(Originally appeared in Lognet 90/2)

A Critique of Loglan Morphology with Suggestions for a New International Language

by Rex F. May

Part 1 — Critique

Utopia is not an option, and even the metric system is imperfect. To find flaws in Loglan is not to condemn it. Quite the contrary, pointing out the flaws is an auto da fe, the fe meaning that one (I, in this case) considers the language to be intrinsically so worthy that it deserves reform and the effort necessary for it. Some flaws I’m going to discuss I have proposals for correcting, some I do not. And some of the corrections are tentative. I hope to get a lot of feedback on this. Am I correct that these things are flaws? If they are, are they serious enough to merit further study? And if that is the case, what merits do my proposed changes have? So first I’ll list the flaws, as I see them, and afterwards make some attempts at correction.

Flaw #1. Leamability of primitives. For me, the presence of ik in likta is of very little help in recognizing the word. It certainly isn’t reflected in the score of 2/3 that the word gets in the evaluation system. And if there is some slight advantage in recognizing the word there is virtually none in learning the word. I think that the word-derivation algorithm is a fascinating intellectual exercise that doesn’t really work in practical terms. We quantify things that really can’t be quantified, and, having done so, mislead ourselves as to the actual outcome.

Flaw #2. Language neutrality. Ever since a bunch of other Europeans jumped all over Zamenhof’s case for not really creating a neutral vocabulary for Esperanto (I say other Europeans because as far as I know, no non-Europeans ever really complained), neutrality has been a bit of a fetish in the artificial language movement. When a group decides they’re going to try to be neutral, right away they find that they have to start making tradeoffs. In Loglan, it would seem that the first tradeoff was to limit the neutrality to eight languages. And this, I instinctively feel, was a tradeoff that indeed did have to be made at some point. It could have been cut off at ten languages or twelve, but a cutoff had to be made somewhere, thus seriously compromising the neutrality. And the next tradeoff was deciding on what basis to select the eight languages. Number of speakers was selected as that basis.

This selection would not have been self-evident to me. My first notion would have been to select the eight biggest (assuming that eight is the handiest number to deal with) language families, and from each, the language with the greatest number of speakers. I can’t say what the eight would be, but among the first few would be English (from Indo-European), Chinese (from Sino- Tibetan) and Japanese (alone in its family, alas). And none of the other languages actually used in Loglan would be used, as they’re all Indo-European. Other languages used might be Swahili (from Bantu), Indonesian (from Polynesian), Turkish (from Ural-Altaic, unless they’ve changed it again), Arabic (from Semito-Hamitic), Quechua and Nahuatl (both from American Indian language families I can’t name), and one of the Dravidian languages from Southern India. And there are many more families. Even if this were done, we’d be leaving a lot out.

I suppose my point here is that neutrality isn’t really possible, and the Loglan attempt at it may have caused more problems than it solved.

Flaw #3. “NG” isn’t a phoneme. This has been driving me crazy for quite some time. Everybody who speaks any of the target languages can make a velar nasal. For some, it’s true, it’s an allophone of “N” used before “G” or “K,” but they can still make it. And for the vast majority—speakers of English, Chinese, Japanese, German, French, not to mention millions of speakers of non-target languages—it’s a full-fledged phoneme. Not having it as a Loglan phoneme severely restricts the importation of words from the target languages, and causes, again, misleading quantification results as far as recognizability is concerned, when an English or Chinese word that contains “NG” has it interpreted as an “N” for primitive-making. I think what we have here may be some unconscious Anglocentrism, or perhaps Roman-alphabet-centrism. “NG” tends to get lost in English orthography, so we forget about it and think of it as somehow less important than other phonemes.

Flaw #4. The admission of y (schwa) into Loglan. I hate this. Schwa is bound to be an allophone of a for English-speakers, and is also bound to be a buffer between consonants for speakers of Japanese and other languages poor in consonant clusters, which is sufficient confusion, without also having it serve as a hyphen, let alone as part of morphemes, as in ytrio. This is somewhat the opposite of Flaw #3, where we lack a phoneme that would be useful without being confusing. Here we have one that is confusing without being all that useful.

Flaw #5. The ciV words. My instincts tell me that cia (for example) is bound to collapse into either ca or ci’a (the same goes for the jiV series), and the fact that ci’a is officially permitted as an alternative pronunciation seems an inelegant solution to me. Giving a word the option of being either monosyllabic or disyllabic is carrying permissiveness too far.

Flaw #6. Ao. Making this letter-combination stand for the diphthong in English cow just seems wrong to me, and I think it has its origin in one or more of the systems of Romanizing Chinese words, as in Mao-tse Tung. I see au as the evident way of representing this sound in Loglan. That is, say ua and then play the tape backwards and you get what is now represented by ao.

Flaw #7. Semivocalic allophones of i, u, and o. This problem is related to what I said above in #5 and #6. For one thing, it is another set of pronunciation rules to learn, and hence a source of mistakes, and for another, it goes against the spirit of the one letter, one sound principle. Moreover, as I said above, there are exceptions allowed to it, a further complication. I frankly don’t think a semivowel should be a permissible allophone of a vowel. Esperanto gets along fine without it, using special letters to represent semivowels.

Flaw #8. Procrustean Primitives. This is related to Flaw #1. I had a big gut reaction against all primitives being required to consist of the form CVCCV or CCVCV in the beginning, though I could see the advantages of the resultant ability to resolve words from the speech-stream and find the morphemes without ambiguity. I didn’t like it, but I saw the necessity for it. Then the borrowings came along, and I got more confused. If iglu and athomi are okay, why restrict composite primitives to the five-letter shape in the first place? For me, at least, the resolving becomes much less self-evident when the borrowings are included, and it would be no more difficult if the five-letter primitives were abandoned altogether. Finally, the frustrating unlearnablily of those final vowels has been commented on by many others. That final vowel seems to have little justification for its existence.

Flaw #9. Affixes. Allophones are one thing, allomorphs are something quite else. Let’s take an example. Bivdu = behaves has four possible affix forms, biv, bid, biu, and bivdy. The fourth one is logically derived from the five-letter form, so there’s no extra learning effort there, but the fact remains that, including the primitive itself, you have to learn four words for behave in Loglan. Few primitives have as many 3-letter affix forms as does this one, but several have two, and many have one. Of course, some don’t have any, but you don’t know that till you look them up. Esperanto certainly doesn’t have this problem, and, for that matter, I doubt if any natural languages have the problem to the extent that Loglan does. English certainly doesn’t. Most allomorph situations in English seem to follow some pretty simple rules, and there are hardly any cases of a morpheme with more than two allomorphs, except in the case of some strong verbs, and even those tend to fall into a small number of categories that do have rules governing them. And the argument that you can use the language without the affixes doesn’t hold water, because you certainly can’t listen or read to someone who does use them without knowing them.

Flaw #10. Esthetics. This is the most subjective. It’s going to be mostly my instinctive feelings, and I’d be interested to see how congruent these feelings are with those of other people.

Trying to classify my esthetic objections, the first would be lack of variety. The CVCCV/CCVCV problem again. Far too many words take these forms, and it’s boring. Despite my other reservations about it, I felt downright refreshed when iglu showed up.

Ugly clusters. This might properly belong up there with the pronunciation problems, but it has an esthetic component, too. Even those of us who have the h phone in our native languages have got to have a hard time with birhu or sofha, and if you’re a native Spanish-speaker, forget it. And a velar fricative can’t be substituted as an allophone, because that sound is restricted to the new x.

R and l as vowels. This looks ugly ortho-graphically (which, I agree, is a very minor point), and has got to sound terrible to a lot of people whose languages don’t have such things—and I think most do not. If hidrroterapi is encountered by someone from Polynesia or Latin America, that first r doesn’t have a chance. It’s sure to get lost or drift into some other vowel. I think I see here some definite Anglocentrism. M and n might be a little more popular as vowels, but have the same disadvantages. In actual fact, I think a lot of vocalic r’s in English are actually schwa-r’s in some idiolects. And that goes for all the liquids and nasals.

Pauses. I don’t want pauses to have any phonemic or morphemic significance whatsoever. Reason being that a pause by its very nature is very hard to avoid when it’s not supposed to be there, and what if it has meaning when you use it unintentionally? And because sometimes it seems quite unnatural to me to insert it in the speech-flow, like after names.

Flaw #11. Names. Very clumsy and contrary to common sense. If we want to use a name a lot, like English, we make a primitive for it, despite the fact that it’s every bit as much a name as La Kraisler, etc. And the names require taking on totally different shapes than the “normal” words in the language, and have that damn pause stuff associated with them, too.

Well, that’s a lot of complaints, and I, like many other people, have been thinking deeply about whether they need solution, and what those solutions might be.

[To be continued.]

Copyright © 1990 by The Loglan Institute. All rights reserved.

Thanks to Mike S. for proofreading the OCR files for this article.