me lu ju'i lobypli li'u 3 moi

From Lojban
Jump to navigation Jump to search

For a full list of ju'i lobypli publications, see ju'i lobypli.
Previous issue: me lu ju'i lobypli li'u 2 moi.
Next issue: me lu ju'i lobypli li'u 4 moi.

Hoi Loglypli *      Number 3 - April 1987

[1993 note: version that follows was actually completed and distributed in June 1987 - some of the events related to the TLI split took place after initial writing, and delayed publication]

* Yep, we've changed our name again. This time it translates loosely as "Hey, Loglan user". We no longer can reasonably claim to be a Washington-oriented publication. So our organization is now the Loglan Users' Group, at least until next issue.

For those new to us, I am Bob LeChevalier, member of the Loglan Institute, but having no official standing in Loglandia beyond that. This publication is not official, and in fact has received the express disapproval of the CEO of the Institute. More on that later. But I speak only for me, and definitely not for the Institute.

Actually, I'll amend that slightly. There will be two Loglanists at my address starting at the beginning of June, and hopefully someday some little Loglanists as well. I am most happy to announce my engagement to Nora Tansky. No date has been sent, but we won't let our Loglan work together wait for a wedding. We'll keep you informed.

This newsletter was founded to bring together the Loglan community, which a year ago was fading into oblivion. I'm pleased to report success. This newsletter is going out to about 70 of you as respondents, and over 70 more who are located in SIG geographical areas. And we are averaging over 1 new recruit a week, primarily people who have never been involved with Loglan before. Loglan lives!

I am publishing this newsletter for anyone who is interested enough to let us know their address. When I have at least one interested person in a community, I attempt to contact other Loglanists in the area via this newsletter in an attempt to establish a local group. We now have over 10 such groups, and the Washington group has over 20 members and meets frequently. Now ongoing is the first Loglan class since the apprentice program 10 years ago. With any luck, we will have fluent Loglanists by year's end. We hope to prepare materials that will help the rest of you achieve what we all have dreamed of - a true Loglan-using community. We're close, notwithstanding the problems that have plagued the language, and those who want to use it, for 30 years.

Any new recipients who did not receive the first two newsletters: we have a few copies left. Just write or call and I'll be happy to send them out while they last.

This issue contains much news - as usual both good and bad - but all exciting. A summary of the contents:

  • Why We've Been So Long Silent / Editorial
  • Institute News: LogNet, MacTeach, Word makers Council, Grammar, NB3 and NB4, L1 Update, Paid Workers
  • Users Group Activities: Public Domain Loglan Parser, LogFlash, Loglan Primer, Loglan Classes, Video, Loglan Bulletin Board, Recruiting, Honor Roll of Contributors and Workers
  • Summary of Logfest 3
  • Loglan-88
  • Plans for Logfest 4 (July-August 1987)
  • Loglan Question and Answer Page(s)

And as a bonus, in an Appendix a complete and up-to-date description of the current phonology and morphology - eventually to be part of a Loglan Primer. This ain't officially approved, but it has been thoroughly checked by several active Loglanists, including pc and Chuck Barton.

And for those who wish to contact me for more information on any topic, to receive appendices or back issues, to volunteer, to attend Logfest 4, and also to contribute money (that $700 just about covers us through this issue, excluding phone bills which are unrepeatably high):

                          Robert LeChevalier
                              Nora Tansky
                             2904 Beau Lane
                            Fairfax VA 22031

                      Home phone (703) 385-0273
                    Bob's Work phone (703) 847-4465

                  Usenet/UUCP: sdcrdcf!slantsun!rjl

 Loglan Bulletin Board (no costs except phone charges to California)
                           (415) 538-3580
(300 baud - type 'help complete' and/or 'read Loglan' and go from there)


Why We've Been So Long Silent

For those who have received this newsletter previously, and especially those who have contributed some $700 towards its continued publication, I owe you an apology. My last newsletter was in August. I had intended, and should have completed, publication of 2 or 3 issues since then. We had a successful Logfest 3 in September, and much good work came of it that was worth reporting. Your support justified better response. Please accept my apologies.

Two things happened to interfere. My father passed away in October - this actually didn't have a lot of direct effect on Loglan work, but kept my emotional energy down when it needed boosting. And in September, JCB (Jim Brown) returned from Europe. As I had reported, much good work had been done - reviewing draft Notebook 3 text, dictionary update work, Nora Tansky's flash card program, and this newsletter itself and the new community it represents. To prepare and report these results to JCB took several dozen pages, and over a month of work.

This newsletter, and my activities, did not go over well with JCB. He has stated that he believes that this publication is competing against Lognet. I held off publishing to dispel this belief. We attempted to negotiate, but those negotiations have broken off unsatisfactorily. Kieran Carroll and Scott Layson have volunteered to continue negotiating. But Dave Cortesi, Nora Tansky, and Ed Prentice, among many others, convinced me that this newsletter was important to the community, and that the language and the community had importance beyond JCB's and the Institute's approval. Besides, I have an implied commitment, to those who have supported me in this effort, to keep publishing. We have now faced the crisis and have passed it. I have talked with over half of you; support for my efforts is nearly unanimous, and this newsletter will continue.

I'll spend a few paragraphs on the issues, and then drop it. Our purpose is not politics.

JCB will be officially stating the Institute's position on copyrights and the language in a forthcoming Lognet. During our negotiations, he sent me an advance copy of this copyright policy. In it, he claims copyright over not only the books and Institute materials published on the language, but on the "lexical elements" themselves. It isn't clear what he believes these are, but JCB explicitly has claimed copyright over all of the words of the dictionary, and the metaphors behind them. And he intends to assume that all work done by members in the past several years belongs to the Institute unless specifically notified in writing. Under his copyright claim, JCB says that we can use Loglan only for limited personal use, specifically writings and speech, and never for profit. All publications must be approved by the Institute in advance.

JCB has refused permission to publish for Nora's and my teaching program series, LogFlash, since we will not sign a legally binding agreement accepting his copyright claims. This is why we have not published or distributed any more copies of those programs with a word list. JCB has specifically forbidden me to distribute copies of his copyrighted word list. So those of you who have my first newsletter (UL1), or test copies of LogFlash/MacTeach: treasure them - since the Institute has not published a version of the primitive list on its own. (I have been told by a member of the Word Makers Council that they are being required to sign a non-disclosure agreement in order to get a copy of the list needed to do their work. The Institute must not value volunteer work enough to ensure its quality. I'm hoping that they refuse - the value of volunteers' time must be learned while the Institute still has supporters.)

JCB's copyright claims are detrimental to the survival and growth of Loglan. I've researched and documented precedents on the legal issues and precedents, which strongly contradict those claims. This data is available on request. But, we will actively avoid a legal fight while seeking ways to promote the language; only lawyers would win, and Loglan would certainly lose.

Editorial on the Copyright Situation

by Nora Tansky and Bob LeChevalier

Why does the copyright issue matter? Because many of us have spent years supporting the Institute with our purchases, and later our dues and contributions. Our only expectation has been that eventually we would have a language we could use. But for two years we have received little for our dues except promises and requests for more money, and the expectation that JCB will leave shortly to disappear into 'The Job Market' for 3 years without the distraction of 'greedy Loglan'. We can't blame him - he's put more than 30 years into this project and has other things to do with his life. But he has not given us what we need to continue without him, and we deserve it. In the process, he has not given us his plans for Going Public Again (GPA), but has strongly implied that this will take place after his Job Market excursion. We guinea pigs of NB3 will hopefully by then have tested the language - enough to allow him to determine what should be published for the mass market. He makes the erroneous assumption that people can and will "test" the language without a capability to build a true language community, which requires that we have a language that we can use freely in all aspects of our life. We intend to teach our children Loglan, but also about the freedoms we hold dear. We don't believe we can properly do so if, in teaching them Loglan, we must teach them to accept prior restraint and censorship. And a final point: the Institute will never achieve the financial security it seeks as long as it sells to a limited audience - limited by its own restrictions on the language.

We look forward to working with all of you who support our work, and we will continue to support both the Institute (as members, though not as volunteers), and the Loglan community (with this newsletter) as long as interest lasts - and that will hopefully be long indeed.

(We will publish signed statements of agreement or rebuttal to this editorial and any other opinions stated in this Newsletter. And as per our original editorial policy, JCB - as founder of the language - is specifically invited to reply. All contributors will please indicate permission to publish with their submission.)

Institute News:

LogNet, MacTeach, Word makers Council, Grammar, NB3 and NB4, L1 Update, Paid Workers

  1. LogNet - Anyone who is an Institute Member, ex-Tler, NB3 or MacTeach orderer, or other hanger-on on JCB's mailing list got a copy of the first issue of the new Lognet, published by Mike Parrish. Excellent job on layout, Mike. So far its pretty much all JCB's writings. And stuff JCB has sent me indicates at least one full issue of his status report remains unpublished. But after that, there will hopefully be contributions from the rest of you in the Loglan community, as well as hard information about some of the new concepts JCB is proposing. We hope JCB has given Mike enough editorial freedom to publish other contributions. Mike has permission from us to use anything from this publication that he feels is of general interest, including from the Primer.

    Also, Mike, please publish Nora and my names, address, and phone. I suggest an easily detachable and returnable note/card for all of us to notify you with similar permission. Contrary to JCB's belief, I've found that nearly all Loglanists are interested in, even desirous of, hearing informally from other Loglanists. And I have yet to receive a single request for deletion from my mailing list, except from permanent Loglan dropouts.

  2. MacTeach - JCB and Glen Haydon now have 3 of their MacTeach series of programs done, which match the first two of Nora's and my LogFlash programs. They collectively teach primitives and affixes. LogFlash is based exactly on JCB's flash card teaching methods; MacTeach presumed no learning method, at least in the original version. JCB has a new pricing scheme that encourages people to try the programs and report their learning methods. As co-authors of LogFlash, Nora and I have an obvious preference for our program, although I did support MacTeach with an order. So, for any who do get the new version - I'm soliciting an independent, and presumably unbiased, review. Ideal would be one of our testers, who could comment on the strengths and weaknesses of each program.
  3. Word Makers Council - There is now a Word Makers Council, presided by Faith Rich, with Jeff Taylor serving in the tough coordinating role. Also on the list are Birrell Walsh, Bob McIvor and Kieran Carroll. The purpose of the Council is to set policy on kibbitzing and to select new words from kibbitzed data. Jenny and Paloma (the new HAGA?) in Gainesville seem to be doing some word-making and policy-setting too, even though they are not on the Council or even the Academy. So there may actually be two councils with overlapping charters.
  4. Grammar - I've heard from Jeff Prothero that JCB says the grammar is finally stable. I've also heard that intermediate versions had major problems, that Trial 55 is the latest version of the grammar, that there are several potential flaws in parsing due to PreParser 'hacks', and that the parser is unstable - it is prone to infinite 'gu loops' when trying to process complex expressions of undetermined validity.

    We've been collecting a set of problems with the Trial.24 grammar that stood for the last few years, with the intent to try them on the new grammar when it becomes available. A major problem with the current grammar is that a good deal of it is embedded in C code in the PreParser and Parser which is not limited to the insertion of the machine lexemes. That portion of the grammar has been particularly full of minor bugs, and YACC won't catch them since it doesn't see that portion of the grammar. Included in the PreParser is some acronym-related logic that seems to contradict the concept of the old resolution algorithm; the listener must determine the lexeme types in order to tell which way an utterance should be resolved into words. Similarly, the PreParser accepts 'lepo' but not 'le po', even though a resolver would break the former into the latter before parsing. A consensus of those I've talked to believes that if the grammar is finally stable, it should be redone without the PreParser kludges. The resulting grammar should then be properly written in BNF - a form used to define computer languages. This form will be much easier to use than the YACC form (though not easily proven unambiguous), Much simpler and better parsers can be designed based on BNF, once the disambiguity proof is settled. (See the note below about the public domain Loglan parser.)

  5. NB3/NB4 - JCB has now announced that NB3 has become two volumes, with an NB4 containing word lists. No word on whether these lists will be available on computer media, the way they'll likely be most useful. I've seen some of the Eaton material, and most of the comma-delimited entries are incomplete. Yet the total data volume is apparently over 400K characters so far, with more being generated daily. And this doesn't include the results from the dictionary reformatting (600K or more) or JCB's SciWords project. For a minimally useful set of lists, they must be sorted in both English and Loglan order, as well as possibly by Eaton number. Thus we can expect several Megabytes of dictionary-like data. At 60 lines per page and an eye-scrunching 120 characters per line, 4 Megabytes comes out to 600 pages. Jim will need to do some major editing to get a printed document down to a size people will buy at 15 cents per page.

    Over the last summer, I and others reviewed the first 30 pages of the then NB3 draft. Finally, after LogFest 3, I compiled about 30 pages of comments. So NB3 needed a lot of work then, primarily because most Loglanists have lots of unanswered questions about the language, and JCB was raising new ones as he answered some of the old. See the Question and Answer pages below for a partial list of the questions and ideas that came up during LogFest. In any event, I've offered to coordinate a re-review before publication.

  6. L1 Update - JCB says that the update to L1 will be trivial after NB3. This seems unlikely, unless NB3 will become the new L1. JCB's writing style in L1 got us interested and taught us plenty, including how to ask questions that aren't answered in L1. He needs a similar review to the one we did on NB3 before he makes decisions, possibly before he starts rewriting. He does not make it clear when this rewrite will take place - before Job Market, or after. Such questions are critical.
  7. Paid Assistants - JCB apparently has a HAGA, now - his half-time Gainesville assistant. Apparently, she is named Paloma. It isn't clear what her duties are, or what will happen when JCB goes sailing. I suspect accountability will determine whether he gets more money to keep her employed. According to one letter he sent me, he also has hired a programmer to do the dictionary reformatting, after 'firing' me from my 'unpaid employeeship' as Dictionary Update Foreman. It isn't clear whether this is someone separate from Paloma.

    In any case, I wish the programmer luck. That reformatting job had become a major editorial chore due to minor inconsistencies in the various dictionary entries, and JCB's desire to have Eaton numbers for each word. For some entries, each of the English translation lines (see UL1 for dictionary formats) requires a separate Eaton number, and JCB's comma-delimited format does not allow for this. Then, the ever-changing prim-list requires occasional reprocessing of the data. And no one has yet succeeded in programming the complete Complex-making algorithm as JCB and I wrote it up (see UL1 Appendix).


Users Group Activities:

Public Domain Loglan Parser, LogFlash, Loglan Primer, Loglan Classes, Video, Loglan Bulletin Board, Recruiting, Honor Roll of Contributors and Workers

You've all been busy, too, Loglypli. And as I said above, none of this is in LogNet. Unless Mike puts it in a later issue.

  1. Public Domain Loglan Parser - Big last minute news just before publication. I have just received from Jeff Prothero his cut at a Loglan parser. Written in C programming language, this parser is a 'top-down recursive-descent, back-tracking, right-reducing' (for those who know what that means) version which should be very efficient with Loglan text. The resulting grammar is expected to be much simpler, too - it does not require special machine lexemes to distinguish ambiguities, as does the current MacGram. It apparently is based on the Trial 55 stable grammar, but has been completely rewritten based on Jeff's knowledge of Loglan and of parser writing. It thus is free of any copyright restriction by the multitudinous precedents regarding copyright of computer languages. Jeff has specifically declared the Parser to be a Public Domain gift to Loglandia. The program will need work. It currently compiles with a special purpose C compiler for a 68020 microcomputer, and Jeff has as yet been unable to get it to compile on any IBM PC compatible compiler. It has a simple form of lexical analyzer (which determines the words from a string of text and assigns them to their corresponding word types) that must be improved to support the full word set. And testing will undoubtedly take a while, since he expects it to help find flaws in the LYCES MacGram that are hidden by the PreParser hacks. We have a host of C programmers in the community, and some have expressed interest. Any more volunteers? Copies are available for free to any who volunteer (but donations of diskettes and postage money may be desirable if too many want it.)
  2. LogFlash - Nora and I finished our teaching program in time for Logfest 3 with several nice bells and whistles to remedy user problems that our testers had found. We were still calling it MacTeach, since it was unclear what the Institute policy would be to two sets of teaching programs. Since then, we have renamed it LogFlash, to eliminate confusion with the other series (and to avoid the confusion that occurs with any program name starting with Mac-). By the time JCB returned, we had versions running under CP/M, IBM PC, and generic MS-DOS versions of Turbo-Pascal. We also had a version of program 2 working, which teaches the affixes, and Jeff Taylor had done a preliminary review in addition to our own testing. So we sent both off to JCB. He told us that the Haydon-Brown MacTeach programs were still going to be sold, though he encouraged our efforts. But he gave little direct feedback on the programs; he had not read my instructions and had tried to run it on the already full floppy disk - so it crashed. In any case, he then raised the copyright issue, and talked of having more people review the program before we could market it, etc. Also at that time, he revised the input primitive list one more time in response to the detailed review Nora and I and the testers performed as part of learning the words. (Those comments appeared in part in UL2.) When we did not immediately agree to his negotiating position, he notified me that I was not to distribute copies of his copyrighted word list. While we couldn't distribute the programs, Nora and I didn't stop working. Since then, she has added I/O error handling for MS-DOS, and has generalized the program so that it can be used for other languages besides Loglan. We intend to market this version to schools after prettying it up, and a test user has successfully tried it with Spanish. Meanwhile, Borland has come out with Turbo Pascal for the MacIntosh. And we have a volunteer with a MacIntosh to convert the program when we again have a word list to distribute it with. At this point, we are planning on distributing object for the Loglan version of the program to User's Group members as Shareware, with an optional charge like $20 per program. However, we cannot supply JCB's word list due to the copyright restriction controversy. Hence, the Loglan-88 effort described below. We will distribute LogFlash with the Loglan-88 word list, which will be public domain and not subject to JCB's copyright claims. This should be available shortly after LogFest 4. After GPA, when we have a package of programs to sell, the price may be higher to cover increased support costs. But until then, the whole purpose of the program is to enable all of you to be able to use the language; this is far more important than the small amount of revenue that it would generate.
  3. Loglan Primer - One of the primary results of Logfest 3 was the realization that the bulk of the Loglan community has no clear idea where the language stands right now. Those of us who are active may have figured out JCB's Notebooks. But the Supplement (TL4/3), TL6/1, and TL7/1, were in no way an adequate description of the language for most of you. Half of the attendees expressed a need for a summary of the phonology and morphology of the language, just so they could understand what the rest of us were talking about. Nora and I promised to write one, and mail it out within a week or two, a truly optimistic proposition. It turned out that the job was half done. Chuck Barton started a rewrite of the Loglan Primer a few years ago, but before GMR was complete. All we had to do was revise it. But with the amount of material I had to prepare for JCB, and then with the paralyzing effects of my father's death and interminable negotiations, the effort ground to a halt. It was, however, at the top of the stack when I restarted work in January. The result is the Appendix for this issue, which pc has called the clearest summary of the phonology and morphology that he has seen. I hope you all agree. Questions and comments are welcome. And for the patient souls from Logfest 3: I hope the quality makes up for the delay. Hopefully, Chuck will take up where we leave off, or assist us in writing the rest of the Primer. We are getting useful data as to what is needed from:
  4. Loglan Classes - Starting in early April, Nora and I, assisted by Jack Waugh, have been conducting a class for new Loglanists. The class has 10 participants, with perhaps 5 or more that have not managed to fit their schedule to our class times so far. We are adding several per month, due to successful recruiting, primarily by Jonathan Tite. We may have a catch-up class or two in another month to allow more students to join in. In the meantime, we are meeting once or twice a month (with allowances for holidays and the time Nora and I will need to relocate her down here). The goal is sustained Loglan conversation by LogFest 4 in July-August, and fluency by the end of the year. Actually, the teachers are learning along with the students. Jack has the best pronunciation of our group, Nora and Jack each know different aspects of the grammar, and Nora and I have the vocabulary down fairly well. When in doubt, we write down questions and get opinions from pc, Chuck Barton, and any others likely to have useful opinions. And we can always check against LYCES and the old reference books. It is surprising how much Loglan knowledge is out amongst the masses, and useful as well. JCB refused to answer, due to our disagreements, when I called him for up-to-date information on a question upon which the dictionary, the old Primer, L1, and LYCES all disagreed. We have decided that we will defer to the Academy for answers to questions like these, if they can answer efficiently enough to support the community's needs. They can start with the list in the back of this newsletter. Or we will derive our own answers when we have to, forming a consensus among the knowledgeable in the community. That widespread knowledge of the language may be coming in handy. Lest those of you who have little time or knowledge of the language believe that you have little to contribute - please stand corrected. The novice students in the class are spending some 6-8 hours per month and are learning quickly, and the questions two brand new students asked at our last class led to two new grammatical concepts and some significant rethought on old grammatical questions such as da series assignment of variables.
  5. Video - A side effect of the Loglan class is a new project that will be done locally. Jonathan Tite and Rebecca Bach will be coordinating an effort to produce a video film about Loglan with information about the language and its development, along with a short sample of subtitled conversation or drama. The Loglan class students will be the stars. We hope to obtain studio (and broadcast) time from the local cable system community access station. The project is still in early planning, but we are hoping to have it produced by the end of the year. No promises on whether copies will be available. We still have to negotiate with the station. We are seeking help from non-local Loglanists for technical advice and script-writing. Birrell Walsh, for example, works for a San Francisco television station and we have asked him to consult with us. With luck, we may eventually be able to produce video lessons in Loglan, but that's a later project.
  6. Loglan Bulletin Board - There was much interest in the Loglan community setting up some type of access to the nationwide computer bulletin board networks. But, after talking to several of you, I have determined that no one network has more than a couple of you as members. We don't yet have the membership to negotiate strongly with the network companies, but we have gotten one generous offer. Bill Ragsdale, who is active in the FORTH community as well as in Loglan, has operated a FORTH bulletin board for several years. He has now set up a conference on that bulletin board for Loglan. It is only 300 baud and has no file transfer capabilities, but he has indicated interest and willingness to upgrade it if people use it. He has the higher speed modem already, and needs Apple software for a BBS to run at that higher speed, and a hard disk will be necessary to support file transfer. But the existing system will do for passing messages. Nora and I have a modem on order that should arrive shortly (to go with our new souped-up AT compatible, also on order). So we'll be able to support it with information and answers. Jeff Taylor is relatively local to Hayward CA, along with Bill Ragsdale, of course. So we together should be able to respond quickly to questions and messages. Bill is providing the phone line and computer for free, so your only costs are phone charges. I have AT&T Reach Out, so I can use it at $7.50 per hour at off-peak times. A little high compared to the network services, but there is no membership fee, so it is probably cheaper overall, especially for those of you in the area. To use it, you need no password. Just dial the BBS number: 415-538-3580. Then type 'help complete' for instructions and 'read Loglan' for messages. It supports about 50 lines of text in a given message, and has a simple editor. Oh, and if you forget the BBS number, just call information. It's listed as 'Ragsdale, Apple' in the Hayward CA directory. In addition to the above, I've wangled a computer address on another project's machine that is tied to Usenet/UUCP. The address is listed with my mailing address earlier in this newsletter. Those of you with similar access, please try sending to me. I also have joined the Capital Heath User's Group which operates a FIDO net. I'll get an address when I get my modem. In any case, I am collecting addresses for any and all networks that people are members of. I'll publish them next issue if you get them to me. Also, anyone who writes - please specify whether your mailing address and/or telephone is publishable. I figure we're about due to put out a directory.
  7. Recruiting - Most new recruiting is going on in the DC area. Last New Years, I went to EveCon, a science fiction convention here. At 3AM in the convention suite, I started talking about Loglan with maybe 50 people in the room. In 20 minutes, I had 6 new recruits. One is Jonathan Tite, who has continued the trend and recruited 5 more since then. It seems that science fiction fans are a good market for Loglan. We are planning to actively participate in Evecon programming next January, which could be the beginning of GPA - at least for Loglan-88. Loglan may not succeed in becoming a universal language, but perhaps within the science fiction community, we will find Loglandia - soon.
  8. Honor Roll of Contributors and Workers - The list grows longer. I won't repeat those from UL2 - this is just new stuff: Rebecca Bach - DC, one of the two co-producers of the Loglan video being planned. Chuck Barton - Boston, assisted in NB3 review and the Primer revision in the Appendix (and has expressed interest in writing the rest). Provided copies of his old materials on primitive making and phonology. A primary advisor on phonology/linguistics for Loglan-88. James Cooke Brown (JCB) - has been busy this fall, winter, and spring - his writings in Lognet tell of his multitudinous contributions better than I can. Jenny Brown - also active again, but I'm not sure what she is working on. Gary Burgess - UK, visited me in April, assisted in developing consistent rules for Loglan-88 phonology, and helped me convince a certain lady to move down here. Nora and I expect to visit him in England later this year and try to regather the British loglanists into a SIG. Kieran Carroll - Ontario, Kieran recently got married; he has volunteered to participate in the Word Makers Council. James Carter (jfc) - LA, has volunteered to serve as a Los Angeles center person. His wife is a native Mandarin speaker, and we will be using their assistance as much as possible in Loglan-88 efforts. Jean Chalmers - by a JCB message in an old Lognet, Jean is the Board of Trustees Secretary this year, and therefore is the one to whom major Institute communications are to be sent. I have an address that may be valid; I have urged JCB to publish the correct one, along with a clarification of who has what roles in the Institute, in Lognet. Bob Chassell - Boston, financial contribution; intends to become more active this summer. Dave Cortesi - SF, reviewed NB3 excerpt for JCB; has provided advice on computer networks and bulletin boards, and helped me get the printer driver that I'm using to produce this newsletter. A much needed phone conversation with Dave in February restarted my productive work. Ken Dickey - Portland SIG center person; Ken is the only other center person to actually start organizing his SIG, though his time has been short the past few months. Bill Dorion - DC, found a letter asking about Loglan in the February BYTE magazine, responded, and brought it to my attention. We thus may have a new avenue of recruiting, and at least one new Loglanist. Paul Doudna - St. Louis, an old contribution, actually, but given to me only recently. Paul did a detailed unpublished study of the primitives, categorized into similar types, that is proving quite useful in the Loglan-88 primitive analysis. Elladan - DC, Loglan class. Bob Hampton - Houston, volunteered to center a Houston SIG; having trouble reaching him currently. Chris Handley - while in Vancouver, BC this year, Chris has been doing another revision of the long-delayed CACM paper on the Loglan machine grammar. Glen Haydon - SF, with JCB, has revised the Haydon-Brown MacTeach programs, and has now produced three of the series. A lot of hard, productive work. Glen also answered the BYTE letter referred to above in a recent issue. Mel Kanner - apparently involved with LYCES/LIP work, possibly doing the same text-processing job that Jeff Taylor also has started. Scott Layson - SF, the other Loglan Academist besides JCB. Scott has again become active, and has stressed his interest in making the Academy an effective organization. Bob LeChevalier - DC, coordinated the NB3 excerpt review, LogFest 3, this newsletter, performed a review of the primitive list that identified many minor problems, assisted Nora in LogFlash 1 and 2 development (and in planning the rest of the series), researched the copyright dispute in attempts to find resolutions, leading the first Loglan class in several years, rewrote the existing Primer material, performing an analysis of LWs, planning of LogFest 4, assisting the video planners, and I am leading the Loglan-88 development effort. JCB has "fired" me as Dictionary Update Foreman. Financial contribution. (Whew! Is that enough? Any volunteers to take some away?) Sheldon Linker - I've heard Sheldon is involved again with the LYCES/LIP work, but I'm not sure of his role. Keith Marshall - DC, Loglan class. Bob McIvor (RAM) - Ontario, Word Makers Council; recently turned around his kibbitzing workload in record time. Emerson Mitchell - SF, Apple programmer; has volunteered to convert LogFlash to the MacIntosh now that Mac Turbo Pascal is out. Buddy Myers - DC, Loglan class, Primer revision review. John Parks-Clifford (pc) - St. Louis, all around advisor to us all; pc has provided me with a wealth of old information, identified new people to contact for SIG activity, advised on grammar and phonology issues, and assisted with my analysis of LWs. pc helped review the NB3 excerpt and the Primer revision. He and his wife Jane hosted me for a few days last October. Kim Pizer - DC, Loglan class, Primer revision review. Mike Parrish - LA, LogNet editor; also did a lot of data entry and formatting of Faith Rich's considerable Eaton page output. Ed Prentice - Boston, has consulted on several issues, and encouraged me to continue this publication at a critical time. Larry Proksch - DC/Wilmington, LogFest 3 and NB3 excerpt review. Jeff Prothero - Seattle, several consultations on grammar issues. Jeff has just completed a first cut at a recursive-descent Loglan Parser, and has donated it as a Public Domain program for the use of and for improvement by the community. Special thanks to Jeff! Bill Ragsdale - SF, hosting the Loglan Bulletin Board, including the financial contribution of supplying equipment and a phone line. Faith Rich - Word Makers Council President; the most productive Eaton worker, Faith has done dozens of pages. Joel Shprentz - DC, Loglan class participant, NB3 excerpt review. Donald Simpson - SF, contributed a new idea on colors that will be used in Loglan-88. Terri Sisley - DC, Loglan class. Steve Smith - DC, LogFest 3, NB3 excerpt review. Nora Tansky - DC, all-around support for me, and much productivity, too. Nora has participated in LogFest 3 and the NB3 excerpt review, completed the first two LogFlash programs and started on the third (as well as written a generalized version for other languages), helped in planning and leading the Loglan class, written a program for automatic C-Prim making which will be the cornerstone of Loglan-88, helped in primitive, grammar, LW, and Primer revision reviews. Contributed to Loglan-88 phonology discussions. Financial contribution. Jeff Taylor - SF, Word Makers Council Formatter/Coordinator/Secretary - he's the one who keeps things going by receiving the word proposals, distributing them to Council members, and collecting and integrating the results, all while doing his own share of kibbitzing. Meanwhile, he has also volunteered to do a major modification to LYCES/LIP to allow it to process chunks of text rather than individual utterances. He has written the closest thing to a Complex-Making program that we currently have. He is also the only person besides JCB to do significant writing in New Loglan. Helped review NB3 excerpt, Primer revision, and LogFlash 1 and 2 programs. Phil Thompson - Portland, financial contribution. Jonathan Tite - DC, Loglan class, Primer revision review, co-producer of Loglan video being planned, and a very active recruiter. Birrell Walsh - SF, Word Makers Council. Jack Waugh - DC, helping lead Loglan class, LogFest 3, NB3 excerpt review, Primer revision review; consulted on grammar and phonology issues. Tommy Whitlock - DC, participated in Loglan-88 phonology discussions, Tommy is relocating to DC at the end of May, and while staying with Nora and me for a few months, he will be our FUWA (FUll-time Washington Assistant) for Loglan-88 work. Tommy is a polyglot with skills in 5 of our Loglan-88 source languages. He also has had some linguistics training. Art Wiener - NY/NJ/Philadelphia, LogFest 3 and NB3 excerpt review. Thanks to all of you. A lot of hard work. Let's keep it going. I figure the above represents over a man-year (2000 hours) of work between September and May, exclusive of JCB/Gainesville work and Eaton work.

Summary of Logfest 3

We didn't have a big attendance at this first LogFest after GMR. But those who participated were both active and interested. We had 8 attendees throughout the weekend, with a couple of others incorporated by telephone. The experience level ranged from total novice, to old-TLer who had forgotten everything, to active current worker.

Our primary activities were completing the NB3 excerpt review, teaching about GMR to those as yet unfamiliar with it, and all-day and all-night discussions on all manner of topics Loglandic. Most of these led to questions that were either incorporated in the NB3 review or left for later resolution. Most of these can be found in our question column below.

The longest discussions were on the definitions of certain primitives - specifically: trime, patce, and matci and their relation to the problem of coming up with a good complex for 'computer'. (We came up with a solution - elegant definitions of the primitives, and a series of complexes relating to different types and ways of looking at computers. But Loglan-88 will have a primitive for computer - our discussions made it clear that the concept as a whole is too general to easily define in terms of other primitives.)

The other major activity was an examination of three Loglan speaking programs that were brought to the meeting. Jack Waugh, Nora Tansky, and Art Wiener each had their own version. There were strengths and weaknesses to each in terms of ease of use, quality of speech, etc. And we improved all programs by cross-fertilizing ideas from the other programs for certain sounds.

But the main weakness/limitation seemed to be the lack of a really good voice generator. They tend to be tailored to English phonetics, and thus do poorly where Loglan differs from English. They have varying quality of pauses between phonemes; these were found to be essential to good clear sounds, especially after stops. The stops themselves tended to be poor; I still can't tell a computer 'k' from a 'p' in some words. We were also limited by speaker quality. These devices apparently need a high quality speaker to sound good.

An essential quality to a good speaker program is the capability to reproduce Loglan stress rules. This is a non-trivial programming effort, and each program had limitations, though Nora's program had the best in this area. Jack's was probably the clearest, primarily because the voice programs are integrated into the Amiga system architecture and it has a lot of system hardware and software support for speech synthesis.


Loglan-88

In late March, when frustration over the failed negotiations with JCB about LogFlash and the copyright restriction issue were very high, several people almost simultaneously suggested a solution that would bypass legal fights, allow the issue to be judged by the proper authority - the Loglan community - and possibly improve the language as well. Specifically, it was proposed to remake the entire set of primitives from scratch, using JCB's algorithm but modifying it to use 1987 language data for language weights. The modified weights would result in a totally different set of words (morphologically, that is), and we would have performed 'significant original scholarly research', which by the copyright law unquestionably bypasses restrictions. The resulting list will be as least as valid as JCB's, although quite different, and we can make it public domain to eliminate future problems with copyrights.

Work has progressed rapidly, and we've set an ambitious timetable. We intend to have the primitives remade, with affixes, by LogFest 4 on August 1. Our philosophy is to make changes to JCB's concepts, which are the core of Loglan, only where those ideas have been questioned by linguists and a significantly better alternative is found. We also intend to rewrite the Primer appendix to reflect these changes that we've made. At LogFest, we will then seek the approval of the attendees on whether to continue. Things will be moving too fast to get feedback on proposals except at that time, although we consult with up to 20 of you on significant changes.

Our goals are a remade public domain word list (computer or printed format), comparable with the L4/L5 dictionary by January 1988, when we would make a limited GPA at the local EveCon science fiction convention, possibly with the planned video. A completed public domain word list comparable with NB3/NB4 (computer or printed format), a Primer or Tutorial, and a BNF grammar and Parser are targeted for Logfest 5 in spring or summer of next year. And a set of tested LogFlash programs, a published dictionary, and a grammar comparable to L1, should be completed by the end of next year. And somewhere in there we may find time to deal with MEX.

Ambitious goals, but we've gotten a good start. With your help, we can indeed have a Loglan for the community in 1988: Loglan-88.


What We've Done So Far

We set out to investigate the problems of word remaking, which proved to be solvable by serendipitous occurrences. JCB had originally taken 1 hour per word to make primitives, then probably repeated that effort for GMR affix-assignment. But computer technology has rendered that easily solved. In two weeks, Nora devised a program to remake C-prims by brute force techniques. We then worked together to make the program 'smart'. On the new AT-compatible we are getting, we estimate it will take only 5 minutes per word - faster than we can generate the data.

At the same time, of course, Nora and I spent enough time with each other to decide that we wished to marry. In addition, Tommy Whitlock decided to accept the use of my home as a base for job-hunting in the DC area for a couple of months. So suddenly, we had three people to work together - two of which will be unemployed for the most critical period of the work and thus able to work nearly full time on Loglan. And Tommy is a polyglot with experience applicable to most of our languages, as well as a linguist by training.

At the same time, I finished the Primer section which is appended to this newsletter, and circulated it for review to pc, Chuck Barton, and others. All of the old phonological issues that decorated TL's pages were brought up in passing. I decided to publish the Primer as a description of Loglan-as-it-is, but to take a pass through the phonological issues to see if answers could be derived that would satisfy all of the linguists - before we committed to rules for word- remaking.

The new language data was surprising, and led to new problems. There is no good current source for second language speakers. But the World Almanac has figures which could be checked. I did an independent check using population data, and information from several books on which languages were spoken in each country, including second language speakers (which I weighted at 1/2 as JCB did in his original work). The results for the two sets of data were reasonably close, so we chose the Almanac data as being on record in an easily obtainable source. But the results introduced new problems.

Since the 1950's Arabic has moved past several languages into 6th place. Chinese and Indian government efforts to standardize language and improve literacy have caused Mandarin and Hindi speakers to increase dramatically, with further increases likely to be major compared to other languages simply because of the populations of the two countries. Meanwhile, French, Japanese, and German, have had little growth - ZPG, the end of colonial empires, and the growth of English as the primary language of science have allowed lesser languages to bypass them. We were thus faced with adding Portuguese, Bengali, and Malay-Indonesian ahead of the other three, making twelve languages in six families.

It was very diverse, but unwieldy. Twelve languages were too many. We did not immediately locate dictionaries for Bengali or Indonesian, so we tested Nora's program with ten languages.

And found that JCB's algorithm broke down. First, it was very slow with so many languages. But most important - with more than three families of languages represented, we found that dozens of trial words could be constructed for some primitives with nearly identical recognition scores. The key to the recognition algorithm is matching consonants and vowels. More than three families meant that each combination of three language families tended to generate its own 'favorite' consonant clusters, that could be combine with the right vowels to get a high recognition score. But we seldom got a recognition score much above 35, which could almost be gotten with Mandarin alone. In a sense, we were spending hours to generate nearly random words with no especial virtue in recognition over other completely different choices.

So we tried cutting out various languages. We tried Portuguese as a 'dialect' second language of Spanish, and we tried returning to the original 8 languages. But we finally decided that scholarly integrity did not allow us to include Japanese, French, and German at the expense of Arabic and Portuguese, each with 50% more speakers. And we never did find a Bengali dictionary here in Washington, even with the international language community that is present for diplomacy purposes.

So Loglan-88 will be developed using six languages, with weights and supporting data as follows:

               World Almanac Data      My Independent Data
                Weight   Speakers   Primary  Secondary   Net       Weight
                         Millions                      P+(S/2)
                                            Millions
Chinese Mandarin33.5%    788        752.1     319.1      911.7        36%
English         18%      420        366.5     322.4      527.7        21%
Hindi           16.5%    382        294       200.3      394.2        16%
Spanish         12.5%    296        264.7     58.2       293.8        11%
Russian         12%      285        164.3     109.7      219.2         9%
Arabic          7.5%     177        155.9     57.7       184.8         7%

Bengali                  171        87        80.8       127.4
Portuguese               164        110.4     45.5       133.2
Malay-Indonesian         128        121.1     39.5       140.9
Japanese                 122        120.1     0.6        120.4
German                   118        105.4     18.3       114.6
French                   114        81.1      75.5       118.9


Notes:

  1. The independent check did not deviate by more than 3% for any of the weights for the 6 languages being used. But for Bengali, Portuguese, and Malay-Indonesian, it can be seen that the language order is reversed and that there is 50% variance in the net number of speakers for Bengali, which is in 7th place by WA data.
  2. We've expressed the World Almanac weights, which we are using, to .5% accuracy to eliminate ties between Spanish and Russian contributions.
  3. When affix assignments conflict or competition between primitives for the same word occurs, we will use the other languages as tiebreakers to help select alternate choices. This algorithm is not yet finalized since we don't know how many primitives will require such resolution.

Finally, in one glorious weekend, four Loglanists who will be doing most of the work (Gary Burgess, Tommy Whitlock, Nora and myself) got together and, in late night session, codified the above language selections, and a set of phonology proposals that answered all linguistic objections that had been raised. Calls to pc, Chuck Barton, and several others approved the proposals, and Loglan-88 had the first Loglan phonology with consensus approval by linguists.

The changes, briefly, are:

  1. the set of languages and weights listed above. We also will be using Chuck Barton's method of word representation - in TL2 he used 'best Loglan representation of phonemes' in word making, as is done for names; JCB represented phonemes that did not match Loglan sounds as hyphens with no recognition contribution, but he often did so inconsistently.
  2. a set of permissible initials and medials based on linguistic principles; this should better support noisy environments and diverse language origins - there are more permissible initials and fewer medials than GMR Loglan;
  3. the letter h with English pronunciation has been dropped. JCB's new x is a full consonant allowable in all word forms. q and w were rejected as not having been sufficiently justified by JCB. y is retained as the schwa hyphen. Buffering is permitted, but some other non-Loglandic vowel must be used, such as the /i/ of 'bit' or the front-schwa sound of 'put' or 'look'.
  4. the vowel set has been fixed and should be resolvable in all languages. There are no vowel disyllables, per se, that can be confused with diphthongs. The set of diphthongs to be used in CVV LWs and affixes is limited to four: ai, au, ei, and oi. In addition, the iV and uV vowel pairs will be pronounced as yV and wV diphthongs in VV LWs and names only; they will otherwise be pronounced as disyllables as follows. For all 25 possible vowel pairs, there will exist a disyllable form with a 'consonant buffer'. This is the Greek 'rough breathing' sound that English speakers hear and pronounce as 'h'. (It has been suggested that any other non-Loglandic unvoiced consonant could be used, such as the theta sound that q stood for, but we will let usage decide. The h sound seems sufficient, especially over the telephone, where Nora, pc, and I now use it in speaking GMR Loglan, pronouncing h the same as x.) The consonant buffer is the reason for eliminating the letter h. The buffer will not be represented by a letter, but by an apostrophe. Thus GMR Loglan matmaa becomes Loglan-88 matma'a and mekykiu becomes mekyki'u. The apostrophe is required to preserve the Loglan dictum of one-sound, one symbol. But it gives the side benefit of expanding our vowel set without adding to the five vowels: ai, au, ei, and oi are easily separable from a'i, a'u, e'i, and o'i. Thus we have 29 vowels for most words, and 10 more in VV LWs where they never follow consonants. And each is uniquely written, without the optional pronunciations that GMR caused.
  5. The final change is in letterals, which no longer need to occupy normal LW space. Consonants will be represented as Cy, vowels as 'V. We will have little words assigned to determine which alphabet is being represented, and upper/lower case. But less than a dozen LWs should cover the alphabets of our source languages and Greek. We expect that letterals will lexemically always be names, and must be followed by a pause, but can be used without la. Thus CIA becomes cy'i'a, pronounced (appropriately?) /shuh-HEE-ha/. (Note that capitalization is used to indicate stress, instead of apostrophe. In Loglan-88 names, irregular non-penultimate stress will be represented by capitalizing the syllable. Nothing else in Loglan-88 is required to be capitalized (although other optional reasons for capitalization exist, such as in acronyms).

The other efforts going into Loglan-88 at this time are part of our effort at 'original scholarly research'. With Mandarin the most frequent language by a wide margin, we felt it desirable to examine that language for primitives, an easy task since those primitives are the pictographs used in Chinese writing. JCB used English and Eaton as his major sources, and gives no evidence of such a study. We also will be adding to the primitive set several words dealing with human emotion - thus helping negate Robert Heinlein's criticism of Loglan in The Moon is A Harsh Mistress, various primitives proposed by Eaton workers, and a few like 'computer' which are post-1940 words that did not get properly represented in Eaton. The resulting primitive set will probably be 50% larger than the current one.

We are also attempting to add data from Eaton word making to our 'power counts'. JCB used these power counts, based on L4/L5 data, to do his original GMR affix assignment, assigning affixes based on how often a primitive word was used in complexes. Adding in as many of the Eaton proposals as we can obtain will enrich this data, and give better affix usage to the new set of complexes. The remaking of many complexes involving the overused madzo and durzo will be included in this effort.

Comments are welcome. We will publish the revised Primer Appendix in the next newsletter. We will also endeavor to clearly distinguish between Loglan-88 and GMR Loglan in material in this publication.


Plans for Logfest 4 (July-August 1987)

LogFest 4 will be held at my house on the weekend of 31 July to 3 August 1987. This is a Friday through Monday period. Most activities will be on Saturday/Sunday, the weekend proper, to maximize participation. But we'll be here the other two days with plenty to keep people busily entertained.

My house and yard have significant lodging capacity - beds/couches for at least a half dozen, floor space for more, and a half acre for outdoors types to pitch a tent or sleep under the stars (if you sleep at all). Others have volunteered to host any overflow, but I can't imagine we'll need it.

For those with families: bring them along. If they aren't Loglanists, there is plenty of convenient sightseeing. I am walking distance from Washington's METRO, and they can get downtown to the monuments and museums with no hassle and no driving. (METRO also serves Washington National Airport and Union Station for any arriving by plane or train. I can pick up any who fly into Dulles Airport, which is about 25 minutes away.) For those planning a summer vacation, I am willing to extend my hostmanship to include either the week before or following Logfest, to allow for you and your family to have non-Loglandic fun together. (But let me know, please.)

Our activities will range from novice level to advanced. We will be discussing technical details of the grammar, as well as conducting mini- classes on aspects of the language. Our primary goal for the LogFest is sustained Loglan conversation among the more experienced and the Loglan class students. This may not be possible due to JCB's restrictions on the primitive list, but we intend to try. Failing that, we intend to have some preplanned Loglan dialogues, written by the Loglan class, to serve as examples. We will record these for distribution to both attendees and non-attendees at cost.

The other major highlight will be a go/no-go decision on Loglan-88. All attendees will have their opinions considered on what exactly Loglan-88 should consist of, and what constraints are necessary.

Guest of honor will be John Parks-Clifford (pc), who will be here from St. Louis. We also have a tentative commitment from Scott Layson in California, and strong interest from Jeff Taylor in California and Bob Jenner in Maine. Thus we may truly achieve the first LogFest that is national in scope.

We hope you all will try to attend. There is no fee for attendance. Advance commitment is not necessary, though we would like to know if you might come, so we can plan for enough people. There will be another newsletter before Logfest in which I'll republish the map to my place for anyone driving, or otherwise needing it.


Loglan Questions and Answers

These are questions, ideas, and proposals, that have arisen in the last few months from the reviews of NB3 and LYCES/MacGram, LogFest 3, the Loglan class, and Loglan-88 work. We are submitting them to the Academy, since both JCB and Scott Layson will receive this Newsletter. But we also are asking the community. If you have an idea, opinion, or even a definite answer to any of these questions, please write or call. We will publish useful responses, and especially will relay any Academy statements on the issues (along with an indication that it is an 'official' answer. The questions are numbered so that you can respond by referring to the question by the number. All items in this issue are prefixed by '3-'. When replies are published, the prefix and number will also be used so that you can know to refer to the question from issue #3.

3-1. You are holding two books, one in each hand, and are talking to another person. You indicate the book in your left hand: levi bukcu ... You then indicate the book in the other hand. How do you refer to this book? Is it also levi bukcu? Leva bukcu implies it is further away, which it is not. Leti bukcu and leta bukcu should be legal, but the consensus is that they would mean 'the this's book' and 'the that's book' implying not that 'this' = 'the first book' and 'that' = 'the second book', but instead that the two books are owned by or in some way related to some other pair of things 'this' and 'that'.

3-2. MacGram must include morphological pauses (e.g. after names, before i between utterances, and after a vowel at he end of a word before a vowel in a separate word) in the grammar definition in order to prove disambiguity. Otherwise, it cannot be stated that a lexemic pause (see TL7/1) will be recognized as such. It is reasonably easy to come up with an utterance which requires a lexemic pause at the same location as the end of a name to be grammatical. Thus a listener will hear the pause as morphological and not as lexemic. We haven't tried, but it seems likely that one could find an utterance that is grammatical both with the pause interpreted morphologically and lexemically, and with different meanings in each case. The obvious alternative is to assign a LW to be used instead of 'PAUSE' for those instances in the grammar where a lexemic pause is now used. Loglan-88 will probably include such a LW.

3-3. Loglan-88 work has determined that the Chinese use several short words as indicators similarly to Loglan's indicators. Some of these should be considered for inclusion since GMR gave a few more VVs and CVVs that may be free still. Loglan-88 will add some of them, or try to find a generalized solution. We can't easily show Chinese tone symbols in this printing, so bear in mind that words which appear the same are not really:

    ai   expresses disagreement
    ai   shows surprise/discontent, also shows grief/sorrow
    ai   shows sadness/regret (a different ai and a different
         attitudinal meaning than the previous one meaning grief/sorrow)
         also regret/annoyance
    ba   indicates a suggestion, a request, or a command
    bei  agreement with reluctance
    bei  simple/easy to understand
    a    sudden realization
    a    surprise/amazement/wonder
    a    pardon?, pressing for an answer
    a    surprise
    an   interrogative

Some of these are similar to existing Loglan indicators; others show that the Loglan indicators are too broadly defined - a way is needed to express subtler shades of emotional feeling. There are more human emotions and attitudes than can be expressed with the current set.

3-4. Add LW haa as an indicator for 'don't take seriously/ tongue in cheek' comparable to the computer network symbols: (:-) or (... :-) which resemble a smiling face sideways.

3-5. There seems to be no problem with (pause)y(pause) (a solitary schwa) as a space filler in the middle of an utterance when one needs to stop and think. That is what we use in English, and it seems that it has no Loglan impact, unless the pauses interfere with lexemic pauses (see 3-1).

3-6. Proposed prims:

  1. x is an abstraction/model of y in properties w of order/degree h (default 1)
  2. x is ordered in property y as determined by rule w
  3. x is a hierarchy of set y as determined by property/rule w

These prims are useful in any MEX environment, and are basic to one proposal to solve the general MEX problem that we are preparing for eventual proposal.

3-7. Kibbitz for the WM Council: takna-muvdo is an awful metaphor for persuade (per L4/L5). Any better ones?

3-8. Per a phone conversation with JCB, I understand this as a definition of Carter vocatives, expressed for MacGram implementation. Is it correct? Have Carter vocatives been officially added to the language?

Upon hoi, stop parsing the main utterance, find the end of the Carter vocative, parse the vocative separately, then resume parsing the main utterance. Later insert the vocative parse in the main utterance parse.

The end of a Carter vocative is a pause, a predicate, or a non-argument LW. But allowed within the vocative are argument-embedded LWs and predicates (in their proper grammatical context).

3-9. The Loglan-88 proposal for letterals, described briefly above, was developed by pc and myself during my visit in October. In addition to the letter words themselves which are caseless indefinite (whether upper case or lower case), we are proposing LWs to indicate upper case lock and lower case lock (each lock for a single word), single upper case and single lower case. Alternatively, the four LWs could represent upper case and lower case with a single letter default, and word lock and indefinite lock. It can be seen that all possible combinations can be expressed with either set of 4, but preference is determined by the type of letterals one might expect to see most common in Loglan usage. Proposed LWs are (for GMR Loglan/Loglan-88) hau/xau, hei/xei, heu/xeu, heo/xeo, for the four of either set, respectively.

Additionally, a LW is used to select the language/character set. No default need be assumed, or perhaps Roman is best since Loglan uses the Roman characters. LWs to cover the major alphabets, especially including all of the source language alphabets, are (GMR Loglan/Loglan-88):

    Greek     gaa/ga'a    Roman     raa/ra'a    Hebrew    jaa/ja'a
    Cyrillic  caa/ca'a    Arabic    baa/ba'a    Devenagari daa/da'a

There remains the problem of assigning LWs to stand for letters and punctuation symbols of the various alphabets and character sets that have no Loglan equivalent. Devenagari and Arabic especially have several such labels needed, but Spanish and French have a lot of special punctuation, too. We suggest reserving an entire set of CVVs for one consonant C to stand for non-Loglandic symbols in a given character set. The definition of each such LW depends on the character set in effect, as selected by a LW from the set above.

3-9a. In addition, the LW maa can then be used to select a MEX character set, where the CVVs each stand for common mathematical symbols. Additional LWs can be used to select special purpose character/symbol sets for special scientific and mathematical fields. With these LWs, and the Loglan-88 proposal for letteral names, it should be possible to read any mathematical expression directly from its international form. Some type of quote pairs like kie ... kiu or li ... lu then serve to set off this expression. This will solve most MEX problems, and gives us at least a temporary MEX for the upcoming GPA.

3-10. Redefinitions of primitives and proposed complexes associated with the attempt to make a complex for 'computer'.

trime     x is a tool of form y assisting w to perform function h.
          The essential concepts is used and assisting. A tool has a
          basic form which may be used for a complex function. Form
          determines, or precedes function. A tool is used because its
          form is suitable to assist w in achieving the intended
          purpose.
          A hammer is a tool. (ba jia mrozu ga trime). A rock can be
          used as a hammer. (ba jia troku ga trime lo mrozu).
patce     x is an apparatus enabling function y with control/controller w.
          The essential concepts are made and enabling. An apparatus is
          made to perform a function. That function determines its form.
          The form is generally more complex than that of a tool, since
          a simple tool form could be adapted to the function without
          being specially made.
          A piano produces music (when played by a musician).
          (li piano lu ga patce lo muzgi proju).
matci     x is a machine performing function y directed by w.
          The essential concept is perform. A machine is directed to
          perform a generally non-basic or complex function. It performs
          this function with internal or automatic control. w is a
          director or initiator, not a controller.
          A radio produces music (when turned on and tuned to a music
          station by someone, who need do nothing more after these
          initiating actions). (ba jia radjo ga matci lo muzgi proju).
CVCCV     x is an entity acting to accomplish goals/purposes/intents y
          by self-direction/control.
          This new primitive is the logical conclusion of the above set.
          The essential concept is acting. An entity acts on its own
          without specific direction or control. The philosophical
          implications of this concept are uncertain but are probably
          significant in defining a Loglandic society. It is not all
          that useful now, but is likely to become significant as we
          advance artificial intelligence, and as we potentially deal
          with other species of varying intelligence.

The following complexes can then be derived:

trime-durzo              tool-do              assist (in the above sense)
patce-durzo              device-do            enable (in the above sense)
matci-durzo              machine-do           perform (in the above sense)
ponda-patce              respond-device       transponder
rulni-matci              rule-machine         computer (general term)
matci-rulni              machine-rule         (computer) program
sanpa-rulni-matci        sign-rule-machine    computer (computer science)
penso-rulni-matci        think-rule-machine   artificial intelligence computer
numcu-rulni-matci        number-rule-machine  digital computer
prase-rulni-matci        process-rule-machine analog computer
rulni-plizo-pernu        rule-use-person      person computing = computer
matci-rulni-pernu        machine-rule-person  (computer) programmer
rulni-matci-pernu        rule-machine-person  generic computer-field person

3-11. Is a sole letter word in an argument position a variable, or is it a letter word? How do you tell? In TL7/1, several corpus samples use letterals to represent some specific argument, somewhat like da-series usage; the letteral sanpa ba (stands for something). But what if you want to use a letteral as le po letra (a specific instance of a letter - representing only the phonetic sound)? If I say Nai hijra, my listener needs to know whether I am saying 'The letter N is here' or 'N. (standing for something/someone) is here'.

3-12. Is it legal to have a name start with 'Le' or 'La'? For example, my name transcribed to Loglan as I say it is Leceva'lier in GMR Loglan phonetics. Used vocatively, it could be broken into le Ceva'lier, which is a different name if le is usable before a name (as Trial 24 suggests). It has no consonant cluster, and has a final consonant, which suggests a name. But it seems that there could be problems. If my name started with La..., there would be ambiguity. And what about embedded kii and kio (backspaces), iy (the buffered dialect hyphen in GMR Loglan), and other constructs. If they are not allowed, how does one transcribe a foreign name that would need them. Using lie to indicate the LW is intended is not sufficient. Liekion could be a valid name in some language. And in any case, it is probably not practical for a speaker to comprehend such potential misunderstandings when de is speaking the metalinguistic operators intentionally.

3-12a. Imagine someone named (for spite?) Bla'nuladjan. If used vocatively, would it not break down as blanu la Djan? The consensus of those I've talked to seems to be that vocatives cause too many problems when used without hoi as a marker. It may be simpler to require hoi, and forbid 'real' metalinguistic operators between the hoi and a pause ending the vocative name. A separate LW must then be used for Carter vocatives that are not names.

3-13. How is multiple stress represented in names, if applicable in transcribing a foreign name?

3-14. Many LWs have been collapsed into the lexeme PA, presumably because the grammar is similar when used in utterances like TL7/1 H12 and H13: Na mi penso da and Na mi, penso da. Substituting sau, one gets sau mi penso da and Sau mi, penso da, which make some sense. But how about H16 and H17? What do Mi sau totco tu and Mi ji sau totco tu mean? And try substituting tie, lui, or neu for real fun. And then compound these tenses as tiesau or pacenoineu or luizu. (And can you insert the infixes -i- and -u-? How?) We've had a lot of jokes about the Whorfian effects of saying such apparent nonsenses. But one must be able to comprehend some logical if nonsensical meaning for Loglan to work with the super tense.

'3-15. How about a LW that turns any primitive or predicate (raba jia preda) into a modal operator? Per L1 pp 177-8, there are "fourteen ... most frequently required ..." which implies that others are possible useful. The proposal would allow flexibility in modal operators without likely grammatical ambiguity. Of course, per 3-14, we may still have to figure out what they mean as tenses.

3-15a. How about another LW to make raba jia preda into a discursive?

3-15b. And another LW to make raba jia preda into a case tag?

3-16. In the acronym USaiAma (USA), how does one tell that the U does not fall off and become a connective?

3-17. For each indicator/attitudinal, there must be a primitive associated with it. Else how can one describe another's 'indicated' statement? If someone expresses hope (ae) as an indicator in a sentence, I can say da spopa le po ... . But others, such as ua and oe, cannot be described to a third person very easily. If these attitudes are primitive enough to express, they should be primitive enough concepts to talk about.

3-17a. Perhaps a LW could be used to convert an indicator into a predicate meaning x indicates emotion/attitude ... to y under condition w. This would eliminate the need for primitives of little use.

3-18. How do you negate an indicator - express the negative of the attitude? There are discrepancies between text descriptions of the language in TL7, L1, and the old L3 Primer. Trial 24 would not allow A09 from TL7/1, since no would negate the sentence mi helba tu, while indicators are 'slurped up' into the previous word. In a variant, try mi oe no helba tu. How does one tell whether no goes with oe or helba, and how does one express the other?

3-19. What is the method to recognize the 'end of the word' for words used as delimiters in the new variable delimited nonce quotes (see TL4/3 Supplement pp52-53)? The Loglan resolution algorithm relies on adjacent words/sounds to resolve words boundaries uniquely. It seems that only lie W(pause)Q W(pause) is needed to be unambiguous, and other pauses between W and Q would be useful to ensure the listener hears correctly.

3-20. In the NB3 draft (pg25, para.3 line 6 from the bottom), the construct vedma la pa Rismi is found. Why la - is pa Rismi a name? It seems that pa la rismi might make sense if one is talking about the former Mr. Rice, but even this seems unobvious. Or is this a typo and it should be vedma le pa rismi? The context gives no clues, and there is no translation given. (This question is for JCB, of course, and is a comment I found after I had already sent off the rest of the NB3 comments to him. But it raises questions of general interest on the meaning of la and the use of predicates as names.)

3-21. Why are dua and due not 5-letter prims, along with bi? By using LWs as primitives it disrupts the language rhythm. A series similar to the preda series is proposed.

3-22.

  • Da blanu de means 'x is bluer than y'.
  • Da blanu means 'x is bluer than something', by implication the same as: da blanu ba. This is taken to mean 'x is blue'. But by some color measurements, green is a combination of blue and yellow, and white has all colors. So it is true to say of a green or white object da blanu; they are both bluer than something, e.g. a yellow or black object.
  • Da no blanu, by previous writings, causes the implied y argument to change from ba to ra ba, implying da no blanu ra ba. This means 'x is not bluer than anything', or 'x is the least blue thing in the universe', which is true of only one object.
  • Da nu blanu still uses ba as the y argument, thus meaning 'x is less blue than something y'; this is closer, but still doesn't mean 'x is not blue'.

So how does one say 'x is blue' or 'x is not blue', without being so extravagant in one's claim. Nora believes that comparatives should not be the form used for primitives. pc suggests loe me ba (the typical or characteristic something) as the generic argument placeholder for both normal and negated predicates. I have modified this to su loe me ba (at least the typical or characteristic something), at least for comparatives, for my own proposal. Or just leave the primitive inadequate for normal use without a second place, and define a complex blanu klesi as 'x is in the class of blue things', which can be negated with no without changing the placeholder rules.

3-22a. At the end of 3-22, I proposed blanu klesi as a complex. Is this correct, or is klesi blanu the proper metaphor? This is a typical case for those who aren't sure how metaphorical modification is supposed to work.

3-23. Another case of metaphorical modification was found in the latest Lognet, and used as an example. Nimla groda was debated as opposed to groda nimla. But both show a cultural bias on the part of the word maker. For those of us who have dealt with fantasy/science fiction or Dungeons and Dragons (tm), a large portion of Loglandia by my guess, the one thing most likely about a monster is that it is not an animal (at least in the sense of being a common Earth creature). It is a non-animal creature whose primary characteristic is strangeness, with size a frequent but not mandatory secondary characteristic. So a defining metaphor would be no(r) nimla gutra (a not-animal type of strange), although groda gutra might be acceptable. But then both are good metaphors. So why not permit both? Why limit Loglandic thinking by insisting that every Eaton concept have exactly one Loglan metaphor. Many Eaton concepts have broad definitions that allow several shades of meaning. If Loglan permits these shades of meaning with its powerful metaphor-making power, let us use it, and not bound thought by rigid and limited metaphor assignments.

3-23a. Another example from the Eaton exercise. Faced with devising a metaphor for 'death' (the event - perhaps the verb 'die' is Loglandically equivalent), two people independently come up with:

         (lepo) morto-cenja                      dead-change
         (lepo) morto-satci                      dead-start

These clearly represent two major philosophies, which Loglan should not choose between. So both should be valid dictionary words for 'death'. And we haven't even dealt with no-clivi (not alive), po-morto (state of being dead), morto-sonli (dead-sleep), and fa-clivi or fiazu-clivi (no affixes for the latter, so it may not be possible to make a complex for it; the former can also be confused with future-life, meaningful to those who believe in reincarnation). All of these are valid metaphors for 'death' as well, especially under certain philosophies; any could be devised by an Eaton worker holding those philosophies, not to mention one working without a primitive list like the current Word Makers Council is doing. Do the Loglan inventors have the right to choose a philosophy, especially given aspirations of testing Sapir-Whorf?

3-24. If there exists a primitive for the measure of something, there should be a primitive for that which is measured. For metro, we have langa. For gramo, we have tidjo. What is measured by volta?

3-24a. If Loglan is to be International, and strongly support the metric system, it should have primitives for the basic units of the International Metric System, and for the things measured by them. All other units should then be expressed as complexes of those basic units in the way they are derived, or alternately should be the ... dugri of whatever they measure. Volts are commonly used as measurements, but are not basic units of the metric system - they are derived. They should not be primitive.

3-25. What is used with jue to access a suteri argument, one which is after the second argument, without specifying the second argument. In L1, we had joe and jae, but these apparently have been dropped. This is a post-TL4/3 change.

3-25a. Was this (and other changes of its ilk that never have seen print as concepts) approved by the Academy? Should the Academy, albeit composed of skilled Loglanists, decide on a proposal of significance without it being written up (in Lognet or wherever) for comment by the community? After all, we have to use the language.

3-26. There seems to be confusion from the new forethought connectives (keks). for kimoi ... ki ... , it seems that the first phrase is the motive, given the word order, yet the motive (indicated by moi) is really the second phrase. This confuses, although there is an English translation that works: 'the motive for ... is ...' The new phrase nuku ... ki ... (or is it ku ... nuki ... ?) means '... whether or not ...'; the English definitely does not correspond in this utterance. Is kanoi ... kinoi ... legal? Does it mean 'not ... or not ...', as it appears?

3-27. Apparently la farfu converts the predicate farfu (father) to a name. The question is, is the use of la veridical? Veridical means that the argument makes a claim. Per pc, le is non-veridical, it makes no claim that the referent exists. Lo is veridical, it does claim that the referent exists, and can be massified. La is not certain; nor are the various other descriptors, such as lie, liu, li...lu, lea, loa, lua, lue, lae, and lio.

3-27a. An example of the veridical question:

la Crlak,holmz no sitfa Lndn
le Crlak,holmz no sitfa Lndn
ne lea Crlak,holmz no sitfa Lndn

In which of these three utterances must Sherlock Holmes exist, in order for the utterance to be true? Those which require existence are veridical.

3-28. Another MEX proposal. We need two LWs for number representation: one for a characteristic, one for a mantissa. This allows them to be expressed in either order, which is sometimes required.

3-29. Does ma mean hundreds, or does it mean two zeros? Does tomane mean 201, 2001, or is it illegal?

3-30. We have CVVs assigned for linguistic items (toa, toi), and for states of affairs (tau, tiu) which corresponds to po. Why not CVVs as counterparts to pu and zo for properties and amounts?

3-31. There are several grammatical forms to express relationship: le tugle pe mi is inalienable possession: it is the leg which I possess.

le tugle pe mi      is inalienable possession: it is the leg which I possess.
le tugle je mi      is my flesh leg even if it is removed from my body
                    and is no longer in my possession

But what do lemi tugle and leda tugle (and le tugle ji mi) mean as compared with the above?

3-32. lemi links a descriptor and a demonstrative pronoun to form a 'possessive' descriptor. levi links a descriptor and a tense to form a different type (non-possessive) of descriptor. Can they be combined as ?levimi or ?lemivi? Which is possessive, if either, and what does each mean?

3-32a. As an example, I wanted to include both a tense and association/possession in an utterance, along with a reason rau. Thus I tried: ?lemi na preda rau mi prede. LIP didn't like this. I tried ?lemi preda ji na rau mi prede. Still no luck. But when I dropped the rau phrase, the latter worked. And what if I had needed to put in je ... jue ... expressions into the argument?

3-32b. Can one build descriptors out of other 'tenses' - members of the PA lexeme - such as ?letie or ?lerau, or out of logically connected tenses (?lepacena, ?lepacevi, ?lepacerau, ?lepacenoitiecevi)?

3-32c. Can one add -zu type suffixes onto levi? (Does le vizu preda mean the same as ?levizu preda? How about onto the longer descriptors in 3- 32? Can one add them onto lemi type demonstratives, and what would they mean? How about the miscellaneous tense descriptors in 3-32b?

3-32d. Can one use the other simple demonstratives in this way? Lomi, lovi, leami, leavi, loami, loavi, laemi, laevi, etc. The lexer/grammar must deal with these complexities, along with combinations of these with any of the allowable forms from 3-32b and 3-32c.

3-33. In NB3, se sorme is described as now legal. Does it mean the same as se ba jia sorme? Can one meaningfully say le se sorme? How would this differ in meaning from the previous?

3-33a. If one can use numbers in this way, what would le re sorme mean? Just plain re sorme or re ba jia sorme?

3-33b. le re sorme would be very difficult for a Japanese speaker to say. How much thought has been given to the l/r problem for a Japanese Loglanist? Is there a solution?

3-34. What is the proper referent of da? How does one count when faced with:

  1. ze groups: in preda ze prede, does da refer to one of the individual predicates, or to the mixed predicate?
  2. subordinate clauses: do arguments in the subordinate phrase count?
  3. in Xai preda Yma Wma da, does it refer to Wma, or to the last argument in the previous sentence?
  4. in Nau da godzi, what does da stand for, since Nau erases all assignments for the start of a new paragraph?
  5. In the example in c: if the following sentence is de prede di do du, it seems as though all variables stand for the same thing (Wma, again). Is this true?

3-34f. pc has suggested that we need LWs for reflexives, to be used in later arguments of an utterance to represent earlier arguments. (In English: 'He shot himself', 'himself' is a reflexive.) This solves the problem of c. above, since a reflexive would be used instead of da to represent Xai, Yma, or Wma.

3-35. In TL4/3, pg. 26 middle - what is the meaning of pia? Is piazu 'a long time ago' or 'a long time period that took place sometime in the past'? Given one, how does one express the other?

3-36. Something like the 'case tags' proposal was apparently used by someone named Fillmore in several papers about AI natural language processing in the 1968-71 timeframe. Does anyone know of this work, and whether it is relevant to Loglan?

3-37. Ken Dickey and others have proposed a project to come up with 'AI frames' in Loglan. Could someone write up a good description of an AI frame, preferably with an English example, for publication (or refer me to one in print already).

3-38. Chuck Barton's primitive making approach uses the best Loglan phonetic approximation of the word, while JCB used hyphens for phonemes with no Loglan equivalent. Thus JCB would use /f-c/ for the English word fish, while Chuck would use /fic/, which is not exact, but gives good visual recognition if 'i' ends up between f and c, and better aural recognition than /foc/, for example. Loglan-88 will be using Chuck's approach.

3-38a. One curious thing about JCB's approach is that he represents diphthongs as hyphens, even when there is a valid Loglan equivalent. Thus, in L2, he represents brain as /br-n/ and not /brein/. Perhaps he, or someone else, can explain why.

3-39. A logic problem: How do we express the phrase 'coffee, tea, or milk' (or worse, 'coffee, tea, milk, or water') so as to make it clear that only one may be selected (true). Neither exclusive or inclusive or works for this combination. In fact, pc has written a program which derives 256 possible combinations of the 3-place logical expression form, and he says that there are 65536 combinations of the 4-place expression form. Loglan does well with the 16 2-place connectives. But his program finds that Loglan can only express 120 of the 256 3-place expressions using only 3 arguments and 2 connectives; some of these can only be expressed using forethought expressions (keks). How do we handle the 136 other cases? pc is working on possible solutions. Others are welcome to try, or may comment when pc sends his proposal to me. We'll submit our best shot to the Academy, or let them select from the various approaches that appear in print.

In any case, it seems that for n-place expressions, it is not practical to use the connective form. Even for n = 3, pc had to do truth tables to prove to himself that there was no way to express the 'coffee, tea, or milk' phrase with the existing connectives, and he does logic for a living. We need a generic form, which could be generalized for complex truth functions such as 'exactly two from column A and no more than three from column B' which can be really bad if A and B each have many members.

3-40. There is still a question on iglu-form borrowings. pc says that iglu itself may be OK, since he recalled that i must always be followed by a pause - apparently this came out of the Carter review. But for V- form LWs other than i, it seems unclear that the resolution algorithm will always work. Let's try an example:

  1. presume the complex ?gramrenu;
  2. put this complex in a compound argument: ?grafumna a gramrenu;
  3. per TL6/1 and the NB3 excerpt, the pronunciation could be: /graFUMna.AgraMREnu/; (the stress on the connective is optional but permitted; after the pause, it seems to be natural for a speaker to stress the connective at least a little)
  4. but this could also resolve as ?grafumna agra mrenu, which is a three-place metaphor using an iglu-form borrowing;
  5. one could go in the reverse direction as well - in at least this instance, it seems that iglu-form borrowings cause ambiguity to the resolver.
  6. for some examples, the required pause between the vowel at the end of a word prior to a initial vowel in a separate word would solve the resolution problem (as in to agra as compared with toagra; the first is pronounced /to.Agra/, the second is /toAgra/. But in the example, the one required pause is the same for both a and agra. And nothing requires a pause between a and gramrenu.

The conclusion seems to be that, except for a possible exception for iCCV words, VCCV is not a valid form for a borrowing.

3-41. A phonology problem: How are the divowels aa, ee, and oo to be pronounced? They must be two syllables, from TL6/1, as opposed to the long vowel of Japanese. But they cannot have a glottal stop between the pair, or toogra would be indistinguishable from ?to ogra (see 3-40 above). one cannot use a secondary stress to distinguish the syllables - all stress combinations are valid: toorveo has both of the pair unstressed, while toogra has the second 'o' stressed, and gratoo has the first 'o' stressed. Try saying all of these resolvably, without using a glottal stop.

'3-41a. Another example: if glottal stop is permitted to separate such pairs, then lii glumrenu resolves also as li iglu mrenu.

3-42. JCB promised in his 5/28/85 announcement that several things would be upcoming in Lognet. It has been two years, and all members have had to renew at least once, or will no longer be getting Lognet. He owes us the promised data in Lognet. Unless someone else knows the nature of these proposals. (Has the Academy approved them yet?) I'll publish them, or we ought to see them in Lognet. (Editorial note - In any case, all those who were members at the time of that announcement should be given the promised data, as well as those who joined in the period shortly thereafter, possibly in response to the promises. Anything less might constitute fraud on the part of the Institute.)

  1. handling of Linnean binomials;
  2. policy statements on borrowing/metaphorizing;
  3. "fast tracks to borrowing"
  4. the Carter changes (and other changes to the grammar).

3-43. On MEX: There have been several proposals since the 1963 grammar in TL2, but no answers. But there are some assumptions that should probably be agreed upon. If so, then we may be able to work on solutions.

  1. The existing mathematical language is international in usage, and has a logical basis, at least within each domain of mathematics. There is no apparent 'universal' form that suffices for all domains of mathematics; MEX must accept this situation as a given. The bounds of mathematics are not much restricted; thus Whorfian arguments are not a sufficient excuse to supplant the international language. Change for the sake of change, or even aesthetics, will not sell to the customers – mathematicians.
  2. We cannot/will not change the international mathematical language to a 'better system' through Loglan. Loglan has no leverage to bring about change at this time, and MEX cannot wait until we do.
  3. It would probably take another 30-year effort like the one so far to come up with a new 'universal' mathematical grammar that might satisfy a willing user. But during any given 30 years recently, several new mathematics, and their associated notations, have been developed. Unlike verbal language, which is relatively stable for centuries, mathematics is seldom stable for even a decade these days. We also would be hard pressed to even assemble a compendium of all existing mathematical domains and their notations to even start a project of this scope. There is no equivalent to a Webster's Unabridged of mathematics. The programmers and engineers of the community can all testify as to the frustration of attempting to develop a system whose requirements are not fully known, and which change faster than the development process.
  4. People write in current mathematical notation. Loglan needs a grammar that allows one to read the current standard notations (any of them), straight from the written notation. Preferably this would be unambiguously grammatical, so it would be useful to speak it in Loglan as opposed to using lie nonce quotes, or a similar construct.
  5. The chemical equation problem is identical to, even part of, the MEX problem. All that I've seen on the acronym proposal makes it seem like it was designed to solve this limited subset of MEX. Such a limited solution improperly constrains MEX, to the extent that usage will cause resistance to a later change that might be needed to make MEX work. The same comment is true for the measurement problem (JCB says that Scott Layson's TL proposal was adopted a long time ago. But I've never seen it so stated in print, and I'm not even sure that it was on the street long enough for proper review before TL folded.) Any attempts to solve these problems independent of a general MEX solution will be of limited use, and of major effect on that general solution.
  6. Thus, MEX must be a generalized way of dealing with notational systems. This is the only true MEX that works universally. Since grammar itself is a notational system, the current Loglan grammar is a subset of the generalized MEX. Thus, to solve MEX, we must generalize Loglan to a greater order of abstraction in grammar, not add constructs to the existing grammar to handle the multitude of special cases.

3-44. In TL4/3, there are defined two conflicting meanings for the LW pua. Pua is part of the Hixson-Bonewits argument tags series. But it is also the tense pa with the infix -u- to indicate habitual. LYCES/LIP only recognizes the former. The -i- and -u- infixes are not yet implemented in Trial 24.




The remainder of these questions and comments are rephrasings of comments on the NB3 draft. They are hopefully fairly clear, even though you do not have the draft text to look at. There is some overlap with previous questions, which simply means that they have come up many times.

3-45. Is it valid for buffered dialects to end names with a buffer? (followed by a pause, of course). French and Japanese speakers don't like to end words on a consonant. It seems that it should resolve.

3-46. Case tags - nobody (except JCB) has seen a list - and while nobody else has publicly taken up JCB's challenge to try their own analysis - there is a general disbelief that 13 types cover all possible argument sets. Perhaps the list of types, even without the specific LWs assigned, would help reduce the questioning.

  1. Also - how universal are they, and (how) do they overlap grammatically with the relative modifiers of L1, to which they seem similar?
  2. Is it possible that any case tag/relative modifier could be attached to a predicate - even those for which the tag has no specifically defined argument to define an otherwise non-existent argument place? The latter would seem to open up the realm of Loglandic thought still further, as ba contemplates the thought of the 'source' of le sucmi, or the 'agent' of le berci.
  3. The idea could be carried still further, with some little word which turns a following predicate word into a special-purpose case tag applicable specifically to the purposes of the speaker. A whole new use for the concept of metaphorizing.
  4. By the way, terms like 'source' and 'agent', etc. seem akin to some work done in natural language processing for artificial intelligence - the subject is called 'case analysis' and first showed up significantly in the 1968-71 time frame.

3-47. The 'se sorme' question. You may find a way to YACC it, but what will it mean in all its ramifications - and what won't it mean? Are you referring to a bare 'se sorme' or is it 'le se sorme'? Wouldn't it mean rather a 'seven kind of sister' - whatever that is? And since quantifiers are of lexeme MEX, what are the implications of MEXing quantifiers together into an expression (of currently indeterminate grammar) in place of se? Ruri sorme? Feni fera sorme?! Ri fefera sorme? (apologies for mangling the L1, Section 4.19 examples.) Will the Loglan for '(2+2) sisters' parse, or perhaps (remember story problem arithmetic) '2 sisters plus 2 sisters'?

3-48. On the apparent two meanings of lo in L1 - does lo to mrenu mean the mass of all instances of '2 men' or a specific mass individual. And how does one specify the other?

3-49. Is there a LW that is the equivalent of 'uh', 'um', and 'er' in English. That is - 'I am not finished with my utterance and do not wish to give up the floor'. We need one, or one will be invented (or borrowed from English) - and the resulting speech may be unresolvable.

3-50. JCB and I dealt on the phone with the question of how a vowel letteral is distinguished from a V LW. He said that Tai u Tai could be distinguished from TaiUTai by the mandatory pause before u in the former. But there seem to remain three other cases that this doesn't deal with:

  1. Djan UTai vs. Djan u Tai
  2. ...ITai vs. ...I Tai at the start of an utterance
  3. UTai as one of a string of acronyms used as several arguments - see JCB's examples in TL6/1.
  4. In short, there is a problem whenever there is another legitimate reason for a pause before the condensed vowel letteral.
  5. We questioned IBaiMai for reason b. above, which seems indistinguishable from . i BaiMai
  6. A comment on the acronym utterance example Kei, e, i, which summarizes the frustration expressed: 'Kei, e, i, - It is not obvious. It's downright ambiguous! Kei, e, i means someone began a sentence " k logical-and ..." and decided not to finish it. The i is the start of a new sentence, and listeners should give the speaker time to form the rest of it.'

3-51. 'nenimei' and 'toXai' - These two examples in NB3 will parse as letterals in LYCES Trial 24, since they end in a letteral. Don't know if this has been fixed in later versions.

3-51a. Nora points out that these are MEX, and that in using them, JCB dictates a MEX solution that may not be desirable. It is not implicit that in Loglan '2X' should mean '2 times X'. Nora and I agree that standard International Mathematical conventions should be able to be written in Loglan as they are internationally, with MEX grammar LWs inserted as needed when spoken (and optionally as written), which would make JCB's examples valid. (This eliminates some of the proposed MEX concepts, such as Reverse Polish). But in any case, this simple idea was controversial in the MEX discussions, and JCB should think carefully before using them in his NB3 examples. A lively discussion at Logfest was choked off so we could get work done. MEX won't be solved in a weekend. It remains to be seen whether it will take the 30 years that the rest of the grammar required.

3-52. My 'borrowing proposal' in the UL2 newsletter has met with very positive responses, although some proposed variations that are not in keeping with the metaphorical forms of the rest, such as putting the type-word first so it wouldn't have to be abbreviated to CVCV form.

3-52a. But there is a strong consensus towards establishing some pattern restrictions in borrowings, at least until we have usage information that justifies using the total borrowing space. And there is strong support for the idea that borrowings and jargon words should be arbitrarily longer than prims and simple complexes. This is partially based on Zipfean ideas - most words that would be borrowings are less frequent than prims and 2-Cpxs. And there is consensus agreement that we want to use Zipf to help Loglan grow. If borrowings are longer than most Cpxs, then people will make metaphors rather than borrow.




There should be more than enough food for thought for everyone in the above comments, questions, and proposals - whether you be expert linguist or novice. Many of those questions were asked by novices, and they may prompt you to ask some that have been bugging you. And some excellent proposals have come from novices who drew upon their non-Loglandic experience to solve a Loglandic problem. (I was virtually a novice a year ago. Most apparent expertise you see is self-taught or derives from lots of question like those above that I thought were stupid - they usually weren't.)

I've heard that many of you became interested in Loglan because it was a chance to help design a language. Here's your chance.


A Not So Quick Glossary

Some of my correspondents have been saying that there are too many unfamiliar words used in Loglan discussions. I remember the feeling well. So I've attempted to select words that are too often used in the community without definition, and I've committed academic sacrilege by defining them. My definitions should generally be accurate, but I'm not a trained linguist, and I've even been wrong once or twice on matters Loglandic. So feel free to correct me.

Oh, and writing a glossary is boring, so beware of the zingers (usually political) I've thrown in. Even Nora can't tone me down when I go manic from glossary writing.


Academy - the Loglan Academy, consisting of Scott Layson and JCB, judges all changes in the language. Appointments are for life. Changes require unanimous consent (thus don't expect rapid changes). They were inactive as a body until about March. It is not clear what changes since GMR must be approved by them, what the procedures for making proposals are, and what notification we get of proposals and of approved changes. See Word Makers Council.

affix - a CVV-, CVC-, or CCV-form abbreviation for a primitive used in making complexes. Some CV LWs also have affixes, made by adding 'r' to the LW.

AI frame - a topic in artificial intelligence. Apparently, it has to do with defining a group of related concepts in terms of their relationship with each other, so that computers can 'understand' them.

algorithm - a series of steps, or procedures, in order to accomplish a process. Several algorithms are used in Loglan, such as the primitive-making and complex-making algorithms. Per copyright law, algorithms are ideas, and not works; hence they cannot be copyrighted.

argument - We have a lot of these in Loglan. They are the one to several phrases which a predicate operates upon in a Loglan utterance. In normal order, there is one argument before the predicate, and any number of them after the predicate. Arguments include names, pronouns, and other phrases often marked with descriptors.

AT-compatible - a type of personal computer. Similar to the IBM PC/AT.

attitudinal - a word which expresses attitude or emotion about what is being said, without affecting the truth value of the utterance. In English, some attitudinals include 'Oh!', 'Ah!', and the colorations in other words/phrases such as 'I dunno'. In Loglan, attitudinals are each unique LWs, usually VV in form.

BBS - Bulletin Board System; a computer network program for several people to independently dial in and leave messages for each other.

BNF - Backus-Naur Form, or Backus-Normal Form - a standardized way of expressing the grammar or syntax of a language clearly, completely, and concisely. Used mostly for computer languages, but applicable to Loglan as well.

Board of Directors - An appointed body, headed by the CEO, which oversees day-to-day policy of the Loglan Institute, Inc. I am uncertain who the members of this board currently are.

Board of Trustees - The legal body, responsible to the State of Florida, which has permanent, long-term policy control over the Loglan Institute, Inc.

borrowing - A word which is borrowed from another language. In GMR Loglan, there are very few restrictions on borrowings, and no clear policy on how to select them. I proposed a set of restrictions in earlier newsletters, which will be strongly considered for Loglan- 88.

buffer - a sound that occurs between two consonants to render them unambiguously pronouncible. The GMR Loglan buffer is a schwa. The Loglan-88 buffer is any non-Loglandic pure vowel. Loglan-88 also has the 'rough breathing' consonant buffer which sounds like English 'h'.

buffered dialect - any spoken version of Loglan in which the speaker is unable to pronounce one or more consonant clusters. As a result, the speaker must insert some sort of vowel sound between (or after, in case of a final cluster) the letters of the cluster. In most languages, this leads to language evolution as the buffering sounds change. In Loglan, the buffer is specified to be a non-Loglandic vowel that cannot be found between consonants in any word. In GMR Loglan, the buffer is schwa (/uh/) and requires that the hyphen be changed from schwa. In Loglan-88, the buffer can be any pure non- Loglandic vowel such as the English vowels in 'bit' and 'put'.

C - a computer programming language, used in LYCES and the Public Domain Loglan Parser. Also used as an abbreviation for a consonant in describing a Loglan word form.

C-Prim - short for composite primitive; these are the primitives that are composed from phonemes that are taken from the source languages and weighted to maximize theoretical recognition to speakers of those languages.

Carter vocative - When a speaker says 'Hey, you!', he is calling you and identifying you with the 'vocative' which is 'you'. Or in 'Hey, girl in the red dress!', the vocative is 'girl in the red dress'. These types of vocatives, in which a phrase is used as a naming device, are called Carter vocatives, after James Carter who first pointed out the need for them. The 'vocative' in a Carter vocative can be any Loglan argument form.

case tag - a late addition to GMR Loglan; case tags are labels for each argument that may be used instead of Hixson-Bonewits operators. These labels indicate what type of argument is being referred to - a unique assignment for any primitive. Thus a speaker can rearrange arguments in an utterance without a thought to their 'nominal' order in the Loglan definition of the primitive.

characteristic - a decimal value used in logarithms; also, sometimes, the number which is raised to an exponent.

comma-delimited - a format that I originally devised for dictionary entries - now adopted by JCB as the standard for all dictionary work. I later discovered that the old dictionary format did not convert well to comma-delimited format, and that people hated to count commas, so I devised the Universal format - which nobody likes either. JCB wants all Eaton words in comma-delimited format, defined in UL1. I'll be taking inputs for Loglan-88 in any convenient and consistent format. I'll be using a database (DBASE III+) to organize and build dictionary entries.

complex - a Loglan word which represents a metaphor of two or more primitive concepts. It is built from the primitives using the complex-making algorithm defined in UL1. Both GMR Loglan and Loglan-88 use this algorithm. I believe JCB has arbitrarily decided on his scoring approach described in the article, with a tie- breaker rule for when two words using CVC affixes have the same score.

compound - more than one LW attached together into a single longer LW; since in general, the resolver will break these down into atomic LWs, there typically is no semantic difference between the component LWs and the compound - the combination into a compound eases reading and reduces pauses in mid-LW-string that could cause confusion.

connective - one of the LWs 'a', 'e', 'i', 'o', 'u' or their derivative forms. The Loglan atomic connectives represent 'and/or', 'and', the sentence connective, 'if-and-only-if', and 'whether-or-not', respectively. They are also called 'eks'. Their cV derivatives are called 'ceks', and their kV derivatives are called 'keks'.

corpus - a set of test utterances used to verify that a version of MacGram is valid. The tester generates the corpus, which may have several hundred utterances, then derives what he/she believes the correct parse to be. This is the compared with the output after LYCES/LIP get through parsing the corpus. Trial 55 (and version 720 of the PreParser) is apparently the first grammar version to successfully parse the corpus that comprises TL7/1.

CP/M - a computer operating system designed for the Z-80 microprocessor. It is now obsolete, but found on many older personal computers.

CV, CVV, CVC, CCV. CCVCV, CVCCV, etc. - these are abbreviations for Loglan word forms, which are rigidly defined as to structure (morphology). The C in a word form stands for any Loglan consonant; the V stands for any Loglan vowel.

default - a computer term adopted for Loglan; when an optionally specified piece of data is omitted from a statement, the reader/listener must make an assumption as to the value of the omitted data. The standard value to be assumed is the default.

delimiter - another borrowed computer term. A delimiter is a symbol or separator, generally occurring both before and after the data to be separated. It typically has either a standard value (such as li ... lu), but may be a variable not found in the delimited data (as in nonce quotes). In English, parentheses and brackets are delimiters.

demonstrative - a pronoun which requires knowledge of the physical situation at the time the pronoun is used. Typically, the speaker will physically indicate (demonstrate) what the pronoun stands for when using it. 'This' and 'that' are demonstratives in English.

descriptor - one of the LWs that labels an argument. They typically start with 'l'. In English, they are called 'articles'.

diphthong - a pair of vowels spoken as a single syllable, generally so quickly that they blend into a single sound. Unfortunately, the blend is weak, and in many cases a listener (especially from a different culture/native language) will hear the two sounds separately, or otherwise mis-hear them. Most English vowels are diphthongs, whereas all single letter (V) Loglan vowels are pure single vowel sounds.

discursive - one of several LWs that clarify semantic relationships and meanings, but which do not affect the grammar or the truth value of the statement. Transition words and phrases, such as 'however', or 'in short', are among the English discursives.

divowel - any vowel pair. In Loglan, all divowels are either pronounced as two syllables, or are diphthongs.

Eaton - a book written by Helen Eaton, and published about 50 years ago. It represents a study of the most frequently used words/concepts in four European languages. As the only major reference on word frequencies, it is a standard for completeness of the Loglan vocabulary. Unfortunately, it is criticized as a standard for several reasons, including ambiguity of the concepts represented, obsolescence (the four languages have changed significantly since the 1920s, when most of the research was done), and cultural bias towards the European, and specifically Romance, language. But it's all we have. The book is out of print, but can be found in libraries.

Eaton interface - a long-term project with the goal of devising words (primitives or complexes, preferably) for each of the words in Eaton. The Word Makers Council coordinates this effort when JCB isn't doing so himself.

Eaton number - Since only one edition of Eaton is extent, a given word is always found at the same place on a specific page. JCB thus decided to use the page and word position on the page as a unique identifier for a concept. By sorting words in Eaton number order, the completeness of the Loglan vocabulary can be easily assessed, and other readers can easily find a word in their own copy of Eaton.

employee - per Webster, someone employed for wages or salary; see 'unpaid employee' and 'volunteer'. The product - 'work for hire' - (and derivative copyrights, usually) from the working hours of employees is owned by the employer in return for the financial consideration. On the other hand, the work of a contractor (person paid by commission for a single or specific task) is shared by the hirer and the contractor - the specific work belongs to the hirer, but copyright belongs solely to the contractor (although this may be modified by specific agreement).

FIDO - a local computer bulletin board that connects into other local boards forming a nationwide network that allows inexpensive mail services. I don't know much about FIDO, but I have access to one board. Anyone else?

forethought - a type of connective that indicates the connection before the first statement, thus allowing the listener to know the relationship between two terms before hearing either. The name derives from the fact that the speaker must have thought out the statement before starting in order to know the connective to be used.

FUWA - FUll-Time Washington Assistant; Tommy Whitlock will be assisting Nora and me in preparing Loglan-88, while he hunts for a job. Nora will also be job-hunting, but for a while both of them should be full time Loglan workers.

glide - a sound made when moving the tongue and/or other mouth parts when moving to or from a vowel sound. The English letters 'y' and 'w' represent the sound of the two primary glides.

glottal stop - the brief pause made by the closing of the epiglottis in the back of the throat. English speakers use glottal stops most frequently at the start of words (note how your throat feels if you start a word with a vowel, such as in 'able'). Loglan treats a glottal stop as a pause. You cannot use one in the middle of a word, such as between the divowels 'ee', 'aa', and 'oo', or they will break up. This makes them difficult to say as two distinct syllables, hence the change in Loglan-88.

GMR - Great Morphological Revision; the result of an effort led by JCB from approximately 1979 to 1982. The result was a complete change in the rules for word making (especially for complexes and borrowings), a set of resolvable affixes for many of the most frequently used primitives, and the remaking of a large number of primitives in order to make things work properly. A more standardized and easy to learn language was achieved, but the manner of its proposal and approval apparently started many of the political (as opposed to linguistic) arguments that abound in Loglan today. JCB continues to make changes to primitives and affixes under the apparently open-ended (to him) charter of GMR. I call his existing language GMR Loglan to distinguish it from Loglan-88.

GPA - Going Public Again; a term coined by JCB for the next major republication of Loglan, which will bring in new people and hopefully money, while greatly expanding the usage of Loglan. GPA has been promised since GMR was 'finished' in 1982. JCB unfortunately chooses to delay it each year, for a variety of reasons. Lately, JCB has referred again to a delay in GPA, possibly for several years. He is instead aiming for a 'Turnover', when he will allow the community to try out his new language for a while before he finalizes it for GPA. The Loglan-88 effort stems partly from our refusal to wait much longer for GPA. Our ability to use Loglan freely in a much-expanded community depends on GPA. We also believe that circumstances seem conducive to successful introduction of a working Loglan into the AI community now (the window may close later); further delay reduces the chances of eventual success. Loglan-88 is so named because it will GPA in 1988, hopefully at the beginning, with articles, a video, and convention appearances by skilled Loglanists. Loglan-88 book publication will hopefully also occur in 1988.

grammar - the set of rules that tells how words of different types (lexemes) may be put together. This rule set is humongous for most natural languages, such that no effective algorithm exists for computer parsing of a natural human language. Computer languages have much simpler grammars, and Loglan's grammar has a rule set on the same order as one of the more complex computer languages.

hack - computer term for a short-term (usually brute-force - hence the name) fix to work around a problem because the 'hacker' doesn't wish to spend the time necessary to think the problem out thoroughly - which might eliminate or redefine the problem. Several Loglanists have become dissatisfied with the course of the language in the past few years because grammar, morphology, and word changes seem to be hacks. The most evident of the hacks has been the collection to be found in the PreParser of the Machine Grammar.

HAGA - HAlf-Time Gainesville Assistant; JCB has been collecting funds for the HAGA for about 1 1/2 years. The HAGA was intended to be his local assistant to expedite GPA and keep the Institute going when he went off to Job Market work. I don't know how close he ever got to his $7000 target for one year support of HAGA, but Nora and I contributed significantly to the Institute. Apparently, close enough, since someone named Paloma is apparently on the Institute payroll, presumably filling the intended role.

Hixson-Bonewits - These two Loglanists proposed the use of argument labels - one for each argument position - as a means to allow efficient, non-confusing, rearrangement of arguments (in place of the nu, fu, ju ... series conversions). As a result, we have the pua, pue, pui, puo, puu series, which label the first through fifth arguments, respectively. Thus, one can say 'pue arg2 puu arg5 pui arg3 pua arg1 puo arg4 preda' as valid and completely understandable Loglan. There is an unresolved conflict between pua as defined for these labels, and pua as the -u- infix form of pa, which means 'habitually in the past'.

hyphen - a morphological concept; the hyphen is used to link two affixes together when morphological problems would cause the linked juncture to be illegal (unpronouncible, unresolvable, or prone to falling apart). The hyphen sound is the schwa, represented by 'y'. In GMR Loglan, when using a buffered dialect, the schwa is the buffer, so the hyphen is modified to the sound of 'iy' or /yuh/ (but is still written 'y' - the sound of 'y' apparently is modified). Loglan-88 does not use schwa as a buffer to eliminate this variation.

I/O - computer term; abbreviation for input/output.

indicator - see attitudinal; these LWs indicate the attitude of the speaker towards his/her utterance in which they are included.

infix - a letter or syllable inserted in the middle of a word to change the word meaning systematically. Compare with English usage of prefixes and suffixes, which are inserted at the beginning and end of words, respectively. Loglan uses a limited number of infixes to modify the meanings of tenses.

Institute - either of two concepts, usually depending on the timeframe referred to; prior to 1975, The Loglan Institute existed and published early books about Loglan. The IRS ruled that this Institute, which published the first version of L1, as well as L2 and L3, is a pseudonym for JCB. At the time when L1, L4, and L5 were last published, JCB formed The Loglan Institute, Inc., a non- profit corporation chartered under Florida law. This organization is distinct from JCB, even though he is CEO (Chief Executive Officer) and member of the Board of Trustees and the Board of Directors. Many of you (and Nora and I) are Members of this Institute, though the Institute has not informed most of us just what Membership means.

JCB - Dr. James Cooke Brown, inventor of the original Loglan, Founder of the Institute (both), author and publisher of all books written about Loglan (and most articles), CEO of the Institute till late 1988, as well as Trustee, Director, and member of the Loglan Academy. JCB also coordinated the GMR effort (and thus, GMR Loglan). He has no role (as of yet) in Loglan-88.

Job Market - a non-Loglan concept invented by JCB in his utopian novel The Troika Incident (published 1969, but out-of-print - JCB is always interested in locating copies of the book to acquire for new readers and classes using the book - contact him if you locate an available copy). The concept is an economic idea which relates to achieving maximum employment. He described it only briefly in the novel, and has long promised supporters of the concept to write a full length exposition of the idea. This project will remove him from Loglan work for about 3 years, starting whenever he starts on it.

kek - a short term for a forethought connective, one that appears before the two terms being connected. The simplest connectives are termed eks; these are called keks because they start with a 'k' and are related to eks.

kibbitz - the reviewing of others' work in making words, hopefully constructive criticism to improve the metaphors made in the Eaton interface project. JCB wants lots of people to volunteer to kibbitz. (I presume he means volunteer, and not unpaid employee, in this case - but one never knows.)

L1 - Loglan 1 - A Logical Language - a descriptive introduction to Loglan written by JCB and revised several times. The last publication was in 1975. It is not in print and few copies exist. It is outdated due to GMR, and has always been incomplete and occasionally confusing, contradictory, and inaccurate. But it was well written as an introduction, and brought a lot of people to a reasonable level of understanding of the language. It needs revision, and JCB plans to do so for GPA, but it isn't clear that he intends to do the thorough job it needs to become the quality document that it could be.

L2 - An out-of-print description of the theoretical principles JCB used in originally devising Loglan, last published in 1968. Most or all of it was reprinted in TL 1 and TL 2. Excellent for understanding what JCB originally intended.

L3 - An out-of-print attempt at a Loglan Primer; this is a programmed text which covered only a small portion of the then-Loglan.

L4/L5 - The Loglan dictionary. L4 is Loglan/English; L5 is English/Loglan. Out-of-print, but JCB has lots of copies in paperback left - if he decides you are worthy of his time to fill your order. The dictionary, is of course, being revised by the Institute for GPA, including the GMR and Eaton changes. We will also be independently producing a Loglan-88 dictionary; it will of course have no resemblance to the other dictionary, since all primitives and affixes are being remade from scratch, and the complexes will be derived from the new primitives and affixes.

letteral - a LW which stands for a letter of the alphabet; a letteral is to a letter as a numeral is to a number. JCB has done extensive but undocumented work in deriving a scheme for massively using letterals and acronyms in the language. lexeme - a grammatical word type, similar to the 'parts of speech' you studied in English. In formal grammars, however, there are many subcomponents of the basic parts of speech - each is a unique lexeme, with its own rules for combining with other words. In fact, many words in natural languages are their own independent lexeme. Loglan has a small number of lexemes (under 100 by MacGram count).

lexemic pause - a pause used as a lexeme, in other words, with grammatical meaning when combine with other lexemes. The inclusion of a pause at certain points in MacGram/GMR Loglan sentences may drastically change the meaning. Loglan-88 will put a non-pause LW in place of the lexemic pause, since many Loglanists believe the lexemic pause to be unusable - a natural speaker inserts pauses semi-randomly as part of speaking; these pauses could create grammatical ambiguity. In addition, it is unclear whether there is conflict between morphological pauses (those required to resolve words correctly) and lexemic pauses.

lexical elements - an undefined term that JCB uses to indicate what he believes he and the Institute hold copyright over. The term isn't part of written copyright law. It seems to include the Loglan words (both primitives and complexes), the metaphors, and the grammar. But copyright law does not allow copyright of ideas, hence the metaphors are out. (Most metaphors have been made by Eaton workers, anyway, and these have not as a group transferred their limited ownership of their creations to the Institute in writing.) Nor may mathematical algorithms be copyrighted (except possibly the exact form or expression thereof that is published), hence the grammar definition is out. The complexes are derived algorithmically from the metaphors and affixes/primitives, so changing either group makes the copyright meaningless (there has never been a word-list published by the Institute with the latest set - presumably this will be in NB4). Incidentally, it isn't clear that a single word is copyrightable under current law, though a word list is.

linguist - a student of linguistics, not necessarily someone who knows a lot of languages (see polyglot). This one has always confused me.

Linnean binomial - the Greek-based scientific names for plants and animals; they have two parts for genus and species, hence are binomial. Carolus Linnaeus invented the system of classification. JCB has reported but never described a system for Loglanizing these terms.

LIP - Loglan Interactive Parser; a computer program written originally by Scott Layson, with contributions from JCB, Jeff Prothero, Sheldon Linker, and several others. LIP is part of the LYCES program set; it incorporates a parser generated by YACC from the MacGram definition. It takes an utterance you type in and generates a parse indicating the lexemes and their relationship to each other for a grammatical sentence (according to the program's definition of grammatical). LIP is the de facto standard of correct Loglan grammar, since no other definition exists. But it has bugs and hacks, and almost no one has a copy (though JCB offered it for sale some years back).

Logfest - a mini-convention of the Loglan community where we try to get together and interact in and about Loglan. There have been three Logfests so far. the first two were held in Boston several years ago. Logfest 3, as reported in this issue, was held last September. Logfest 4, the biggest and best yet, will be at my house (and yard as necessary) the first weekend in August this year.

LogFlash - the current name for Nora Tansky's and my teaching program set. The name was changed to eliminate confusion with MacTeach. We have two programs working and well tested for IBM PC, MS-DOS generic, and CP/M (and a volunteer to convert it for the Macintosh), but there is no sense distributing it without a word list. JCB will not allow us to distribute a word list of GMR Loglan, and is not distributing one himself without a non- disclosure agreement. So Loglan-88 will render his restrictions moot.

Loglan - the name of:

  1. several different versions of an artificial language being developed since 1955;
  2. part of the name of a nom-profit Institute for linguistic research;
  3. part of a former pseudonym for JCB (see Institute);
  4. part of the name of an Academy intended to standardize the language (see Academy);
    But, most important, it is:
  5. an idea, and a dream that has inspired hundreds of supporters and workers for parts of the last 30 years - inspired them enough to fight for that dream.

The language was first described in print by JCB in 1960, and has been mentioned or discussed by several other authors over the years, including science fiction author Robert Heinlein and novelist Robert Rimmer, as well as by pc and many contributors to TL, Lognet, and some non-Institute publications, such as this one. At least 7 versions of the language can be identified, versions by JCB and his ex-wife in 1960, 1962, and 1968, as well as the 1975 edition, GMR Loglan, MacGram, and Loglan-88. The basic ideas behind the language haven't changed, but the details have evolved considerably over the years.

Loglan-88 - the newest version of the Loglan language under development, and the first being developed independently of JCB (he will first hear of it when he receives this issue). Its sole original purpose is to eliminate silly fights over legal ownership, and place a usable version of the language solidly in the public domain, where it hopefully will prosper. The 88 refers to our target of GPA in 1988, and the name has been varied from the generic name to ensure that you can clearly tell the difference between it and GMR Loglan when we (and others) discuss both.

Lognet - the printed bulletin board of the Members of the Institute, nominally a publication of the 'Members Council' which has never been constituted, and specifically forbidden by contract and bylaw to be controlled by JCB. However, it currently is the house organ of the Institute, each page claims copyright for the Institute (not the Members), and it is distributed as JCB sees fit. Mike Parrish is the Editor, and he did a lot of work composing JCB's writings into a neat-looking newsletter.

LW - abbreviation for 'little word'; one of the many short words in Loglan. LWs are shorter than four letters long, or are compounds of other shorter LWs.

LYCES - Loglan-YACC Corpus Eating System - a set of computer programs used in developing MacGram and LIP. Written in C, the latest versions run under MS-DOS, and are copyright by the Institute. Contact JCB to determine if they can still be obtained as advertised a few years ago.

MacGram - short for machine grammar (nothing to do with the Aplle MacIntosh); MacGram is the formal language definition for Loglan, as formatted for input to the utility known as YACC. YACC verifies that the resulting grammar is unambiguous and generates a parsing program for it (which is used in LIP). MacGram is composed of a formal language definition in a form similar to BNF, and a set of algorithms in the PreParser which analyzes words and assigns them to lexemes. In addition, it performs several hacks which avoid or solve ambiguity problems. These result in MacGram being slightly different from the human-speakable grammar of Loglan, a highly undesirable state. As a formal language definition, MacGram is considered a mathematical formula. The PreParser code expresses an algorithm for lexical analysis. MacGram is thus explicitly not copyrightable under current law, and Loglan-88 and the Public Domain Loglan Parser (and you) can use the resulting grammar without violation of copyright.

MacTeach - short for machine teacher (nothing to do with the Aplle MacIntosh); This is the long-promised teaching program series being produced by JCB and Glen Haydon. Three programs have been produced that cover the same ground as the first two of Nora and my LogFlash program series. Early test versions of LogFlash programs were entitled 'MacTeach' as well. The name change to LogFlash has been made to eliminate confusion with the Haydon-Brown programs that are 'official' (but we feel are less effective than our own).

mantissa - the exponent part of a logarithm or a number in scientific/exponential notation.

medial - middle; in Loglan morphology, primitives and complexes often have consonant pairs or triples in the middle of the word (medial consonants). There is a restricted set of permissible medial consonant pairs. See the Appendix for further details.

metalinguistic operators - refers to a small number of LWs that are used to deal with linguistic elements of the language, as opposed to grammar and semantics. These include the backspace operators kie and kio and the various quotation/parentheses such as li...lu, among others.

metaphor - JCB uses a definition of this term in Loglan that is not the standard linguistic definition. A Loglan metaphor, also called a 'binary metaphor', refers to a pair of concepts (each expressed as a predicate in Loglan) that are paired together; their combined meaning is derived from or suggested by their separate meanings, the combined definition need not exactly match, however. Three or more predicates may be grouped together. Conceptually, the metaphor-maker will build such longer metaphors from pairs of existing predicates and shorter metaphors. The analysis of metaphors must always be performed by determining the two components, and then determining their relationship to each other. Loglan has rules for expressing complex groupings of metaphor pairs - this is the 'pretty little girls school' (PLGS) problem raised in L1. But there are less specific rules for determining the effective meaning of a given metaphor pair.

MEX - short for mathematical expressions; there has been much debate over how mathematical ideas should be expressed in a logical language such as Loglan. JCB has proposed several ideas over the years, as have others. The problem is currently considered unsolved/unresolved, (and JCB has stated that it will not be dealt with prior to GPA), although JCB has put a few MEX concepts into GMR Loglan that may constrain an overall MEX solution.

modal operator - the Loglan equivalent of a preposition; there are 14 modal relative phrases per L1 section 5.10 that can be used to attach extra arguments onto a predicate. By the evolution of MacGram, JCB has made these operators the grammatical equivalent of tenses. They may thus now be used anywhere that the more standard tenses can (though the meaning of some such usages is unclear).

morphological pause - Loglan requires some pauses in speaking in order for the word resolution algorithm to properly separate words. These pauses include the pause after names and those between vowels in separate words, as well as the one before i at the beginning of an utterance. These mandatory morphological pauses are distinguished from lexemic pauses (see above), and optional pauses that a speaker inserts naturally for rhythm, or as he/she stops to think, stutters, or takes a breath.

morphology - the set of rules for forming words in a language, or the study thereof.

MS-DOS - an operating system for the IBM PC and compatible microcomputers. Most microcomputer software is written to operate in an environment that is controlled by MS-DOS.

NB1 - Notebook 1; this notebook reported on the research that led to the first version of MacGram which YACC found unambiguous, and on the later version that parsed the first corpus. This version of MacGram was Trial 19, which was obsolete within a month after NB1 came out.

NB2 - Notebook 2; this notebook reported on the results of GMR. Most people thought GMR was complete when it came out, but it also was obsolete within a month. Although GMR was approved as complete several years ago, JCB continues to make fine tuning modifications to primitives and affixes as 'part of GMR'. This is one reason why no version of the GMR Loglan primitive list is accurate and up-to- date.

NB3 - Notebook 3; this notebook has been promised by JCB for about a year and a half as the description of 'New Loglan', which I call GMR Loglan to distinguish from Loglan-88. It was originally intended to contain all information necessary for a Loglanist to learn the current language, but I believe he has reduced the scope partly in response to our comments on the first 30 pages. In any case, he has removed most word lists from NB3, and put them in NB4.

NB4 - Notebook 4; this notebook is a recent addition promised to come out eventually. It will contain multiple word lists occupying several hundred pages by my calculations. But conceptually it is the other half of NB3. We are hoping NB4 will be made available on computer disk, where it can be effectively used.

operator - a type of LW that 'operates on' the semantic meaning of other words that it is tied to. Modal 'prepositions' are operators, as are connectives, tenses, and descriptors. Indicators stand alone, semantically and grammatically - they are not operators.

parser - an algorithm, usually in the form of a computer program, which breaks up a string of symbols (speech or text) into lexemes, categorizes them, and determines the grammatical structure that underlies the input. If the input is invalid grammatically, the parser should detect it. If the grammar is ambiguous, the parser must be given rules to choose which alternative is correct.

pc - John Parks-Clifford, professor of Logic and Philosophy at University of Missouri/St. Louis; editor of TL while it was alive; President of the Board until politics intervened; currently, advisor and supporter to the Loglan-88 effort. pc is one of the most skilled Loglanists outside of Gainesville; he has linguistic and AI experience in addition to his academic standing, hence he can credibly serve across the spectrum of Loglan efforts.

permissible initial - in Loglan morphology, specifically in CCVCV primitives and their complexes, there is a limited set of permissible CC pairs that are allowed to start the words. See Appendix for details. See also medials.

phoneme - an individual unit of sound; the Loglan ideal is audio-visual isomorphism - i.e. one phoneme corresponds to exactly one symbol (letter, punctuation, etc.). As originally designed, this was true of Loglan. But lexemic pauses (see this term and morphological pause), optional disyllabization of vowel pairs, the variations in pronunciation of i and u in diphthongs, and confusion over the pronunciation of eV divowels such as eo have made GMR Loglan suspect in this area. Loglan-88 will attempt to return to the fundamental principle.

phonology - the rules for pronouncing sounds in a language, or the study of those sounds. Phonology, morphology, and grammar are the three areas of Loglan linguistics that are most talked about. Semantics remains the most unstudied area of Loglan linguistics. pictograph - in ancient languages, one symbol pictured an abstract form representing an object or concept that the symbol stood for. This picture-writing lasted until the invention of the alphabet in the West. In the Orient, pictographic writing evolved so that in Chinese, for example, one symbol still represents an idea, but no longer resembles a picture of the referent.

polyglot - a speaker of several languages, such as Tommy Whitlock, our new FUWA.

power - besides the obvious innuendos I might make about power and Loglan in this era of politics, JCB has used the term power in morphology to indicate the usefulness of a primitive in complex- making. Primitives with high power thus received priority selection of affixes over lower power words during GMR. Some primitives have so much power, they received two, or even three, affixes. The GMR power assignments were based on L4/L5 dictionary entries. Loglan-88 will use a similar power-based affix assignment, but will use as many proposed metaphors from Eaton work and other word-making efforts as we can gather. We are, after all, trying to assign affixes based on the final form of the language rather than the current form; there is no way to exactly know the eventual power of a primitive in making metaphors, but increasing the statistical base should reduce the error.

predicate - in English grammar, this is the verb plus object that makes a claim about the subject of a sentence. The logical basis of predication has been made explicit in Loglan - the predicate still makes a claim, but it does so functionally (as in a mathematical function). Like the computer language PROLOG, Loglan statements consist of a predicate and a set of several arguments. (In Loglan, one of these - the first, usually, corresponds to the English subject.)

PreParser - a portion of the LIP computer program that handles various portions of the grammar that cannot be unambiguously processed by the YACC-generated Loglan parser. This includes some necessary preprocessing like the breaking up of compound LWs into their components, as well as some possibly necessary processing like the insertion of 'machine lexemes' which allow YACC to more easily achieve disambiguity. But it has lately collected a lot of hacks, some of which may be concealing ambiguity in the language.

Primer - a project that has been started several times, but never completed. A Loglan Primer would teach a novice, non-linguistically oriented person to some degree of fluency in Loglan, starting from scratch. The earliest Primer, L3, used programmed techniques and several hundred pages to teach the first 2 or 3 lessons. More recently, Chuck Barton started a Primer during the GMR process, which was changing the language as fast as he could write about it. The first part of Chuck's work has been rewritten so as to be current for GMR Loglan- it is found in the Appendix to this newsletter. Your comments on its usefulness are appreciated.

primitive - in Loglan, one of the approximate 1000 root concepts that are considered basic to the language. Other predicate concepts are generally expressed as metaphors from these primitives.

Public Domain - something published and not subject to copyright constraints. It thus may be used by anyone, and can be modified, reprinted, etc. as needed. Something in the Public Domain is thus less controllable. However, as in the microcomputer software marketing technique known as shareware, it can still result in income to the author. In general, the freer the access, the easier it is to spread usage, and the less formal marketing and publicizing is needed to gain acceptance.

reflexive - a pronoun that refers to an earlier argument in the same sentence. In English, this usually is an object reference to the subject, such as 'George hurt himself' (himself reflexively equates to George). In Loglan, with unlimited arguments, it is possible to want to refer to any of the previous arguments reflexively, which is a more complex proposition. GMR Loglan does not clearly support reflexive pronouns.

resolver - an algorithm to resolve sounds into words, and hence into statements. This algorithm could be programmed on a computer for Loglan - the language is unambiguous in morphology (although this hasn't been proven using YACC, it seems obvious - but see 'phoneme' for some problems in GMR Loglan). The algorithm is intended to be simple enough that it can be performed subconsciously (but without error) by a listener. This also has not been proven with GMR Loglan, although earlier Loglans were speakable and understandable with minimal effort.

Sapir-Whorf - a hypothesis that can be expressed as 'the limits on human thought are constrained by the language used to express thought'. This hypothesis was in vogue in the 1950s, and Loglan was originally proposed as a test of this hypothesis. Arguments about its form of expression, its implications, and its usefulness have caused the hypothesis to become a footnote in linguistics nowadays, but the original goal for Loglan is still cited as a reason for or against proposals for changes. The idea of a test is still theoretically valid for when we have a large enough community that fluently speaks Loglan as a primary tongue.

schwa (/shvuh/)- the unstressed neutral vowel for English and many other languages; it is the sound of 'u' in 'but' and of 'a' in 'sofa'. In GMR Loglan, schwa is used for both hyphenation and buffering (though not at the same time). In Loglan-88, schwa is used only for hyphenation.

sentence - a complete expression that states (statement), asks (question), or commands (imperative) something. A sentence in Loglan has a predicate and optionally one or more arguments. (A bare predicate is an imperative form.)

shareware - a concept, often under other names, rapidly promulgating a marketable idea (usually a computer program) without prohibitive marketing and advertising costs. Versions of the program are freely distributed (with sufficient documentation to enable them to be easily used) via on-line computer bulletin boards and free copying. The recipient of a copy is invited to try the program out. If it is useful, and especially if the user wants to receive notice of updates to the program and other continuing support, the user is asked to send some amount to the author in payment. If the support is valuable, and other related materials are for sale, the author makes considerable income with minimal expenditure, and exposes his ideas to the maximum number of people. If others can improve on the author's product, they of course can market it independently. But since the original product is public domain, any such improvement must be significant to allow the reviser any control. It has been proposed to spread Loglan via shareware concepts to maximize our audience. Since only an organized effort such as the Institute, or our SIG could provide meaningful support for the language (and then only with significant preplanning), putting Loglan in the public domain via shareware does not reduce control over the language (To make a change, one must spread it to all users, and the original 'sharer' who is contacted by users has the most organized access to those users. Also, others cannot compete until they can offer comparable levels of support, i.e. dictionaries, teaching materials, etc.)

SIG - special interest group; one of the possible names for our as-yet- unofficial organization. Many technical societies, especially in the computer field, have SIGs for the several special disciplines that subsets of the membership are interested in. They are often informally organized. Loglan is a special discipline for computer people as well as for linguists, so the concept is appropriate to our group. We have attempted to break our organization into geographic SIGs, since the most logical subdivision of Loglan work is the community that can gather together and use the language with each other.

stop - what I'd better do with this glossary, which has gotten too long. But also, in phonology, a phoneme which interrupts air flow completely, thus causing a break in sound. The letters 'p, t, b, d, g, and k' are also stops. The glottal stop is a phoneme where the epiglottis, in the throat, does the stopping.

stress - a form of relative emphasis on one or more syllables. Stress can affect meaning in some languages. In Loglan, however, stress is used as part of the resolver - it helps determine where word breaks are in uninterrupted speech.

subordinate clause - a clause with a predicate which is a subsidiary part of a sentence; in an argument, a subordinate clause identifies the argument more completely to ensure understanding. The clause may be a complete sentence with its own subordinate clauses, ad infinitum. A subordinate clause can be either non-restrictive or restrictive. The difference depends on whether the qualification must be true (or may be false) in order for the main sentence to be true. The difference is shown in:

The hat which you bought is green. (restrictive) The hat, which you bought, is green. (non-restrictive)

The first one is restrictive. It identifies the hat in question as being restricted to the one which you bought. The second is non- restrictive. It claims that you bought the hat. In addition, it claims that it is green. In GMR Loglan, the first type of clause is connected with jio. The second type of clause is connected with jia.

Other types of subordinate clauses include relative clauses such as 'after work' (fa turka), and qualifying clauses such as 'Because he was a teacher' (inurau da pa ditca).

Supplement - another name for TL issue 4/3. This issue was entitled as a supplement to L1, including all changes to the language up until about 1980. It was written by pc. JCB considers it to be an official statement of changes to Loglan. But pc points out himself that much of the document summarizes pro and con arguments on various issues without giving a final ruling. It thus is unclear which proposals were officially adopted. (Some weren't, such as cua. And some are contradictory - see Hixson-Bonewits.)

tense - nominally a word which modifies a predicate by placing it in time. JCB has expanded the concept in Loglan to include place location (vi, va, vu), and more recently has incorporated modal operators as tenses.

TL - The Loglanist; our predecessor publication. TL was a journal for the purpose of allowing Loglan workers to exchange ideas about the language. It followed too closely the academic journals, in that it was frequently too technical for the average reader. We have a similar purpose, but are unofficial. Hopefully, techniques like this glossary make it useful to the many who aren't fluent in all aspects of the language.

triphthong - a series of three vowels pronounced as a single sound. These are not found in Loglan (except possibly in names), but are found in English. See the Appendix.

Turbo-Pascal - a version of the computer language PASCAL that was written by Borland, Inc. It is available to run with CP/M, MS-DOS, and the Apple MacIntosh computers. It was marketed by a technique similar to shareware - it was priced extremely low compared to other versions of PASCAL, especially for its high quality, and was made available without copy-protection. Many people violated the copyright and spread copies of the language (and programs using it) to others. But when people found it was good, they bought an official copy. Then came a series of improved versions of the language, tutorials, a useful users' manual. Now, Turbo-Pascal is the standard version of the language for microcomputers and has supplanted the original UCSD Pascal almost completely. LogFlash is written in Turbo-Pascal.

unpaid employee - a contradiction in terms that is often used by JCB (see employee). If called one, it is probably because the caller wishes to assert control over your efforts, or ownership of the product thereof. Most of us are instead volunteers, who work on Loglan at our pleasure, (generally in our limited spare time beyond our real employment), and for our own perceived benefit - the chance to have and use a language that we helped build.

unvoiced - a consonant which is made by shaping the mouth without vocalizing. Unvoiced Loglan consonants are : p, t, f, s, c, and k. GMR Loglan h is now unvoiced, though it was formerly indeterminate. GMR Loglan q is also unvoiced.

Usenet/UUCP - a computer bulletin board and mail network which operates nationwide. It has nodes at major universities and industry sites that operate under the UNIX operating system. Anyone with access to a UNIX-based computer (such as a VAX or SUN) probably can access Usenet. The Defense Department and its contractors also have access through Arpanet, a separate network that ties into Usenet. Network usage is free if you have computer access, which is better than the commercial networks. But many of you do not have access, so we've set up the Loglan Bulletin Board and are investigating other options to ensure as much of the community is served as possible.

utterance - the smallest unit of Loglan speech or text that can be said to be grammatical.

vocative - a portion of an utterance which identifies and calls to the listener but is grammatically isolated from the rest of the statement. 'Hey, John!', 'Betty, go to the store!', and 'O, ye of little faith!' all contain vocatives.

voiced - A consonant that is made while vocalizing. Loglan voiced consonants are: b, d, v, z, j, and g, (voiced versions of the unvoiced consonants) and l, m, n, and r. X is the 'velar fricative', which can be voiced or unvoiced in Loglan-88 or GMR Loglan. Most English speakers will voice it.

volunteer - as opposed to employee, one who works by choice, and not for payment. Work done by volunteers is not explicitly covered under copyright law, since it is not 'work for hire'.

veridical - relating to truth or reality. A statement, or portion thereof is veridical if it has a logical truth value.

VV - see CCV

Whorf - see Sapir-Whorf; Benjamin Whorf was the co-inventor of that hypothesis. Loglanists often speak of Whorfian effects to refer to the unbounding of thought caused by certain features of Loglan.

Word Makers Council - an official body of the Institute; actually, per JCB, it is a subunit of the Academy, though members are not full Academy members. The Council has been delegated by the Academy to decide on new Loglan words and metaphors that do not involve morphology changes. The latter is a fine line, that I can't explain - JCB has accepted some of Nora and my criticisms of the primitive list in UL2 without requiring anyone else's approval, has told us to refer others to the Council, and has labelled other changes as morphology proposals to be handled by the Academy. In any case, the Council's primary current job is the kibbitzing and decision-making on Eaton interface metaphors.

YACC - Yet-Another-Compiler-Compiler; a UNIX operating system tool which is written in C. It generates language parsers for a type of grammar called LALR 1 (which I can't easily explain). It incidentally verifies such grammars as unambiguous. MacGram is an attempt to define a Loglan grammar that meets the test of YACC.

Zipf - a reference to Zipf's Law, which I've never seen clearly stated. It can be stated as KISS (keep it simple, stupid). But for Loglan purposes, it states that the easiest or shortest words will be used most. Long words that must be used often will be jargonized or made into acronyms, or a shorter word will be modified in meaning to equate to the longer word, basically because people are lazy. A language designer must keep in mind Zipf's Law, or the shortcuts people will take will cause subgroups of people to rapidly modify the language so that it is unintelligible to others. Read a scientific journal or the Congressional Record for further examples, or look 'run' up in your English dictionary.

ZPG - Zero Population Growth; a worthy goal, but certain European countries and Japan are close, so their languages are not spreading as fast as third world countries'. Thus the Loglan-88 language weights differ greatly from the 1950 data, and we must sadly exclude some familiar (to Americans) languages that have been overtaken by such languages as Bengali, Hindi, and Malay- Indonesian.

HL3 Loglan Questions and Answers by pc and jp and others

3-1. You are holding two books, one in each hand, and are talking to another person. You indicate the book in your left hand: levi bukcu ... You then indicate the book in the other hand. How do you refer to this book? Is it also levi bukcu? Leva bukcu implies it is further away, which it is not. Leti bukcu and leta bukcu should be legal, but the consensus is that they would mean 'the this's book' and 'the that's book' implying not that 'this' = 'the first book' and 'that' = 'the second book', but instead that the two books are owned by or in some way related to some other pair of things 'this' and 'that'.

Answer 3-1. If the book in the left hand is levi bukcu what is the one in the right? Ans. levi bukcu, with contrastive stress - just as in English: "this book and THIS book." Your stated objections to leva, leti and leta are all correct. Note that the two levi bukcu's will pick up different anaphorics (dVs) and so be distinct after this introduction (if we can figure out how to use anaphorics.) Of course, you could use leva, say, since le is not veridical and all that counts is that the hearer picks up the right book (if it is a book - nonveridicality again.) pc

3-2. MacGram must include morphological pauses (e.g. after names, before i between utterances, and after a vowel at he end of a word before a vowel in a separate word) in the grammar definition in order to prove disambiguity. Otherwise, it cannot be stated that a lexemic pause (see TL7/1) will be recognized as such. It is reasonably easy to come up with an utterance which requires a lexemic pause at the same location as the end of a name to be grammatical. Thus a listener will hear the pause as morphological and not as lexemic. We haven't tried, but it seems likely that one could find an utterance that is grammatical both with the pause interpreted morphologically and lexemically, and with different meanings in each case. The obvious alternative is to assign a LW to be used instead of 'PAUSE' for those instances in the grammar where a lexemic pause is now used. Loglan-88 will probably include such a LW.

Answer 3-2. I think rather than a single PAUSE Loglan may need a number of words to do the wok of pauses in various places. But getting rid of all those problem removing but undetectable pauses is bound to be on the right track. pc

3-3. Loglan-88 work has determined that the Chinese use several short words as indicators similarly to Loglan's indicators. Some of these should be considered for inclusion since GMR gave a few more VVs and CVVs that may be free still. Loglan-88 will add some of them, or try to find a generalized solution. We can't easily show Chinese tone symbols in this printing, so bear in mind that words which appear the same are not really:

    ai   expresses disagreement
    ai   shows surprise/discontent, also shows grief/sorrow
    ai   shows sadness/regret (a different ai and a different
         attitudinal meaning than the previous one meaning grief/sorrow)
         also regret/annoyance
    ba   indicates a suggestion, a request, or a command
    bei  agreement with reluctance
    bei  simple/easy to understand
    a    sudden realization
    a    surprise/amazement/wonder
    a    pardon?, pressing for an answer
    a    surprise
    an   interrogative

Some of these are similar to existing Loglan indicators; others show that the Loglan indicators are too broadly defined - a way is needed to express subtler shades of emotional feeling. There are more human emotions and attitudes than can be expressed with the current set.

Answer 3-3. Actually, this makes it look like Loglan is pretty good. But, yes, some subtle - and not so subtle: reversal, say - modifiers are needed for some cases, mainly the uv set. pc

3-4. Add LW haa as an indicator for 'don't take seriously/ tongue in cheek' comparable to the computer network symbols: (:-) or (... :-) which resemble a smiling face sideways.

Answer 3-4. Yuck! Such comments lose their force if flagged. And people that don't get 'em deserve what happens to 'em. pc

(I agree in general, but unambiguous speech requires that we be able to make a joke - a statement not necessarily true, but somehow funny - and be able to disavow our seriousness in claiming its truth. We won't necessarily be required to do so. But politeness may demand it, and to not allow identification of humor may bound some people's willingness to entertain it. rjl)

3-5. There seems to be no problem with (pause)y(pause) (a solitary schwa) as a space filler in the middle of an utterance when one needs to stop and think. That is what we use in English, and it seems that it has no Loglan impact, unless the pauses interfere with lexemic pauses (see 3-1).

Answer 3-5. Although I think that this will develop naturally and outside the Loglan phonlogical system, just as it does in natural languages, we will need a way of representing it. But then, following a basic Loglan principle of isomorphism of print and speech, it will have to be represented by an unused letter. Finally! A place for q! Another famous proposal in this area was to use r. That then was a vocative, presumably invoking the god R, lord of well constructed sentences endings. Let it go til needed. pc

(It is needed now! We are preparing for GPA. There is no maybe that something of this sort is needed in human speech. Therefore, let us solve the problem, if we can. Why must the 'uh' be a sound not found in Loglan? We use 'uh' in English without confusion, and schwa's abound. rjl)

3-6. Proposed prims:

  1. x is an abstraction/model of y in properties w of order/degree h (default 1)
  2. x is ordered in property y as determined by rule w
  3. x is a hierarchy of set y as determined by property/rule w

These prims are useful in any MEX environment, and are basic to one proposal to solve the general MEX problem that we are preparing for eventual proposal.

3-7. Kibbitz for the WM Council: takna-muvdo is an awful metaphor for persuade (per L4/L5). Any better ones?

3-8. Per a phone conversation with JCB, I understand this as a definition of Carter vocatives, expressed for MacGram implementation. Is it correct? Have Carter vocatives been officially added to the language?

Upon hoi, stop parsing the main utterance, find the end of the Carter vocative, parse the vocative separately, then resume parsing the main utterance. Later insert the vocative parse in the main utterance parse.

The end of a Carter vocative is a pause, a predicate, or a non-argument LW. But allowed within the vocative are argument-embedded LWs and predicates (in their proper grammatical context).

Answer 3-8. Carter vocatives are something Jim has added to the language, in the wake of the great "Carter Review" by JCB, PC & I. (Scott Layson was invited but declined...didn't respond). The proposals accepted as a result were:

  1. ige ice as sentential connectives
  2. modal sentence connectives: ina etc.
  3. reverse gi
  4. nuge, noge for long scope
  5. "bio as prenex bua marked" (I can find the results but not the review text proper. Hmm! hoi not on list! But I am sure JCB regards it as part of the language.) jp

3-9. The Loglan-88 proposal for letterals, described briefly above, was developed by pc and myself during my visit in October. In addition to the letter words themselves which are caseless indefinite (whether upper case or lower case), we are proposing LWs to indicate upper case lock and lower case lock (each lock for a single word), single upper case and single lower case. Alternatively, the four LWs could represent upper case and lower case with a single letter default, and word lock and indefinite lock. It can be seen that all possible combinations can be expressed with either set of 4, but preference is determined by the type of letterals one might expect to see most common in Loglan usage. Proposed LWs are (for GMR Loglan/Loglan-88) hau/xau, hei/xei, heu/xeu, heo/xeo, for the four of either set, respectively.

Additionally, a LW is used to select the language/character set. No default need be assumed, or perhaps Roman is best since Loglan uses the Roman characters. LWs to cover the major alphabets, especially including all of the source language alphabets, are (GMR Loglan/Loglan-88):

    Greek     gaa/ga'a    Roman     raa/ra'a    Hebrew    jaa/ja'a
    Cyrillic  caa/ca'a    Arabic    baa/ba'a    Devenagari daa/da'a

There remains the problem of assigning LWs to stand for letters and punctuation symbols of the various alphabets and character sets that have no Loglan equivalent. Devenagari and Arabic especially have several such labels needed, but Spanish and French have a lot of special punctuation, too. We suggest reserving an entire set of CVVs for one consonant C to stand for non-Loglandic symbols in a given character set. The definition of each such LW depends on the character set in effect, as selected by a LW from the set above.

3-9a. In addition, the LW maa can then be used to select a MEX character set, where the CVVs each stand for common mathematical symbols. Additional LWs can be used to select special purpose character/symbol sets for special scientific and mathematical fields. With these LWs, and the Loglan-88 proposal for letteral names, it should be possible to read any mathematical expression directly from its international form. Some type of quote pairs like kie ... kiu or li ... lu then serve to set off this expression. This will solve most MEX problems, and gives us at least a temporary MEX for the upcoming GPA.

Answer 3-9a. And generally on MEX. There just ain't no way to read off standard mathematical notation. That notation is, in many cases inherently two dimensional and speech ("reading it off") is inherently linear - one dimensional. So, we either need a "new" system (not really new of course, since, despite all this talk about standard notation, most branches of mathematics are almost as bad as logic in having a variety of forms available) or we need a set of tricky conventions. Both of go far beyond just getting a word for each symbol. For two examples, the vinculum (or firmus) that is said to mark division in fact also is a grouper for both top and bottom. so just a word for it (even if we could figure out how to say it and its two arguments in parallel) would not be adequate: we would need grouper devices as well. Second, try to say a long division computation, with columnar line-ups, right shifts, etc. etc. pc

3-10. Redefinitions of primitives and proposed complexes associated with the attempt to make a complex for 'computer'.

trime     x is a tool of form y assisting w to perform function h.
          The essential concepts is used and assisting. A tool has a
          basic form which may be used for a complex function. Form
          determines, or precedes function. A tool is used because its
          form is suitable to assist w in achieving the intended
          purpose.
          A hammer is a tool. (ba jia mrozu ga trime). A rock can be
          used as a hammer. (ba jia troku ga trime lo mrozu).
patce     x is an apparatus enabling function y with control/controller w.
          The essential concepts are made and enabling. An apparatus is
          made to perform a function. That function determines its form.
          The form is generally more complex than that of a tool, since
          a simple tool form could be adapted to the function without
          being specially made.
          A piano produces music (when played by a musician). (li piano
          lu ga patce lo muzgi proju).
matci     x is a machine performing function y directed by w.
          The essential concept is perform. A machine is directed to
          perform a generally non-basic or complex function. It performs
          this function with internal or automatic control. w is a
          director or initiator, not a controller.
          A radio produces music (when turned on and tuned to a music
          station by someone, who need do nothing more after these
          initiating actions). (ba jia radjo ga matci lo muzgi proju).
CVCCV     x is an entity acting to accomplish goals/purposes/intents y
          by self-direction/control.
          This new primitive is the logical conclusion of the above set.
          The essential concept is acting. An entity acts on its own
          without specific direction or control. The philosophical
          implications of this concept are uncertain but are probably
          significant in defining a Loglandic society. It is not all
          that useful now, but is likely to become significant as we
          advance artificial intelligence, and as we potentially deal
          with other species of varying intelligence.

The following complexes can then be derived:

trime-durzo              tool-do              assist (in the above sense)
patce-durzo              device-do            enable (in the above sense)
matci-durzo              machine-do           perform (in the above sense)
ponda-patce              respond-device       transponder
rulni-matci              rule-machine         computer (general term)
matci-rulni              machine-rule         (computer) program
sanpa-rulni-matci        sign-rule-machine    computer (computer science)
penso-rulni-matci        think-rule-machine   artificial intelligence computer
numcu-rulni-matci        number-rule-machine  digital computer
prase-rulni-matci        process-rule-machine analog computer
rulni-plizo-pernu        rule-use-person      person computing = computer
matci-rulni-pernu        machine-rule-person  (computer) programmer
rulni-matci-pernu        rule-machine-person  generic computer-field person

3-11. Is a sole letter word in an argument position a variable, or is it a letter word? How do you tell? In TL7/1, several corpus samples use letterals to represent some specific argument, somewhat like da-series usage; the letteral sanpa ba (stands for something). But what if you want to use a letteral as le po letra (a specific instance of a letter - representing only the phonetic sound)? If I say Nai hijra, my listener needs to know whether I am saying 'The letter N is here' or 'N. (standing for something/someone) is here'.

Answer 3-11. A single letter word is a way of saying a single letter. What that letter means in a given context is something else again. In some contexts, it may mean a the letter itself, for spelling, say; in in others it may be a shorthand for a name or some other term; in yet others it may be a sign for some unit. It seems to work, in short, just like the corresponding words in English ("Kay needs 160k on the K-90, 'kay?") We may think that we need to separate these out free from contextual clues and maybe we do, but this is a late and minor matter. Unless the anaphoric use gets to be well-established, rather than just a suggestion with little support. [A letter word can stand for le letra but not, except in the abbreviatory sense, for le po letra - for the letter itself but not for the event of being a letter. I don't understand your gloss - as I take it to be - about a specific instance of a letter. That - the token-type distinction is something that Loglan has not - but probably should - developed yet.] [sic!] pc

(Perhaps I meant lae le letra (what the letter stands for) - but that is ambiguous. Does it stand for something semantic, or phonemic? In any case, Nora and I have discussed the concept of using a LW to distinguish between the letteral, and its use as an assigned variable. Treat them like we do numbers. Perhaps, we can even use the same LWs for both letterals and numerals: -ra, -ri, lio. The bare letteral works like the bare numeral - very useful in MEX - tony poi fo (2n + 4). When suffixed by -ra or -ri or something comparable, it can be used as a predicate. When descriptor lio is applied, you are referring to the letteral itself as a token.) rjl

I think we need variables much worse than letter-names -- I am interested in quantitative reasoning more that discussing text. I'd like to see letter variables largely replace da...du. jp

3-12. Is it legal to have a name start with 'Le' or 'La'? For example, my name transcribed to Loglan as I say it is Leceva'lier in GMR Loglan phonetics. Used vocatively, it could be broken into le Ceva'lier, which is a different name if le is usable before a name (as Trial 24 suggests). It has no consonant cluster, and has a final consonant, which suggests a name. But it seems that there could be problems. If my name started with La..., there would be ambiguity. And what about embedded kii and kio (backspaces), iy (the buffered dialect hyphen in GMR Loglan), and other constructs. If they are not allowed, how does one transcribe a foreign name that would need them. Using lie to indicate the LW is intended is not sufficient. Liekion could be a valid name in some language. And in any case, it is probably not practical for a speaker to comprehend such potential misunderstandings when de is speaking the metalinguistic operators intentionally.

3-12a. Imagine someone named (for spite?) Bla'nuladjan. If used vocatively, would it not break down as blanu la Djan? The consensus of those I've talked to seems to be that vocatives cause too many problems when used without hoi as a marker. It may be simpler to require hoi, and forbid 'real' metalinguistic operators between the hoi and a pause ending the vocative name. A separate LW must then be used for Carter vocatives that are not names.

Answer 3-12a. The L1 answer is pretty clearly NO. But no one has held to it for years. I don't know whether the resolution algorithm (if there still is one) has kept up with this change of habit (or, rather, failure to get the habit right in the first place because we early on had all these people with names that didn't fit the ideal pattern. I suppose that the "solution" had something to do with the pauses again, but I don't know how it goes. And if we can get, e.g. metalinguistic operator-shaped syllables into names without screwing up, how will we use metalinguistic operators on names? How will we delete a name we know is going wrong? [Current case: Sara regularly starts to call Princess Eilonwy in the Black Cauldron books "Princess Ellowyn". So how does she stop off at "ello", delete it, and pick up at "Eilo"? Probably throw in a nonce consonant, pause, kio, and start again "Eloooooor. kio Eilonwis"] pc

(That sounds like a good idea. And there is no requirement that one be able to use metalinguistic operators everywhere. In fact, the language didn't have them until a few years ago. The standard seems to be that one can use anything in a Loglan name, so long as it ends in a consonant. We need to make the formal algorithms fit this standard, or change the standard. In any case, the question must be resolved before GPA. rjl)

Why a separate word for Carter vocatives? Or have I forgotten what they are? Aren't they just predications (in argument form, i. e., with linked subordinate arguments? In any case, the end of one would be just the same as the end of a le argument, just as the end of a regular vocative is the same as the end of a name. pc

(No problem with this explanation, as long as one can clearly end a Carter vocative in the same way as an argument. But the Trial N grammars have never been able to handle the Carter vocatives. The PreParser identifies them and removes them. If they were just a form of argument, surely a way could be devised to identify one with real grammar rules (in BNF form), and to have them processed by the Parser rather than the Preparser. It appears, though, that I have been misusing the term 'Carter vocative'. These are vocative argument-like structures other than names. I guess that I mean to propose that all vocative names use Hoi.

3-13. How is multiple stress represented in names, if applicable in transcribing a foreign name?

Answer 3-13. How is SINGLE stress represented? It looks like you are using the proposed single quote trick - with some assumed position if unmarked? (Ought I have reviewed the Primer at this point?) So, use more that one quote in a name - maybe even double quotes for a subordinate accent still not neutral. But Loglan is unlikely to take kindly to multiple stresses in a word. pc

(I actually was inconsistent in the Newsletter and text. As part of the Loglan-88 changes, we have pre-empted the apostrophe. There have been two methods of indicating stress: capitalization and the apostrophe. So I have attempted to become consistent in using capitalization. Other reasons exist for capitalization, though (names, start of sentences, etc.) that, while not required in Loglan, are used by some. The result was that I didn't catch myself always in editting. rjl)

Names must be preceded & followed by pauses... a name is identified by the ending consonant, (i.e. a consonant followed by a pause) and extends back to the first previous pause. jp

In my opinion, metalinguistic operators are not practical in names. Written/spoken isomorphism breaks down in lie...lu quotes and names, if hyphenated names and such are allowed. (might as well) jp

This is and example of the inevitable tendency of neophyte Loglanists to push written/spoken isomorphism too far. Some sounds and phonemes have no standard written representation in Loglan, just as some symbols have no standard pronunciation in Loglan. People who import such constructs must simply live with this. Names and quotations are the perennial sources of these problems: when making such and import one must simply choose between isomorphism and fidelity to the original. jp

3-14. Many LWs have been collapsed into the lexeme PA, presumably because the grammar is similar when used in utterances like TL7/1 H12 and H13: Na mi penso da and Na mi, penso da. Substituting sau, one gets sau mi penso da and Sau mi, penso da, which make some sense. But how about H16 and H17? What do Mi sau totco tu and Mi ji sau totco tu mean? And try substituting tie, lui, or neu for real fun. And then compound these tenses as tiesau or pacenoineu or luizu. (And can you insert the infixes -i- and -u-? How?) We've had a lot of jokes about the Whorfian effects of saying such apparent nonsenses. But one must be able to comprehend some logical if nonsensical meaning for Loglan to work with the super tense.

Answer 3-14. Tense and aspect is one of those areas that needs some serious work and not just for Loglan. Even without locatives and then all the semantically unrelated argle-bargle, the Loglan set is either inadequate or inadequately explained. With the new "tenses", I don't know what to do, since they have such different semantic structures behind them. I see no reason to think that the infix i goes with any but the central tenses, since it is derivational not syntactic (It gives new LWs, not conjugated ones.) The suggested corresponding forms of the CVV "tenses" are not new words but - at best - compounds, and not even compounds involving the original words: *tiie is nowise related to tie. So far as I know, the infix -u- was never accepted, partly because it conflicted with an H-B tag, and partly because it didn't seem to signify anything for which there was a place in the Loglan tense/aspect (etc.) structure that had been developed or was emerging at the time (or has emerged since, for that matter). I {???} sounded like intentional distribution aspect, of which we have nothing else - but maybe should. pc

It is not necessary to assign a logical meaning to every syntactically correct Loglan utterance, and it does not pay to try. It is sufficient that the grammar supply the correct structure for accepted Loglan usages, without attempting to forbid all currently unused constructs. It is much more important to have a simple grammar. Semantics is a whole new can of worms. jp

(I wasn't asking for a useful meaning - perhaps only a native speaking Loglanist will be able to determine that. By logical meaning, I mean a semantic interpretability. To say 'fox brown the quick slyly the jumped dog lazy' is grammatical requires only the naming of rules permitting it. To figure out how one might assign semantic meaning to it is a separate problem. Somewhere in between is a determination that we want the power(?) implicit in this grammatical structure to be present in the language. And there is the other intermediate problem of determining how these

3-15. How about a LW that turns any primitive or predicate (raba jia preda) into a modal operator? Per L1 pp 177-8, there are "fourteen ... most frequently required ..." which implies that others are possible useful. The proposal would allow flexibility in modal operators without likely grammatical ambiguity. Of course, per 3-14, we may still have to figure out what they mean as tenses.

3-15a. How about another LW to make raba jia preda into a discursive?

3-15b. And another LW to make raba jia preda into a case tag?

Answer 3-15. Yuck! People keep saying this but when pressed to come up with a predicate that they might want to modalize, they either cannot or give one we already have but they'd forgotten or give facetious answers. I suspect there may be a few more, but probably not over ten or so. [raba jia preda doesn't mean "any predicate" or the like, since preda does not mean "is a predicate of .. meaning.." (second place defaulting to la Loglan), though we probably need such a word (kakpua?) Preda is a predicate variable, implicitly universalized and/or a store of nonce predications "is whatchamacallit".] All comments apply to parts a and b as well. pc

(I have used preda throughout my writings, since there is yet no approved word for predicate. I don't think any complex could do the Loglandic predicate justice, or most other basic notions of Loglan, and intend to propose them as primitives. But preda was, I understood, a predicate variable. So I put it in with the intention of allowing proper substitution later. rjl)

A perennial suggestion. It will probably happen. jp

3-16. In the acronym USaiAma (USA), how does one tell that the U does not fall off and become a connective?

3-17. For each indicator/attitudinal, there must be a primitive associated with it. Else how can one describe another's 'indicated' statement? If someone expresses hope (ae) as an indicator in a sentence, I can say da spopa le po ... . But others, such as ua and oe, cannot be described to a third person very easily. If these attitudes are primitive enough to express, they should be primitive enough concepts to talk about.

Answer 3-17. I don't remember what ua and oe mean but I do remember that Carter showed how to say all of these in third-personally ways. He was making a wrong-headed point, but I recall that his identification all looked correct. Alas, Jim has those notes now, along with the rest of the TL stuff (unless it is in some one of Carter's books.) pc

3-17a. Perhaps a LW could be used to convert an indicator into a predicate meaning x indicates emotion/attitude ... to y under condition w. This would eliminate the need for primitives of little use.

Answer 3-17a. LW's are more precious than prims. jp

3-18. How do you negate an indicator - express the negative of the attitude? There are discrepancies between text descriptions of the language in TL7, L1, and the old L3 Primer. Trial 24 would not allow A09 from TL7/1, since no would negate the sentence mi helba tu, while indicators are 'slurped up' into the previous word. In a variant, try mi oe no helba tu. How does one tell whether no goes with oe or helba, and how does one express the other?

Answer 3-18. Not quite sure what the negation of an attitude is in some cases but in others - those where it clearly makes some sense - L1 does give a list. Does TL7 supplant that? (L3 is hopelessly out of date.) On the final question, could there be an order distinction: oe no vs no oe? I'm not sure which would be which, but we could just decide it, I think. pc

(Per the logic of JCBs grammar pronouncements recently - and accepting TL7/1 as being a relatively current statement of his intent, indicators group to the next leftmost word (or to the entire sentence if on the left). This means that no oe would cause the indicator to group with no, leaving no untouched to modify the word following. I suspect that if JCB has a solution, it uses lexemic pauses. rjl)

oenoi was one approach...parallel to enoi anoi etc. jp

3-19. What is the method to recognize the 'end of the word' for words used as delimiters in the new variable delimited nonce quotes (see TL4/3 Supplement pp52-53)? The Loglan resolution algorithm relies on adjacent words/sounds to resolve words boundaries uniquely. It seems that only lie W(pause)Q W(pause) is needed to be unambiguous, and other pauses between W and Q would be useful to ensure the listener hears correctly.

Answer 3-19. I'm not a fan of lie but it seems that the pause after the W and before it after Q as well are sufficient for resolution.. Remember that, by definition, W cannot occur in any way in Q. pc

Sounds good, as long as W is vowel terminated. jp

3-20. In the NB3 draft (pg25, para.3 line 6 from the bottom), the construct vedma la pa Rismi is found. Why la - is pa Rismi a name? It seems that pa la rismi might make sense if one is talking about the former Mr. Rice, but even this seems unobvious. Or is this a typo and it should be vedma le pa rismi? The context gives no clues, and there is no translation given. (This question is for JCB, of course, and is a comment I found after I had already sent off the rest of the NB3 comments to him. But it raises questions of general interest on the meaning of la and the use of predicates as names.)

Answer 3-20. I think that it was once decided (after Carterization) that la was essentially a member {allolex} of LE {lexeme}, so lapa was going to be legitimate. But la, pa rismi (with a clear break between la and pa either PaRismi, a real name, or "Former Rice", a description (and a pretty odd one) doing duty as a name. {I added the comma after la - JCB's standard indication of mandatory pause. rjl} And apparently the latter, since the former lacks the final consonant. [As the former Mr. Clifford. I can assure you that is lapa Klifrd; pa la Klifrd means "before Clifford (in time)".]pc

(But I don't think lapa Klifrd is correct. Pa as a tense must occur before a predicate string. And Klifrd isn't a predicate. So the Parser is likely to reject this construct. Would you believe le pa me la Klifrd? rjl)

le and la have moved erratically towards being as parallel as possible,

              le = 'the unique instance I have in mind'
              la = 'that which is called'

jp

3-21. Why are dua and due not 5-letter prims, along with bi? By using LWs as primitives it disrupts the language rhythm. A series similar to the preda series is proposed.

Answer 3-21. Dua and due aren't primitives, but predicate "variables" (in JCB's strange use of that term), predicates on a par with ti and tu in the realm of arguments. You can't make compounds out of them (or, if you can, they are a very different kind of compound. The situation with bi and cie and so on is less clear. They probably should be primitives - and indeed all are. Why we keep the LW duplicates is unclear linguistically, but is obviously part of the MEX subsystem. (they are part of "reading off" the standard notation.) pc

(The answer to MEX should not be dependent on allowing pre-calculated Zipfean considerations to overwhelm the well-thought out morphological structure. (In short, I agree.) Else it is not clear when one should use bi and when one should use samto. This has already come up in our pre- Logfest attempts at spontaneous Loglan conversation. rjl)

I would prefer to see bi a prim, with no special grammatical privileges, and moved in this direction during MacGram. (It seems to have swung back since then, if I am not mistaken.) Will Zipf allow a two syllable bi? jp

dua and due may need special grammatical privileges in prenex position. Ask PC on this one. jp

(pc?)

3-22.

  • Da blanu de means 'x is bluer than y'.
  • Da blanu means 'x is bluer than something', by implication the same as: da blanu ba. This is taken to mean 'x is blue'. But by some color measurements, green is a combination of blue and yellow, and white has all colors. So it is true to say of a green or white object da blanu; they are both bluer than something, e.g. a yellow or black object.
  • Da no blanu, by previous writings, causes the implied y argument to change from ba to ra ba, implying da no blanu ra ba. This means 'x is not bluer than anything', or 'x is the least blue thing in the universe', which is true of only one object.
  • Da nu blanu still uses ba as the y argument, thus meaning 'x is less blue than something y'; this is closer, but still doesn't mean 'x is not blue'.

So how does one say 'x is blue' or 'x is not blue', without being so extravagant in one's claim. Nora believes that comparatives should not be the form used for primitives. pc suggests loe me ba (the typical or characteristic something) as the generic argument placeholder for both normal and negated predicates. I have modified this to su loe me ba (at least the typical or characteristic something), at least for comparatives, for my own proposal. Or just leave the primitive inadequate for normal use without a second place, and define a complex blanu klesi as 'x is in the class of blue things', which can be negated with no without changing the placeholder rules.

Answer 3-22. I'm sorry you don't like the loe solution to all of the problems with blanu and similar words. It really is a solution, you see, and it has the added advantage of being the solution for corresponding problems in natural languages (English and German actually cited) that had found the widest acceptance in the linguistic community (indeed, it is the only one with any significant degree of acceptance beyond its proposers and their students.) Two features of your presentation - and one of Jim's - make it look weird, One is the marginally intelligible line you give for the loe, loe me ba, "the typical thing related to X", if X is not yet assigned, which is dead wrong. In either case, this is not the way to say it. Indeed, there is no general formula for saying it, which can be applied by filling in gaps in an expression form (beyond the loe, that is.) The choice is a matter of pragmatics more than even semantics. The second place refers to the typical thing of the whatever genus (maybe even species) the first referred to thing is being conceptualized under. This may be as broad as "thing" or as narrow as "pretty little girls school house door knob screw thread" (however grouped). With descriptors, it is usually that description, or, in modifying position, the head, In other cases it is less clear - as far as a rule goes, though we are rarely in any disabling doubt of the appropriate type. [If you don't like this, then take it to be the bluest non-blue thing of appropriate sort. But I suspect the problem is with the "appropriate sort". not with the "typical".] Jim's problem is that, despite his own usage from time to time, he says that the default is always ba. It probably usually is ba, but there are many cases in which this is clearly wrong and where Jim even cites other defaults back in L1 (and more subsequently.) Comparatives are the most obvious one (and here Jim has muddled things worse by actually talking for a while as though he meant the default to be ba and by trying to make it make some sense. He failed, of course.)

(I like the straightforward idea that there are no defaults. Not specifying an argument makes the statement ambiguous. The listener can assume a default that makes sense to him/her, in accordance with the interpreted meaning. If the speaker does not want the user do have a choice, it must be more exactly phrased. rjl)

(I don't quite follow your analysis of loe me ba. It seemed the only way to put loe in that place without specifying a predicate. In comparing blues, you do not wish to constrain the predicate domain in any way - even by using bekti. So ba is the only appropriate choice to mean 'something'. The grammar requires that it be turned into a predicate to be preceded by loe, which I did with me. What is wrong with this logic? rjl)

Thinking in terms of 'implied arguments' is cute, but raises endless conundrums of this type. I think we will eventually regard 1-argument blanu and 2-argument blanu as logically separate predicates, much as APL gives independent meaning to 1-argument and 2-argument forms of a given operator. I expect most people will want to bang their head on this wall for a good while yet, however..... probably until we get serious about Loglan semantics. jp

3-22a. At the end of 3-22, I proposed blanu klesi as a complex. Is this correct, or is klesi blanu the proper metaphor? This is a typical case for those who aren't sure how metaphorical modification is supposed to work.

Answer 3-22a. blanu klesi, if either: "blue kind of class". The other is a class kind of blue, which won't do. The whole idea is needlessly complex, though, since it already involves the notion wanted, blueness, in its definition. Why not just use the basic notion and be done.

[Another possible route is to incorporate fuzzy logic in some way and then whip in its "adequately" operator: deep enough into the probability aura of the concept to count as fully in". The problem with that is 1) that that is defined as being further in than some typical object and 2) ignores the fact that the distance in which is adequate is different for different things (blue skies are bluer by and large than blue houses - at least the adequacy point is farther from true blue for houses than for skies. "call that a blue house? Why, if it was a sky it'd be buttermilk." This latter consideration is central to the acceptance of the "hidden loe" thesis for natural languages.)

(If loe is the answer, it's probably OK to translate da blanu as 'x is blue'. But problems still abound with da no blanu and da no nu blanu. The upshot of it all is the contention that when we call something 'blue', we are not necessarily comparing it to anything. We have an internal definition of a range of spectral colors that constitutes 'blue'. Something falls in that category, or it does not. We also have a separate notion that some color may be a more 'pure' blue than another, as well as concepts for intensity, etc. My contention is therefore that the primitive notion is 'x is/appears blue by standard y (under conditions w)'. If someone wants a comparative, let them make a complex with mordu, and possibly with one of the conversions (nu, fu, ju) of kolro to express it. In fact, I argue that a comparative is inherently a complex of a basic concept. Otherwise, why would the natural languages generally have developed comparatives and superlatives as modifying suffixes. rjl)

3-23. Another case of metaphorical modification was found in the latest Lognet, and used as an example. Nimla groda was debated as opposed to groda nimla. But both show a cultural bias on the part of the word maker. For those of us who have dealt with fantasy/science fiction or Dungeons and Dragons (tm), a large portion of Loglandia by my guess, the one thing most likely about a monster is that it is not an animal (at least in the sense of being a common Earth creature). It is a non-animal creature whose primary characteristic is strangeness, with size a frequent but not mandatory secondary characteristic. So a defining metaphor would be no(r) nimla gutra (a not-animal type of strange), although groda gutra might be acceptable. But then both are good metaphors. So why not permit both? Why limit Loglandic thinking by insisting that every Eaton concept have exactly one Loglan metaphor. Many Eaton concepts have broad definitions that allow several shades of meaning. If Loglan permits these shades of meaning with its powerful metaphor-making power, let us use it, and not bound thought by rigid and limited metaphor assignments.

Answer 3-23. Yeah! Let the hundred flowers bloom! And usage pick (and redefine for Loglan needs, not Eatons's. pc

(Note that I am not saying that we should drop Eaton as a standard for completeness of our word set. It would be great to find something with more recent data than from the 1920's. But apparently Eaton-like work has not been re-performed since, so we are stuck with it. I am just saying that we should not be limited to the concepts given, or to assume that the most frequently used word is the primitive form. Primitives should be selected on the basis of their usefulness in making other words (i.e. complex making) more than on a frequency basis. There are some laws more important than Zipf's. rjl)

3-23a. Another example from the Eaton exercise. Faced with devising a metaphor for 'death' (the event - perhaps the verb 'die' is Loglandically equivalent), two people independently come up with:

          (lepo) morto-cenja                      dead-change
          (lepo) morto-satci                      dead-start

These clearly represent two major philosophies, which Loglan should not choose between. So both should be valid dictionary words for 'death'. And we haven't even dealt with no-clivi (not alive), po-morto (state of being dead), morto-sonli (dead-sleep), and fa-clivi or fiazu-clivi (no affixes for the latter, so it may not be possible to make a complex for it; the former can also be confused with future-life, meaningful to those who believe in reincarnation). All of these are valid metaphors for 'death' as well, especially under certain philosophies; any could be devised by an Eaton worker holding those philosophies, not to mention one working without a primitive list like the current Word Makers Council is doing. Do the Loglan inventors have the right to choose a philosophy, especially given aspirations of testing Sapir-Whorf?

Answer 3-23a. Loglan is modifier-modified morto-cenja almost certainly means 'kill'. jp

(I don't see this one. Remember that cenja means x becomes/changes into y. morto-ckozu might mean kill. But given your logic and the definition of cenja, x dies ... is the reasonable equivalent. rjl)

3-24. If there exists a primitive for the measure of something, there should be a primitive for that which is measured. For metro, we have langa. For gramo, we have tidjo. What is measured by volta?

3-24a. If Loglan is to be International, and strongly support the metric system, it should have primitives for the basic units of the International Metric System, and for the things measured by them. All other units should then be expressed as complexes of those basic units in the way they are derived, or alternately should be the ... dugri of whatever they measure. Volts are commonly used as measurements, but are not basic units of the metric system - they are derived. They should not be primitive.

Answer 3-24. Why? Sometimes the measurement is more primitive (for us at least) than the thing measured: volta measures voltage, what else? (zo volta maybe). It may be that somewhere in some scientific arcana, voltage is explained (the pressure part of the water analogy?) but all we need and use is the meter reading. I am also not a great fan of the SI, especially of its frequently changed formulations of fundamental and derived measures. Most of these critters are interdefinable and so what to take as basic is a matter of convenience, not something profound. [Entirely appropriate for a system that has been based on high flown principles and practical incompetencies for two centuries: they mismeasured the meter to begin with and misboiled the water and have been covering up ever since, now being stuck with a claim about the metaphysical priority of some very peculiar multiple of the wavelength of an odd line in the spectrum of some flaky gas. Of course it is French, so you can't expect too much] Anyhow, why not just have a predicate for each unit? There is no wy you'll get predicates of acceptable length that accurately reflect the relation of say volts to kilograms and meters (are those the right guys or do we need seconds in here too?) nor would anyone care if you did. they would still want something like "volt", the international name.

{The French are the people who gave us Louis Pasteur et. al., who made some rather important advances in the sciences and medicine. tlw} {uo - rjl}

(It was not my intention to translate volts into an agglomeration of basic metric units like kilograms and meters. I am saying that either we should have a primitive both for the thing measured and for the unit, (for the basic metric units, only, as far as I'm concerned), or make them both complexes. EMF can then be lenki-fosli, and volts lenki-fosli- dugri, with appropriate affixes to shorten the words. That way it remains clear what volts is measuring. If you don't care what it is measuring, don't put units on it: le merli-patce gi cutse lio netonira. (gi seems to have absorbed the old ga in the latest grammar) rjl)

Don't get carried away here. Zipf over-rules all issues of logical derivation. A name is not a definition!! Volts may be derived, but can you build a useful metaphor short enough to satisfy Zipf's Law? jp

(But there are dozens of units. Most are used only by specialized scientists and engineers. Which justify occupying primitive space? Certainly Eaton is not a useful basis - certain measurements, such as volts are more commonly used now than then. Zipf's law applies when one needs a short word for a frequently used concept. A reasonable case is that the basic SI units are most useful in complexes, and at least liters, meters, and grams, are the most common SI units that occur, especially when supported by the full range of metric-prefix prims. Of these, only liter has not been mmade a primitive; I don't know why? We will fix it for Loglan-88. (I do admit to not having an obvious use for two of the eight basic SI primitive units - candela and steradians.) rjl)

3-25. What is used with jue to access a suteri argument, one which is after the second argument, without specifying the second argument. In L1, we had joe and jae, but these apparently have been dropped. This is a post-TL4/3 change.

Answer 3-25. I think that JCB was going back to the earlier version that used just je and numerals. The first argument to appear is linked je followed by te if it is the third rather than the second, and so on. Subsequent arguments are linked by jue, assumed to follow in order. If they do not, then a number is put in to indicate the correct place (or the correct displacement from the last mentioned place? Surely the first.) pc

(Seems workable, but it would be more consistent and less ambiguous to use pua-series tags than numbers, if it parses. Else does je te le sorme refer to the second argument with 'three of the sisters' or the third argument with just 'the sisters'. je pui le sorme seems like it would be unambiguous. rjl)

Aha, now I remember more of this. Loglan 1 specified je ...jue...joe...jae..., but in YACCing i discovered problems which L1 didn't solve. (This wasn't unusual!)

* The L1 construction didn't extend to more than 4 trailing arguments. It is clearly unrealistic to architecturally limit the language to +5-place predicates.

* L1 didn't explain how to resolve

              ...je (preda jue prede)

correctly -- missing-first-argument version of

              je (preda je predi jue prede)

As was my custom at the time, I plugged in the simplest pro-tempore solution I could and moved on. (I didn't have time or mandate to attack fundamental design issues.) The ...je...(jue...)*gu... provides a minimal mechanism for attaching an arbitrary argument string to an arbitrary predicated. It was not intended to address ordering issued at all -- je jue in this interpretation are pure glue. The intended semantics are:

Any valid top-level preda ba pue da type pre+argument string can be glued together for lower-level use by inserting ... je ... jue ... jue ... (gu) appropriately. This solution may ultimately prove too verbose, but it is better to start too simple than too complex. jp

3-25a. Was this (and other changes of its ilk that never have seen print as concepts) approved by the Academy? Should the Academy, albeit composed of skilled Loglanists, decide on a proposal of significance without it being written up (in Lognet or wherever) for comment by the community? After all, we have to use the language.

3-26. There seems to be confusion from the new forethought connectives (keks). for kimoi ... ki ... , it seems that the first phrase is the motive, given the word order, yet the motive (indicated by moi) is really the second phrase. This confuses, although there is an English translation that works: 'the motive for ... is ...' The new phrase nuku ... ki ... (or is it ku ... nuki ... ?) means '... whether or not ...'; the English definitely does not correspond in this utterance. Is kanoi ... kinoi ... legal? Does it mean 'not ... or not ...', as it appears?

Answer 3-26. It may be that the problem with kimoi is back at imoi. How is that defined this week; it has changed at least three times that I know of? Is it now the "because or the "therefore" word? Your case suggests that it is the "therefore" one nowadays. Why is nuku ... ki a new phrase? and it ought to mean the same as the second thing, disregarding the first {which referent? rjl}. kanoi ... kinoi ought to be legal and mean what you suggest, the kekked form of noanoi. To ruin your day, what are the kekked forms of noa and of anoi? It's a nasty rule but it is the only way to make forethought really work. pc

(I don't claim anything as the 'nowadays' definition. I never understood keks before, and still don't. The term 'new phrase' referred to the forethoughts being reversed from their L1 grammar, which would not have attached the connective to the first half of the connective. rjl)

3-27. Apparently la farfu converts the predicate farfu (father) to a name. The question is, is the use of la veridical? Veridical means that the argument makes a claim. Per pc, le is non-veridical, it makes no claim that the referent exists. Lo is veridical, it does claim that the referent exists, and can be massified. La is not certain; nor are the various other descriptors, such as lie, liu, li...lu, lea, loa, lua, lue, lae, and lio.

Answer 3-27. la, like le is not veridical. la farfu, like le farfu, does not claim that the person referred to is a father. [note: it is not a question of whether the referent exists but whether it has the property apparently being attributed to it. That is where the question of veridicality arises. There is a question about existence, which comes up in a variety of peculiar places, but not specifically here. The referent of le exists - barring a peculiar kind of mental problem on the part of the speaker - simply because he is in effect pointing to it with this expression. We skip over the problem of works of fiction viewed from outside their dzabi for the moment.] The quotes are no problem, since they present the object in the saying, and so give both its character and its existence. loe and lea are veridical, about the real sets of things that really have the property mentioned, and I think the same is true for the rest, but can't be sure, because I can't remember what some (many) of them are - and there may be a le-like set marker - and even mass marker. pc

(If loe is veridical, what are the properties of loe li unicorn lu. Or if that comes too close to the fiction question, loe gandi when expressed by a devout theist to an atheist? rjl)

3-27a. An example of the veridical question:

la Crlak,holmz no sitfa Lndn
le Crlak,holmz no sitfa Lndn
ne lea Crlak,holmz no sitfa Lndn

In which of these three utterances must Sherlock Holmes exist, in order for the utterance to be true? Those which require existence are veridical.

Answer 3-27a. Negation is one of the places where the question as to whether the referent exists arises. Assuming that all of your samples are grammatical (as at least the first and third would be if you put a la in front of Lndn - maybe even without, but then we are invoking {vocative} the city - or the singer {oops! I intended the la - rjl}) and that subj -no- pred means exactly the same as no-subj-pred, then none of these would require that any Sherlock Holmes exists. I doubt that the second condition is true, however, and so think that all of them require something to exist which does not inhabit London and which can usefully be referred to as "Sherlock Holmes", whether as a real name (I think that la with a name is veridical to the extent that it requires that the referent actually be called by that name {it is - by the speaker! rjl}) or at least as an implicit description which will direct the hearer to the right person. But, none of them require that London exists. In short, nothing in the scope of a negation need exist. (This is an extreme position, but none of the halfway positions seem to make consistent sense.) No denies truth to all that follows it and one way that a claim may be false is just that the referred to object does not exist. Not a common way, but one of the obvious possibilities. [The standard raba kanoi preda ke prede does not require that there are predas, yet we often do want to be sure that there are. How do we say that in Loglan? I used to use raba jio preda ga prede but I am not sure it will still fly. pc

(How about su preda gi dzabi .ice ra preda gi prede The claim of existance is a separate claim if the descriptors are non-veridical. Hence, it should be separately stated. rjl)

3-28. Another MEX proposal. We need two LWs for number representation: one for a characteristic, one for a mantissa. This allows them to be expressed in either order, which is sometimes required.

Answer 3-28. ? Does this mean we need two ways to represent numbers? Surely not, since the number is the same whichever role it plays. I suppose this is just a request either for a LW for "characteristic" and one for "mantissa" or else for two words, one to join c to m and one to join m to c (nu the other, probably.) pc

(Right! So only one is needed. rjl)

"We need another LW"... an incessant refrain. Each such claim should be supported by frequency data -- else the number of LW's "needed" will quickly exceed the space available. jp

(Thought must also be given to freeing up LW space by eliminating un- needed LWs. Obviously, choices must be made, and a new LW must be as useful or more useful than existing ones (and/or have no effective workaround, even if rarely used). rjl)

3-29. Does ma mean hundreds, or does it mean two zeros? Does tomane mean 201, 2001, or is it illegal?

Answer 3-29. Seems always to be used as "00", never as "hundred" in any meaningful way. pc

(Exactly opposite from the answer I've gotten from others verbally. Which is why these questions need asking, and resolving. rjl)

3-30. We have CVVs assigned for linguistic items (toa, toi), and for states of affairs (tau, tiu) which corresponds to po. Why not CVVs as counterparts to pu and zo for properties and amounts?

Answer 3-30. dua, due for pu. Do we really need them for amounts? pc

(Please elaborate - I thought dua and due were variables that stood for the predicates (with all arguments previously specified attached). How does this relate to the concept of pu. As for zo, we don't use it much in English, so its hard to think of examples of much of anything using zo. But symmetry demands the zo abstractions be treated equally, since zo exists, and no doubt mind-expanded native Loglanists will figure out how they can be used. rjl)

3-31. There are several grammatical forms to express relationship:

le tugle pe mi      is inalienable possession: it is the leg which I possess.
le tugle je mi      is my flesh leg even if it is removed from my body and is no longer in my possession

But what do lemi tugle and leda tugle (and le tugle ji mi) mean as compared with the above?

Answer 3-31. You have "alienable" and "inalienable" reversed. The other expressions are flat ambiguous in cases where inalienable possession is possible. Or we could decree that all follow pe or that all but one did (what the fatal fandango is le tugle ji mi? "the leg which is I"???? Odd appositive identification - some late JCB mishagash?) pc

(JCB defines ji as a local modifier in TL7/1, not an identifier. It would seem that this means that any type of modification is permitted - certainly he allows le tugle ji pa, (which doesn't mean the leg which is 'identified as the past'), as well as ji followed by a described argument. So why not a pronoun type argument. I thought it means 'in some way related' - or is that pe? Since he tried to combine ji with jio (unsuccessfully, it appears), ji appears to be a modifier that is non- veridical - modifier which identifies. But I'm now not sure how it is (was ever?) intended to be different than jio except that the latter takes an predicate utterance, and the former a tense or an argument. rjl)

"lemi tugle' etc are, quite simply, vague. It shocks may to find vague expressions in a "logically precise language", but the price of infinite precision is infinite verbosity, which we are not willing to pay. The Loglanist must be free to select how much precision of expression da wants, and then be willing to pay the verbal price. Every statement is ambiguous -- it leaves unsaid things which could have been said. Lemi tugle is shorter, and vaguer, than le tugle pe mi or le tugle je mi the speaker has decided not to pursue the relationship in detail. jp

3-32. lemi links a descriptor and a demonstrative pronoun to form a 'possessive' descriptor. levi links a descriptor and a tense to form a different type (non-possessive) of descriptor. Can they be combined as ?levimi or ?lemivi? Which is possessive, if either, and what does each mean?

Answer 3-32. Both are legal. Both are to some extent possessive - and to some extent locative. A plausible but speculative answer on meaning:

{ mine   thine
   x       x
 levami levatu} {?? rjl ??}

 mine    thine                     mine x lemiva
   x       x
 levimi levitu                     mine x lemivi
       me                               me

{I am assuming the above are spatial representations. rjl}

3-32a. As an example, I wanted to include both a tense and association/possession in an utterance, along with a reason rau. Thus I tried: ?lemi na preda rau mi prede. LIP didn't like this. I tried ?lemi preda ji na rau mi prede. Still no luck. But when I dropped the rau phrase, the latter worked. And what if I had needed to put in je ... jue ... expressions into the argument?

3-32b. Can one build descriptors out of other 'tenses' - members of the PA lexeme - such as ?letie or ?lerau, or out of logically connected tenses (?lepacena, ?lepacevi, ?lepacerau, ?lepacenoitiecevi)?

Answer 3-32b. I don't know about "tenses" like tie. I suppose letie has to be legal if tie is in PA (which I doubt {It is! - rjl}), but Lord knows what it means. But the logically connected tenses are surely ok and clear; remember la Artr ji lepacefa braga or god ol' la Zog ji letopa braga je la Albanias. pc

3-32c. Can one add -zu type suffixes onto levi? (Does le vizu preda mean the same as ?levizu preda? How about onto the longer descriptors in 3- 32? Can one add them onto lemi type demonstratives, and what would they mean? How about the miscellaneous tense descriptors in 3-32b?

Answer 3-32c. Yes, we can add -zu etc. to levi etc. Apparently, levizu is the only proper way to set TENSE before a predicate in a description, i.e., le vizu is just a variation of it. As far as I can figure out, zu and the like are tensors on vectors, so they can only be added where there is a vector to attach to. There is one inherent in proper tenses, in some locators (Wait! Is vi the "Here" form? That is clearly not vectored), and some aspects (physical contour extensive.) But it is hard to see vectors in most of the new "tenses". It may be legal in some of JCB's grammars to letiezu, but I think it is semantically indefensible. By the way, I think that, if you are going to write your own grammar as well as vocabulary, you would be well advised to go back to somewhere around trial 19 or so, and some of those earlier corpora, rather than trying to catch up with the latest version.

(Our intent with Loglan-88 was to change as little of JCB's grammar as was necessary. In short, if he and the Academy state clear policies, we will try to follow them. If no policy is stated, or it is ambiguous, or contradictory, or incompatible with other changes we've been forced to make, or (of course) likely to be challengable in court as a copyright violation, then we will change the grammar. This should be rather seldom on the whole. But then, if we attempt to rederive a grammar which doesn't rely on PreParser hacks, we may end up going back to a more primitive grammar. The idea, though, is to keep this language within the boundaries of the Loglan gestalt. Else we should change the name, as many have suggested. (The specific words for the primitives are not part of the Loglan gestalt, else JCB wouldn't be so fast and loose with changing, adding, or subtracting them, as he has done repeatedly since GMR was nominally complete several years ago. Also, he must feel that they are expressions of an idea/ideas rather than the idea(s) themselves in order to have a meaningful copyright claim. You cannot copyright an idea; only the expression thereof may be protected. rjl)

3-32d. Can one use the other simple demonstratives in this way? Lomi, lovi, leami, leavi, loami, loavi, laemi, laevi, etc. The lexer/grammar must deal with these complexities, along with combinations of these with any of the allowable forms from 3-32b and 3-32c.

Answer 3-32d. Yes, any descriptor can be modified by personal or tense affixes. I am not sure whether all lv(v) words are in this class. For example, I don't think that the various quote devices fit: lae, li lu, and lie. I forget what some of the others mean, e.g., loa. pc

3-33. In NB3, se sorme is described as now legal. Does it mean the same as se ba jia sorme? Can one meaningfully say le se sorme? How would this differ in meaning from the previous?

Answer 3-33. I think that, at its first appearance, se sorme means the same as se ba jia sorme (or is it jio, I can't remember how those go anymore - it should be the restrictive one.) It is not clear what would on second occurrences. {I don't understand this statement - rjl} What seems to be clear is that they would have different anaphora (pronoun pickup) : se sorma would be picked up by dv, while se ba jia sorme would of course repeat the ba. Both of these differ from le se sorma in at least one and probably two ways. 1). The le form is definite while the unmarked form is indefinite, about the some difference as that between English "the seven sisters and seven sisters" (=` "some seven sisters") standing alone. 2) le is not veridical but ba jia is veridical. The status of se sorme is undecided on this issue, but I suspect that the referent of se sorme have to be sisters and there have to be seven of them. [By the way, there is another kind of pronoun we need, reciprocals "each other, one another" - for the assumption is that the sisters are sisters of one another, not either that each is the sister of some other or that all are sisters of some other. This is also another case of implicit place fillers and notice that it is not the ordinary default value but one imposed by context. Like Carter, you are apt to end up having to read Austin and Grice.]

3-33a. If one can use numbers in this way, what would le re sorme mean? Just plain re sorme or re ba jia sorme?

Answer 3-33a. le re sorme means the same as neither re sorme nor re ba jia sorme, which probably mean about the same thing (if not exactly.) Again, it means about the same as "the several sisters" (or maybe just to stress the plurality of sisters) but again is not veridical. pc

[I should note that the whole notion of definiteness is open to some question as being a semantic, rather than a grammatical category. {Does this mean that definiteness is indefinite? - rjl} Many languages do very well without it, and it does seem to arise only in three kinds of contexts: follow-ups on indefinite ("a man" is introduced but later references are to "the man"), contrastively with a further specifier ("the man on the left, not the one with the handkerchief"), and where uniqueness is presupposed or guaranteed ("The High Mucky-muck.") So it may be that veridicality is the only significant difference here between the forms with and without le.] pc

3-33b. le re sorme would be very difficult for a Japanese speaker to say. How much thought has been given to the l/r problem for a Japanese Loglanist? Is there a solution?

Answer 3-33b. The problem with le re sorme for Japanese is that however they manage to say it, it may well not be distinctively different from their utterance of re le sorme, another and rather different expression ("several of the sisters".) pc

There must be endless such cases... a serious international language would start by finding a phoneme set distinguishable in all target languages, and adopt a phonology also in the common denominator. Try Hawaiian... jp

(Polynesian languages have also been suggested by Jack Waugh. But I think major changes to the phonology must wait for another language. We should at least be considerate of major problems for our target population in choosing words. It takes little effort, for example, when I am choosing a new prim for Loglan-88, to check for and avoid conflicts such as ?cinra and ?cinla, or even ?cilra and ?cirla, for separate primitives. rjl)

3-34. What is the proper referent of da? How does one count when faced with:

  1. ze groups: in preda ze prede, does da refer to one of the individual predicates, or to the mixed predicate?
  2. subordinate clauses: do arguments in the subordinate phrase count?
  3. in Xai preda Yma Wma da, does it refer to Wma, or to the last argument in the previous sentence?
  4. in Nau da godzi, what does da stand for, since Nau erases all assignments for the start of a new paragraph?
  5. In the example in c: if the following sentence is de prede di do du, it seems as though all variables stand for the same thing (Wma, again). Is this true?

Answer 3-34. When the rules were clear (even though even then unbelievable), the rule was that dV picked up arguments in backwards order of left ends. Thus,

  1. immediately after le preda ze le prede, da picked up the referent of le prede, de picked up le preda, and di for the whole mass (which begins just to the left of the beginning of its leftmost member. I do now wonder what would have happened with le preda ze le prede ze le predi: would di pick up le preda or le prede ze le predi or, since ze is presumably left associative, wouldn't do pick up just le preda ze le prede and only du get the whole expression. pc [Note, please, that ze connects arguments, not predicates, and that dV anaphorizes arguments, not predicates, so your a. question is ill-formed and misdirected. My answer is to what I take your point to have been] pc
  2. Of course subordinate clauses, linked arguments, and whatever else comes in line (except parenthetical remarks) counted.
  3. Wma (assuming, as always, that da has not already been assigned.
  4. Does nau really do this? If it does, da becomes a free referrer, essentially a ba.
  5. Again, by the old rules (and they have been rejected but not yet replaced), once a pronoun had been assigned, both it and its assigned term are taken out of the ordering. Thus, once da picks up Wma, de gets Yma and di gets Xai. pc

3-34f. pc has suggested that we need LWs for reflexives, to be used in later arguments of an utterance to represent earlier arguments. (In English: 'He shot himself', 'himself' is a reflexive.) This solves the problem of c. above, since a reflexive would be used instead of da to represent Xai, Yma, or Wma.

Answer 3-34f. Typically, reflexives in English just pick up the subject in some objective position. It might be useful - looking toward a new set of rules for anaphora in Loglan - to have a set that covers all prior places (one basic LW and a number?) within a single sentence, saving the dv series for previous sentences. pc

3-35. In TL4/3, pg. 26 middle - what is the meaning of pia? Is piazu 'a long time ago' or 'a long time period that took place sometime in the past'? Given one, how does one express the other?

Answer 3-35. Pia is not a problem; it places a predicate as holding for some extended period at a past time (literally, for some time on either side of a point at which it held.) But piazu is a problem for it contains two vectors to which the tensor might attach and so it is not clear which one it does attach to. Further, we need both kinds of attachments, so we need expressions for them. I recall a proposal (it even smells a bit like one of mine) to distinguish among piazu (ongoing a measured time), piaziu (a measured time ago ongoing for a measured time.) NOT OFFICIAL ever, maybe not even published. pc

3-36. Something like the 'case tags' proposal was apparently used by someone named Fillmore in several papers about AI natural language processing in the 1968-71 timeframe. Does anyone know of this work, and whether it is relevant to Loglan?

Answer 3-36. James Fillmore, "The Case for Case", in Fillmore and Terence Langendoen, but I forget the name of the book, but before 1971. I have read the article. It is relevant to Loglan but not very decisive. He is just, at that point, trying to explain some similarities and differences among different case systems. He is not trying to set up a universal system of cases, though he does discuss that briefly. pc

3-37. Ken Dickey and others have proposed a project to come up with 'AI frames' in Loglan. Could someone write up a good description of an AI frame, preferably with an English example, for publication (or refer me to one in print already).

3-38. Chuck Barton's primitive making approach uses the best Loglan phonetic approximation of the word, while JCB used hyphens for phonemes with no Loglan equivalent. Thus JCB would use /f-c/ for the English word fish, while Chuck would use /fic/, which is not exact, but gives good visual recognition if 'i' ends up between f and c, and better aural recognition than /foc/, for example. Loglan-88 will be using Chuck's approach.

3-38a. One curious thing about JCB's approach is that he represents diphthongs as hyphens, even when there is a valid Loglan equivalent. Thus, in L2, he represents brain as /br-n/ and not /brein/. Perhaps he, or someone else, can explain why.

3-39. A logic problem: How do we express the phrase 'coffee, tea, or milk' (or worse, 'coffee, tea, milk, or water') so as to make it clear that only one may be selected (true). Neither exclusive or inclusive or works for this combination. In fact, pc has written a program which derives 256 possible combinations of the 3-place logical expression form, and he says that there are 65536 combinations of the 4-place expression form. Loglan does well with the 16 2-place connectives. But his program finds that Loglan can only express 120 of the 256 3-place expressions using only 3 arguments and 2 connectives; some of these can only be expressed using forethought expressions (keks). How do we handle the 136 other cases? pc is working on possible solutions. Others are welcome to try, or may comment when pc sends his proposal to me. We'll submit our best shot to the Academy, or let them select from the various approaches that appear in print.

In any case, it seems that for n-place expressions, it is not practical to use the connective form. Even for n = 3, pc had to do truth tables to prove to himself that there was no way to express the 'coffee, tea, or milk' phrase with the existing connectives, and he does logic for a living. We need a generic form, which could be generalized for complex truth functions such as 'exactly two from column A and no more than three from column B' which can be really bad if A and B each have many members.

Answer 3-39. We should note that Loglan in fact does not cover two of the 16 two-place connectives. Admittedly, these are pretty useless: they make a true or a false sentence out of any components. On my program, it has one important restriction: it assumed that the items were mentioned in a fixed order. If we allow the order to be shifted (or invented a device for indicating such a shift within a fixed physical order), we could cover many more cases and could get rid of all the kek-only cases. I haven't a programming language with me, so I haven't pursued it further, but I do know that there are still some cases which are not covered. All the ones I have found so far (working by hand) can be described as "so-and-so many of this list" - "exactly one", "at least one", "at most one" and so on. pc

Now you are getting into Nary {?} logical operators. I've been there before, and the basic answer is: you really don't want to live there. As PC began enumeration, there are 22n n-place logical connectives: specifying an arbitrary one will take 2n bits of information, phonetically encoded in some fashion. Unspeakable and (worse) incomprehensible. Humans can handle 7 + 2 'chunks' in short-term memory -- just about the number of situations a 2-place logical connective presents the listener with. A general n-place logical connective presents the user with 0(22n) possibilities to track -- the speaker is asserting that one of the possibilities holds, but not saying which one. So you don't want compact forms for all such connectives, just the minority of comprehensible ones. (Note current Loglan can express any n-ary boolean function, it just gets very verbose.) jp

(I don't think I asked for compact forms for all logical connectives, especially if they mean complex truth trees. We do need to be able to specify order in sets, and selections from an ordered set - preferably without a lot of heavy duty connective use. Unfortunately, Jeff, I think this will take more LWs. But in this case no workaround is obvious. As you said, logical connectives won't work. rjl)

3-40. There is still a question on iglu-form borrowings. pc says that iglu itself may be OK, since he recalled that i must always be followed by a pause - apparently this came out of the Carter review. But for V- form LWs other than i, it seems unclear that the resolution algorithm will always work. Let's try an example:

  1. presume the complex ?gramrenu;
  2. put this complex in a compound argument: ?grafumna a gramrenu;
  3. per TL6/1 and the NB3 excerpt, the pronunciation could be: /graFUMna.AgraMREnu/; (the stress on the connective is optional but permitted; after the pause, it seems to be natural for a speaker to stress the connective at least a little)
  4. but this could also resolve as ?grafumna agra mrenu, which is a three-place metaphor using an iglu-form borrowing;
  5. one could go in the reverse direction as well - in at least this instance, it seems that iglu-form borrowings cause ambiguity to the resolver.
  6. for some examples, the required pause between the vowel at the end of a word prior to a initial vowel in a separate word would solve the resolution problem (as in to agra as compared with toagra; the first is pronounced /to.Agra/, the second is /toAgra/. But in the example, the one required pause is the same for both a and agra. And nothing requires a pause between a and gramrenu.

The conclusion seems to be that, except for a possible exception for iCCV words, VCCV is not a valid form for a borrowing.

3-41. A phonology problem: How are the divowels aa, ee, and oo to be pronounced? They must be two syllables, from TL6/1, as opposed to the long vowel of Japanese. But they cannot have a glottal stop between the pair, or toogra would be indistinguishable from ?to ogra (see 3-40 above). one cannot use a secondary stress to distinguish the syllables - all stress combinations are valid: toorveo has both of the pair unstressed, while toogra has the second 'o' stressed, and gratoo has the first 'o' stressed. Try saying all of these resolvably, without using a glottal stop.

3-41a. Another example: if glottal stop is permitted to separate such pairs, then lii glumrenu resolves also as li iglu mrenu.

3-42. JCB promised in his 5/28/85 announcement that several things would be upcoming in Lognet. It has been two years, and all members have had to renew at least once, or will no longer be getting Lognet. He owes us the promised data in Lognet. Unless someone else knows the nature of these proposals. (Has the Academy approved them yet?) I'll publish them, or we ought to see them in Lognet. (Editorial note - In any case, all those who were members at the time of that announcement should be given the promised data, as well as those who joined in the period shortly thereafter, possibly in response to the promises. Anything less might constitute fraud on the part of the Institute.)

  1. handling of Linnean binomials;
  2. policy statements on borrowing/metaphorizing;
  3. "fast tracks to borrowing"
  4. the Carter changes (and other changes to the grammar).

3-43. On MEX: There have been several proposals since the 1963 grammar in TL2, but no answers. But there are some assumptions that should probably be agreed upon. If so, then we may be able to work on solutions.

  1. The existing mathematical language is international in usage, and has a logical basis, at least within each domain of mathematics. There is no apparent 'universal' form that suffices for all domains of mathematics; MEX must accept this situation as a given. The bounds of mathematics are not much restricted; thus Whorfian arguments are not a sufficient excuse to supplant the international language. Change for the sake of change, or even aesthetics, will not sell to the customers – mathematicians.
  2. We cannot/will not change the international mathematical language to a 'better system' through Loglan. Loglan has no leverage to bring about change at this time, and MEX cannot wait until we do.
  3. It would probably take another 30-year effort like the one so far to come up with a new 'universal' mathematical grammar that might satisfy a willing user. But during any given 30 years recently, several new mathematics, and their associated notations, have been developed. Unlike verbal language, which is relatively stable for centuries, mathematics is seldom stable for even a decade these days. We also would be hard pressed to even assemble a compendium of all existing mathematical domains and their notations to even start a project of this scope. There is no equivalent to a Webster's Unabridged of mathematics. The programmers and engineers of the community can all testify as to the frustration of attempting to develop a system whose requirements are not fully known, and which change faster than the development process.
  4. People write in current mathematical notation. Loglan needs a grammar that allows one to read the current standard notations (any of them), straight from the written notation. Preferably this would be unambiguously grammatical, so it would be useful to speak it in Loglan as opposed to using lie nonce quotes, or a similar construct.
  5. The chemical equation problem is identical to, even part of, the MEX problem. All that I've seen on the acronym proposal makes it seem like it was designed to solve this limited subset of MEX. Such a limited solution improperly constrains MEX, to the extent that usage will cause resistance to a later change that might be needed to make MEX work. The same comment is true for the measurement problem (JCB says that Scott Layson's TL proposal was adopted a long time ago. But I've never seen it so stated in print, and I'm not even sure that it was on the street long enough for proper review before TL folded.) Any attempts to solve these problems independent of a general MEX solution will be of limited use, and of major effect on that general solution.
  6. Thus, MEX must be a generalized way of dealing with notational systems. This is the only true MEX that works universally. Since grammar itself is a notational system, the current Loglan grammar is a subset of the generalized MEX. Thus, to solve MEX, we must generalize Loglan to a greater order of abstraction in grammar, not add constructs to the existing grammar to handle the multitude of special cases.

Answer 3-43.

  1. I don't think there is an international mathematical language - just a collection of essentially unrelated and conflicting notations, largely unpronounceable.
  2. If you solve MEX, you can chuck the rest of the language, which will be a small, redundant fragment then.
  3. Strysmi, the language I've been working up, was born of MEX study, MEX enthusiasts may want to think about it.
  4. In practice, you're probably wasting your time. jp

(On 1, I'm coming to agree with you. But some portions are related, and I believe expressible in Loglan grammar. On 2, I believe that MEX grammar is a superset of the current grammar - so I agree in a limited sense. But not redundant; it will be integral, or it won't be Loglan. On 3, I will make available material jp has sent me on Strysmi (assuming he does not object), to any who are interested. On 4, I hope not. I believe a major application of Loglan is formal verification of computer software. The researchers in that field may have already solved MEX without knowing Loglan, and Loglan may solve their problem of designing a language to support their mathematical-language grammar. I am in contact with some who work in this field. Anyone else out there knowledgeable? rjl)

3-44. In TL4/3, there are defined two conflicting meanings for the LW pua. Pua is part of the Hixson-Bonewits argument tags series. But it is also the tense pa with the infix -u- to indicate habitual. LYCES/LIP only recognizes the former. The -i- and -u- infixes are not yet implemented in Trial 24.

Answer 3-44. So far as I can remember, -u- never was official, precisely because it interfered with H-B tags. But we probably do need something to do this job and an infix looks right (since this is just the intensional analog of -i-.) pc




The remainder of these questions and comments are rephrasings of comments on the NB3 draft. They are hopefully fairly clear, even though you do not have the draft text to look at. There is some overlap with previous questions, which simply means that they have come up many times.

3-45. Is it valid for buffered dialects to end names with a buffer? (followed by a pause, of course). French and Japanese speakers don't like to end words on a consonant. It seems that it should resolve.

3-46. Case tags - nobody (except JCB) has seen a list - and while nobody else has publicly taken up JCB's challenge to try their own analysis - there is a general disbelief that 13 types cover all possible argument sets. Perhaps the list of types, even without the specific LWs assigned, would help reduce the questioning.

  1. Also - how universal are they, and (how) do they overlap grammatically with the relative modifiers of L1, to which they seem similar?
  2. Is it possible that any case tag/relative modifier could be attached to a predicate - even those for which the tag has no specifically defined argument to define an otherwise non-existent argument place? The latter would seem to open up the realm of Loglandic thought still further, as ba contemplates the thought of the 'source' of le sucmi, or the 'agent' of le berci.
  3. The idea could be carried still further, with some little word which turns a following predicate word into a special-purpose case tag applicable specifically to the purposes of the speaker. A whole new use for the concept of metaphorizing.
  4. By the way, terms like 'source' and 'agent', etc. seem akin to some work done in natural language processing for artificial intelligence - the subject is called 'case analysis' and first showed up significantly in the 1968-71 time frame.

3-47. The 'se sorme' question. You may find a way to YACC it, but what will it mean in all its ramifications - and what won't it mean? Are you referring to a bare 'se sorme' or is it 'le se sorme'? Wouldn't it mean rather a 'seven kind of sister' - whatever that is? And since quantifiers are of lexeme MEX, what are the implications of MEXing quantifiers together into an expression (of currently indeterminate grammar) in place of se? Ruri sorme? Feni fera sorme?! Ri fefera sorme? (apologies for mangling the L1, Section 4.19 examples.) Will the Loglan for '(2+2) sisters' parse, or perhaps (remember story problem arithmetic) '2 sisters plus 2 sisters'?

Answer 3-47. Se sorme doesn't mean "seven kind of sister" primarily because se is not a predicate, so this is not a Pred-Pred structure of the sort that gives the "kind of" semantic structure. On the other hand ruri and fera are clear "enoughth sister", "pentadic sister", etc. But the predicates are not part of the MEX, though they contain them. And, I don't see that there will be any special problems with using more complex expressions - once we figure out how to do them - in quantifier positions. pc

Se sorme is clearly intended as a straight import from English, and means just what "seven sisters" would mean. (Nobody has ever tackled the 'meaning' of Loglan utterances in general. Loglan-to-date is an entirely trivial project compared to this) jp

3-48. On the apparent two meanings of lo in L1 - does lo to mrenu mean the mass of all instances of '2 men' or a specific mass individual. And how does one specify the other?

Answer 3-48. I'm not sure this right. Lo to mrenu ought to mean "the mass of all men, of which there are two". To in this position is, apparently, not restrictive, that is, this does not mean "the mass of some two men" nor "the mass of all men pairs". I think that one of the new lVV's is for the first of these "the ones I have in mind" (ahah! another function of le which may be distinctive.) The second would be lo mrenu tora. All of these refer to specific mass individuals, i.e., individuals composed of a single mass. pc

Lo is clearly intended to be totally alien to the western mind-set, and virtually unintelligible in terms of it, The point, I think, is that one is completely failing to distinguish between members of a set -- every rabbit is Mr. Rabbit. Holistic 'bigotry'? Who knows? I've come to conclude that higher mathematics is the most alien, exotic, mind- expanding stuff around, hence the best source of inspiration for would- be mind-expanders. jp

3-49. Is there a LW that is the equivalent of 'uh', 'um', and 'er' in English. That is - 'I am not finished with my utterance and do not wish to give up the floor'. We need one, or one will be invented (or borrowed from English) - and the resulting speech may be unresolvable.

3-50. JCB and I dealt on the phone with the question of how a vowel letteral is distinguished from a V LW. He said that Tai u Tai could be distinguished from TaiUTai by the mandatory pause before u in the former. But there seem to remain three other cases that this doesn't deal with:

  1. Djan UTai vs. Djan u Tai
  2. ...ITai vs. ...I Tai at the start of an utterance
  3. UTai as one of a string of acronyms used as several arguments - see JCB's examples in TL6/1.
  4. In short, there is a problem whenever there is another legitimate reason for a pause before the condensed vowel letteral.
  5. We questioned IBaiMai for reason b. above, which seems indistinguishable from . i BaiMai
  6. A comment on the acronym utterance example Kei, e, i, which summarizes the frustration expressed: 'Kei, e, i, - It is not obvious. It's downright ambiguous! Kei, e, i means someone began a sentence " k logical-and ..." and decided not to finish it. The i is the start of a new sentence, and listeners should give the speaker time to form the rest of it.'

3-51. 'nenimei' and 'toXai' - These two examples in NB3 will parse as letterals in LYCES Trial 24, since they end in a letteral. Don't know if this has been fixed in later versions.

3-51a. Nora points out that these are MEX, and that in using them, JCB dictates a MEX solution that may not be desirable. It is not implicit that in Loglan '2X' should mean '2 times X'. Nora and I agree that standard International Mathematical conventions should be able to be written in Loglan as they are internationally, with MEX grammar LWs inserted as needed when spoken (and optionally as written), which would make JCB's examples valid. (This eliminates some of the proposed MEX concepts, such as Reverse Polish). But in any case, this simple idea was controversial in the MEX discussions, and JCB should think carefully before using them in his NB3 examples. A lively discussion at Logfest was choked off so we could get work done. MEX won't be solved in a weekend. It remains to be seen whether it will take the 30 years that the rest of the grammar required.

3-52. My 'borrowing proposal' in the UL2 newsletter has met with very positive responses, although some proposed variations that are not in keeping with the metaphorical forms of the rest, such as putting the type-word first so it wouldn't have to be abbreviated to CVCV form.

3-52a. But there is a strong consensus towards establishing some pattern restrictions in borrowings, at least until we have usage information that justifies using the total borrowing space. And there is strong support for the idea that borrowings and jargon words should be arbitrarily longer than prims and simple complexes. This is partially based on Zipfean ideas - most words that would be borrowings are less frequent than prims and 2-Cpxs. And there is consensus agreement that we want to use Zipf to help Loglan grow. If borrowings are longer than most Cpxs, then people will make metaphors rather than borrow.

Answer 3-52a. Strongly agree -- have argued this for years... jp