lerfu se tcidu" ("lefselci'u") to be absolutely clear, if "tcidu" comes to have broader meanings than just the reading of text. Chinese ideograms probably are included in these definitions for "text", since they are "read" and they are "lerfu" of a sort. This finding, if agreed upon, will remove this proposal from the table.
  3. Bob wants to revisit the issue of whether we need a gismu to separately indicate "day" in the sense of "daylight". The current meaning of "djedi" is the time interval of '24 hours'; the separate term was omitted in the original gismu effort because nearly all languages use the same word for both concepts, although most languages recognize that the meanings are distinct. The concept 'daylight' can be represented clearly as the tanru "sun-above" or "sun-high" ("solga'u" or "solgau"), but Bob believes that there will come to be many concepts built from the concept "daytime", enough that it should have a rafsi. Otherwise those compounds will be 3-part lujvo at minimum, and probably much longer. Bob notes that we have gismu currently for "evening", "morning", and "night", all of the same semantic group as "daytime". (There is also a question as to how to create culturally neutral definitions of these terms - for example, the word for "morning" describes a different time-range in Sweden and Germany than it does in the U.S.. This will also be discussed at LogFest.)

To avoid confusion between meanings, and possibly bad tanru as a result, Bob also proposes changing the keyword for "djedi" to "diurnal" or "diurne".

An argument against a word for "daytime" is that the same words are used for both meanings of "day" in all of our languages. Thus the algorithm will spit out "djedi" as the preferred form. But there are lower score forms on the list, possibly similar to "djedi". Bob notes that "djedi" currently has 3 rafsi, and if the word for "daytime" was close enough, one of these could be transferred to the new word.

4.-5. Bob proposes that "dawn" and possibly "twilight" (the evening equivalent of dawn) be considered based on similar analogy, noting that these are primitive concepts in many cultures. At the very least, we need good lujvo for each.

6. A similar situation to the "diurne"/"daytime" pair exists for "good" and "virtue". With the exception of English, the same words are used for both of these concepts. Tommy Whitlock thus argued successfully for not including "virtue" in our original gismu list. The corresponding was not true for "bad" and "evil", these were distinguished in the source languages (as shown by the widely differing words "palci" and "xlali").

As a result of the negation analysis, these concepts have again surfaced for discussion. A few concepts are being proposed for gismu as a result of that analysis, on the basis that the concepts are not well represented as negatives of their perceived 'opposite' in English, or that there is significant difficulty with assigning one to be 'positive' and the other 'negative'. In each of these cases, with the exception of "good"/"virtue", languages generally have words of different roots for each of the opposing pairs.

There are two related arguments for adding "virtue". One is based on place structures. The place structure for "xamgu" and for "xlali" are parallel: "x1 is good/bad for x2 by standard x3". "palci", on the other hand, has the place structure "x1 is evil by standard x2" - "evil" is an absolute concept in that it doesn't depend on who it affects.

Clearly, while the opposite of "good" is "bad" and vice versa, the opposite of "evil" is not "good"; this will confuse people who don't carefully read place structures, and lead to the same murky semantics that exists in the natural languages.

Tommy Whitlock has argued for eliminating "palci" and making both "virtue" and "evil" as tanru: "marde xamgu" ("madyxau") and "marde xlali" ("madyxla"). Bob feels that the place structure argument shows that this will only weaken the distinctive relativity of the concepts "xlali" and "xamgu", wherein a single event can be good for one person and bad for another, even by the same standard. It is easy to imagine philosophical and theological arguments where the place structure distinction is vital, and distinctive place structures are the core of Lojban's identity as a predicate language.

There are uncountable other examples where place structure arguments could be used to justify adding a gismu, but the dependence of entire fields of thought on the distinctions implicit in these particular place structures suggest that this is one case where a change is justified.

7. Another proposal based on the example of other languages is "tears", referring to droplets expelled in weeping. The concept has apparently been traced back to proto-language times. Lojban has gismu for "weep" and for "drop" and for "liquid". Thus all needed aspects can be expressed as lujvo. In English, of course, we have numerous connotations of tears, and a particular shape called 'teardrop'; it is unlikely that these connotations are universal, and tears are no more likely to take the shape named after them than any other kind of droplet on a surface. Thus we cannot claim any particular use in lujvo as a justification. The only solid argument is historico-linguistic. Is this sufficient?

8.-11. Four other 'opposites' are proposed based on the negation analysis. These are "ugly" as opposed to "beautiful", "diffuse" as opposed to "dense", "decrease" as opposed to "increase", and "deficient" as opposed to "sufficient". These, of all scalar negatives, seem especially common concepts in natural languages that justify gismu assignment. The first two are further justified by having separate roots in most languages. The second pair is further justified by analogy with "ckamu", as the 'less than' equivalent of their opposite.

12.-16. Lojban has assigned gismu to the major grains and staples of the world, and especially those of the 6 source languages. Various animals, plants, metals, and substances common enough in various cultures to have been assigned connotations and/or to be heavily used in metaphor in their languages have also been assigned gismu (in Lojban, cultural, as opposed to functional, connotations should not be referenced in tanru, unless figurative speech is explicitly marked - and even there should be avoided except in poetic usage).

Unfortunately, it has been pointed out over the last 2 years that three major staples not commonly referenced in English were omitted from the list. These are "buckwheat" which is historically and culturally the staple grain of Russia (one of our 'Big 6'), "sorghum" the staple grain for most of Africa used as animal feed in the U.S., and "cassava" which is the major staple of South America and part of Africa, and the third most planted crop after rice and wheat in the world, most familiar to Americans as the source of tapioca.

Considering the importance of staple foods to human life, and their use as ingredients in a wide variety of foods, Bob believes that these omissions should be remedied.

Two other grains have been mentioned, but are not as strongly recommended for gismu. A weaker case exists for "triticale" and "alfalfa", which are not major grains in any culture. But these should be considered for completeness of the set. In addition, readers familiar with other cultures should propose any other basic foods that are widely used outside the U.S. but are not well-known to Americans. Because most early Lojbanists are English-native Americans, Lojban's image of cultural neutrality requires that we 'bend over backwards' to avoid favoring concepts that are familiar to Americans while omitting correspondingly important concepts with which we are unfamiliar.

From John Hodges, on Readability - In a phone conversation with Bob, I said that I had noticed that JCB's Loglan looked a lot more like English text than Lojban did. JCB has periods at the ends of sentences, and capital letters at the beginnings, and so forth. I felt the appearance of Lojban was unaesthetic, and was casting around for possible changes to improve it.

More generally, the issue is 'readability'. I have some further thoughts.

I would like to retract the suggestion I made on the phone. I suggested switching the functions of the period and comma, because "it is less jarring to encounter a comma in the middle of a sentence". Looking over the text in JL10, I think it would be even more jarring to see a comma at the beginning of a sentence. Periods are visually neater than commas; we can get used to them.

I also suggested capitalizing the first letter of certain 'punctuation cmavo'. In the "Guidelines for Understanding Unfamiliar Lojban Text" on page 6, you start with "Put brackets around each sentence in the text." Capitalizing ".I" would do the same thing - though just putting two spaces before the ".i" would probably do just as well. By the same logic, we might capitalize the following:

"Ni'o", "Lu", "Li'u", "Zo", "To", "Toi", "Sei", "Se'u", "Mo", "Ma", and any word following "la" or "doi".

Anything in lexeme COI might optionally be capitalized as well. Punctuation exists to help the eye break up the text into meaningful units. I am suggesting that if punctuation may optionally be capitalized, just as 'r's may optionally be trilled, it would improve readability without seriously affecting audiovisual isomorphism.

Enclosed is a bit of text from JL10, before and after ...

Bob responds: First a bit of history. I used to capitalize the 'I' at the beginning of a sentence before the period was put in. Nora pointed out, though, that the sans-serif typeface used in The Loglanist rendered capital 'I' and small 'l' indistinguishable, and that this had occasionally caused problems in understanding.

Shortly thereafter, when work on Lojban got started, the visual pause (period) was proposed, along with the use of capitalization to indicate abnormal stress. The period is found at the beginning of every sentence. I also use a double space between sentences as a convention, though in our current compressed typeface, this may not be obvious. For a long time, periods were seldom found anywhere else, though as Lojban use has gotten more sophisticated, the periods have been multiplying.

I have to admit a preference for non-capitalization, since it means I seldom have to use the shift key when typing - hence fewer typos and inconsistencies. Typing Lojban text is not easy, because of its unfamiliarity, and I am especially worried about Lojban typos, other errors, and inconsistencies, appearing in the stuff I produce, since people depend on me for 'good' examples.

I don't, however, see any major problem, other than my own fumble-fingers and troublesome typefaces, with capitals on most of the cmavo proposed. There is the aesthetic consideration of having capitalization mean stress in some cases, and not in others; I find consistency more aesthetically appealing than a resemblance to English text. (Note that German capitalizes all nouns, both proper and improper, without confusion, and American writers do not capitalize after a colon or semicolon; capitalization is clearly not necessary to recognize sentence structure.)

For names, capitalization would be ambiguous, and lead to people with vowel initial names having stress on that first vowel. Thus a certain former US President would be pronounced /AH,yee,zehn,khau,r./ if written "Aizenxau,r." instead of /AI,zehn,khau,r./, resulting from "AIzenxau,r." And sometimes the default would not be capitalized at all, when penultimate stress were desired, as in some pronunciations of the English "Alicia" (Lojban ".alicas.")

As another historical note, there have been several times in Loglan history when people have proposed using punctuation marks as a substitute for cmavo, justified by 'readability'. Thus JCB favors French quotation marks for his equivalent of "lu" and "li'u" (French quotations, << and >>, unlike English quotes " are directional, as required for the two separate words). The list of proposals got quite long, and thus complex, though no clear approval was given to any of them.

I am opposed to such substitution, in that I think many people would have trouble learning to speak punctuation marks as words if they aren't made explicit. Forgetting the verbal punctuation, of course, leads to ambiguity. (We do allow symbols instead of words in MEX constructs, including periods, commas, and colons, as well as numerals. But in MEX, ALL of the words are replaced by symbols, not just some.)

Having stated this, as an alternative to John's proposal, I would suggest using the punctuation marks in addition to the words to enhance readability.

Specifically:

  • "zo" and "zoi" would use individual and paired quotes, respectively:
    zo "mi
    zoi .kuot." non-Lojban text ".kuot.
    (Athelstan devised the 'name' .kuot. as an obvious mnemonic for delimited quotes. Any single Lojban word can be used as the delimiter.)
  • regular quotation would use French quote marks:
    <<lu .... li'u>>
  • parenthesis of various types would use the parenthesis marks:
    (sei...) if the "se'u" is elided or
    (sei...se'u) if it isn't (to...toi)
    etc.
  • "ma", "mo" and other question words would be marked with a question mark, which I would prefer before the word to help remind English speakers that Lojban differs from English in not requiring a rising tone of voice to express a question, and to make it less likely that an unnatural 'end-of-sentence' pause will be inserted:
    ?ma
    ?mo
    ?pei
    ?xu

Because I've previously used a single question mark before a construct to indicate questionable grammaticality, I would have to change that convention (probably to a double question mark), but would prefer that to the question mark at the end of the word.

As an alternative for beginning of sentence or paragraph, how about something totally new to mark these (making the period unnecessary - remember that all periods are optional in Lojban, being that they can be inferred from the morphology rules.) A hyphen or equal sign would be distinctive.

Whatever the conventions, they should be optional, and a reader's understanding should not be reduced for not knowing them.

Having presented an alternative, let us look at some current text, John's alternative, and the one I've just proposed. I'm using a different text than John supplied, one from Athelstan's translation of Saki which uses more varied features:

Current style:

.i fo'e bacru lu la selke'i rirme'i cu roroi krici lenu ko'a ba xruti ko'a ca lo ba djedi .i ko'a noi se kansa le cmalu je bunre pangerku poi ke'a se cirko fa'u lenu ke'a kansa ko'a cu dzukla levi nenri fo leva canko ta'i le purci po ko'a .i la'edi'u cu krinu lenu le canko cu kalri ranji fi ro le vanci pagbu pe ze'o le ctebixtei

.i la selke'i je selnei rirme'i goi ko'u puta'e tavla mi lesu'u cliva ne pu'e lenu ge leko'u speni goi ko'e cu kansa ponse leko'e blabi je jacnalgre gackosta noi ke'a dandu leko'e birka gi la ranis. po'u leko'u citrai bruna goi ko'i cu sanga lu doi brtis. mu'i ma do plipe li'u noi roroi se sanga semu'i lenu zdifanza ko'u ku mu'i lenu ko'u xusra lenu le nunsanga cu fanza ko'u .i sei ko djuno be la'edei so'uroiku ca lei bifcau je smaji vanci poi ke'a simsa ti ko'u mi piso'aroi pencauji'i lenu ro ko'a ba dzugre leva canko li'u

John's proposal:

.I fo'e bacru Lu la Selke'i rirme'i cu roroi krici lenu ko'a ba xruti ko'a ca lo ba djedi .I ko'a noi se kansa le cmalu je bunre pangerku poi ke'a se cirko fa'u lenu ke'a kansa ko'a cu dzukla levi nenri fo leva canko ta'i le purci po ko'a .I la'edi'u cu krinu lenu le canko cu kalri ranji fi ro le vanci pagbu pe ze'o le ctebixtei

.I la Selke'i je selnei rirme'i goi ko'u puta'e tavla mi lesu'u cliva ne pu'e lenu ge leko'u speni goi ko'e cu kansa ponse leko'e blabi je jacnalgre gackosta noi ke'a dandu leko'e birka gi la Ranis. po'u leko'u citrai bruna goi ko'i cu sanga Lu doi Brtis. mu'i Ma do plipe Li'u noi roroi se sanga semu'i lenu zdifanza ko'u ku mu'i lenu ko'u xusra lenu le nunsanga cu fanza ko'u .I Sei ko djuno be la'edei so'uroiku ca lei bifcau je smaji vanci poi ke'a simsa ti

ko'u mi piso'aroi pencauji'i lenu ro ko'a ba dzugre leva canko Li'u

Bob's alternative:

=i fo'e bacru <<lu la selke'i rirme'i cu roroi krici lenu ko'a ba xruti ko'a ca lo ba djedi =i ko'a noi se kansa le cmalu je bunre pangerku poi ke'a se cirko fa'u lenu ke'a kansa ko'a cu dzukla levi nenri fo leva canko ta'i le purci po ko'a =i la'edi'u cu krinu lenu le canko cu kalri ranji fi ro le vanci pagbu pe ze'o le ctebixtei

=i la selke'i je selnei rirme'i goi ko'u puta'e tavla mi lesu'u cliva ne pu'e lenu ge leko'u speni goi ko'e cu kansa ponse leko'e blabi je jacnalgre gackosta noi ke'a dandu leko'e birka gi la ranis. po'u leko'u citrai bruna goi ko'i cu sanga <<lu doi brtis. mu'i ma do plipe li'u>> noi roroi se sanga semu'i lenu zdifanza ko'u ku mu'i lenu ko'u xusra lenu le nunsanga cu fanza ko'u =i (sei ko djuno be la'edei) so'uroiku ca lei bifcau je smaji vanci poi ke'a simsa ti ko'u mi piso'aroi pencauji'i lenu ro ko'a ba dzugre leva canko li'u>>

Other alternatives are possible. What do our readers think?

Jamie Bechtel raises an issue on conventions and terminology which may be related to the above:

How do you hyphenate in Lojban if the hyphen is used to show a buffer vowel? Can you use capitalization to indicate that a word has been cut in half? For example:

...selji-
tro ... 

would be written:

...selJI-
tro ... 

Bob responds: I'm hoping there is no terminology problem here, so let me be sure, and remind readers as well.

Hyphen - in Lojban, the letter 'y', or the vocalic 'r' or 'n' between rafsi used to solve junction problems in lujvo-making caused either by impermissible consonant clusters, or a possibility of the lujvo breaking up into two smaller words.

There is also a 'hyphen symbol': "-". This has no formal purpose in Lojban, though I may have suggested it as a possible way to explicitly show pronunciation of a 'buffering sound' between consonants that a speaker has trouble pronouncing together.

'Buffer' is used for the 'consonantal buffer', represented by the apostrophe, and for the buffering sound between consonants just mentioned. The latter sound has no letteral representation in Lojban, since it is by definition "some vowel sound not otherwise found in Lojban" (usually for English speakers the sound of 'i' in English "bit").

It is my belief that visual representation of buffers should be rare, and a word in which they are explicitly used should never be split across lines (I currently try to avoid hyphenating Lojban words at all, because no one reads the language well enough to easily put "seljitro" back together if it is split over two lines.) This would resolve the problem Jamie points out. As for explicitly marking the stress in such a split word - this is already legal in any word, although only in marking abnormal stress in names has it been found useful enough to justify the jarring appearance of capital letters in the middle of a word.

As for hyphenation at the end of lines, another alternative is the use of underscore. I recently received a letter from Italy (in English, but presumably using Italian conventions), where underscore was used for 'hyphenation', and there was no particular attempt to break words at syllable boundaries. Whether this is desirable for Lojban or not, clearly the English model is not universal.


John Hodges on Lojban representation of Time of Day: [This discussion is based on material in draft lesson 3 on telling times and dates, and may not be too understandable if you haven't read that lesson. John first wrote it last summer while the Blacksburg class was conducted, and recently updated it for LogFest consideration. Karen Stein also contributed to this proposal, and wrote on slightly different aspects, but John summarizes the issues well enough to enable people to prepare for LogFest discussions.]

John: Re. telling time: Your "modified base-12" system was not popular [with the Blacksburg class]. One problem is that it is not truly base 12, since it contains "gai", which is the digit for "12" in a base larger than 12. "12" in base 12 would be "li pano". Your names for the hours go from "la gaicac." to "la papacac." without any "la panocac." When we hear "la pacicac." how are we to know what base it is in? On the face of it, it could be any base larger than 3. Given the global use of base-10, even for the numbers of 12-and-24-hour clocks, I think it would be better to go with the crowd on this one. Long Live La Recicac. (la daucac. lir. and la gaicac. lec. and their kin are OK but I think are likely to be used only on flowery and formal occasions, the ones where we use Roman numerals today.)

Bob: Based on this argument, it is clear that a base-12 clock system should go from 0 to B (B is the base 12 digit for '1110'), and not from 1 to C as proposed in the draft lesson. The 24-hour clock system should go either from 0 to 23, or from 00 to 1B, base-12, which I strongly prefer.

The problem of knowing what base it is exists in English as well, but the difference between 12-hour and 24-hour clocks goes unnoticed, since people generally use the time form that others expect. You also have to know what time zone is referenced. I sure have been confused at times when someone has responded to "What time is it?" in Greenwich time. The answer is convention. If people use one standard all the time they won't get confused. When they don't use one standard, you need to flag base and time zone, just as we flag them in English. (Whenever I call someone in another time zone and leave a message on their recorder as to what time I called, I state 'your time' to indicate the time zone. Otherwise the time I give would be ambiguous.)

As for 'going with the crowd', which way does the crowd go?. Remember that in English we effectively use a base-12 system for numbers verbally - we say "twelve o'clock", not "one-two o'clock", or "twoteen o'clock" (this may not always be true for 24-hour clock users, who sometimes say "one-two-hundred" for "12 noon", which is clearly wrong). Only in numerical form do we use base-10 numbers, and then you can't do base-10 arithmetic with them. If we drop base-10 completely, as per the preferred form in the draft textbook, there is no confusion as you've indicated, arithmetic is straightforward, and there is no effective difference between the 12-hour clock and the 24-hour clock. If this is 'bucking the flow', it is only being simple and logical, as with the rest of the language.

John: Re. telling the date: The system you present on page 3-19 seems to break down because months typically begin in the middle of a week. What do you do with the fractional week at the beginning? Is it the first week of the month? e.g. 1 August 1989 is a Tuesday; is it the first day of the first week of the month? You give an example 2:2:9:5:1989, "the 2nd day (Monday) of the 2nd week, (which is the 9th), of the 5th month (May) of 1989". If you look, 9 May 1989 was a Tuesday. (It could be worse. 1 July 1989 was a Saturday. How do people count "weeks of the month" nowadays anyway? When a "month" was the time between one New Moon and the next, it was easy; see which quarter the moon was in. It might give you some 8-day weeks, but what the hey.) I believe people say "the third Tuesday" rather then "Tuesday of the third week."

I suggest not attempting to count "weeks of the month" at all. 3:9:5:1989, a set of four numbers, is easier to hear than a set of five. (Karen said five was too long a string.) 3:9:5 and 9:5:1989 would usually be understood. Sets of two still leave you guessing between day-of-week/day-of-month and day-of-month/month. A bare set of two numbers could also be a time of day. You mention this problem on page 3-15 and offer possible approaches but set no standard.

HEY, I'VE GOT IT! Declare a convention that:

2 digits:2 digits is a time
08:56

1 digit:2 digits is a Day-of-week/Day-of-month
4:06 =Wednesday the Sixth (default assumption)
This may be overridden if the two digits are obviously a year
5:89 =May, 1989

2 digits:1 digit is Day-of-month/Month, using d,f,g for dau, fei, gai
05:9 =Fifth of September

1 digit:2 digits:1 digit is Day-of-week/Day-of Month/Month
7:03:f =Saturday the Third of November

2 digits:1 digit:2 digits (4 digits optional) is Day-of-Month/Month/Year
02:2:92 =Ground Hog's Day, 1992
08:56:3:09:5:89 =Eight-fifty-six A.M. Tuesday the 9th of May, 1989

3 digits:2 or 4 digits is day-of-year/year
001:95 =New Year's Day 1995

08:56:3:129:89
Clear enough.

Bob: Yes, weeks of months normal include partial weeks. We don't often express such dates in numerical form, as is possible in Lojban, but we do count and use them. Anyone who ever has meetings on the '4th Tuesday of every month' knows to count all Tuesdays, and the week of month convention was added to deal with such problems. I will agree that the existing proposal is cumbersome and confusing, though. I don't like some aspects of John's specific proposal, though.

  • After the year 2000, it will not be clear for many years whether two digits are a year, a month, or a day, so this convention is doomed during our most critical growth period. I think we should break with English a decade early.
  • Most speakers will dislike leading zeroes - this is 'bucking the system' as much as my base 12 proposal, only more complex, because there are a lot of patterns to remember. Unlike with time, a given person will have uses for all of these forms at some time, and cannot limit himself/herself to just one.

I will suggest a skeleton counter-proposal for consideration. Come up with a canonical form, leaving out the week-of-month element.

Try: Day-of-Week:Day-of-Month:Month:Year

If less than 4 terms are given, the stated ones are the leftmost of this set. You can omit one of the early terms leaving the colon "pi'e". The behind. Special cases of week-of-month and day-of-year would be inserted at any logical place when desired, but must be marked with a lerfu "jy." for weekday (jeftu djedi) or "ny." for day-of-year (nanca). If we use "ny.", it should replace both the day-of-month and month fields in the canonical form, since it is month independence that is the intent of the usage; this prevents us from having to express two useless colons every time. I'll also add "dy." to stand for an arbitrary non-year-based Julian Day (by whatever convention - like time zones, you need to know the convention in use to interpret the date. Astronomers use one standard, and various businesses and military projects each use their own separate one; the lerfu stands either for "djedi" or "djulien.")

John: Postscript 4/27/1990... looking at this nine months later, I would support giving names to the days of the week, so that they can be abbreviated with letters instead of digits. Padjed. redjed. cidjed. vodjed. mudjed. xadjed. zedjed.... so the long string of digits above is broken by a lerfu in the middle, e.g.,

16:37:x:27:4:90 4:37 P.M. Friday 27 April 1990

This would be easier to read than a string of six digits, and would reduce confusion over which digit meant what. Bob: This would be grammatical - you can have a bare lerfu in a numeral position, but would be difficult to use. Remember that in even a moderately noisy environment, lerfu are nearly indistinguishable, so we can't use them too heavily in verbal communications (their main purpose is for spelling, which is slow and enunciated, and for numerical variables, which tends to be few in number or are subscripted - John Cowan has noted the need for long-name forms of the ALPHA-BRAVO-CHARLIE genre for noisy environments, and names couldn't be used inside of a numeral string.) Requiring a difficult to distinguish day-of-week letter every time you want to state the day-of-week seems to be asking for resolution difficulties. I still prefer my null:dayofmonth... idea.

If you want to use time/date combinations, you need to label the final position of the time, or you can't tell whether it is minutes, seconds, etc. You could use "t." for "tcika". I think combined date/times will be used only in specialized circumstances, just like the other two conventions I've proposed that require lerfu, and thus won't cause confusion. The four lerfu I've proposed are probably close to a maximally diverse phonetic set, which is convenient for the noisy environment problem, though I think varying position of the lerfu-label within the various colons will provide redundancy for noisy environment resolution.


Lojban as a Language Template

by James A. Yorke and Celso Grebogi

Abstract: We propose creating a series of languages using Lojban as a template. These "image" languages would be identical to Lojban except that the predicate words would be adapted so as to be based on the language of the target population. For each and every predicate in Lojban there would be exactly one predicate in the image language, and it would have exactly the same meaning as in Lojban. These "image languages" would be trivially translated into each other trivially by substitution. Continued work on Lojban would of course be critical since Loglan is the template. No development work would be carried out for the image languages. All official changes in Lojban would by implication be immediately adapted to the derivative languages. The advantages would be to minimize memory of those parts of the language which do not carry structure, allowing concentration on grammar and short words.

The arithmetic of memorization:

Lojban: 300 cmavo, 1000 difficult gismu, 2000 easy rafsi

Image language: 300 cmavo, an easy 1000 gismu.

These languages would be immediately translatable into each other and would be stepping stones, allowing the students to get more quickly to the point where they can think in the language, the next step being the learning of the Lojban predicates. The easier entry should allow many more people to get involved in Lojban's logical structure, and the goal would be to rapidly swell the number of people interested in Lojban/Loglan. A major goal of Lojban should be to build up public awareness of the language and increasing the number of people involved in Lojban (and minor variants).

Technology continues to make communication around the world easier. A person with a PC or MAC on a computer network types out a message to a correspondent and it is received around the world, sometimes with a delay of a day but often in seconds. Such ease of communications with an electronic interface creates opportunities.

Automatic computer translation between languages does not appear feasible in the foreseeable future. One particular problem is that words in the source language have many meanings and the speakers and listeners are able to select one from the context. Another problem is that the source language and the target language have completely different structures. Undoubtedly the occasional ambiguity of the grammars creates additional problems.

Suppose a group decided to establish a family of languages designed so that translation between them would be easy. What would that family look like. We are going to call the version for speakers of English "Logenglish" and other versions might be Logmandarin, Logportugese, etc. We will call these languages "images" or "image languages" since they are exact images of the original. The vocabularies of Logenglish would be selected based on English in order to ease the learning problem. On the other hand the words should not be English words since their grammatical use and exact meanings would be different and the user should not be lulled into thinking that he or she was using a trivial variant of English. (A problem with the image language approach is that the student might confuse what is a legal Logenglish predicate word and what is not, but when eventually learning the full language, it is unlikely that he/she would confuse Logenglish words with Lojban, because if the word is recognizably like English, it is almost certainly not Lojban.) The vocabularies of other image languages would be based on the native language. In principal anyone could create his or her own set of words to correspond to the Lojban. However, it is likely that a group such as a high school class would speak the language to one another and so would have to have a single image.

The initial goal would be have the structures of Logenglish (i.e., Logical English) and Logportugese so similar that a computer could translate from one to the other trivially so that a Logenglish message typed into one terminal could be translated into a universal form by the terminal and transmitted and the receiving terminal could virtually instantly translate the universal form into Logportugese or Logswahili. The sender would not have any reason to know the dialect of the receiver.

A second more distant goal would be to have these languages created so that when spoken they would be easily parsed into words by computer so that when a sentence in Loghindi was spoken into a computer controlled telephone, the computer would translate the sounds into letters and words of the universal language and transmit the universal form, which would then be transmitted to the listener after being transformed into Logspanish of Lognavaho or whatever was needed. If the family of languages had a one to one correspondence between words and the word ordering was the same, the spoken words could be transmitted word by word (with electronic indicators of tone and pitch and inflection and timing) and the listener would here those words translated into his or her own variant as they were spoken by the speaker, word by word.

Assuming these goals were set for a family of languages, then what would the design look like. The universal form would be a template. Words would have meanings that would be invariant for all the dialects. We would like to suggest that Lojban is an excellent candidate for such a template. The image languages would not have to have the same set of letters and sounds as in Lojban but predicates should have the Lojban prescribed pattern of vowels and consonants, (e.g. basic predicates of the form CCVCV or CCVCV). In our efforts here, we have assumed the permissible vowels are the same as for Lojban and have the same pronunciation. Such a choice makes it easier to use the image language as a stepping stone in acquiring Lojban. The image languages should be designed to maximize the ability to learn the vocabularies. The truly important constraint is that there must be a clear one-to-one correspondence between words, including compound predicates. Hence the universal language (Lojban) would continue to be developed and each addition or modification would be reflected in the image languages.

For the beginner, the terminal might be quite helpful, with an on-line dictionary suggesting Logenglish translations of English words, and it could parse the sentences and warn of ungrammatical constructions that violate the rules of Lojban. Such programs would have to created. Some, such as the parser, would be the same for all the languages, since it could be executed after the translation into Lojban. The basic translation program would just substitute words from one language to the other, based on a list of pairs it has in memory. It would be much simpler than the existing parsing programs.

The creation of Lojban from Loglan (a registered trademark of Dr. J. C. Brown) has demonstrated how the words can be changed, and how harmless it is to substitute new spellings. Lojban has gone much further in continuing the development of the language. No such development would be possible for the image languages since it is critical that they be exact template projections of Lojban in order to guarantee trivial translation between the image languages.

The following predicates indicate how one might proceed, but each image language should have rules of thumb describing how it handles certain situations such as how to expand a single syllable word in English into a two syllable predicate, from "bulb" to "bulbu" and "grow" into "groro" for example. Here we need to create a final vowel since the original words have only one and we have chosen to repeat the existing vowel. We suggest that when possible, Logenglish should stay close to the spelling of English (within the constraints of Lojban's rules for C/V patterns) even when the pronunciation changes significantly. In Logenglish, all letters are pronounced including a final "e".

Hence:

"cattle"
becomes
"catle".
"The pretty little girls school"
might become
"le preti litle girli skolo"

Of course the Lojban grammar is unambiguous so that "preti" modifies "litle", which modifies "girli" which modifies "skolo". We have retained the short word "le" (from Lojban), since we believe that little would be gained in trying to get little words that correspond to English. In fact the little words represent structures that are quite different from English. In learning short words, the student would be putting the effort into learning aspects of the grammar that are fundamentally different from his/her native language. The effort in learning the image language would make it easier to progress to Lojban, having already learned the grammar and short words.

In choosing words for the image languages, we have chosen to keep the spelling as similar as possible to the natural language, keeping in mind the pronunciation would probably be following Lojban rules. That would emphasize to the student that this is not his/her native language. Notice in any case the spellings in the image language are rarely the same as for the natural language.

SAMPLE ELEMENTARY PREDICATE WORDS (=gismu IN LOJBAN)

The words below are from the beginning of the "baselined gismu list" and hence all begin with 'b' since gismu begin with consonants. The table illustrates that many gismu can be created in image languages with spellings recognizable to a speaker of the language. Others like "badna" ("banana") are not easy to improve on. The Logmandarin words were created with the assistance of M.-Z. Ding and Y.-C. Lai.

Parentheses show the word or phrase being approximated.

Lojban     Log-   Logpor- Log-     Log-                    
    english	      tuguese spanish  mandarin                
------     -------	      -------  -------	 --------      
bundle/package containing ...                              
bakfu      pakje  fardo   bulto    bogva                   
           (fardo)	      (bulto)  (bogwa)                 
cow/cattle/ox/beef					   
bakni	   catle  gadjo	  gnado	   nuxnu		   
	   (gado) (ganado)    *(female                     
			  cow)				   
chalk in form of ... from source of	...                    
bakri      tcaka  jitsi   tizja    fenbi                   
           (giz)  (tiza)  (fenbi)                          
                                                           
bucket                                                     
baktu      bukte  balde   balde    tongo                   
           (balde)	      (balde)  (tong)                  
                                                           
plant bulb of species ...                                  
balji      bulbu  bulbo   bulbo    falgi                   
           (bulbo)	      (bulbo)  (hui gin)               
                                                           
balcony                                                    
balni      balko  balko   balko    yangta                  
           (balcao)	      (balcon) (yan tai)               
                                                           
blade of ...                                               
balre      blade  filjo   filjo    davpi                   
           (fio)  (filo)  (dao pi)                         
                                                           
future of ...                                              
balvi      futre  futro   futro    junla                   
           (futuro)	      (futuro) (jung la)               
                                                           
protect...from...by...under                                
bandu      prote  prote   prote    bavfu                   
           (proteger) (proteger)   (bao hu)                
                                                           
*"mumnu" is	"female	cow" while "nu"	is non-sexed "bovine"  

In Logenglish the letter 'x' could be reserved to indicate where the constructed word does not correspond to the English. Hence English "apple" might be "xaple" in Logenglish. We used "tc" to represent "ch" since the spelling of "ch" in English is "tc" in Lojban.


COMPOUND WORDS (lujvo) FROM PARTICLES OF PREDICATES (rafsi)

Since Lojban is the template, it remains critical to continue development of compound words in Lojban. While users could make up compounds on the fly as they are encouraged to in Lojban, the only official compound words in any Log-language would be those that correspond to those officially sanctioned words in Lojban. For the purpose of minimizing the required memory, we propose that there be only one acceptable rafsi for each gismu in derivative languages, namely the fragment formed by dropping the final vowel. To go from a compound Lojban word to the corresponding word a derivative language, break the compound into rafsi, e.g., R1 R2 R3, and then list the corresponding predicates, P1 P2 P3, and translate them into the derivative language, yielding D1 D2 D3, cut the final vowels off D1 and D2 and put the result together, as one would in Lojban, adding hyphen 'y' as required [ed.: which is between each pair of rafsi, as the proposal is stated - the canonical 4-letter form is best described as replacing the final vowel by a 'y'].

The paucity of fragments will make pronunciation a bit more awkward but it will drastically cut the list of items to memorize. The objective here is to minimize the memorization of parts of the language that do not correspond to really new grammatical structures.

By providing the beginner with predicates that have a rather recognizable spelling, we would making easy to get into the first level of the language. And it would be easy for people of every language group. If someone wanted to make Lojban available in Tibet, they would undertaking a major project, since they would have to produce instructional material in the image language. It would a relatively easy extra step to create a translation list of 1000 basic predicates.

While many people will tell us it is not difficult to learn the almost 1000 gismu, apparently not many have in fact mastered all the gismu and the short words including the digits and so on. There is another thousand plus rafsi to memorize in Lojban, while all of the rafsi would be automatic in the image languages.

FINAL THOUGHTS

Lojban helps us understand the nature of thought and the nature of language. It is a wonderful opportunity to explore these ideas that are so fundamental to the nature of human beings, and we feel Lojban should be of interest to an immense audience for this reason.

The first question about the usability Lojban is whether people can think in the language. Users would really like to know what thinking in Lojban is like. We suspect that limited vocabularies prevent most people interested in the language from ever finding out and the image language idea should allow people to expand their vocabularies rapidly. Logenglish and Logportugese and other image languages could help people find out. When people learn something, they like a learning curve that gets them to a operational level quickly. (It will take some time to see if it is possible to restrict each word to a single meaning).

The learning curve should allow them to progress incrementally, so that learning a bit more allows them to do a bit more. Currently Lojban presents a huge memory requirement that has little to do with the required learning of the grammar. If people want to learn to speak the language, we should make it as easy as possible, (and in particular we should not burden them with technical terms like "bridi", "valsi", "brivla", "gismu", "kunbri" (obsolete), "cmavo", "glico", "lujvo", "rafsi", "tanru", "selbri", and undoubtedly others).

By making relatively easy to get fluent in a dialect of Lojban, we could get a larger number of people who could experience the mind-bending aspects of Lojban (as opposed to the purely painful aspects). By having people communicating between Logenglish and Logmandarin, we could rapidly see what the differences are in the way the language is used. We could more rapidly demonstrate that there is something to test. We would greatly increase the number of candidates who might want to take the next step of learning the Lojban predicate vocabulary.

By having an array of these image languages, we would keep the supremacy of Lojban clear.

It would seem wise to limit those criticisms of image languages that are made on the grounds that the similarity to native languages would cause confusion, since it is often claimed that Lojban and Loglan and both highly recognizable. While those two are clearly less recognizable to the student, the requirements of testing Sapir-Whorf suggests that any similarity is scientifically problematical.

We would like to thank Bob LeChevalier for his comments which have helped us clarify the issues.


Bob: I'll try to respond with brief summaries of some points that I've raised in objection in the comments they mentioned (my actual responses to their drafts were several pages long). First a general objection, though:

The basic assumption behind this proposal is that memorizing all of the Lojban cmavo, gismu, and rafsi is necessary to get "into the first level of the language", in their words. It isn't so. For the "first level" we would never recommend anyone learn other than the long-form rafsi, which are made the same way as in their proposal. Also, to learn the basic grammatical structure required to 'think in the language', you don't need more than several dozen gismu.

However, conversation and letter writing are not 'first-level' skills. Nor would learning the basic structures or even a thousand gismu allow you to 'think in the language' for real. You can only get a taste of the experience until you can remember or conjure up words for the entire variety of human experience that you wish to deal with; the average English speaker has a vocabulary between 20000 and 100000 words. You can 'think' in a foreign language with a frac-tion of those numbers, but it isn't a tiny fraction. For thinking, conversing, or writing in a new language, it isn't practical to 'look up in the dictionary' to find suggested Logenglish words for your English words, even when it would be a straight word for word substitution - which it usually wouldn't be.

Thus we have two conflicting goals. If you want the 'quick gratification' of a survey of the grammar - enough to learn about the perils of English grammar and the nature of language, you can use Logenglish and the other images, or you could just as easily use Lojban with real English 'predicates' - the specific meanings of the words is not as important for such a purpose, and English speakers could get used to a pidgin that did not distinguish parts of speech easier than they could memorize any set of words well enough to communicate.

But if you want to use Lojban (or a relative) as a language, for the purposes of a language, you will need to work quite a bit harder, and the advantages of an intermediate image language are lost.

Specific (and minor points):

  1. Many of the specific proposed image words do not meet the Lojban rules for permissible initials an medials. Making these lists up will not be trivial. But if someone wants to volunteer, and then get a few people to volunteer learning the words to try an image language experiment, fine. But I doubt if we can justify any major effort by la lojbangirz. as an organization along these lines in the near future.
  2. Teaching to people in Tibet will still require writing a textbook and dictionary in Tibetan, whether image languages are used or not. I therefore don't see any savings due to image languages; you actually need to do more work, since you additionally have to invent new gismu and redo all the Lojban words in the books to fit the new image.
  3. People who know only the image languages can talk only to others who know the same image, unless they are computer aided. Computers are not yet sufficiently ubiquitous, especially outside the U. S., to make intercultural computer communication anything but a toy of the elite.
  4. There is a reference objecting to the Lojban terminology we use in teaching the grammar, and Dr. Yorke has proposed using metaphorical English instead (such as "scene" for "bridi", "script" for "selbri", and "role" for "sumti"). I personally think more people would be confused by this, since the might not see the analogy, but we can mention such analogies in the textbook as alternative teaching aids, if readers think they help (can readers determine the analogy without further explanation - does it clear anything up for your understanding?). It is certainly not Lojbanic to use figurative metaphors like these - Lojban tanru are as analytic as pragmatically possible.
  5. Regarding recognition of Logenglish apart from Lojban, they say: "If a word is recognizably like English, it is almost certainly not Lojban". Truly a cynical observation. Many Lojban words remind people of English words, but not always the one closest to their meaning.
  6. The only way to talk intelligently about this proposal is with an example involving more than just isolated words. I'll use the same Saki excerpt used for examining John Hodges writing convention proposals. The following are three versions of the same passage, using alternatives forms for the Logenglish brivla (and some of my readability proposals from that earlier text).
    1. the text using Logenglish words derived as closely as I can guess to the algorithm suggested (which was loosely stated, and not followed exactly in all of the sample words), including using 'x' where English words do not have a necessary consonant (or a 'v' 'z' or 't' if 'x' cannot be used and a voiced or unvoiced consonant is needed), reusing the vowel if the English doesn't have two, and otherwise stressing visual recognition; I'll also use only the key word to maximize the reader's chances of guessing.
    2. a variant version, still maximizing visual recognition - but trying to devise less structured ways to deal with recognition problems of each word. In some cases, I use a synonym of the key word. (In neither a. nor b. did I check carefully for words that would conflict if final vowels are dropped. In a., however, I had to use "accompany" instead of "with" for "kansa" in order to get a non-conflicting word without great gymnastics.)
    3. using English words, tied together with a visible hyphen which is to be pronounced as /yuh/ (this syllable is one proposal for hyphenating le'avla), or prefixed with a hyphen so they don't get read as Lojban words. Nora's Loglan translator involved a step something like this, but added standard suffixes to convert between nouns, verbs, and adjectives.

Now, how many of the Logenglish substitutes can you identify without going to the third passage? Do any of the substitutions help you recognize the meaning of the passage more easily? Anyone who wishes is welcome to try doing better at coming up with good Logenglish words for each gismu in the passage.

a.:

.I fo'e xutra <<lu la selpitxi parnysiste cu roroi belvi lenu ko'a ba retru ko'a ca lo ba daxta .I ko'a noi se xacpo le smala je bronu spanydogvo poi ke'a se losxe fa'u lenu ke'a xacpo ko'a cu valkycomxo levi xinsi fo leva vindo ta'i le pasta po ko'a .I la'edi'u cu resno lenu le vindo cu xopne conti fi ro le xevne parta pe ze'o le nigvybetnytimxe

.I la selpitxi je selxinsi parnysista goi ko'u puta'e talka mi lesu'u levza ne pu'e lenu ge leko'u sposu goi ko'e cu xacpo xonxu leko'e xitxe je vatrynalpatru covrycotxo noi ke'a xanga leko'e xarma gi la ranis. po'u leko'u xongysupre brote goi ko'i cu singi <<lu doi brtis. mu'i ma do lipxi li'u>> noi roroi se singi semu'i lenu xamzyxanxo ko'u ku mu'i lenu ko'u xasre lenu le nunsingi cu xanxo ko'u .I sei ko konvo be la'edei so'uroiku ca lei brezyvitxo je silna xevne poi ke'a simli ti ko'u mi piso'aroi tinkyvitxyxopni lenu ro ko'a ba valkypatru leva vindo li'u>>

b.:

.I fo'e speka <<lu la selpitli parnysista cu roroi belvi lenu ko'a ba retru ko'a ca lo ba dande .I ko'a noi se copni le smalu je brona spanydogji poi ke'a se losna fa'u lenu ke'a copni ko'a cu valkycomgo levi ninsi fo leva vindo ta'i le pasta po ko'a .I la'edi'u cu justi lenu le vindo cu kopne conti fi ro le nevne parta pe ze'o le nignybetnytimne

.I la selpitli je selninsi parnysista goi ko'u puta'e talka mi lesu'u levna ne pu'e lenu ge leko'u sposa goi ko'e cu copni ponsu leko'e blino je vatrynaltrugo covrycotsa noi ke'a xanga leko'e barma gi la ranis. po'u leko'u zungymosta brote goi ko'i cu singe <<lu doi brtis. mu'i ma do jumpa li'u>> noi roroi se singe semu'i lenu muzdyrirta ko'u ku mu'i lenu ko'u serta lenu le nunsinge cu rirta ko'u .I sei ko konvo be la'edei so'uroiku ca lei brizyvitno je silna nevne poi ke'a simla ti ko'u mi piso'aroi tinkyvitnynopni lenu ro ko'a ba valkytrugo leva vindo li'u>>

c.:

.I fo'e -utter <<lu la sel-pity -parent-sister cu roroi -believe lenu ko'a ba -return ko'a ca lo ba -day .I ko'a noi se -accompany le -small je -brown -spanish-dog poi ke'a se -lose fa'u lenu ke'a -accompany ko'a cu -walk-go levi -in fo leva -window ta'i le -past po ko'a .I la'edi'u cu -reason lenu le -window cu -open -continue fi ro le -evening -part pe ze'o le -night-between-time

.I la sel-pity je sel-in -parent-sister goi ko'u puta'e -talk mi lesu'u -leave ne pu'e lenu ge leko'u -married goi ko'e cu -accompany -possess leko'e -white je -water-nal- pass_through -cover-coat noi ke'a -hang leko'e -arm gi la ranis. po'u leko'u -young-superlative -brother goi ko'i cu -sing <<lu doi brtis. mu'i ma do -leap li'u>> noi roroi se -sing semu'i lenu -amusing-annoy ko'u ku mu'i lenu ko'u -assert lenu le nun-sing cu -annoy ko'u .I sei ko -know be la'edei so'uroiku ca lei -breeze-without je -quiet -evening poi ke'a -similar ti ko'u mi piso'aroi -think-without-opine lenu ro ko'a ba -walk-pass_through leva -window li'u>>

Letters, Comments, and Responses

from David Morrow

By the way, has anyone tried to train animals using Lojban? Perhaps it would turn out more effective then natural languages for communication maybe with highly developed types like primates or those who are intelligent and very different like whales.

Bob: Not really, although I talk to my cat in Lojban, and he sometimes seems to know a few words - but it may be my tone of voice or the similarity to English words that does the trick. There aren't all that many people working seriously on training primates and dolphins, which are likely to be the only animals intelligent enough to deal with grammar at all.

David: Also, it seems to me that history can show whether Whorf-Sapir is right, though I'll admit the evidence is open to interpretation. However, the evidence seems to show that language's effect on thought is not so ironclad. Notice that highly inflected Latin and West Teutonic (ie, English) changed during the Dark Ages to predominantly analytical grammars that remain today. This might be ascribed to the changed mindset between Classical and Medieval civilizations, but German in its homeland remained inflected, and non-Indo-European languages like Basque and Hungarian (or rather their speakers) all participated in Medieval culture. Further, Modern civilization differs profoundly from Medieval and even Renaissance culture despite the fact that we could probably converse with Chaucer or Shakespeare. Indeed, our civilization probably resembles Roman and Hellenistic civilizations - at least in attitudes and philosophy - much more than it does the worlds of Alfred or Luther. Of course, we could view the attempts to develop artificial languages as a result of our outlook...

Bob: I read a history of English not long ago. Researchers believe that English inflection was lost for a much different reason than you propose. Specifically, England was conquered twice in less than a century, first by the Danes, then by the Normans. The Danes spoke a North-Germanic language which was quite similar to West-Germanic Old-English in vocabulary and grammar. But the inflection pattern was different for most words. As the two peoples merged, the Danes learned English easily, but not the declensions. The same thing happened when the Normans conquered England - they generally didn't learn English, so English borrowed heavily from Norman French. But French also had different, incompatible declensions, and differences in gender, thus reinforcing the confusion when English speakers learned French. These two languages were so mutually incompatible that eventually Anglo-Normans reverted back to English when their ties to France weakened.

David: Still, I don't think thought is as constrained by language as might appear. Poetry struggles to overcome that limitation; most likely social factors constrain thought, like being burned at the stake for saying the Earth is a sphere.

Bob: I think the Sapir-Whorf constraint, if it exists, is of a different nature. I think that the constraints of grammar may hinder our ability to manipulate concepts that are encoded into particular grammatical boxes. Where the hindrance is too severe, we avoid pursuing those lines of thought unless they are particularly fruitful and useful in the short term. Individuals, such as poets and geniuses like Einstein, may break through the barriers for one or two concepts, but the society as a whole cannot follow in their footsteps.

from Jamie Bechtel:

"Nomic" is a game designed to mimic the way a government works. All the rules of Nomic are subject to change, and the initial set of rules for any game of Nomic simply tell how to go about changing the rules. After a short while of playing Nomic, the game can easily be changed to the point where it is no longer the same game. Whether a person is playing by the rules or not, is determined almost entirely by judgement, which need not be bound by rules of precedent. Any game of Nomic can quickly expand to include all aspects of life. A description of Nomic can be found in the article "Nomic: A Self-Modifying Game Based on Reflexivity in Law" in the June '82 Scientific American. It can also be found in Metamagical Themas by Douglas Hofstadter.

A possible way to test Sapir-Whorf would be to create an artificial "game-culture" and test the influence of various languages on its development. It seems to me that a culture can be thought of as an extremely complex game that extends to all aspects of social interaction. Of course, culture differs from most games in various ways. Cultures have little distinction between the rules that compose the "games", and the rules of skill (social skill). The rules of cultural-games are also constantly under change, (especially between individuals). Also, in natural cultures, it is not always clear when one is acting outside the cultural rules. But Nomic does have these traits.

To test Sapir-Whorf (at least its influence on culture) you might create a flexible self-modifying game (like Nomic) that closely mimics most aspects of culture. You could then have several small populations of different linguistic-cultures, each starting out with the same game-culture, and then eventually developing their own. (The games will probably change the languages too - maybe just pragmatics). Testing the effects of various languages on these artificial game-cultures would be simple, since you would be able to see the change of the game's rules (since you could know the rules they started with).

from Jim Carter

[Editor's Note: These comments were received and replied to via the network mail, and have been slightly revised to make them more clear to the general reader. In some cases there is non-standard technical vocabulary, and references to things Jim Carter has worked on apart from Lojban. Enough of the comments are relevant that we've chosen to print them with minimal editing, and without trying to add much extra explanation.
Some of the comments are in response to things in this issue (specifically the LogFest plans), and some are responses to items in JL10. Bob responds to some, but not all, of the comments.]

Jim:

1. Responding to Bob's: "What short term applications for Lojban are worth highest priority?"

  • As a substitute for Pidgin English in linguistically impoverished areas (not exactly short term).
  • Developing a parser-organizer -- not just a parse tree, but identify the antecedents of pronouns, which phrase ends up in which case, etc.


This has been extremely valuable to me in developing - gua!spi [ed. note: Carter's derivative of Loglan], as it reveals usage problems which are not at all obvious to an unaided human.

  Approximate distribution of code lines:		   
Data structures, link-list subrt., etc.	  1000		   
Word-related basic subroutines		  1000		   
Reading	dictionary, lexer, parser,			   
  caselinks				  800		   
Arguments, infinitives,	compounds,			   
  replication				 1200		   
Modal phrases				   300		   
Pronouns, modal	antecedents, proper names 1400		   
Printing output				  1000		   

2. Replicated cases -- does Lojban have them? Does Lojban want them? For example, in "I want to swim", who's going to swim? You or I? This is an example of a replicated case. Replicated cases have turned out to be much more important for -gua!spi than I originally anticipated.

Bob: If I understand your referent, these are usually ellipsized in the natural languages, and are optionally so in Lojban, although in Lojban it is far more obvious that an ellipsis is taking place. Thus "I want to swim" is "mi djica lenu limna", where limna is a bare selbri with the first place 'mi' ellipsized, meaning literally "I want the event of '... swims ...'", or "I want the swimming". You can of course be explicit with "mi djica lenu mi limna", and we tend to do so in writing for new people. If the replicated place is more complicated and you don't want to repeat it, we have a couple of anaphoric ways to back reference the earlier place (or forward reference the later place - a very un-English thing to do).

Jim: Yes, Loglan abstractions lack arguments. But when you look carefully at what you (I) actually say, I find that literally 99% of the abstractions need an argument. Some gismu have cases which are usually occupied by abstractions, like djica x2, and these almost invariably need a copy of a main phrase argument (x1 for djica). Of course variants occur, as in "mi djica lenu do limna". But the variants are much less common. Even rarer are open first arguments; in -gua!spi I tell the user to say "^:i !ji /daw !suy !jl", where "jl" is the anonymous variable. But this most rare form is the default in Lojban.

In modal phrases it turns out that most (maybe 80%) also need a replicated argument, but it's much more variable than in abstractions; it's about equally divided between "previous sentence" (in discursives), x1 of restricted phrase (in restrictive sub-clauses and many subordinate assertions), and "me" (for the rest of the sub-assertions). I have some rules worked out that are about 98% effective at getting the right replicated argument, with very little cueing from the user.

Bob: I may be missing something here - I think we're saying the same thing. In normal conversational use in Lojban, people use a lot of abstractions. But normally, the x1 place of those abstractions is totally obvious from context; in fact, even without context, generally it is the x1 of the main bridi. The speaker is always permitted to omit 'obvious sumti' as ellipsis, provided that the listener will understand the implied reference. Thus in Lojban usage, the replicated x1 place will often be left out. (But it can be referenced easily with a single cmavo pro-sumti if you really want to be explicit.)

Now, as to other-than-x1 places of the abstraction, these tend to be specified about as often as they would be in the main bridi, and therefore might use replicated places. But again, our pro-sumti system handles this.

In Lojban, discursives and 'modal phrases' (our sumti which are tagged with a sumti tcita, I think), are grammatically distinct from simple abstraction sumti, and a difference in the default interpretation of the ellipsis is less likely to be a problem. For subordinate assertions in relative clauses, we have the pro-sumti "ke'a" to refer outward to the main assertion, but this can usually be ellipsized, too, just as in English ("The house that Jack built." = "The house that is built by Jack." = "The house such that Jack built it." The latter is the translation of the Lojban phrasing of the colloquial English.)

By the way, Institute Loglan does allow places on abstraction - they just make a grammatical distinction between an abstraction optionally with sumti, and one without sumti. This distinction is made in violation of audio-visual isomorphism ("lepo [mi] godzi" vs "le po godzi"), where in oral speech you cannot tell the intended grammar of the "po".

Jim:

3. Responding to Bob's: "We also will have the goal, by the end of the weekend, of deciding on several last minute proposals regarding the language and declaring a conditional or unconditional baseline (freeze) on the grammar."

- Hooray! At last, an official Loglan grammar! (Now you can proceed to work on the organizational syntax level.)

Bob: If you mean the syntax of greater than sentence length constructs, it is already in the grammar. If you mean questions of anaphora interpretation, they're already defined.

Jim: You wouldn't believe how hard I had to work to get "the speaker" and "the listener" identified consistently, as well as "the reply sentence". And that was only a small part of the organization work.

Bob: Ah! I think you are talking about computer recognition of anaphora in a multi-party conversation, given no clues as to who is talking. That isn't a grammatical problem, but rather an artificial intelligence problem.

With live speakers, this isn't a problem - in any given utterance, we know who 'I' am and who 'you' are, and if there is any question in our minds, we can identify either with a vocative or a relative clause. In narrative recording of dialog, the narrator has to identify the speakers, too.

Jim:

4. Responding to Bob's: "There will be a discussion of Jim Yorke and Celso Grebogi's proposal to have Lojban image languages using gismu based on single languages to make them easier to learn."

- Hiss, boo! I am of two minds on this. First, the main benefit to me of Loglan, and now -gua!spi, is its grammar and organization, and so it makes a lot of sense to use Lojban grammar on English vocabulary. I do this in the "English" output option of my parser-organizer. However, when you've read as much of that swill as I have, you quickly learn that Loglan demands 1-1 correspondence between meaning and word mapping; "un-only" just doesn't cut it for "barely".

Second, case defaults, switches, replication, etc. are very specific and aren't present in English. Thus you're really learning Lojban or -gua!spi vocabulary, just without the words, and you might as well do the whole job right.

5. Responding to Bob's: "Side issues may include applications such as teaching Lojban to kids (including using such teaching to help them learn English better)"

- See above. I freely teach my kid -gua!spi grammar since it works pretty well on English and since it's more coherent than what he's getting in school.

6. Responding to Bob's: "Carter's work, in particular, will take a fair amount of tanru and lujvo re-making; he used a lot of Institute Loglan words that are based on no longer acceptable tanru."

- Plus playing fast and loose with the organizational compounding rules, such as they were under the Institute. This is an area that you should look at very closely: regularities in compound formation.

Bob: We want to see what large numbers of Lojbanists do in compound formation before making decisions. 'Let usage decide.' Your experiences were relevant, but are only one set of data points. Besides you, only JCB ever made a great number of compounds, and we both know that his were awful (He still defends Institute Loglan "mormao" for "kill" even though it requires a different interpretation of "madzo" in a compound than separate. Oh, and Faith Rich made a lot, too. But she (and most others who make compounds) made some bad ones too, based on JCB's patterns, and often overdefined or used English idiom. She also never wrote text, so what work she did with place structures is untested. In fact, almost everyone who makes lujvo forgets about place structures.

Jim: My point here is that compounds ought to have a specific meaning, and therefore need to be prescribed, not described.

Bob: Agreed that compounds must have a specified meaning, but who is doing the prescribing. Every linguist and lexicographer will tell you that a dictionary is-and-must-be descriptive of what people actually do, not prescriptive. The main difference in logical languages is that whoever does the dictionary must gather enough information to form a consensus on what actual usage dictates a word's meaning to be. To make this process easier, we'll quickly develop guidelines as to how to algorithmically derive the place structure by combining terms - probably quite similar to what you proposed several years ago for Institute Loglan. But if people don't choose to follow these guidelines when they make a new lujvo, it's their language, not the dictionary writers'.

Jim:

7. Responding to Bob's "In addition, the current gismu list has more expressive power than earlier lists."

- Right on! Most if not all of your added gismu are in - gua!spi.

8. Responding to the Institute's proposal to divide gismu into "nouns", "verbs", "bodyparts", etc. sounds rather familiar. In Old Loglan JCB defined all the motion words with the same cases (x1 = mover, x2 = destination, x3 = origin x4 = route), and many such categories were apparent that had somewhat uniform cases. I was able to adjust the definitions of a relatively small number of words and get the great majority of Loglan words into uniform case categories. This was the basis of my Loglan thesaurus, which I have continued into -gua!spi. HOWEVER, it was always my intention that the thesaurus categories were merely a learning aid, just as JCB says in L1 of the motion words, and if a category member seemed to need a variant definition I gave it one. For example under Clothes, "x1 is a boot"; "x1 is a coat"; etc.; but also "x1 is a curtain over x2" and "x1 is a garment worn by x2 on its body part x3".

As for "nouns", I'm not averse to using that terminology for single-case gismu, as a tie-in for people who do have nouns in their native language. But of course the student has to be reminded that Lojban doesn't have special words for "is a" in sentences like "George is a horse"; "horse" is very verb-like in Lojban.

9. Responding to Athelstan's comments on Esperanto "rules": - a. How does Esperanto get away with not having all the article variants that Lojban has? For that matter, how does English? Close attention to articles is one big benefit Loglan (and progeny) has brought to me.

Bob: The distinctions made by Lojban aren't often consciously made in English and other languages, although different ones are used that are somewhat comparable. Thus le/lo correspond approximately to definite/indefinite articles in English; lo/loi corresponds approximately to the difference between 'count nouns' and 'mass nouns' - though nearly all English words are categorized as one or the other, whereas any Lojban word can be either. Names in English have no article, except for 'The Donald' of recent news tabloids, and a few archaic usages surviving from when British explorers used them for faraway places like 'The Transvaal', and rivers like 'The Mississippi'. In any case, the capital is enough to mark a name in English in most circumstances. Set descriptors are relatively un-useful in English. Now of course, all this is excuses - English is just a vague language, but it does have some complex structure to its articles and their equivalents.

Jim:

b. I suggest that you nail down official terminology for "case", "sumti place", and so on. In Latin there are a finite number (six?) of official cases, which are used semi-idiosyncratically on each word, much like Lojban's x1, x2, x3, x4, x5 (or Loglan's X, Y, Q, H, W). As I use the words, a gismu is a relation between one or more "cases", much as a function takes arguments. However, what you're talking about is places into which actual parameters can be substituted. The actual parameter can be called a "case occupant". Here are some examples:

sin(x), x being an angle in radians. "x" IS the first case of sin. (Or "x" is the symbol representing the first case of sin.) sin(pi). pi OCCUPIES the first case of sin, not "is" or "represents". "x1 is the house of resident x2". x1 is the first case (or represents). "That is my house". "That" occupies the first case, and "me" occupies the second.

Thus the "accusative case" or "x2 case" is actually a set of what I would call "cases", one per word, so you would more naturally say "the accusative case of 'flecto'". The relation "case" isn't very useful without its x2 case being occupied by a specific word. (This despite the Institute's idea to codify definition regularities by designating specific meanings for specific cases, e.g. x1 is "always" an actor, except that isn't true.)

Bob: A 'bridi' is a relation among 'sumti places'; a 'gismu' is only a word form - the five letter brivla (bridi-valsi = bridi-words) that can be combined into lujvo compounds using their rafsi affixes. The relationship of a bridi is indicated by the 'selbri' (erroneously called 'kunbri' for a while). A selbri may be an individual gismu or other brivla, or it may be a tanru metaphor. We avoid 'case', along with other linguistic terms that have different definitions for different people (do we mean accusative/nominative/genitive, or actor/beneficiary). Thus, what JCB calls 'case tags', and I think you call 'modals', we call 'sumti tcita' or 'sumti tags' that label a sumti as to the relationship it holds to the rest of the bridi, or in a relative clause, to the sumti to which it is attached. (All this confusing and conflicting grammar terminology is why la lojbangirz. uses Lojban words in its teaching materials. Any English word we use gets attacked by experts who say that we are using the wrong jargon word for the concept as it is understood in their specialty. We cannot afford to misuse jargon if we want to gain academic respectability.)

Jim: 10. My comments on Bob's answers to questions from students in Dr. Gorsch's class on semiotics: - Without question, different languages slice up the pie differently. My own favorite example was when my wife and I were buying a picture frame. She was chattering with the storekeeper in Chinese, choosing which one, and finally decided: "|mai /hong -kuang". Now I know a little bit of Chinese, and "/hong" means "red". So he takes down a BLACK frame. It turns out that whatever the luminance, if the hue is red then you can call the thing "/hong", whereas in English, things below a certain luminance are "black" whatever the hue.

The same is true on phonemes. I have to hear a word repeated several times before I can distinguish between Chinese 'q' and 'x' (matching English 'sh'), whereas they have trouble distinguishing our 'l' and 'r' even though they have their own versions of both phonemes.

I despair of ever writing down in a dictionary an authoritative statement of the denotation of a word. "Red" is so obvious: 670 nanometers. Yet you can trip up on a little detail of luminance. Now try to define "liberty". Connotations, as noted by the student, are even more slippery. I solve the problem by slicing the pie with connotation on the outside: the kinds of connotations mentioned by this and other class members are simply too complex for 20th century logic and database software to handle, and so, connotations will have to wait for major advances in the study of language artifacts.

On the third question, I find that my referent boundaries and connotations were not changed by learning Loglan -- since, of course, JCB didn't emphasize referent boundaries -- but the major effects were that the predicate logic definition of the words makes them much clearer to me, and certain grammar elements, particularly the articles, the abstractions and the tenses, fill gaps in English that I feel a need for.

Bob: Learning to write good dictionary definitions is non-trivial. Even if you feel that the best modern dictionaries convey definitions well, which Jim suggests is not the case, the people who write those definitions are bound by space limitations. They also spend years learning to write such definitions, and the results still go through several editing passes. Lojban words, each referring to a single concept, should be easier to define, but I still expect that our first attempts will fall far short of the state of the art.

Right now, I'm going through an intensive self-teaching in lexicography right now, so the problems are clear to me. But neither Jim, nor anyone else who isn't concentrating full time on dictionary work, should expect to be able to write good definitions for words.

As for colors and the like - we cheat in Lojban. "Red" means 'what you want it to mean'. You add a sumti tcita indicating things like the standard you are using (which could be a wavelength standard, or might simply be labelled 'the Chinese standard'). You also may throw in sumti for 'against background' and 'to observer', since colors are perceived. But since most people don't clearly mean any of these when they mention colors, we make all of the places optional.

Jim:

11. Regarding le lojbo se ciska pe la maiky'elsym: His experience with learning Lojban sounds very similar to mine learning Loglan -- though I think he has benefitted greatly from the lessons and the live feedback that you have provided.

12. On page 45 bottom, you say that the glue in a compound is always 'r' unless the second rafsi begins with 'r'. I believe I tangled with a similar Institute rule. 'n' has to be lexically valid in place of 'r'; why not allow the speaker to choose what sounds most pleasing?

Bob: No inherent objection, but most people don't want a choice, especially one based on personal aesthetics, and it sure makes the computer program harder (and which goes into the dictionary?). The expandable forms are justified by the pragmatics of redundancy - longer lujvo contain more information for a noisy environment (a criticism of - gua!spi and other affix-primitive varieties is that they reduce redundancy too far in a language already pushing redundancy requirements on the lower limit. But we may discuss this proposal at LogFest.

Jim:

13. Regarding comments from jyjym.: In -gua!spi I also decided that nationalities and national money units should not have primitive words (gismu), for the same reasons. Also, I define "gua" to mean "x1 is the language of people x2", and "spi" is "x1 is a member of the local culture", where "local" is defined relative to the group the speaker is (currently) in. Thus -gua!spi means, literally, the local native language.

The idea is that by speaking this particular language the speaker places himself into a particular culture, from which self-referential words have a unique meaning. I should post this definition in sci.lang -- but my mailer would overflow with people telling me what a dumb idea this is!

14. Regarding changing "ckamu" to "mleca": The whole issue of gismu negations is a can of worms. For example, should you say "evil" or "un-good"? I think that as a matter of policy, negated gismu should be discouraged (shades of 1984) unless they clearly are used frequently. Go through all the text you have so far -- admittedly not too expert, from the students -- and identify all compounds ever used with zmadu, and judge whether, had ckamu had a decent rafsi, a compound with ckamu would have been more felicitous. Make the decision by the resulting numbers. The main problem will be to decide in each case if it's too wordy to use the conversion of "zmadu". I'll bet, actually, that you get a decidedly uneven bias towards positive gismu (where "zmadu" is the natural choice).

Bob: This is partly addressed in the gismu proposals section above. Looking at usage so far will tell nothing useful. English has an irregular split between words that are negated by un- and words that have an apparent opposite that is a different root. Lojban usage so far has exactly mirrored English, since English speakers have written almost all of the Lojban that exists (which isn't much). NO ONE yet writes a Lojbanic seeming idiom. (Helsem tries in his poetry, but in several cases you can tell that he thought of an idea in English and translated it word-for-word in to Lojban. The results have been pretty bad grammatically at times, though he learns.) But he is the best writer outside of DC. I suspect that I can write a Lojbanic idiom, but I have not yet started writing in Lojban on my own - my evidence is my translation of Arabian Nights, which I am doing in a very non-English idiom in places (and I don't know Arabic).

In any case, we do use 'more' and 'less' in English in some pattern, though I'll agree that the positive comparison is more frequent than the negative one. We also use 'not less than' and 'not more than' (which are of course not quite opposites, since they include the equality middle). Presumably the usage patterns have some metaphysical basis - there are times when people want to express 'less than' as a positive comparison. Since Lojban tries to remove restrictions rather than add them when possible, we should make some effort to model this feature of natural language. Why should Lojban restrict negative thinking?

The several other gismu being proposed for consideration at Logfest include some where we don't want to assume that the negative of the existing gismu is identical to the positive of its 'opposite'. Thus we are proposing 'ugly' to correspond to 'beautiful', 'diffuse' to correspond to 'concentrated', etc. Since we have other pairs of this type in Lojban, we have to at least consider these.

From John Cowan:

Herewith a few tidbits that may interest Lojbanists.


In describing Lojban to a group of friends, I mentioned as a virtue the default tenselessness of Lojban sentences. An English-speaker and an English/Spanish-speaker both expressed puzzlement: "Why would you want to say something without mentioning tense?" Rather than replying directly, I chose to describe a little bit about the Navajo language. I don't know Navajo, so what I say here is subject to correction.

In Navajo, it is necessary to mention all sorts of things that English finds unnecessary to specify. The sentence "You eat blueberries", for example, must be rendered as "You [pl.] eat separable objects one at a time." The vagueness of English about singular/plural in the case of "you" is impossible to render in Navajo, as is the vagueness of "eat". The Navajo fence-rider cannot simply report "Fence broken"; he must mention whether the breakage appeared to be accidental, deliberate, or the result of an animal's act. Likewise it is impossible to say "They went thataway!" in Navajo. The direction of motion must be nailed down, as must the mode of travel, as must the distinction between going >to< a place and going >toward< a place, or going to a place and passing it on the way to another place, etc. etc. etc.

By comparison, English is downright sloppy. The Navajo-speaker's reaction to English is "How can I know what the >belagana< are talking about?" From within English itself, this sloppiness seems more like a freedom. It is not necessary to pin down all these details to make a grammatical English sentence. Lojban, although it can be as precise as Navajo, can also be even vaguer than English, leaving even the details English thinks are fundamental unspecified. This gives Lojban an additional range of expressiveness not present in any natural language.


It seems there has been much dispute recently about the ' character, which the Lojban materials say is pronounced like the English "h". "Why not use the letter 'h', then?" In JL10, lojbab gives a number of reasons for not using a letter. It seems to me that confusion might be avoided by explaining the role of the buffer sound differently. (I have checked this with lojbab.)

Consider the word "co'o". An English-based view of this would be that it contains four sounds, "c", "o", an h-like sound, and another "o". This is also the view of the current Lojban material. Another view, however, says that the four sounds are "c", a standard "o", an >unvoiced< "o", and then another standard "o". In other words, the two "o" sounds are separated by an interval of time when sound is being produced, the vocal tract is in the position for "o", but voicing has been "turned off". In the sequence "o'o", the three sounds are separated only by boundaries between voicing and non-voicing.

What about words where the ' separates two distinct vowels, for example, "ko'a"? In this case, there is the same pattern of voiced vowel + unvoiced vowel + voiced vowel; the unvoiced vowel may be either "a"-like or "o"- like -- I myself tend to make it "a"-like.

This makes for a nice symmetry between . and ' . . signifies a period of no sound production at all -- voicing turned off, exhaling turned off. During ' , voicing is turned off but exhaling continues, producing a voiceless vowel. Exactly which voiceless vowel does not matter, since Lojban does not distinguish between the different possibilities.

Note: I am not proposing that this long-winded explanation replace the basic instruction to beginners of "Pronounce an 'h'"! It is merely available in reserve, to be trotted out when the student asks "If it sounds like an 'h', why not use the letter 'h'"? Then you can reply, "It's not >really< an h-sound at all, it's a voiceless vowel."

Bob: I like the explanation, which explains the intent. One note on "ko'a". I suspect that in the 'voiceless time' in between vowels that are different, you are expressing a 'voiceless vowel glide'. If you don't turn off the vocal cords, the sound you get is /kowa/, where the /w/ is the glide. It is this glide that we are trying to 'devoice' with the ' , because if the glide were audibly voiced, it would be impossible to distinguish "*koa" from "koua" from "ko,ua".

John: Mathematical nit:

If 'tanjo' means 'trigonometric tangent', how do we say 'geometric tangent'? Why are the trigonometric functions given gismu rather than being handled by something in MEX? If 'tangent', why not 'Riemann zeta', or 'factorial'? -------- Bob: We could generalize "tanjo" to include both, but I suspect that you can do a short tanru for 'geometric tangent', but not for 'trigonometric tangent'. The primary reason for including both "sinso" and "tanjo" was so the concepts, as opposed to the mathematical functions/operations themselves could be talked about. Trigonometric functions are basic to mathematics and many other fields - not all purely scientific - as wave-forms (shapes), yet in English we can only talk about these in terms of their mathematical function 'names'. You could make a tanru like square-triangle-ke-opposite-ce'o-adjacent-ke-side-ratio (no, I won't make the lujvo), which captures the original intent. But a non-mathematician - say an artist - can talk in Lojban about 'tangent curve' (tanjykru) without knowing the mathematical basis.


As for the others that are excluded - lack of usage frequency, or any metaphorical advantage outside of one field.


John: Another book for the list, too simple for you probably but I would recommend it to people who want to understand Lojban's underlying basis: Semantics, by Geoffrey Leech. (Penguin Books, 1974, no ISDN.)

It's a simple introduction to the subject in plain language, and much of the explication in the central chapters reads like a point-by-point explanation of Lojban grammatical structures -- though without reference to Lojban, of course! --------

John: Here's my proposal for a super-simple le'avla algorithm:

  1. Lojbanize the word to be 'borrowed' by the methods used for cmene.
  2. Convert all y's to some other vowel or to a vocalic consonant.
  3. Modify the ending to be a vowel, either by dropping a final consonant or by adding an extra vowel.
  4. Modify the beginning to be a consonant, either by adding a extra consonant or dropping an initial vowel.
  5. Choose a gismu (not a rafsi) that categorizes the le'avla into a "topic area". Replace the final vowel of the gismu with a vocalic 'r'.
  6. Prefix the modified gismu to produce the final le'avla.

Examples:

spaghetti -> cidjrspageti
maple tree -> tricrmeipli
maple sugar -> saktrmeipli
mathematical integral -> cmacrnintegra or cmacrntegra
brie -> cirlrbri
cobra -> sincrkobra
quark -> saskrkuarka
iambic -> pemcrniambo

Bob: Good proposal, and it appears to generate good words, so people can use it for now (note that in step 5., you may have to use an "n" instead of an "r", though, to avoid a doubled letter). It takes only minor modifications to allow this to be used with rafsi (I think). You glue the rafsi on in almost the same way, but always use a gluing vocalic consonant with a 3-letter rafsi.


Well, this ends our biggest issue yet - and in only 6 weeks turnaround!

co'o

Map to LogFest 90

You need to get to Interstate 66, on the West side of Washington D.C., just outside of its 'Beltway' freeway as shown on the map below. For those who attended in previous years, our exit from the freeway is now a cloverleaf, so the exit from the freeway is different, and Nutley St. has been shifted near our house, so it will look much different.

From the West on I-66, take Exit 17, Nutley St. going South (to your right). This exit did not really exist last year, though it was possible to play games going through the Vienna Metro Station parking lot.

Taking the southbound exit to your right you want to enter Nutley St. (not the Metro parking lot). You need to get over to the left lane fairly quickly.

There is a traffic light one block away from the overpass (Swanee Lane on the left). You will see a grassy slope on your left ahead, and a large office building behind some trees ahead on your right. A very short block later (opposite the office building), there is a left turn lane. You turn left here on Hermosa Dr. Go to the end of this street - only about 3 houses, and turn left on Beau Ln. Bob & Nora's house will be the first non-corner house, directly opposite the cross-street Suteki Dr.

From the East - Exit the Beltway for I-66 Westbound 'Front Royal'. You will go about 1 1/2 miles and take Exit 17 - Nutley St., which is the first exit possible after leaving the Beltway. There is a very long ramp - stay on it, and take Nutley St. South which exits the ramp just after the overpass. You will join with Nutley St. and must move left quickly to merge. Thereafter, get in the left lane, and follow the instructions in the last paragraph of From the West section.

Some More Logo Proposals