Copyright, 1988, 1991, by the Logical Language Group, Inc. 2904 Beau Lane, Fairfax VA 22031-1303 USA Phone (703) 385-0273 lojbab@lojban.org All rights reserved. Permission to copy granted subject to your verification that this is the latest version of this document, that your distribution be for the promotion of Lojban, that there is no charge for the product, and that this copyright notice is included intact in the copy. The following article, by logic and philosophy professor Todd Moody, may be useful to those who want to know how Lojban compares with other artificial languages. ___________________________________________________ Lojban in Perspective by Todd Moody Lojban, whatever else it may be, is a fascinating project. It represents at once some old and new ideas within what Andrew Large calls the "artificial language movement," in his book of the same name (Basil Blackwell: 1987). For example, it partakes of both a priori and a posteriori characteristics. In the jargon of artificial, or planned, languages, an a priori language is one that is constructed around some central set of philosophical principles, instead of being modeled on existing languages. An a posteriori language, on the contrary, takes one or more existing languages as its model, and aims to "improve" upon them in certain ways. Needless to say, no planned language can be entirely a priori a posteriori. Even an a priori language, for example, must look to existing languages to discover the actual functions of language. And each "improvement" found in an a posteriori language is likely to be based upon certain philosophical principles. Still, these terms are useful, if not precise. Esperanto, for example, is clearly an a posteriori language, in that almost all of its lexical material is drawn from existing languages, mainly French, German and English. Its grammar, though simplified and regularized to a great extent, closely resembles the grammars of other European languages. Zamenhof did not hesitate to refer to the familiar "parts of speech" in his early writings on Esperanto. He was evidently unaware that these concepts do not always have clear counterparts in non-Indo- European languages. Zamenhof's naivete may be striking, but it should be remembered that although he wanted Esperanto to be a neutral language, the sort of neutrality that he wanted was political, not linguistic. Esperanto is not the cultural property of any nation, not a "national" language at all. Many critics of Esperanto, however, have charged that Esperanto is not truly politically neutral because of the extent to which it is clearly not linguistically neutral. It may not be a national language, but it is surely a European language, they complain. This charge obtains its force from the undeniable fact that there is no clear line of demarcation between linguistic non-neutrality and political non- neutrality. I happen to think that Esperanto can, to some extent, be defended against these charges, but this is clearly not the place to do it. The purpose of these observations is to point out the built-in risks to any a posteriori language. There have been numerous a priori planned languages proposed. These have generally involved some classificatory schematism which would allow the names for things to be derived from the metaphysical category in which the things belong. A recent attempt is the language "aUI," which I would venture to guess is unknown to most who will read this. aUI is interesting enough to warrant a brief description. The language aUI was created by W. John Weilgart, late of Luther College in Decorah, Iowa. Weilgart held doctorates in psychology and philology, from Munich and Heidelberg, respectively. Originally, aUI was intended to be used as an aid in psychotherapy. The idea was to ask people to represent their difficulties in this language, forcing them to reconceptualize and, ultimately, to overcome them. aUI begins with an alphabet of some 31 symbols. These symbols are in fact ideograms, but Weilgart also provided a system for using the Roman alphabet. The main thing is that each letter of the aUI alphabet represents a basic aUI concept. Thus, aUI words are spelled according to what they mean. Alternatively put, each phoneme is a morpheme. Lower-case 'a' (pronounced as in "mama") means "space." Upper-case 'A' (a similar sound, but longer, like the 'A' in "fAther") means "time." The letter 'U' as in "lUnar") means "mind," and 'I' (as in "machIne") means "sound." Thus, the word "aUI" itself means "space-mind-sound." Well, "mind-sound" is an aUI word for "language" so, more idiomatically, "aUI" means "language of space." Weilgart saw aUI as a kind of language of the cosmos. And so it goes. Clearly, the lexical material of aUI will bear no non- coincidental resemblance to any other language. Translated into Esperanto, "aUI" is "spaclingvo." This kind of a priori language is only as good as its classificatory scheme, obviously. If you can't classify things properly, you can't derive words for them. I can't comment on the success or failure of aUI in this regard, although Weilgart did produce a dictionary of several thousand aUI words. Others may judge how well he succeeded in his project. It is fair to say, however, that philosophers have mostly abandoned hope of devising an exhaustive classificatory schema for everything there is. We know better than to expect that all people will find any given system "natural" or complete, or even adequate. We certainly know better than to suppose that classification is an activity that is somehow above or beyond cultural arbitraries. This brings me, finally, to lojban. Where does it fit into this taxonomy of planned languages? My answer, based as it is upon a very rudimentary grasp of the language's central ideas, is that lojban leans toward the a priori end of the spectrum. First, consider the lexical material. It is produced by algorithm, using the most widely spoken existing languages as a source. This, on the face of it, is an a posteriori move. Still, the lojban gismu are essentially composites, not copies. Classically, a posteriori languages such as Esperanto attempt to find the most "international" lexical material and copy it into the language, adapting it to the new orthography. Esperanto borrows words more or less intact; lojban borrows them in order to change them substantially. I have no data on the recognizability of lojban words for various language populations, but my guess is that in no population would it be very high. By subjecting lexical material to its statistical and morphological constraints, lojban puts-- to borrow a politically fashionable word these days--a strong a priori "spin" on a basically a posteriori approach. Where lojban shows its true a priori stripes, though, is in its syntax and semantics. Although I have not in any sense mastered these, what is clear from the outset is that lojban is modeled upon the predicate calculus, or quantificational logic. Predicate logic, suitably enhanced, is the scaffold upon which the language is built. I say "suitably enhanced" because predicate logic alone is clearly not going to do the job, not even with a large dictionary of predicates and arguments. To take a trivial example, the English words "and" and "but" differ in meaning, even though they are both used as truth-functional conjunctions (and "even though" is another). Sentences that differ only in these words will have the same truth conditions, but different meanings. Simple predicate logic makes no distinction between them. As far as it is concerned, they are equivalent. Indeed, it would be fair to say that predicate logic was developed as a tool to help analyze and keep track of certain relations between propositions, and so to screen out other factors. That is, predicate logic is useful precisely because it is not a complete language, but only the abstraction of certain aspects of one. I have taught courses in formal logic more than once, and I have been guilty of referring to logic as the "skeleton" of language, as if it were somehow encased in language, waiting until the twentieth century to be extracted. It's a pretty thought, but quite silly. It suggests that logic is somehow more important than other aspects of language, that it somehow holds everything together. As far as I can see, such thoughts have no basis. We need to consider what language is for, and what we do with it. The adequacy of any given planned language must be judged on how well it does what a language is supposed to do, rather than on how well it reflects our favorite aspect of language. If lojban is in fact a "logical language," that is in itself neither a virtue nor a vice. We use language to perform speech acts. Alternatively, speech acts are the things that we do by means of language. Some speech acts have English names. Examples of speech acts are: insulting, promising, lying, cajoling, threatening, assenting, demurring, slighting, and so forth; the list is long. The successful performance of speech acts requires a certain amount of standardization, so that the listener can recognize, or make a good guess at, the speaker's intentions. Speech acts are, to some extent, culture-bound. There is no reason to expect a one-to-one correspondence between speech-acts across languages. In plainer English, different populations may use language to do different things. So, specific languages are embedded in cultures, and the features of languages will naturally be adapted to the efficient achievement of the speech acts that are important. To take an example, the Japanese language is very simple in certain respects, but it has a rather elaborate and complex system for representing levels of politeness and social standing. It seems fair to say that this system makes possible a number of speech acts that could be accomplished in English only with a certain amount of difficulty and awkwardness, if at all. It is tempting to conclude, therefore that an artificial language--any artificial language--simply misses the point, because there is no super-cultural set of speech acts for such a language to hook onto. This, I think, is a mistake, for two reasons: First, it seems likely to me that there is a set of speech acts approximately common to all languages, even though it may be a subset of the speech acts available in any given language. Second, this conclusion overlooks the considerable ability of people to improvise speech- acts. I believe that language use inevitably involves a substantial amount of improvisation. A useful international language, therefore, will be flexible enough to support this improvisation. It will give people the room to do the things that they need to do. If the language forces people to focus too much of their attention on other details, they will find it difficult and tiresome to use the language. To a certain extent, we all go through this phase when we learn any second (or third) language. We find that our attention is caught up in grammatical details and that, until we have "mastered" them, using the language is exhausting. So, an optimum international language will offer the resources for performing speech acts while minimizing details that are not particularly relevant to them. I simply do not know how lojban would measure up, according to these criteria. It is possible that lojban's isomorphism to predicate logic may, in the end, be little more than a distraction and a nuisance. As a teacher of the subject, I know that formal logic does not come "naturally" to many people, and the reason for this may simply be that it is for the most part extraneous to the successful performance of speech acts. The members of the Logical Language Group need to ask themselves why it is important, or desirable, for a language to be "logical," in the sense in which lojban purports to be a logical language. If the primary purpose of lojban is to serve as a neutral international language, then this fact must figure prominently in any attempt to answer this question. The linguist Otto Jespersen was much involved in the artificial language movement. He was familiar with Esperanto, and later became president of the organization that formed to support Ido, the variant of Esperanto that was created under circumstances that are sometimes described as "schismatic." Even later, he was the author of yet another artificial language project, called "Novial." He offered the following utilitarian principle for assessing planned languages: The best language is the one that offers the greatest ease of learning to the greatest number of people. On the basis of this principle, for example, he criticized Esperanto's compulsory "agreement" of adjectives and nouns. It make sense to ask whether lojban's logical character--the fact of its being modeled on predicate logic--conduces to the greatest ease of learning for the greatest number. This is an empirical question, and I do not know the answer. My guess, based upon my teaching experience, is that lojban's logical structure will make it harder, not easier, for most people to learn. But that is only a guess. So, if we accept Jespersen's rule for assessing planned languages, the first thing to find out is whether lojban deviates in principle from it. As for the issue of lojban's neutrality, Ralph Dumain has already made the point that predicate logic is not, in any real sense, "culturally neutral." On the contrary, it is the highly sophisticated artifact of the scientific culture that originated within European culture. Predicate logic was developed in the nineteenth and early twentieth centuries as an accompaniment to the positivistic worldview that was also achieving ascendancy at the time. That is, there was a distinct cultural need to find a formalization fo the language of science and mathematics and to provide a basis for the unification of these disciplines. Principia Mathematica was not written just for the hell of it. Naturally, Esperanto is equally susceptible to these criticisms based on neutrality. It may turn out that neutrality is a red herring. I want, therefore, to register these notes of (I hope) constructive skepticism about lojban as an international language. I still find it a fascinating project, as I stated at the outset, and I am very interested in seeing where it goes. ___________________________________________________ Response by Bob LeChevalier Lojban is not necessarily intended to be an international language, though it is an 'artificial' one. Todd occasionally uses the two concepts such that they seem interchangeable, which they aren't. The 'logical' of Lojban is not especially relevant to Lojban's promise as an international language. I have rarely used the logical connectives when I do speak the language; this is partially due to the fact that I was a dunce in logic class (my only 'D' in college, primarily because I couldn't get into doing the homework), and partially because I don't yet know the cmavo that well. When I take time to look them up, I do fine. But logical connection is apparently not necessarily a part of conversation. The first significant Whorfian effect of a logical language might be that people don't imply a logical connection unless they really want one. The other part of the 'logical', the predicate grammar, is simply different, not necessarily hard to learn. I don't know of anyone who has had real trouble with the concept given a little explanation. In addition to its grammar, Lojban is definitely a priori in its words, given Todd's definition. We presume that everything can be covered as compounds of the classification scheme implied by the gismu. The difference with Lojban is that we don't claim that the classification scheme is optimal. There is redundancy in the gismu; there may even be a few gaps that will eventually have to be filled in - though not many. We haven't, though, tried to impose a system on the universe like most a priori languages have. Instead, we have tried to broaden gismu flexibility so that multiple approaches to classifying the universe are possible. Our rule is that any word have one meaning, not that any meaning have one word. There is no 'proper' classification scheme in Lojban. The use of existing languages is not an attempt to make Lojban seem a posteriori. It is part of our solution to the 'ease of learning' problem. The rest of the solution is Lojban's ability to accommodate the structures of other languages. For example, English speakers can use Lojban's SOV and AN order easily. French speakers can use NA if they choose. Speakers of languages with other orders are also able to easily use the language, though their utterances might be somewhat different than those of native English speakers. They will still be understandable to any Lojban speaker. Esperanto (and English) speakers must phrase their sentences in particular ways which are more comfortable for Europeans than for non-Europeans. That Lojban is presumably no harder for a Chinese speaker to learn than for an English speaker (possibly easier - they are used to the concept of tanru, and Lojban's nominal method of asking questions resembles the Chinese) is Lojban's major reason to aspire to be an international language. Todd's comments on non-truth-functional conjunctions caused us to rethink the matter. These have been carried in prior versions of the language by modal operators, causal operators, and discursives. Based on Todd's insights and some of Jim Carter's concepts, Athelstan and I worked up a more systematic approach to discursives, and I was able to tackle the modals and causals myself. The machine grammar work helped simplify things further. The result is that all of these can be used as connectives, but are optional. You can use bare predicate logic, or more colorful connections. It will just be clear which you are doing at any given time. Incidentally, one of the few things we haven't yet tried to accommodate are the Japanese social forms Todd mentions. This is example of a possible use for the 'experimental cmavo' I mentioned above. These cmavo can also be used for unofficial 'local' purposes, as long as it is recognized that they are unofficial and optional. If adopted, they will remain optional. This is an example of a change that might show up after the five-year baseline. I think I will leave it to an eventual Japanese Lojbanist to develop the approach, though. You are correct that Lojban is not neutral with regard to scientism, as I said. We have to be aware of the biases we build into the language in designing and testing it.