Return-Path: Message-Id: <9201161924.AA18627@cunixf.cc.columbia.edu> Date: Thu, 16 Jan 1992 20:02:25 +1100 Reply-To: cbmvax!uunet!MULLIAN.EE.MU.OZ.AU!cuvmb.cc.columbia.edu!nsn Sender: Lojban list From: cbmvax!uunet!MULLIAN.EE.MU.OZ.AU!cuvmb.cc.columbia.edu!nsn Subject: dikyjvo essay. X-To: lojban@cuvmb.cc.columbia.edu X-Cc: nsn@ee.mu.oz.au To: John Cowan , Eric Raymond , Eric Tiedemann Status: RO X-From-Space-Date: Thu Jan 16 15:29:29 1992 X-From-Space-Address: cbmvax!uunet!CUVMA.BITNET!LOJBAN The following is intended for inclusion in the soon-to-be-published proto-dictionary. Comments requested, particularly from Carter, Cowan and LeChevalier. Lujvo place structure guidelines ("dikyjvo") In constructing lujvo without the aid of a dictionary (something likely to occur for a long time), the novice is likely to be at a loss in coming up with a place structure. In doing so, these four patterns may be considered: For, say, the lujvo {sondji}, the relation between the two component gismu, {sonci} and {djica}, may be: {je}: ko'a sondji .iva'inai ko'a sonci je djica. Since tanru modification is often restrictive, this relation can also be denoted by {poi} .i co'e le sondji .iva'inai co'e le djica poi sonci. *Usually* the place structure will be that of the last gismu. Places of the other gismu can be incorporated at will, but preferably after those of the last gismu. eg. xaupre: le prenu noi xamgu. cambarda: carmi je barda (but see below). badzda: big house: le zdani poi barda {be}: ko'a sondji .iva'inai ko'a djica be lo sonci. The place of the modiicatum (rightmost gismu, herein cmugi'u - le gismu poi jicmu [le tanru]) given by the modificant (other gismu - herein gafygi'u) may be omitted in the place structure, which will usually be that of the cmugi'u, possibly with the places of the gafygi'u appended. The place of the cmugi'u described by the gafygi'u will usually be the 2nd, but the 3rd is feasible. Obviously the 1st is not. For example, in omitting the gafygi'u place in zdazba, we are left with mi zdazba lo mudri .ita'unai mi zbasu lo zdani lo mudri. In retaining the place, we'd have mi zdazba lemi zdani lo mudri .ita'unai mi zbasu be lo zdani fe lemi zdani lo mudri .iva'inai mi zbasu lo zdani kujo'u lemi zdani lo mudri .iva'inai mi zbasu lemi zdani no'u lo zdani lo mudri. Obviously, this information is redundant, and {zbasu} is adequate as distinct from {zdazba} in the latter sentence. But these decisions must be made pragmatically. eg. lo cipyzda cu zdani loi cipni .i lo pregri cu girzu loi prenu .i lo xlictu cu ctuca loi nixli .i ko'a xlictu lenu karce litru - a Girls' school teacher. i ko'a ctuca lei nixli pe leva ckule lenu karce litru {belenu}. Properly a special case of {be}-lujvo, this is the most productive pattern. Its use is already widespread with {rinka}. "I" cause that the cloth be dry of milk. itu'a mi rinka lenu le bukpu cu sudga loi ladru .ita'u tu'a mi sudri'a le bukpu le ladru "I" dry the cloth of milk. The gafygi'u is the selbri of a {nu} clause which is an argument of the cmugi'u; the arguments of this selbri replace the clause in the list of lujvo sumti. Later arguments of the cmugi'u stay in place: [the following assumes an unclefted place structure for simlu: event x1 seems to be true to x2 under conditions x3] It seems to me that lojbab is identical to LeChevalier. i lenu la lojbab. cu mintu la lecevali,er. cu simlu mi .ita'u la lojbab cu mi'umlu la lecevalier. mi (This lujvo, incidentally, has been proposed to take over the current [1990] place structure of simlu). The {nu} clause can be the x2 of the cmugi'u (sudri'a), the x1 (mi'umlu), or the x3: I used the rings to climb the mountain .i mi pilno lei djine lenu [mi] cpare le cmana .ita'u mi parpi'o lei jdine le cmana. As the last lujvo shows, repeated (replicated) sumti can be omitted from the place structure; for cases other than rinka, this is in fact the norm. Thus under this interpretation, sondji becomes djica be lenu sonci .iki'anai da djica lenu da sonci: a would-be soldier. Examples of such lujvo construction abound. {fu'i}. Dump the arguments of the ga'irgi'u after those of the cmugi'u. Useful when none of the above analyses seem to work, but can lead to excessively long and counterintuitive place structures. eg jdaselsku: le selsku cu jdaselsku le cusku le te cusku le ve cusku le lijda le se lijda. Note that the place structure given for this word in JL7 is (at least by these rules) erroneous: the x2 of jdaselsku should be the x2 of selsku, which is {le cusku}, and not, as in JL7, {le te cusku (to no'u la jegvon. toi)}. The structure given in JL7 violates one of the two presupposed bases of these guidelines: that the places of the lujvo be a subset of the combined places of the component gismu (gi'urpa'u), and that the order of sumti in the gi'urpa'u not be altered in the lujvo (as in JL7, where x1 x2 x3 is turned into x1 x3 (x2?). It is interesting to note that Jim Carter proposed a similar inversion in {batykla}. I disagree. Obviously much caution must be exercised in applying these guidelines. Omission of sumti in the lujvo which results occurs freely when these are redundant - but this is a matter of pragmatics and semantics, not syntax (which these guidelines, being transformational, can be said to be). Additionally, these guidelines do not carry the official sanction of the LGG, and have not hitherto become widely known. Specific problems. * The fact that there are four basic structures, subdividing into at least a dozen instantiations, means that, even with the application of pragmatics, and possibly future gismu classifications (see below), a combinatorial explosion of interpretations results for lujvo with more than two components. The fact that scoping cmavo (ke, ke'e, bo, je) will usually be omitted in frequent such lujvo, for Zipfean reasons, will further complicate matters. * That these rules are primarily a syntactic, rather than semantic device, and thus do not render obsolete the lexicographer's work, is shown by these two points: a) Often in lujvo construction, the cmugi'u will be the modificatum semantically, but not most conveniently syntactically. Thus the afore- mentioned mi'umlu will often be seen as mlumi'u, which maps the semantics more nicely ("seemingly-same"), but which has no transformational inter- pretation by the above guidelines which is as convenient as that for mi'umlu. This is yet another demonstration of the fact that that which is transformed from (the "deep structure") is not necessarily an accurate representation of mind processes. It may become convenient in the future to incorporate in such guidelines a transformation inverting the lujvo tanru in such cases, but this depends greatly on evolution. Analogous gismu to {simlu} in this respect are, I suggest, mutce, carmi (cambarda (= badycai: carmi be lenu barda) - but note Helsem's use of -cai final), mabla, simxu, cumki, suksa etc. b) LeChevalier complained of the ambiguity of zasyspo, proposing the alternative deep structures "destroy with temporary effect" {le za'i da spofu ja'e lenu ba'o se daspo cu zasni} and "temporarily perform acts of destruction" {lezu'o da ca'o daspo cu zasni}. These guidelines are powerless to choose between {za'i} and {zu'o}. All they can do is suggest a place structure, and they cannot quite do that, because as can be seen from the deep structures, the "syntactic" lujvo is spozasni (cf. mi'umlu for place structure). This does not mean that zasyspo is in any way to be condemned; it is, after all, much more natural - in that it is conceptually simpler. It does mean that caution is to be exercised. (It is interesting to ask whether, as a result of the publication of this essay, more "syntactic" lujvo will become more common. An informal analogy from Esperanto's history suggests that it will, but also that entrenched "irregularities" in the scheme (like "duon-" ({xab-}) in Esp, {mal-} in Lojban) will not be displaced.) (What I intend as the "syntactic deep structure" of an expression in Lojban, empirically, is the most concise single-jufra statement of a concept in the language, such that it is devoid of tanru, ellipses, and (at least for this purpose) conjunction (rather than subordination) of clauses. A transformational approach based on this will probably be the basis of Lojban-language grammatology, and in fact is already current: see the recent Cowan-Fine exchange on the elliptic places of a jufra containing a logical connective). * Before picking from between the four options too casually, users must beware that there is a possibility (however incongruous with current LLG policy) that in the future an aposteriori analysis of lujvo will be developed based on the classification of gismu- namely, that certain gismu, when used as cmugi'u (or gafygi'u) are primarily associated with one of the three main types of lujvo place structure. (This seems to be already true of rinka: I know of no use of {-ri'a} which analyses as rinka je broda, or rinka befi lo broda). Such a classification is already being performed (without LLG sanction) by Jim Carter, who first formulated the aforementioned guidelines in his Loglan variant guaspi. The above makes me personally question forms such as {cevrirni} (JL7); I suspect that {rirni} will be mostly used in {be} rather than {je} lujvo (eg. sperirni - parent in law - rirni be le speni; cf. Theotokos - cevma'a: the formal Greek appelation of the Virgin Mary). However, givenhow complex and exception-ridden such aposteriori rules are likely to be, and that they will bw based on the natural usage which also give rise to a form like {cevrirni}, this warning is not to be taken as binding. CONCLUSION: One of the hitherto unstated base hypotheses of the language, one in my view much more important that Sapir-Whorf, is that all concepts expressable in human language can be expressed in predicate form, and more- over that these predicates are the result of some combination of the language's gismu sumti, cmavo sumti and sumti tcita. It is as unprovable and as challenging as anything Whorfian. To put it in terms more familiar to you, it is the thesis that all tanru are expandable. What tanru are expanded into is what I have termed the syntactic deep structure - Basic Lojban, if you will. To put the thesis one these terms: that Basic Lojban is as expressive as any human language. As presented here, these guidelines are an approach to this problem. They waddle and stumble, but they are extremely productive and valuable, and deserve the attention of the Lojban-speaking public. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Nick Nicholas, Melbourne Uni, Australia. nsn@{munagin.ee|mullauna.cs}.mu.oz.au "Despite millions of dollars of research, death continues to be this nation's number one killer" - Henry Gibson, Kentucky Fried Movie _______________________________________________________________________________