(From Lognet 95/2. Used with the permission of The Loglan Institute, Inc.)
Letters policy: Unless otherwise stated, letters addressed to logli in general, to The Institute, JCB, or any editor of Lognet will be considered as offered for publication. But it would be good if the writer explicitly offers. We reserve the right to edit letters, mostly just to drop material that has to do with ordering books, etc. Sometimes a given correspondent will have several letters in the hopper, so to speak, and we will combine them into one to save space. If you are not on e-mail and your letter is a long one, we’d be grateful if you’d enclose a soft copy on a diskette. We can translate most word-processors into the one we use and having your letters on disk could save us a lot of typing.
This first letter comes from a New Member living in Taiwan, obviously a “sinophone”, a juntaa (“Chinese-talker”). We have, among our active logli, very few speakers of this second most populous language on Earth...actually, the most populous if we count only primary speakers. So each new juntaa who arrives in our imaginary country deserves, and gets, a very special welcome as we strive, in Loglandia, to become as internationally diverse as we can be.
I wonder if you could reply to any following question whose answer will not be more-or-less evident from Loglan 1: (Also, questions based on the 1960 [Scientific American] article may no longer be applicable due to subsequent revision.)
(1) Perhaps I have misunderstood, but Da cortu de means X is shorter than Y, Da farfu de means X is the father of Y, and Da donsu de means X [gives] Y? Using an identical structure, three different relationships—comparative, genitive, and accusative—result...depending on some intrinsic quality of the predicate in question?
(2) Da vedma de di do means X sells Y to Z for [price W]. What happens if du is added on after do?
(3) If a Han speaker is uncomfortable with the word order resulting from the use of a relative pronoun, can he say Donsu de ne mreni instead of Ne mreni ke donsu de?
(4) Perhaps under the influence of Yale Romanization, the 1960 article spells certain elements from Hanyu Putonghua (“Chinese”, “General Hannic”, “Mandarin”) containing unaspirated [ts] as dz-; would ds- not be closer to the borrow[ed] sound, and be less likely to auditory confusion with z-?
(5) What are the Loglanized words for China (state/area) and Chinese (language)? Djunguo[+C]? Dcunguo[+C]? (Or, is the metaphor “central+state” replicated?) [Or] Xan’iw[+C]? Han’iw[+C]?
(6) There is (or used to be) a class of persons regarding “mere formal logic” as a “flatland”. They make claims for a “Hegelian” or “dialectical” logic in contradistinction to the “formal”. Does/Would Loglan satisfy them in its expressiveness?
What follows is a slightly edited version of the p-mail (“paper-mail”) reply sent to Taiwan:
Hoi Ivan Aimat, fie (which means ‘O Ivan Aymat, friend/comrade’):
Your order has been sent. ... Now for your questions:
1) Da corta de di = X is shorter than Y by quantity Z (note the phonemic change in *cortu as well as the addition of a third place to its place-structure); Da farfu de di = X is a/the father of Y through mother Z (also with a 3rd place); and Da donsu de di = X gives gift Y to recipient Z. All three of these predas are, from the logical point of view, “3-place predicates”; that is, they require a specification of a triplet of “designata” (designated things) for their claims to be completely expressed. The “intrinsic quality” shared by these 3 predicates is that each of them takes exactly 3 designations, no more no less, to produce a sentence from which it can be ascertained whether its claim is true or not. Thus one can’t know that X is shorter than Y unless one also knows that X is shorter than Y by some measurable/perceptible amount Z; similarly one can’t know that X is the father of Y (in a sexual species) unless one also knows that a mother Z exists; and so on.
2) Given Da vedma de di do, you ask what happens if du is added. If the logically complete sense of the preda (in the sense above) is X sells Y to buyer Z for price W, then adding a fifth place will announce to the reader/listener that the writer/speaker does not think the standard 4-place structure is complete and so has added a 5th place—at first metaphorically—to make it so. Eventually your view of the matter might prevail. In that case vedma would be defined in the dictionary, and presumably be used by others, as having a 5th place. (What is the 5th place you had in mind, by the way? Or is this a hypothetical example?)
Sometimes this does happen. For example, years ago someone pointed out that “seeing” is always a figure-ground operation: one sees some figure/thing only against some ground/background. This insight added a 3rd place to vizka, the present word for see, which now means X sees Y against background Z.
3) Can a Han speaker—who would evidently feel more comfortable with ?Donsu de ne mrenu as a variant of the relative clause construction Ne mrenu ji donsu de = A man who gives Y— use this variant and be understood? No. But let us first note that *mreni is now mrenu and that who/which/that, once ke, is now either ja or ji depending on whether the relative clause is “non-restrictive” (predicating) or “restrictive” (identifying). You probably have the identifying sense of E who in mind; so I’ve replaced 1960’s ke with ji. But the answer to your question is in either case no, for Donsu de ne mrenu already has a clearly defined meaning. It is an imperative meaning Give Y to a man! Relative clauses in L always immediately follow the argument they modify...which, among other things, makes room for imperatives as unmarked predicates. Additionally, and as apparently doesn’t happen in Chinese, logli must specify whether each relative clause they use is restrictive or non-restrictive. We do this by choosing ji (or one of its kin) or ja (or one of its kin) to attach the relative clause to the term being modified.
4) Yes; the source of this error, if error it was—and I accept your judgement that it probably was—was the Yale Romanization scheme commonly used by Western scholars in the 1950s and ‘60s. This was itself an effort to correct what were evidently the many errors of the Wade-Giles system of transcription used—at least by E-speaking scholars—in earlier decades. I’d be interested to have your judgement of the influence of this error on our selection of “clue-words” for Chinese-speaking learners of L. To make that assessment for us, you’ll need a dictionary: either the old, 1975 one (still available from us in paperback) or the current and much larger computer-based dictionary, LOD (“Loglan Online Dictionary”). If you want the latter, I will need to know which version—Mac or IBM—your computer requires.
5) The name of China is Djunguos in modern L (the final /s/ being added to make a L name-word out of /djunguo/). The L predas meaning what E Chinese means in its four predicative senses, i.e., Chinese language/Chinese territory/Chinese person/Chinese culture, are, respectively, junga/junge/jungi/jungo. These endings are now standard in L for all “ethnic” predicates. Thus dotc-a/e/i/o mean E German in these same four senses, the L words having obviously been derived from G Deutsch/deutsche; and glec-a/e/i/o are the four senses of E English. There are now dozens of such “ethnic” quadruplets in L.
6) I cannot answer this one because I do not know what sorts of logics are currently considered “Hegelian” or “dialectical” by logicians. I’ll get one of our lodtua (“logic-workers”) to field this one.
Kerju! (‘Take care!’)
Hue Djim Braon (‘Said Jim Brown’).
The next letter is an answer to Ivan’s Question (6) and comes from Emerson Mitchell, one of our two currently active lodtua:
Hoi Ivan Aimat, fie:
Re your question: “There is (or used to be) a class of persons regarding “mere formal logic” as a “flatland”. They make claims for a “Hegelian” or “dialectical” logic in contradistinction to the “formal”. Does/Would Loglan satisfy them in its expressiveness?”
I am not an expert in the dialectic, but I understand that it proceeds historically, chronologically, in time, using the sequence Thesis then Antithesis then Synthesis (T then A then S). All the examples I have read are in English and use ordinary natural language to express T, A and S. There is no universally agreed way to mark T, A and S that I am aware of. All of the dialectical language I have read of could be translated into Loglan (as easily or with as much difficulty as any other specialized jargon) assuming we had words for introducing T, A and S. I do not think we have words for T, A and S. Would Ivan Aymat like to suggest some?
So the short answer is, Loglan would satisfy Hegelians no better and no worse than any natural language.
The next letter comes from Alex Leith, recently but no longer a New Member. This letter reports his experience with our MacTeach programs.
Dear RAM and JCB:
Having completed M1 and M2 I can report that it’s delightful. Makes learning easy and fun. The amount of repetition with the three Rungs, and the extra repetitions in the Error Box, seems to be about right (using the default settings of 4 in M2 and 2 in M1). I’ll send in the stats files separately. To JCB or whom? [To JCB, eo.] ...
Re MacTeach 2: The strategy I evolved was to select 25 new words at a time into Rung 1, and move them up 25 at a time, having established 200 in Rung 3 and 100 in Rung 2. So they were well mixed. I only used Review mode in Rung 1, and for the Error Box. This seemed to work well, with few errors. So the sequence was: from the 200 in Rung 3, move 25 into Top. Review Rung 1. From the 100 in Rung 2, move 25 into Rung 3. Move all 25 from Rung 1 to 2. Select [a] new 25 into Rung 3 and Review. I’d let 20 or so accumulate in the Error Box, then clear them into Rung 3 all together, and add a few new ones. The Error Box was reviewed from time to time while it was filling.
When working on Rung 3, if I noticed hesitation, or felt I didn’t know a word well enough, I would deliberately let it fall into Error. Having started to do a bit in Brush-Up mode, [I find] there are still some errors and confusion (e.g., cundo/condi).
[RAM invited Alex to make changes in the MacTeach input files.]
Changes made: With the animals, I found it confusing (in L>E) to have to remember to put male, female, baby first, and to choose between baby, infant, larva. So I’ve changed these so that the response is species first, then type. So lion male, tiger female, butterfly young, walrus-like. Specific words like tadpole, mare, caterpillar are kept as alternates. Young is the default for all immature forms. [Can I still respond female tiger to tigra and get it right?—JCB]
With the nationalities, I have changed [the input file] to require [Italian] person, district, language or culture. So it won’t accept just Italian. Latin is accepted as an alternate to Roman language.
[Alex also made the program accept many more E expressions as “correct responses”. Here are some examples.]
Additional E variants:
pozfa -> added resist (to better distinguish from bufpo)
kraco -> added squash (to better distinguish from zakra)
fatru -> added bother (conform with M1)
femdi -> added feminine
mendi -> added masculine
[Feminine, of course, only translates femdi when it means pertaining to females, as in feminine politics. Often E feminine means like a female, and that is femcli in L. Remember that it is the Loglandian habit to build these semantic distinctions into the words themselves, that we are not content to guess even closely-related meanings from context. Is there any way, Bob and Alex, we can require the M2 user to qualify the responses feminine and masculine so that the program will know that the sense intended is pertaining to females/males and not female-/male-like? I suspect that what we may need here is an entirely new teaching program, a MacTeach 4, that will teach people to decipher and use our complex predicates with all the subtlety they deserve. For it is now with CPXs that Loglandians make the fine-grained semantical distinctions that E lacks. Any takers? Alex himself seems a lively prospect for this role.]
Re MacTeach 1 : [My] strategy here was to take a few [utterances] at a time and move them straight up to Rung 3, then let a bunch accumulate there and clear them to Top now and then.
Personally I’d still rather have Review mode give only the full [utterance]. Perhaps at Rung 1 [I’d] show the word-blanks. But [I’d rather] not ... have to go through the whole backward build-up. This latter is fine for Test mode at Rung 1. ([I’ve] mentioned this before to RAM, and I’m content to leave it as is, since a change would demand recoding & compiling.)
In E>L it’s annoying to have to do a sentence twice, for the literal and idiomatic English versions, though these are good to have when doing L>E. Solution would be a separate file for each direction; but [that’s] not worth doing, [as] it only applies to a few sentences.
There are a few places where, in L>E, a perfectly good E expression was rejected. But covering all the possibilities would be unduly complex, and one gets used to watching for the acceptable expression.
Changes made: Piu, niu [were] changed to pio, nio (in jolkeo [“clock-time”] utterances); cue [was changed to] geu (to RAM: by the way the amended LOD doesn’t contain geu).
Animal [word]s [were] changed to their generic form[s], [e.g.,] horma -> hormu [and] gotca -> gotcu ...
[In] Ui la Keit, no dupma mi [I] ... inserted [pa] as the English has didn’t. I feel it makes more sense to regard each utterance as separate, rather than [part of] a story. [Yes indeed, especially as, in M1, the user should randomize the utterances for a final test of u’s “context-free” understanding of the material.]
I don’t quite know what to do about [the following items]. In view of the decision about no+UI [see RAM’s report on these new compounds in this issue], [it] seems to me that Ai no is [still] fine [as I intend to not (do or permit something)], but now *No ai [will be heard as the single word Noai] express[ing] a lack of [firm] intention: I don’t strongly intend to (do or permit that). So [Noai] is not a refusal at all. [That’s right; it isn’t.—JCB]...
No oi is [still] translated correctly [by the program], I think, as You-need-not/You-don’t-have-to/It’s-Ok-if-you-don’t. [although of course that phrase will now be compounded as Nooi].[Nope. Nooi—pronounced [NOH-oy], by the way; remember the “pair from the right” rule?—will now mean the negation of the permissive sentiment: I’m not giving you/them permission (to do that). This is “refusal” in one of its senses, namely “not giving permission”.—JCB]
[Here are some other attitudinal-teaching items, some changed to teach the new No-compounds:]
Ai [Empasize the strength of the intention.]
I+will./shall/intend-to.(Similar to the nautical reply ‘Aye, aye, sir’, a signal of an unconditional intention to comply.)
Ai no [Emphasize the negativeness of the in-tended action.]
Certainly-not!/(No)I=will-not/won’t. (An equally firm refusal to comply.)
Noai [Emphasize the lack of firm intention.]
I’m not certain I’ll do that..
Noai no [Emphasize the lack of firm intention and the negativeness of the intended action.]
I’m not certain I’ll not do that.
[All this works.—JCB]
Djori Aleks has been doing some much-needed work on our MacTeach programs as well as becoming one of Loglandia’s most active purmao (word-makers).
The next letter comes from Germany. As always we try to make our overseas newcomers feel especially welcome. For any X not E, we hope to make our X-taa logli kumce grow.
About one year ago I first heard about your project on Loglan. This appeared to be interesting for me as I at that time had lessons in language philosophy, formal logic (G. Frege etc.), and phonetics, and I’m since long interested in language planning and planned languages such as Volapuk, Esperanto, and Interlingua.
After reading several articles about Loglan in some linguistic magazines, among them your introducing article in Scientific American of 1960, I borrowed the book Loglan 1 from the Universitaetsbibliothek in Franfurt am Main. The reading increased my interest in the project.
Some weeks ago I tried to get all the books from this series Loglan 1...5 through my bookstore, and my bookstore got a list from you which prescribes what you are offering now, also the magazine La Logli and the software about which I didn’t know before.
Therefore, I decided to join your small community and order your complete package of “everything we make” [which he forthwith did] ....
But I have to admit that my studies at Goethe-Universitaet Frankfurt are my second ones. I’ve already got an academic degree as an engineer in 1988 at Karlsruhe University, and I’ve been working for years. Then I decided to study linguistics in Frankfurt. Today I’m part-time working and studying. ....
I’m looking forward to hear from you soon.
We hope Loglan sets Cninu Djori Felman alight with new ideas, and that he becomes an active member of the doctaa logli munce (German-speaking loglanist community).
The next letter comes from Roy Bigelow. It is a continuation of a correspondence among Roy, Dr. McIvor, and me that began in LN 95/1.
Dear Jim and Dr. McIvor:
I read with great interest your comments concerning my letter ... in Lognet 95/1. Since you collaborated on the comments, I am sending both of you this letter.
For Jim: ... I wonder if you have any current documentation explaining the abbreviations and conventions used in the presenting the definitions in LOD? [RAM will answer this in his letter below.—JCB]
For both of you: My thanks for the comments. I learned quite a bit about both the Grammar and Loglan. The comments concerning the grammar per se provide confirmation that the Grammar is nicely intended to be as mute as possible. I also want to thank you for clarifying the uses of the words nor- and kin, nu/fu/ju and kin, [and] nue and zea. I like the words ganlea and damlea for superscript and subscript. [They have now been added to LOD.—RAM] And I certainly agree with your approach to constructing [complex predicates]. My discourse at the end of my letter was only intended to tell you about my “quick fix” to one of my problems in applying the older version of the language, and as an illustration of a reason that the newer version is superior.
However, to continue in regard to the Grammar:
Your comments concerning the word no only made me do some thinking about why I was bothered about the lack of distinction between its logical and non-logical appearances. [Are there any “non-logical appearances” of no?—JCB] I remembered that, in relation to semiotics, I (perhaps foolishly) employ the concept of “context-free” to differentiate syntax and semiotics [see JCB’s Note 1 below], and it distresses me to find the same word appearing both as a syntactic constant and as a syntactic variable. Such gibberish! Then I thought about the fix. Pragmatically, my “quick fix” is to eliminate the word no from the rules for <freemod> and <uttA1>, specify use of one of the words which are listed in LOD as synonyms for the word negative (or some alternative word), and add that word to the UI Lexeme. As argument: Is the language “context-free” when semantic differences are determined by context? [See Note 1 again.] Isn’t the word no the only word in the language to have this peculiar characteristic? [No; but for the answer to this, see RAM’s letter.] And isn’t it “logic-facilitating” to make a clear distinction between its two meanings? [It would be if it were not for the extraordinary ability of humans to recognize different “species” of linguistic “genera” from their linguistic settings alone. Thus, the difference between No, ta breba and No ta breba, in which two different kinds of no are being used, is clear from the presence in the first of the pause-comma and its absence from the second. In the first, No is a “sentence negative” and has the sense of It is not the case that. In the second, No is equally clearly an “argument negative” with a sense, in this sentence, of Not that (but something else) is bread (That’s not bread!) We humans have no difficulty making such syntactic distinctions from context; so far, at least, machines, with their shorter “lookaheads”, do.]
Your comments concerning dio were interesting but not completely satisfying. To put the question another way, are there any instances in which one might want to include arguments with differing argument tags in a string of connected arguments? [LIP doesn’t check case tags for “appropriateness” to the local predicate, so the grammar would accept this. But usage rules might well dictate against this practice....at least for a while. Consider Da madzo sau de, e pou di (X makes (something) out of material Y, and (something else) into product Z). Would you be likely to hear this pair of claims in the compact original? I don’t think I would...at least not for the first time or two. But it is possible that logli of the future will make these elegant place-filling transformations with blissful ease!]
Sorry, but my example sentence Jack, Jane, Joe, walked, ran, flew, from Paris, Rome, London, to Vienna, Moscow, Cairo did not communicate. I had intended, by the superfluity of commas and the absence of a conjunction, to convey the fact that it is an abbreviation of a collection of 81 different English sentences obtained [by] using [one] of the subjects, one of the verbs, one of the direct objects, and one of the indirect objects. [In that case, and assuming the connections intended are all e’s, the L rendering is: La Djek, e la Djein, e la Djos, pa dzoru, e pa prano, e pa maifle la Vin, e la Maskvys, e la Kahiras, la Pari’s, e la Romas, e la Lyndyn (I’ve put the 2nd and 3rd connected arguments in L-standard, “to-from” order in order to avoid case-tags), and this is, indeed, shorthand for 81 separate claims.] However, your comments that the purpose of gu and kin is limited to making the parser work as desired made me think some more. I mistakenly understood gu and kin to be punctuation marks rather than grammatical operators. [But they are “punctuation marks” as well as “grammatical operators”! That is, they make a difference in the structures, and so ultimately in the meanings, of utterances.] I therefore withdraw both my suggestions regarding “alogical” connectives and my suggestion regarded the word ?igue. On the other hand, I think you will agree that I identified an area not covered as yet: a well thought-out Loglan style manual ... [I agree that we need such a manual; see my Note 2.] Surely there is someone in the Loglan community who has been sufficiently interested in semantics to have listed those instances in which punctuation is useful in clarifying the semantics of text in a variety of languages and, therefore, would find the generation of such a manual a trivial exercise. [I hope so; and if not a “trivial” at least an eminently worthwhile one for a logical language.]
With respect to my comments regarding the words cao and siu, I now find them in the dictionary. In trying to figure out how I could have generated the comment, I repeated the exercise ... checking the meaning of all the UI words. Now the word ii doesn’t appear. In the past, I have attributed all the funny things that LOD does to the fact that I run it under Windows. [See RAM’s letter below.] Is ii really missing, or is it a configuration problem? One of the funny things that my LOD consistently does to me when I use the “-n” option is to shift the definiens (but not the rest of the display) off the screen to the left either one or two characters. Is it a configuration or [a] software problem? [See RAM’s letter.]
With respect to my comment regarding nui, I apologize for a typo. ... Let me change the comment to: Neither the Grammar nor LIP considers the word nuu, which is defined in both LOD and Loglan 1 to mean whether. [See RAM.]
With respect to Lexeme NI. I am revising my rules and will forward a copy when finished. The revision occurs because Loglan 1 and NB3 talk only about concatenated words; I read the Grammar as not permitting gaps between them; and LIP concatenates them on output. However, don’t expect any great insights because, basically, I follow the suggestions made in my last letter: Differentiate between the conversational uses of numbers (counts, measurements, ranges, etc., where the digit zero is a word and not a number and the mathematical symbols do not appear) and the mathematical ones (where counts and nulls do not appear). [I don’t understand this. Surely zero is another name for “null” and does appear in many mathematical expressions. Moreover, any integer can be a “count”, that is, a measure of the numerosity of some set.]
With respect to bi and kin, I note: In mathematics there are many symbols denoting “equivalence” which might appear in an “identity sentence”. [See Note 3.] For many of them there is not, as yet, an associated Loglan word. In mathematics, the “equals” symbol may be interpreted to mean is computationally the same as and the “identity” symbol may be interpreted to mean may, wherever it appears, be substituted for. In logic, the “identity” symbol denotes tautological equivalence. [Not usually, see Note 3; but perhaps I don’t understand this paragraph.] The symbol has approximately the same meaning as its mathematical counterpart. But the “equals” symbol is a symbol which is reducible to, or is defined in terms of, tautological equivalence. It denotes the equivalence of the qualities or characteristics of primitives rather than the primitives themselves. In Loglan, there is only one word denoting equivalence. With the exception of its last paragraph, the discussion in Loglan 1 concerning bi hits the mark. However, the last paragraph denies the existence of more than one of the interpretations of equivalence and thereby engenders a semantic ambiguity. Ah, such gibberish! As argument: How can you say you have made logic speakable when the language does not distinguish between the [several] logical meanings of “equivalence”? Is it not making mathematics “speech friendly” to include in the language words for all the mathematical symbols for equivalence? [See Note 4.]
As a small prolegomenon, some maunderings:
The survival advantage of humans over other forms of life is the ability to communicate experience. That ability requires ... a part of the brain which is unique to humans and language. A by-product of that ability is the ability to reason. [Note 5.]
Text exists because it extends the ability to communicate experience by permitting it to be recorded and played back. A by-product of the existence of text is that it permits the ability to reason to be exercised in the absence of communication. That is, it permits the extended attention span needed for the manipulation of closely related but complicated ideas ... . [Note 6.]
From experience, I claim that Loglan text so enhances one’s ability to reason and effectively communicate that it or some analogous language will always exist. [Note 7.]
Fortuitously, I ran across a thing called the predicate calculus and the Backus-Nauer notation, which were of great help to me in my work with computer languages and compilers. Fortuitously, you, Mr. Brown, used the predicate calculus as a template in the design of Loglan [Note 8.] and I ran across it in Scientifiic American. To find out whether or not Loglan was of value, I tried translating parts of the technical documents I was generating into Loglan, doing the editing in Loglan, and translating the Loglan text back into English...with phenomenal results, both good and bad. Since the documents lacked the usual soothing filters, the people who read them found them to be abhorrently clear and concise. [Note 9.] It was then that I knew I was on to something important, so much so that I worked out my own syntax. Fortunately, you worked out the Grammar in YACC [Note 10] and I came across your advertisement in the Scientific American . ...
Again, thanks for your attention.
Robert A. McIvor’s (RAM’s) answer to some of Roy’s questions follows. Some of my responses to Roy’s letter have already been inserted in it. Other, longer responses—mini-essays on interesting topics broached by Roy—appear as a set of Notes immediately following RAM’s letter.
Thank you for your interesting letter of May 31. At the time, I was in the midst of preparations for the next issue of LOD, and now ... I’ve found some time ... [to] answer [it].
As for the format of LOD, on the L-E side, the first line of a predicate entry gives the predicate word, its affixes (if any), its etymology enclosed in < ... > , its derivational type, e.g., Cpx, C-Prim, or S-Prim, and its “recallability score” if it is a composite primitive (C-Prim). The remaining elements are the source of the word—this is given either by the initials of the wordmaker or by ‘L4’ if it came from the 1975 edition of the dictionary—the year of its acceptance, and finally its “Eaton use-frequency rank” in 1,000s or fractions of a 1,000. E.g., ‘1.1’ means that it is one of the first 1,100 most commonly used concepts tabled in Eaton’s 1940 An English-French-German-Spanish Word Frequency Dictionary. This final number should not be taken as gospel, however. It is especially likely to be inaccurate if it is for a recently added word. I would be in favour of dropping the use-frequency datum entirely. But currently the first line in a predicate entry always ends with this use-frequency number.
The first line of an L-E predicate entry is always followed by one or more English definition lines. The first of these always translates the L-word into English as it is used “completely”, that is, with all its possible argument-places filled; e.g., Da farfu de di = X is a father of Y through (mother) Z . This translation of the most complete usage of the L word may then be followed by translations of its various incomplete usages. E.g., the role of farfu in Da farfu peebru = X is a paternal uncle is explained by the line ‘(a) paternal, pert. to fathers, q.v.’ Finally, there may be one or more lines that translate various phrases involving the L word. In these, the em-dash stands for that L word; e.g, ‘po — ’ represents the phrase po mrenu if the entry word is mrenu. Between the L and its E translation there is always a coded abbreviation in parentheses that shows the English grammatical type of the translating word or phrase. The grammar code in the first and most complete translation will always include the maximum number of arguments that can be used with the L-word; thus ‘(3v )’ means ‘a verb phrase that takes a maximum of 3 arguments’, e.g., ‘madzo (3v) K makes state/object P out of S’. As in this example, the translation may contain uppercase Latin letters (K, P, S, etc.). These identify the “cases” of those places for that predicate; Table 4.1 on p.247 of L1 gives the meanings of these case letters and the “case tags” to which they correspond. (You should make the notation on this page that M(ao), once derived from cmalo = small has been replaced by J(ui) <= junti = young. Mao was confusingly like the affix -mao from madzo = make.) Assigning case-tags to all the places in the predicate words in LOD is a long-term project that is just now starting. Eventually all L predicate words will have their most complete E translations so-marked; e.g. ‘K go to D from S via V’. Note that case tags may be considered as either singular or plural; so the E words in the translation appear in the form in which they would appear on the E-L side of the dictionary. (This is because a single master file is used to produce both sides.) Hence ‘K go’, but ‘B is a box’.
On the E-L side of the dictionary, a line containing only the E word or phrase to be translated into L is given first. On each of the indented sutori lines, an L translation is given and then explained or translated back into E. Between the L translations and the E back-translations there is always a parenthetic code giving the grammatical type of the principal word in the translation, e.g., ‘(2a)’ = ‘2-place adjective’; ‘(af)’ = ‘affix’; ‘(i)’ = ‘imperative’. After the grammatical code, a derivation of the L word often appears in <...> , and this is followed by a list of case tags if any have been assigned, e.g., ‘[K-B]’; in these expressions the en-dash indicates the position of the predicate. The E back-translation follows, often with case-tags in place.
Turning to your next question, NO is not the only lexeme in the human dialect of Loglan that is context-dependent. A, KA, ZE, PA as well as NO have subscripted forms in the context-free machine grammar. Not all of the subscripted forms are necessary to make the language context-free, but they are necessary to give sufficient redundance so that many right-hand closures may be omitted with impunity. This is the way most humans prefer to speak.
I interpreted your point about Jack, Jane, Joe, walked, ran, flew, from Paris, Rome, London, to Vienna, Moscow, Cairo as intending just three combinations. At least I assumed that interpretation, although I don’t think the E would necessarily be interpreted in that way. As JCB indicated, the way to say this in L is still under consideration, other than the much more understandable La Djek, dzoru la Pari’s, la Vin, ice la Djein, prano la Romas, la Maskvys, ice la Djos, maifle la Lyndyn, la Kahiras. It seems to me that the e-conjunction, as in La Djek, e la Djein, e la Djos, dzoru ce prano ce maifle la Pari’s, e la Romas, e la Lyndyn, la Vin, e la Maskvas, e la Kahiras, gives the required 81 combinations.
Ii is really missing from the dictionary, or rather it got unintentionally combined with one of the others (I believe io) This has been fixed in the version of LOD that has just become available. That shifting you mention doesn’t happen in my DOS program as simulated on the Macintosh. I can’t speak for Windows, but we would very much like a Windows programmer to volunteer to make Windows versions of all our software.
Thanks for pointing out that nuu didn’t get into LIP,which interprets it as nu u. A new version of LIP may be out soon that will deal with this, as well as one or two other minor points.
As was pointed out, we have no official MEX expansion. Also, because we have no definite stress rules for little words in combination yet, LIP assumes on a trial basis that all sequences of little words unbroken by punctuation marks are compound little words. Where the little-word grammar is sufficient to tell LIP what to do with these trial compounds, the compound is then either preserved or rebroken in various ways. Currently all MEX words remain compounded, but when a grammar of <mex> is developed, it will be added to the grammar that runs LIP.
I agree that equivalence may need to be looked at some more.
Robert A. McIvor
Those of my own answers to Roy’s questions which were too long to insert in his letter, and so became mini-essays, follow as numbered “Notes”:
Note 1: Re your remark that you “employ the concept of “context-free” to differentiate syntax and semiotics”, I don’t see the relevance of “context-freeness” to this distinction. In the first place the relation between syntax and semiotics is a part-whole relation. Semiotics is that part of the general theory of signs (see, for example, Charles Morris, “Foundations of the Theory of Signs” in the International Encyclopedia of Unified Science, vol. 1, 1938, for the seminal essay on this topic) that deals with “constructed signs”, which of course includes human language. Syntax is one part of semiotics, namely the part that concerns itself with the relations among the elements of a constructed sign; the other two parts of semiotics are generally called “semantics” and “pragmatics”. Context-freeness or -dependence can occur in both.
The word grammar is a funnily ambiguous term. It was once used by linguists to mean the entire rule-structure of a language, but in L studies, as increasingly elsewhere in modern linguistics, we confine it to the syntax of a particular language. Thus, in our usage, the “grammar of Loglan” is its syntax. Interestingly enough, the grammar of “machine Loglan”—the language that LIP parses—is context-free; that is, its grammar instructs a parser that is limited to one “lookahead”. But machine Loglan is different from “human Loglan”, the language that you and I write and occasionally speak. The chief difference between machine and human Loglan is, in fact, that human Loglan is “context-dependent”. It requires a Preparser to process each utterance before delivering it to the Parser. The Preparser does this mainly by examining the linguistic settings in which certain “ambiguous lexemes” in the human language occur. For example, both A and NO are ambiguous in human L. But their syntactic roles are easily determined from their lexemic settings. So the Preparser looks at those settings and attaches a “subscript” to each of their instances; that subscript then “disambiguates” that lexeme for the Parser. For example, there are four types of human Lexeme A; but as each type is recognizable from its setting alone, the Preparser can examine those settings and then relabel each instance of A it comes across as one of the machine Lexemes A1, A2, A3, or A4. As one of the Preparser’s tasks is to recognize these machine lexemes from their contexts, the Preparser is a “context-dependent” part of the Loglan grammatical system. Once all the As and NOs (and other ambiguous lexemes) in a human-generated L utterance have been transformed into “machine lexemes” in this way, the Parser can then read the transformed utterance—which is now an utterance in machine Loglan—and use the Preparser-generated information it finds in it to parse it. The notion of context-dependence is thus primarily a grammatical notion in L studies, although it does have some limited use in semantic studies of the language...for example, in differentiating the several uses of the same lexeme string by their syntactical contexts.
Note 2: Re your suggestion that someone should write a “Loglan style-manual”, I certainly agree. What you have in mind would be very similar to what I have been proposing for several years now, namely that we compile a “Catalog of Loglan Usages”. Such a catalog would eventually contain all the usages, logical and extralogical, that we have decided are not only grammatical L but “good Loglan” as well. For example, it would contain the “observative” (Lo simbu!) and rules for interpreting it; it would contain the “unwinding rules” for connected arguments, modifiers, and predicates (A, e B preda <=> A preda, ice B preda); it would contain the “elimination rules” for all “non-primitive” operators (e.g., nuo as in A nuo preda B C <=> A preda A B C), thus isolating the set of primitive ones; and it would ultimately contain every usage we can uniformly interpret. In addition, the catalog would specify the context of a usage if it is context-dependent. It would also give an “interpretive expansion” of each usage showing what it “really means”. This last point really needs doing. For until we logli can interpret our usages formally in this or some other way—that is, eliminate them by saying exactly what they mean without recourse to them—it will not be strictly true that “we mean what we say and say what we mean”. For in the case of an uninterpreted usage, we won’t know what we mean!
Note 3: Equivalence and identity are two quite different relations. An identity sentence simply equates two names—two different designations, to be more precise—of the same thing. The Evening Star is the Morning Star is the classic identity sentence...and a surprising one it is, too, to those who do not know this interesting bit of local astronomy! It was made a “classic” specimen by Gottlieb Frege’s use of it (in German, of course) in 1892 to make precisely this point (in “Ueber Sinn und Beteutung”, translated in Feigl and Sellars, Readings in Philosophical Analysis, 1949). Most philosophers of mathematics would still say that “2 + 2 = 4” is an identity sentence in just Frege’s sense, ‘2 + 2’ and ‘4’ being just two different ways of designating the same number. If so, this is all that “mathematical equality” means. (To some mathematicians that’s going to seem just too “nominalist”, “empiricist”, or “positivist”—in more modern language, too anthropological or evolutionary—to tolerate without rebuttal. I invite any such readers to write letters to this column expressing their different viewpoints.) The word equivalent, on the other hand, while wonderfully ambiguous in E, is generally reserved by logicians for the biconditional relationship between sentences: S1 if and only if S2. S1 and S2 are said to be “equivalent” if and only if (there it goes again!) their truth-values are the same under all possible conditions. We also say (rather loosely) that equivalent sentences “mean the same thing”, but linguistically they do not. “Truth-value meaning”—which is what most logicians are concerned with—is only one of the many kinds of linguistic meaning. For example, (1) John hit Joe and (2) Joe was hit by John are logically equivalent sentences; but they do not “mean the same thing”. This is because in (1) that John did the hitting will be psychologically more prominent to the listener, while in (2) that it was Joe who was hit will be the more prominant feature. A speaker—say a lawyer—who could have said (1) but chose to say (2) could well have had good reasons for doing so. As it is all of the intentions of a sign-producer—including p’s rhetorical goals—that constitute the meaning of p’s sign, it is more accurate to say that a pair of equivalent sentences “make the same claim” than that they “have the same meaning”.
Note 4: There is no specifically mathematical symbol for equivalence that I know of...at least not in the sense just developed. Lines in a mathematical proof or calculation, for example, are related to some preceding line or lines by either the conditional or the biconditional relation, that is, by either “implication” or “equivalence”. But these are logical relations of quite general application. In L, for example, the equivalence relation may be expressed in any sort of utterance by any of the o-bearing connectives o, co, ico, or ko...ki, while the identity relation, whether between the designata of two mathematical expressions (To pio to bi fo, for example, which is normally written in L, as in E, as 2 + 2 = 4) or of two verbal designations (La Djan, bi la Djek, which could be written La Djan = la Djek in L, or as John is the same person as Jack in E), is expressed by bi. At one time in the history of L it was thought that an identity sentence involving two mathematical expressions—such as To pio to bi fo—required a special identity sign; but someone—I forget who—satisfied him- or herself and then the rest of us—by which I mean those of us who were deciding what was “official Loglan” at the time—that it did not. Again, To pio to bi fo is one of our usages. Examination of the linguistic context in which it occurs can provide a “correct expansion” of it into Lio to pio to, bi lio fo.
Note 5: Formulating evolutionary hypotheses about the biological connections between language and thought has been a compelling activity for many recent thinkers, including linguists Steven Pinker (whose book on this topic was reviewed by Alan Gaynor in LN 95/1) and Noam Chomsky (Language and Mind, 1968). All such hypotheses bear strongly on the Sapir-Whorf hypothesis, which is perhaps the most famous of these scientific models and the one that sparked the Loglan Project. As L was developed specifically to test such notions, your interest in them is quite natural.
A few years ago biologist Bill Greenhood and I put together what is, to date, the longest biological scenario purporting to explain the language/mind phenomenon—by which I mean a string of causally-linked evolutionary hypotheses that reaches over the longest interval yet covered by such language/mind scenarios—which attempts to explain the biological connections among language, mind, and other evolved human behaviors...especially the peculiar ones, such as our sexual behavior, our fathering, our tool-making, and our singing and joking. You might be interested in having a look at this still fairly recent work. Unsurprisingly, our thesis is closely related to the idea behind the Loglan Project. For among other hypotheses, we propose that human language—especially its grammar—may be viewed as a “disambiguation machine”. That, of course, is what L supremely is. (The last published version of our scenario appeared in the Journal of Social and Biological Structures in the fall of 1991 and was called “Paternity, Jokes, and Song: A Possible Evolutionary Scenario for the Evolution of Language and Mind”. If you’re interested, reprints are still available from TLI.
Note 6: Yours is an interesting view of the role of text in human mental development. I’m sure it has one, and this may be it. But there is an impressive body of evidence (marshalled in PJS, above, but also in other places) suggesting that the biggest change in human inventiveness—or at least a very big one—occurred during the transition from the Middle to the Upper Paleolithic, which was of course long before the appearance of writing. This blossoming in human tool-forms occurred nearly everywhere between about 40,000 and 30,000 B.P. (“Before the Present”) and represents a quantum leap in human inventiveness. The movement into the Upper Paleolithic apparently changed the slope of the “hominid culture accumulation curve” forever. It changed from something that had been nearly horizontal in all previous epochs into something that was to become—and stay—nearly vertical...as of course it is today. Moreover, this extraordinary change in the archeological record followed very closely on the heels of a suggestive paleontological one: the “sapientization” of the hominid chin and forehead. All scholars agree that these skeletal features are signs of the arrival of characteristically human tongues, pharynges, and frontal cerebral lobes. These developments are telling us that human speech was developing into a critical new stage, that it was probably speeding up, and that these now “sapient” humans were probably producing longer utterances. These, in turn, would have required more powerful grammars to disambiguate them, and out of this spurt in the evolution of human gammatical capacity might well have come that “quantum leap” in human inventiveness that the archeologists observe. How? Through this sudden increase in the variety and flexibility of utterance forms, the “world of ideas” might just as suddenly have emerged. Out of new ideas arise (among many, many other things) new tools.
Human writing, on the other hand, appears very late in the Neolithic, the earliest signs of it appearing about 5,000 B.P. So it appears that the earliest episodes in human prehistory that suggest “the manipulation of ideas” were Paleolithic ones that came long before the invention of writing.
Note 7: Your view that writing “Loglan text so enhances one’s ability to reason and effectively communicate that it or some analogous language will always exist” is certainly a most encouraging prediction. I have often felt such enabling effects myself...or ones that I tentatively classified as enabling. But one’s own life is such a poorly designed “experiment” that one can never know the causes of its singularities. In any case, as the inventor of Loglan, I am certainly pleased that someone else has such confidence in its long-term survival...or in that of some other “predicate language” (as linguist Steve Rice calls L and all its progeny).
Note 8: I hope not quite fortuitously, soi crano. The project of discovering whether a wide sample of natural language structures could be rewritten in the predicate calculus—or in the extension of that calculus that I developed in 1956—was the very first “engineering test” to which the baby language was put. It was during the summer of ’56, scarcely 6 months after the first descriptions of L grammar and morphology had been set down, that I put the idea of making the predicate calculus the grammatical foundation of a “tiny, experimental language” to an empirical test. I used Otto Jespersen’s Analytic Syntax, a challengingly diverse catalog of natural language specimens, as my “test-bed”: I attempted to recast every specimen in it in the predicate calculus, or in that extension of it. By the end of that summer I had succeeded; I had rewritten all of Jespersen’s specimens in what had become “juvenile Loglan”. The foundation of L grammar had not only been laid, but had been found capable of carrying a fairly heavy load.
Note 9: Re your observation that some people found your back-translations of L text into E to be “abhorrently clear and concise”, I expect we should pay more attention to these aversive phenomena. The very existence of such effects suggests that there may be logli—people who are drawn to L not for its clarity but for some other compelling reason (but what?)—who are actually appalled by its “option of clarity”...or, at any rate, by the decision of others to use it. I too have found logli saying—indirectly, to be sure, but nevertheless unmistakably—that some standard ways of saying things in L were “abhorrently clear and precise”. Is there, then, a style of mind to whom imprecision, vagueness, ambiguity, and equivocation are perhaps the very soul of language? Or is this, perhaps, an expression of the poetic impulse in us all? I admit it is hard to see how such a “taste for unclarity” could have been a product of any evolutionary process that was driven, as Greenhood and I have proposed this one was driven, by a need to disambiguate increasingly complex signs. Yet “poetic ambiguity” is also involved in word-creation, and in perhaps critical ways. It may have been the long utility of metaphor—and metaphors are necessarily ambiguous—to those among our ancestors who were our word-makers, that now makes our Loglandian pursuit of more and more exact ways of saying things “abhorrently clear and precise”...at least to some of their descendants, soi crano. In L, we try to feed the poetic impulse as well as the clarifying one. But we may not have succeeded with the one goal as well as with the other.
Note 10: Your natural assumption that it was I who “worked out the Grammar [of Loglan] in YACC” is false. I helped; but it was primarily the work of five other logli: (1) Sheldon Linker, who first called the existence of the YACC algorithm to my attention, and who, by casting out the parts of the existing grammar that didn’t “yacc”, wrote the first “core grammar” of L that did; (2) Douglas Landauer, who built the first version of LYCES, the tool with which we brought a corpus of L utterances to bear on our grammatical work, an old linguistical trick; (3) Jeff Prothero, who used Doug’s tools and our growing corpus of test sentences—it was I, I confess, who played the role of the “native informant” who supplied the items in the corpus—to extend the scope of the yaccable grammar; (4) Scott Burson, then Layson, who worked with me to perfect LYCES, and who used it with me to parse larger and larger parts of the corpus...leading to that unforgettable, shared moment in February 1982 when we parsed it all; and (5) our present Takrultua, Bob McIvor, who took over from me as Chief Grammarian in 1987, and who has kept Loglan grammar on that pinnacle of “total yaccability” ever since. These five people, with an occasional hand from me, “worked out the grammar in YACC.” —JCB