Monday, June 9, 2014

Lojban is monoparsing

Someone has compiled the following list virtues of Lojban
Lojban is designed to be used by people in communication with each other, and possibly in the future with computers.
Lojban is designed to be neutral between cultures.
Lojban grammar is based on the principles of predicate logic.
Lojban has an unambiguous yet flexible grammar.
Lojban has phonetic spelling, and unambiguously resolves its sounds into words.
Lojban is simple compared to natural languages; it is easy to learn.
Lojban’s 1300 root words can be easily combined to form a vocabulary of millions of words.
Lojban is regular; the rules of the language are without exceptions.
Lojban attempts to remove restrictions on creative and clear thought and communication.
Lojban has a variety of uses, ranging from the creative to the scientific, from the theoretical to the practical.
Lojban has been demonstrated in translation and in original works of prose and poetry.
These are put forward as reasons to learn Lojban.  But, in fact, unless you are looking or a language that is free of syntactic ambiguity, none of these virtues obviously favor Lojban, nor do they collectively, unambiguity aside.  In fact, many constructed languages, well-known or not, have the feature listed here that are not connected with unambiguity in as high a degree as, or higher than, Lojban. For most cases, Esperanto will do as an example.  

Esperanto has been used by people in communication with each other for well over 100 years and has a vast literature in translation as well as in original works, far more of both that Lojban does (and had that much already before it was as old as Lojban is now).  Esperanto is as likely to be available to computers as any natural language, but Lojban has the advantage of monoparsing: that there is only one grammatical reading of any sentence, a clear advantage.  This is why, of course, monoparsing is the essential Lojban virtue.

Esperanto is a SAE language in many ways, also andrist, heterosexist etc. etc., so not neutral in many respects.  Lojban is also SAE in the grammatical sense, but not in vocabulary (as Esperanto is) nor in most other respects.  However, most constructed languages are SAE in this way and many are as neutral as Lojban in the ways Lojban is neutral.  Neutrality means primarily that no cultural categories are compulsory (and, secondarily, that all are about equally permitted).  So, for the usual examples, Lojban (and any number of other constructed languages) do not force gender distinctions grammatically but permit them to be made semantically as wanted and similarly for singular/plural and for tense/aspect (Esperanto does force singular/ plural and tense and makes a clear gender distinction without actually making it compulsory.  It also has a case system, which is somewhat suspect, for some reason.).  This is mostly a matter of vocabulary and an analytic syntax, which many conlangs can provide.  Lojban's deriving its vocabulary from a collection of languages from various parts of the world does little to improve this situation (the claim that this derivation process makes the vocabulary easier to learn across cultures has never been tested but flies in the face of considerable anecdotal evidence at least.  I note that this internationally based vocabulary is not mentioned as a virtue.).  Even toki pona -- not usually an exemplary language for these discussions -- can match all Lojbans advantages here,

The claim that Lojban's grammar is based on symbolic logic is often made but never really demonstrated, A logician looking at anything beyond a very simple Lojban sentence would be hard pressed to see what the connection was and the simple sentences look pretty much like simple sentences in any SAE language (but then, where did logicians from England and France, Germany and Italy get there models, after all?).  The virtue of being "spoken predicate logic" (JCB) is exactly that predicate logic is monoparsing and, if Lojban really is that, however disguised, then it has achieved that goal.  Unfortunately, it is not clear that it is that, since the grammars written for Lojban have not be attached in any formal way to the language of Logic, but tend to be linear.  As tfor the other supposed virtue of being a language based on Logic, that learning it will make you (or help you) be logical, one has to ask whether learning French will make you a better cook or whether logicians as a classs are more logical than other folks (go to an ASL meeting before you answer).

I suppose that Lojban's flexibility is demonstrated by its use, in which case, Esperanto's is even better demonstrated.  Again, toki pona, in about half the time, as a reasonably good demonstration available, as do many other conlangs.  Lojban would be appreciably more flexible and fluid did it not have always to check back that it had not violated the monoparsing rules.  Monoparsing is very expensive in terms of rules and vocabulary, since very minor deviations can either ruin the whole or at least give the wrong results (dropping a right end marker completely changes a sentence, even when the intended -- but now missed -- meaning remains clear).

Phonetic spelling is pretty much a given in conlangs.  It is required for Lojban -- along with a very restrictive set of morphophonemic forms -- for monparsing to work from the bare speech stream, not just from the word level.  The resolution into words is also a part of this process.  Other languages work more or less from the word level, perhaps allowing (even occasionally encouraging) some uncertainty about word boundaries (puns in Lojban work at a different level).

It is a little hard to compare Lojban's grammar with those of other conlangs (let alone natural languages), because Lojban (and Loglan) are virtually unique having actual full-scale grammars in a linguistic (or closely related) sense.  Since the aim is monoparsing, a good deal of Lojban grammar is devoted to ensuring that.  It has been estimated that, were monoparsing not required -- merely,say, as good a job of combating ambiguity as English --  Lojban's grammar would be half its current size.  But that estimate is pulled out of nowhere. The same can be said for estimates of how long it takes to reach what level of competence (however defined) for various languages -- the measure of simplicity that seems to be left.  Anecdotal reports, however, make Lojban much harder than Esperanto or toki pona at both the basic and more advanced levels (indeed, places basic Lojban at near the same level as advanced forms of the others).  Partly this is surely due to the vocabulary issue, since toki pona vocabulary is tiny and Esperanto vocabulary is largely familiar, while Lojban vocabulary is totally new (has its old well hidden).  It may be, of course, that there is a level of Lojban that can be reached in a reasonable time that can be reached in other languages only with great difficulty, but it is not clear that this is so nor what it might be.  No one clims that you can be more profound in Lojban than in Esperanto, I think (though maybe than in toki pona).

Lojban does have a well developed (albeit complex) system for word construction, perhaps better worked out than other languages.  But other languages seem to manage to deal with more notions than are in their basic vocabularies:  Esperanto has a variety of word formation techniques as well as compounding (the main Lojban system) and borrowing (also in Lojban but constrained by the morphophonemics. needed for monoparsing) .  Even toki pona creates new compound phrases freely and allows them attain the status of semantic units.  Esperanto can clearly go as far as Lojban; toki pona can apparently go as far as it needs to -- not so far, but then it starts with only 10% of the vocabulary.  

Exceptionless regular rules of grammar are a norm in conlangs, so that is not Lojban's uniqueness.  On the other hand, monoparsing means that there have to be separate rules for various cases which might not otherwise have been separated.  thus what in one language might be covered by a single rule (folding together the differences in two parallel sentences, for example) might in Lojban take a number of separate rules, depending on the exact character of the pieces involved (NP is different from VP is different from S is different from String is...). Thus, while the rules have no exceptions, there are still a greater number of cases to cover.  

The last three claims are again about what has been done in the language.  As in such cases mentioned earlier (in fact, these are essentially repetitions and look like padding the resume), any claim Lojban has here was met and bettered by Esperanto in the beginning of the 20th century. toki pona has more modest results, of course, but much more modest claims.  Other languages come somewhere in between, largely a matter of the amount of work their users have put into them (and, of course, just the number of users).  But many have the potential to do as well as Lojban, given the effort.

In summary, then, Lojban is nothing special among conlangs, if you ignore monoparsing.  If you are not interested in monoparsing, there are many other langugees which will give you Lojban's other virtues and generally more simply than Lojban does, since monoparsing is expensive in terms of grammar and vocabulary, of easily forgotten little word and very finely differentiated rules.


No comments:

Post a Comment