(From Lognet 90/1. Used with the permission of The Loglan Institute, Inc.)

Lo Lerci (Letters)

Letters policy: Unless otherwise stated, letters sent to the editor 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 purely for the sake of clarity. —Ed. 

Dear Editor:

RAM’s More Declensions for Loglan [LN89/1:6-8] [presented] an intriguing, original idea. At first reading, I had a negative reaction, but could not articulate my objection easily. Now that I’ve given it some thought, I believe I can. I thank RAM for helping me "unpack" some hidden as-sumptions I was nursing about how Loglan should be, and crystallizing for me what I like best about Loglan.

Adding declensions seems to increase the information density of the Loglan predicate, but at the expense of the ability of Loglan to express ideas. This troubles me. I can see how increasing information content is an advantage, but I much prefer preserving Loglan’s unique ability to express a wide range of ideas.

I think declensions would hurt Loglan in this way because they would reduce the analyticity of the language. What do I mean by this? My objection might be better expressed in terms of some descriptive linguistics. (My linguistics was limited to one semester in college, so I apologize in advance for any mistakes I might make due to my amateur status.) I understand that natural (and for that matter constructed) languages can be grouped according to their degree of inflection. Inflections are changes in word form used to express concepts such as case, number, person, mood, voice, tense, etc. Declensions are traditionally inflections for nouns, pronouns, and adjectives. (RAM uses declension to mean inflection of primitive predicates.) On one hand are languages such as Latin or Turkish or Volapük, which are very highly inflected (in the case of Turkish, I think the term is agglutinative, but it’s nearly the same). The relationships between the words in these languages are expressed in inflections such as  infixes, prefixes, and suffixes. On the other hand are languages such as English, Chinese and Loglan, where individual words are atomistic and largely uninflected. These languages are called analytic. Relationships between words in these languages are expressed in the word order [and by particles such as articles and prepositions which tell you what's going on—JCB]. It’s possible to say a lot more in a given space in a highly-inflected language like Latin. But the flexibility in word order in analytic languages means that more can be said, at least if you accept the Sapir-Whorf hypothesis. [The editor butts in to insert that Wes probably doesn’t mean flexibility here, but rather the fact that analytical languages tend to have morphemic word order. Right, Wes?—Ed] It may be my Western, English bias, but it seems one reason English has emerged as a world language, especially in scientific and technical areas, is its analytic nature. In any event, I have a strong prejudice in favor of analyticity.  I’d prefer not to see Loglan have any declensions. I don’t want a Loglan that’s even a little bit like Latin or Esperanto (which has noun suffixes galore and even inflects the direct object). I suggest Loglan can do without the compactness of information provided by declensions. As JCB has observed, it often takes more words to say something in Loglan than in English, but that’s because Loglan offers a fuller description of the utterance. And I would add also because of the greater degree of analyticity in Loglan.

Consider computer languages. Loglan, having a fully-specified formal grammar, is much more like a computer language than any natural language. I think this is a good thing. It’s quite an achievement to create a syntactically unambiguous human language, a feature Loglan shares with most computer languages

Now, computer languages rarely allow declensions. A few declension-like examples come to mind: in Lisp, cdr, cddr, cadr, cdar, cddar, etc. In Pascal, write and writeln, and read and readln. These examples are just barely declensions. I think that the great need for non-ambiguity and analyticity in computer languages is the reason they are uninflected. 

I doubt there is any conscious conspiracy among authors of computer languages to eschew inflections. It’s certainly possible to write a compiler for an inflected computer language. But it’s not done because it would inhibit analyticity. By the same logic, we should keep inflections out of Logan.

It appears that using word order to express ideas will always allow a greater range of expres-sion  than  declensions.  Take the current declension of hinda, hindi, hindo, which refer to Indian or Hindu language, people, and culture. The a, i, and o represent the triad of language, people, and culture while the root hind-  expresses something like ‘Indianness'. I don’t like even the language/people/culture declension, minimal as it is. Proceeding analytically, a language planner should create a primitive predicate such as hindu, which would have the same meaning as the [present] root hind-, and then add another predicate to represent language, people, or culture. Thus, hindu lengu would represent Indian language, Hindi; hindu piplo, Indian people; and hindu kultu, Indian culture. Instead, Loglan has the declined root hind- plus vowel for this concept. Now, hind+a/i/o is a lot more compact than hindu lengu, etc., but I still like hindu lengu better because it is analytic. I can immediately see the atomic predicates that make it up. If necessary, I can use the combining affixes of the primitives to create a new Loglan predicate.  [For example,  hinleu.—JCB] Moreover, I can create other predicates, such as, say, slove, for Slovenianness, plus bivdu for behavior or manner. From this comes slove bivdu [or slovybiu.—JCB], for [behaving in] the Slovenian manner, or hindu bivdu for the Indian manner. 

Using predicates, I can put concepts together using the very powerful Loglan rules governing predicate modification. With declensions, I am limited to the particular concepts expressed by the declension. I can’t say ‘Indian manner’ using the declension system, but I can using traditional Loglan grammar. Even if an ending  were created for ‘manner,’ there would inevitably be other concepts that declension would not allow.

In sum, it appears to me that any linguistic relationship could be handled by either combining predicates or by using declensions, and that predicate-combination will always be more powerful.

Now that I’ve harped on how analytic languages like Loglan allow da better to express daself, an inflectionist might point out, “But these inflections actually increase the number of possible productions in Loglan by increasing the word pool. There’s more that can be said.” Not really. The inflections offered by RAM duplicate productions that can be achieved, at greater length, through ordinary predicate expressions. The real harm of allowing declensions is when they  replace analytic predicate-combinations, which RAM doesn’t suggest. Still, I don’t like the declensions suggested because they taint Loglan with non-analytical rules. Loglan is the most analytic human language in existence and should stay that way.

Let’s keep Loglan analytic. Hinda/hindi/hindo and the related "ethnic" declensions are harmless and seem to arise more out of the sparsity of [properly discriminated source words] for a language, people, and culture than out of any commitment to declensions. If we absolutely have to express these concepts in one word, let’s do it through combining predicate affixes, Loglan’s standard procedure for these situations. For now, I suggest discarding the ethnic declensions and rejecting any other declensions. Declensions undercut Loglan’s great advantage as a language —its analyticity—and aren’t needed.

Wesley R. Parsons

Dear Wes:

Perhaps my error was in selecting the word 'declension' as a metaphor, even though I was aware that it was not technically a declension.  The main object of the proposal is to ease learning, add redundance by relating the meaning to the word form, and to provide shorter forms to counteract to a certain degree the tendency of natural language speakers to contract longer words (e.g., veggie).  Since it does not change in any respect the grammar of the language, it does not affect its analyticity, or what can be said, any more than the [the existence of the] word cow in English prevents one from saying female bovine animal if one so desires.

By increasing the number of ways in which an idea can be expressed, since traditional complexes would still be valid, there would be more scope for poetic expression, developing subtleties of meaning, and additional opportunities for observing Whorfian effects.  With regard to the "ethnic declension”, one can, should one desire, say hindybivdu, hindypernu, hindylengu, hin-dylilfa, or if hinde were added as the general adjective, hinde bivdu, hinde pernu, hinde lengu in addition to the shorter forms.  Likewise, one can [already] say any one of cmalo hasfa, cmahasfa, cmalyhaa for euphony or extra redundance, even though the canonical form is cmahaa.           Robert A. McIvor

Editor kibitzing again:  First off, I agree that picking the word ‘declension’ led to a lot of this.  Declensions are grammatical.  This problem isn’t grammatical. It does have to do with inflection. Inflection can take place on the grammatical level, as in Russian, Latin, etc., and also on the word-building level, as in English (‘photograph’, ‘photography’, ‘photographer’, ‘photographic’ are on the word-building level, while ‘photographing’ and ‘photographed’ are grammatically inflected). Or in both, as in Esperanto. If we restrict the meaning of ‘analytical’ to the grammatical level, English is very close, and Mandarin and Loglan make it 100%. Do we want Loglan to lack inflections on the word-building level?Wes says yes. But  one of Esperanto’s most attractive fea-tures is its ability to so build words from a store of roots. This is a feature that appeared earlier in Volapük, and was a large factor in its success before Esperanto supplanted it. Esperanto 'mortig’ means kill, from 'mort’ (dead) and 'ig’ (cause). It’s similar to mormao which is built the same way and means nearly the same thing. So if you consider this phenomenon to be an instance of inflection in Esperanto, you have to consider it also to be inflection in Loglan, which would be very interesting.  The difference between Loglan and Esperanto here is on another level: whereas '-ig' is a suffix primarily (though it can also act as a root, as can many Esperanto affixes), mao is derived from a full-fledged primitive, madzo, as are all Loglan affixes (with the exception of the handful derived from little words). So here is the real problem. Allowing that final vowel to vary to make new words is a seeming departure from the otherwise pristine dichotomy in Loglan between little words and predicates. We seem to have a new animal: a morphemic final vowel...unless we go at it (as RAM suggests) simply as a learning tool, a mnemonic. Just as the fact that almost all the "the-ish" words in Loglan begin with l-  assists us to learn them, when we're trying to learn an ethnic word, we are helped by the fact that if it ends in -a, it means that kind of language, if in

-i, then that kind of person, and so on. So these final vowels need have no more morphemic status in the language than that initial l- does. So let's keep talking about extending the psuedo-declension system until we decide whether it works or not, but let's not interfere with its current use in the ethnic realm, where it is indeed harmless, and rather handy.—Ed.