[YG Conlang Archives] > [engelang group] > messages [Date Index] [Thread Index] >
Mike: > > Regarding Mike's SEMANTIC PRECISION goal, I found his > characterization > > a bit unclear. It could be construed as any of these: > > (i) Avoidance of semantic ambiguity, so that every sentence has > exactly > > one encoded meaning, even if it is a vague one. > > This seems to refer to avoidance of homonyms, which I feel > is more related to either MSS/syntactical ambiguity or to > logicality as you suggested. Kaliputra has described the > essential differences between precision and non-ambiguity. > I have been endeavoring to use these terms, well, precisely. I agree that avoidance of homonyms would be avoidance of syntactic ambiguity. I was thinking more of things like "the singing of X", where X may be singer or song. > > (ii) Avoidance of vagueness. > > Well of course this goal as stated is simply quite impossible. > What I had in mind was selecting the degree of vagueness, > or, if you will, the minimum level of precision. Partially, > I think, the designer will have an impact in this area when > the lexicon is first assembled--the more primitives with > finer distinctions, the more precision. For example, we > could choose eleven primitives for color terms or only three. Fair enough. But avoidance of vagueness should be distinct from precision, given that you could give the lg the capability both of being vague and being precise -- which is the ideal, IMO. So, to take your colour ex, besides having 11 basic terms, one could also have 3, which carve up colour space into more general areas (warm colour, cool colour, etc.) > However, what I really had in mind were specific situations > regarding syntax. My idea is that if a modifier is connected > to a head, and the relationship is not only crystal clear > semantically (a case role for example) but also the most > frequent construction form, then it makes sense to prescribe > this minimal level of precision for the briefest/simplest > construction. Another construction can and should be made > available in the event that the default case is too preceise, > but this should be the positively marked construction, > insofar as it is the less frequent construction. Okay, but this is too mired in specifics to be elevated to a general principle. And if a putative modifier-head construction always expressed a certain case-role relation, why should it be called a modifier- head construction rather than an argument-predicate one? > Mulling my comments about the lexicon, perhaps I should have > called this category "Semantic Precision & Expressiveness". Would that be different from what I called Compendiousness -- having extra words to express rare or highly nuanced concepts that could not easily be as precisely expressed periphrastically? > > (iii) Making it possible to express very precise meanings. > > I also had this in mind. But as I just described, I think > the language can not only make possible but actually encourage, > to some extent by offering conveniences in the way of > simplicity and brevity (when it makes good statistical sense > to do so), and primitive lexicon. Fair point, and well-made. John Woldemar Cowan is fond of saying "the price of infinite precision is infinite verbosity", which is true, but a lg can definitely strive to make it possible to buy a certain amount of concision without having to pay the price of an equivalent amount in vagueness. > > VI. Non-goals. > > LEARNABILITY. > > NATURALNESS. > > GLOBAL/FUNCTIONAL SIMPLICITY. > > > > --And. > > A very interesting and revealing list. > > What surprises me most about your list is the placement of > naturalness in the non-goal slot. Although I have no > intention of ever offering my loglang as an IAL (I'm afraid > I don't really have much interest in the "language problem" > --better called, IMO, the "language inconvenience"), I did > want to build-in a goodly amount of IAL character simply to > make the language more widely accessible in case I ever > achieve a baseline and decide to publish it. For me, part of the interest of engelanging is to ignore Naturalness and then see how (un)natural the results turn out to be. (E.g. Livagian to my eyes has been getting spontaneously/emergently more natural in recent phases of development, as the basic structures have been augmented by others to add concision and flexibility.) Even if you don't exclude Naturalness as a goal, then if the other goals take priority, I don't think there's much room left for Naturalness. So for Naturalness to be a meaningful goal, it has to outrank some of the others. > Regardless, as uncompromisable logicality is also a design > goal for my conlang (although I may ultimately be tempted to > introduce clunkiness in order to achieve minimal naturalness) > I have to say my curiosity in your creation could not be more > piqued, and I would be very grateful to see how you have > implemented uncompromising logicality as I myself continue > to study the problems of quantifiers, natlang syntax, etc. Once I have other professional commitments out of the way by the end of July, I hope to be able to go some way to satisfying your curiosity. However, as I've said elsewhere, uncompromising logicality is not so hard to achieve if that is your starting point in the design process. Just take a logical formula, put it in Polish notation, lexicalize it, and hey presto you have something syntactically and logically unambiguous. The challenge is to make the result flexible, concise and user-friendly. --And.