Friday, August 28, 2015

The Role of Errors in the History of Loglans 4

Maxim 4: Loglans are unambiguous.

Officially, that means that they are free of syntactic ambiguity, that the sentences cannot be analyzed in more than one way. This does not prevent any other sort of ambiguity: ambiguous words or phrases, ambiguous references, and so on. Nor does it prevent vagueness, a lack of precision in the given situation. Much of the history of the vocabulary of the Loglans has, however, been directed against these problems as well.

The basic claim of anamphiboly flows from the fact that FOPL++ is easily shown to be to be such and that the Loglans, as derived from this base, with care, preserve this feature. The immediate problem is that the Loglans were not derived systematically from the base but were created by different processes which, nonetheless kept that base more or less in sight. These processes centered on grammatical theories and presentation that were in several ways different from that employed in FOPL++. So the claim of freedom from syntactic ambiguity rests on the products of these theories, not on FOPL++. There are open (and unexplored) questions whether every grammatical sentence according to these can be automatically traced back to a well-formed formula of FOPL++; whether, if so, the formula is the right one, given the meaning of the sentence; whether every well-formed FOPL++ sentence gives rise to a grammatical Loglan sentence (and the right one); whether there are alternate, equally valid, grammars (from other theories perhaps) that yield different results on Loglan sentences (giving them different structures or even reversing their grammaticality). There are in particular problems, more or less on the line between grammar and semantics, about pronoun reference and the scope of quantifiers, two systems from FOPL++ rejected early in Loglan history. Part of the problem is simply that the transformation of FOPL++ to a Loglan using familiar sort of rules has never been attempted . For now, though, it seems that the Loglans are pretty much ambiguity free according to their own grammars. And this is no small things, since no other languages, natural or constructed, can make that claim or offer that thorough a grammar.

Taking that as settled and despite repeated advice against it, much attention in the Loglans has been directed at other forms of ambiguity and vagueness. So, for example, figurative meanings are to be marked to distinguish them from literal ones and, typically, several level and types of figurative meaning are also distinguished and marked (optionally, of course). But even literal meanings have been refined considerably. Part of this is, of course, just part of getting technical terminology for various fields. But many of the discussions of new words suggest ordinary applications are intended. Some of this is surely fed by L1 preconceptions, which find any other way of parceling out the universe to be inherently wrong (never mind neutrality) and so want to divide broader concepts (“vague”) or combine separate ones, rather than leaving them stand on their own and working with them. (Oddly, nothing seems to have ever been done with {klama}, which covers two verbs in most familiar languages: “come” and “go” in English, for example.) This is in addition to the natural expansion of the vocabulary to meet the needs of the modern world: words for pizza and smart phones and dick pix and whatever else in todays headlines. So, the vocabulary of the Loglans (I admit I really only know about Lojban here) has grown enormously and in a fairly uncontrolled fashion. No authority seems to check words to see if they are actually new or well formed or fit to their intended use and no on seems to guarantee that all words coined get recorded in a central file (of which there seem to be several, run by different groups).

While the expansion of the predicate vocabulary has been large, it has at least some natural roots. The expansion of the particle vocabulary is proportionately even greater and less clearly motivated. The phonological definition of a particle (in Lojban, again) has been expanded several times to meet the growing number of new forms suggested. While many of these new forms are cases of preciding old forms and so make no new grammatical categories, many others seem to be new grammatical functions (not always terribly clear) and thus to require new grammatical rules. And that requires another look at the whole grammar to see whether it still coheres. It is not clear that this has been done for several years. The overall effect has been to complicate a once fairly straightforward grammar into something quite Rococo with relatively little motivation other than someone's “brilliant idea” and a plausible appeal to some language somewhere maybe doing it on some level.

Since the grammar is both complex and minutiae-laden, there is thus the practical question whether a speaker can be sure that what they say is grammatical and what they meant to say. The omission or misplacement of tiny particles can materially change the whole structure (and so, presumably, meaning) of a sentence. Loglans generally have low redundancy, so mistakes of this sort are not as readily spotted as they might be in natural languages, which offer added clues, nor is backtracking allowed.

But even in the area of fixed items, there has been considerable uncertainty and change and lack of settlement. To take the longest running case, {lo} was inherited by Lojban from Loglan with a very unclear – indeed, diverse and contradictory – definition. It is a term maker, a mate to {le}, itself a somewhat problematic version of the logical definite descriptor (the unique object with the property if there is exactly one but otherwise – the usual case – the thing the speaker has in mind and calls by this name, regardless of whether it has the property). The one sure thing about {lo} is that {lo brida} is actually a brida – in some sense. After a long period of dispute about the implicit quantifiers with the descriptors which decided there were none, the question turned to what the expressions actually referred to: the obvious things with the property (one or several – which creates its own problems), the mass of such things, the archetype of such things, and so on, usually trying to reproduce the superficial expressions of, say, Trobriand Islanders or the Piriha~. Eventually, a reasonable solution appeared to have been found in the notion of plural reference (or Lesniewskian sets) and of saliency. However, even after the dust settled, an informal survey suggests that now two Lojbanists has the same idea about what the official definition is and almost none have it nearly right. Similar, though less virulent, discussions have clouded the issue of quantifiers, resulting in a variety of basic ones in place of the original two. In the case of anaphoric pronouns, there are several systems, formal and informal, but no guarantee that they are collectively adequate, short of actual repetition or overt assignments (not a part of ordinary language). Some of the systems are also impractical to apply, requiring remembering the exact grammatical role a word played in an earlier sentence, for example. And failure here does reflect back on syntactic ambiguity, since, if we cannot be sure whether what goes in one place is the same as in another, we have lost the logical form.