(Originally appeared in Lognet 90/3)

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

Rex F. May

Part 2 — Suggestions

Here are some proposals [for correcting the flaws discussed in Part 1 of this article], organized in as logical an order as I can come up with:

A: Sounds & Letters

First, adopt the “X-Rating Loglan” proposal described in a previous issue of Lognet [LN89/1]. (Quickly, X has no value, but is really just a diacritic. It changes the meaning of the letter that follows it. xt is the English sound [th] in thin, for example.) This removes X from the list of real letters, leaving us with the 25 others. There is no doubt that the no-diacritics idea is a sound one, so we’ll stick with those 25 other letters. There will be some changes:

A - No real change. As in father. Schwa as allophone strongly warned against. Schwa to be restricted to use as a buffer between consonants.

B, C, D, E, F, G - No changes.

H - As in English, with German CH or Spanish J permissible allophones. I do this because all the target languages besides German simply do not have these as separate phonemes. This eliminates Loglan X, with its separate [kh] sound, and collapses it with H.

I - only the full vowel sound in machine. No longer ever a semi-vowel. That role to be served by Y.

J, K - No changes.

L, M, N - Always will behave as consonants, as will R.

O - only a vowel. Will not be a semi-vowel in diphthongs.

P - No change.

Q - Big change. Now this letter represents a velar nasal, the sound in sing or singer. Not the sound in finger, which is a velar nasal followed by a voiced velar stop. This sound will come in handy, as the target languages are loaded with words including it.

R - No longer vocalic. Any language’s R-sound will do, but trilled as in Spanish or Italian is the preferred allophone.

S, T - No change

U - No longer a semi-vowel. Always a vowel. See W.

V - No change.

W - The semivocalic version of U. Kaw is pronounced [cow]. Twa is French trois, pro-nounced, as it often is, without an R-sound.

X - Now a control letter, used to modify pronunciation of the following letter in a standard way. Not to be used in any Loglan words in the usual sense, but only to transcribe “foreign” words.

Y - Now the semivocalic version of I.

Z - No change.

With this alphabet, we retain our freedom from diacritics, and regain a complete one sound-one letter system, by eliminating the situations where a letter can be either a vowel or a semivowel. Inclusion of Y and W also will serve to increase word space, that is, tua and twa are now separate words, the first disyllabic, the second mono-syllabic.

B: The new predicate shape

The old CCVCV/CVCCV shape had the in-spired function of insuring that speech could be broken down into words. That function must be preserved in any new predicate shape. What I propose here is a development of my old “RMR” suggestion of several years ago ["Rex May's Revolution".—Ed.], and it requires redefinition of letters to some extent.

This next is hard to express clearly, so bear with me. Liquids and nasals, now R, L, M, N, and Q (like [ng], remember), are for word-shape determination purposes reclassified as vowels, as are the new semivowels, Y and W. That is to say, they still behave as consonants in the sense that they don’t form syllables, but for determining word-shapes, we consider them vowels. Then, the rule is:

All primitives will have the form nCnV, that is, any number of C’s followed by any number of V’s. The immediate problem here is that little words also fit that limitation. So the rule actually is:

All morphemes have the form nCnV. There is then a special case of morpheme, the little word. Little words, then, are morphemes, but they have a more limited shape.

All little words have the form CV or CVV, with liquids/nasals excluded. I exclude the liquids/nasals from little words for a few reasons. First, I want them to be as similar as possible to the current little words. Second, allowing the liquids/nasals to serve as vowels in the little words would make far more little words than we need, while making many nice tidy monosyllabic primitives impossible.

This system will make for a much easier analysis than the current system does. Any morpheme you hear or read will begin with a consonant. After that consonant you keep going past x number of consonants, knowing that you’re still in the same morpheme. Then you come to a vowel, and then some more vowels. You’re still in the same morpheme. But when you encounter the next consonant, you know that the morpheme has ended and you’ve just started on a new one.

Disadvantages. Oh, yes. First off, all mor-phemes beginning with vowels are gone. Including attitudinals. Likewise all morphemes beginning with liquids/nasals. This has the unfortunate effect of eliminating all those neat little words that begin with L, R, M, and N. A quick count of those little words affected from the list starting on page 511 of Loglan 1 shows a loss of the following: 5 V’s, 22 VV’s, 20 CV’s, and 40 CVV’s, for a total of 87 little word losses. However, little words now in the form CuV and CiV will largely remain and become disyllables, while the monosyllabic forms will now be spelled CwV and CyV. And a quick run-through of the CVV list shows about 65 new little words will be gained that way, and also a number of Cao’s will now be disyllables, with the old pronunciation being represented by the Caw form. Likewise, current CVi’s will split into disyllabic CVi’s and monosyllabic CVy counter-parts. In short, we’ll probably end up with just as much little-word space as we had to begin with. And the obvious disadvantage would be that we would have to remake all primitives and most of the little words.

But there are heaps of advantages, the main one of which, in my estimation, is that primitives can be lifted entire and whole from the target languages. A tremendous number of words with the nCnV shape are right there waiting for us. Here’s how it should be done:

Let’s say you want the word for red. You don’t make it, you look for it. You look in this order:

Mandarin 806 (million speakers)

English 426

Hindustani 398

Spanish 308

Russian 287

Arabic 182

Bengali 175

Portuguese 166

Indonesian 132

Japanese 123

German 118

French 115

Punjabi 74

Korean 67

Telugu 64

Italian 63

Tamil 62

Marathi 61

Cantonese 60

Wu 59

Vietnamese 53

Javanese 53

Turkish 52

Min 45

Thai 45

Ukrainian 44

Polish 41

Swahili 40

That’s 28 languages with 40 million or more speakers. You look in order for the word. You look for red, and sure enough, there it is in Man-darin, hong, which becomes Loglan hoq. I’ll bet that over 90% of the primitive vocabulary can be located this way before we get down to Russian. (And, of course, you can keep going down the list of the remaining 4,768 languages if you haven’t found it by Swahili.) This should be done system-atically, of course, but just casting about in my own memory for words, I can find a load of them right off...

From English:

Tree > tri

Brown > brawn

Green > grin

Banana > banana (Yes! Multisyllables will be permitted, but the vast majority can and will be monosyllables. In fact, if the word you want isn’t there in Chinese, and is a 2-syllable in English, look a little further to get the word, especially if it’s a common one that will be used in compounds a lot.)

Spoon > spun

Stem > stem

Feel > fil

From Mandarin:

Shuei > cwey (water) (note that this is CVVV, and therefore not a little word.)

Chi > tci (eat)

Ban > ban (carry)

Jin > jin (gold)

From Hindustani:

Dil > dil (heart)

Janna > djan (know) (of course, in inflected languages, endings like the -na for the infinitive here, will be left off)

Bhalu > balu (bear)

Billi > bili (cat)

Kala > kala (black)

From Spanish:

Tener > ten (have)

Gran > gran (large)

Cama > kama (bed)

Perro > pero (dog)

Tomar > tom (take)

Poner > pon (put)

But wait. This solves some of the problems above, but does it solve the problem of distin-guishing between a modifier-modified phrase and a compound? I think it can be made to:

For the time being, I’m replacing all the little words beginning with l with some beginning with d. So we have:

de blufawl (the bluebird) and

de blu fawl (the blue bird)

Okay in writing. If they’re written together, two morphemes make a compound, but how are the two cases distinguished in speech? Two ways. First, and most common, will be stress:


[deh -bloo-fowl]

That should suffice for most speech. If we have a two-syllable situation in one or more of the primitives, it might be harder to distinguish. Oh, time out—the rule is that the stress of a word will always fall on the first syllable, okay? So it’s [BAH-nah-nah]. So let’s say the black bird and the blackbird:

de kala fawl = [deh-KAH-lah-FOWL]

de kalafawl = [deh-KAH-la-fowl]

Then black berry and blackberry:

de kala beri = [deh-KAH-lah-BEH-ree]

de kalaberi = [deh-KAH-lah-beh-ree]

Actually, in English, there’s more to it than stress. There’s also a different juncture. But I think stress will suffice, and stress of some sort is used in all languages that I know of.

Okay, if that ever doesn’t work, or we’re not completely sure that we’re communicating, or if we’re reading it to a computer and want to make it absolute, we can do a non-Zipfean thing:

de kala sa fawl = de kala fawl

de kala zu fawl = de kalafawl

Sa says the word is over, zu says it’s not done yet. This is optional, and for most union of monosyllables situations the stress will do it. Sa/zu would almost never be actually written down, but serve as a “pronunciation” of the orthographic space or non-space between primitives.

One possible final variation would be to use sa and not zu. That is, two primitives together would make a compound unless separated by a little word. If no other little word is appropriate, sa is used. Feedback?

Another thing... Current Loglan has a lot of common affixes that tend to be used in compounds a lot. Neo-Loglan primitives that are going to be found in compounds a lot should be selected with that in mind. I’ve already mentioned that they should be monosyllabic. Also, they should not have consonant clusters if they’re going to be used as anything other than the first element. Moreover, such primitives should wherever pos-sible begin with a consonant that will interfere as little as possible with a terminal nasal in the beginning primitive—in short, they should begin with a fricative followed by a V. As an example, I’ll choose out of Hindustani an ending, van, which I think doesn’t have a separate existence as a word, but means cause or make, much like the Esperanto ending -ig. With van, then, we could make hoqvan (to redden, make red) granvan (to make big, cause to be big), drimvan (to make, or cause to, dream). When you have a word that does have interference, you can of course insert a buffering schwa. Maybe, say grinkala (greenish-black) could be pronounced [GREEN-uh-kah-lah], so the N won’t turn into a Q.

So we really ought to end up with a list, much like the list of Esperanto affixes, of primitives that are used in compounds a lot.

This is an example of what Loglan must do if it’s to get anywhere—be as good as Esperanto in every way possible. One thing Esperanto by damn does not do is lay allomorphs on us.

I think if we adopted this sort of change, we’d end up with a language easier to learn and more pleasing to the eye and ear. We’d have at least as much neutrality as current Loglan, and maybe more, since we’d occasionally dip down into Indonesian or Swahili or Mayalayam for a word. And, practically speaking, the learning of the affixes in current Loglan isn’t as hard as it looks, despite my fussing about it. But the point is that it does look hard, and has no doubt frightened a lot of newcomers away into Esperanto or Basic English or aUI.

And now I’m going to take on the name prob-lem...really sticking my neck out. Since I’ve made la impermissible, I replace it with da. Names will now have the same form as predicates, save that they’re introduced with da instead of some other “the”-word.

da djan = John.

de djan = the knower, he who knows.

Before I go any further with this, I have to introduce my number reforms. Since we have to remake the little words anyway, why not fix the numbers up and give them some redundancy? I thought of this when I remembered way back when I was first learning Esperanto, and I was bothered by the pronouns all sounding alike, being mostly in the form Ci. Esperanto numbers have pretty good redundancy. Loglan has the opposite problem, enough redundancy in the pronouns, not enough in the numbers. So I propose the following numbers:

1 - key (pronounced [kay], the Hindustani eyk backwards, sort of)

2 - du (from Esperanto, and basic Indo-European languages generally)

3 - twa (French)

4 - fo (classic Loglan and English)

5 - kwi (Romance languages)

6 - ze (German)

7 - sye (Russian, Spanish)

8 - vo (classic Loglan, Russian)

9 - dye (Russian)

0 - zi (English zero)

00 - zio (zero + another zero, so to speak, two syllables)

000 - zia (no particular reason, again, two syllables)

This gives us plenty of redundancy, except ze and zi might be too much alike. That can be changed. Also, learnability isn’t much hurt, and it’s at least up to Esperanto or Interlingua standards for Europeans. Now back to names:

Da means that the following morpheme is a name. Any morpheme following the name will be considered to be modified by that name.

De da djan fawl = John’s bird

If we want to make the da cover more than one morpheme, we follow da with the appropriate number:

da djan = John

dadu djan fawl = John Byrd

de dadu djan fawl fawl = John Byrd’s bird

de datwa djan kalafawl fawl =

John Blackbird’s bird

Now, here’s the thing. Names coming into Loglan will operate the way I think it happens in Chinese. If the name happens to fit Loglan morphophonemics, like John does, it’s just borrowed and spelled in Loglan, no problem. If it doesn’t, it can either be translated, like Byrd, or cut-and-pasted till it does fit, like say:

George Bush = dafo djordjo ciawtri

Dafo means we have a 4-morpheme name coming up. Djordjo can be viewed as a cut-and-paste or as a total borrowing from Italian. Ciawtri [SHEE- ow-tree] is a translation, small-tree, from Mandarin and English. Again:

John Danforth Quayle =

dakwi djan denforte kweyl

Thusly, all names in Loglan would end up having meaning in Loglan, sometimes deliberate, with translations, sometimes accidental, with borrowings and cut-and-pastes. Sometimes, of course, the latter morpheme sets will not yet have Loglan meanings assigned. But just speculating for fun...

Djor doesn’t mean anything in Mandarin or English, but lo and behold, in Hindustani it means to join or unite. Djo, right off, means several things in Mandarin, of course, and my eye is caught by one definition, state. Is this an omen, or what? Djordjo means united-states!!

So we dispense with all the pauses and rules about ending with a consonant in dealing with names. I think these changes, taken together, would pretty much eliminate the need for any meaningful pauses in Loglan altogether.

Another thing about names...I’ve been wanting to use names in compounds, and maybe this sys-tem will make it possible. I’m really going out on a limb here...

da fran = French(wo)man

dadu franstan = France

Now fran might well have a meaning as a regular primitive, but when it is da fran it is a name, and there may or may not be a relationship between its meaning as a primitive and its meaning as a name. De stan does mean the country. Dadu franstan means France. De da fran stan means the French country, which may or may not be France. Quebec, for example. Thus, we’re distin-guishing a 2-morpheme name from a primitive modified by a name. Similar things happen with vil for city, boli for language, etc.

dadu franboli = French, the language

de da fran boli = the French language

that I’m talking about

though I don’t know if the distinction is useful.

This would give us a lot more leeway. All other words in the language can be used as names. A little word can still be used as a converter of names into predicates, as before. Can’t use me, so supposing it’s fa

de pom = the apple

da pom = the Apple (computer)

fayda pom = is an Apple (computer)

If these proposals are adopted (or others that solve the same problems) it will finally be possible to consider Loglan a potential interlanguage, one that can compete with Esperanto. As it is, there’s not a chance. Except for a few rarified intellectuals (and not all of them), everybody initially turned on to current Loglan will be lost the instant he encounters Esperanto. [This can hardly be true, forso many of our most passionate loglanists are ex-Espernatists! But we leave Rex on that resounding if somewhat dubious note.—Ed.]

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