From cbmvax!uunet!CUVMA.BITNET!LOJBAN Sat Feb 8 02:26:36 1992 Return-Path: Received: by snark.thyrsus.com (/\==/\ Smail3.1.21.1 #21.19) id ; Sat, 8 Feb 92 02:26 EST Received: by cbmvax.cbm.commodore.com (5.57/UUCP-Project/Commodore 2/8/91) id AA14137; Fri, 7 Feb 92 20:36:15 EST Received: from cunixf.cc.columbia.edu (via [128.59.40.130]) by relay1.UU.NET with SMTP (5.61/UUNET-internet-primary) id AA26666; Fri, 7 Feb 92 15:27:57 -0500 Received: from cuvmb.cc.columbia.edu by cunixf.cc.columbia.edu (5.59/FCB) id AA04867; Fri, 7 Feb 92 15:25:46 EST Message-Id: <9202072025.AA04867@cunixf.cc.columbia.edu> Received: from CUVMB.COLUMBIA.EDU by CUVMB.COLUMBIA.EDU (IBM VM SMTP R1.2.1) with BSMTP id 6340; Fri, 07 Feb 92 15:24:18 EST Received: by CUVMB (Mailer R2.07) id 6173; Fri, 07 Feb 92 15:22:50 EST Date: Fri, 7 Feb 1992 15:49:32 GMT Reply-To: cbmvax!uunet!bradford.ac.uk!cuvmb.cc.columbia.edu!c/j/fine.maildb Sender: Lojban list From: CJ FINE Subject: proposed extension to grammar X-To: Lojban list To: John Cowan , Eric Raymond , Eric Tiedemann Status: RO I propose two extensions to the grammar for the next baseline - both are for greater generality. Lojban is canonically SVO (subject-verb-object) and AN (adjective-noun or at least MH - modifier-head), but allows you to vary either of these orders for style, effect, or concordance with the shape of another language. There is one structure which is compulsorily HM (head-modifier), and that is relative clauses/phrases. le zdani pe mi le dinju po'u la kreml le zdani noi lemi rirme'i cu xabju ke'a etc. [Actually, there is one slight exception - "lemi rirme'i" is a MH version of "le rirme'i pemi". I have always found this an admittedly useful, but very ad hoc and awkward construction, particularly when the modifier is more than just a KOhA.] I would like to suggest that relative phrases/clauses be permitted before the selbri of a description or name: le pemi zdani le po'u la kreml. dinju le noi lemi rirme'i cu xabju ke'a [ko'u] zdani. I have not looked at the formal grammar, only the bnf, and I haven't a parser, but it seems to me that this extension is simple and does not cause ambiguity. I would change sumti_tail = [sumti-4] selbri to read sumti-tail = [pre-relative] selbri and add pre-relative = sumti-4 | relative-clauses Note that the existing sumti-4 expression just becomes a zipfian contraction ("lemi zdani" for "lepemi zdani") rather like "ze prenu" for "ze lo prenu". I am not claiming this allows you to express anything more than at present, merely that it removes an unnecessary restriction for no significant cost - and it allows translation from certain languages to maintain the structural order: In Japanese, adjectives are a special kind of verb, and a (verb-final) relative clause precedes its head just like an adjective. omosiroi hito interesting person boku ga kinoo mita hito I (subj) yesterday saw person ("the person I saw yesterday") le poi mi viska cale prulamdei ke'a [ku'o] prenu 2. (This suggestion is less strongly motivated, but still worth considering). Extend this pre-relative to all sumti. This will need a cmavo - I have picked "na'a" from the currently unused ones, but I have no particular motivation for this choice. The cmavo then introduces a relative-clause(s) which stands before the modified sumti. The only question is what level of sumti it attaches to. My suggestion would be to replace sumti-2 = sumti-3 [KE etc] by sumti-2 = [NAhA pre-relative] sumti-3 [KE etc] Thus: na'a poi xabju co lamji mi ku'o la fred. na'a ne le ckule la djenis na'a noi mi gleki lenu tirne ke'a ko'u ra na'ami la mam. (again the last could be thought of as a contraction of "na'apemi la mam" The main use I have thought of for this is epithets and restrictions before names: na'a pe la bakin,mcr la xodjez "The Buckinghamshire Hodges" na'a noi misno la ringos. star. "The famous Ringo Starr" na'a poi nolraitru la xusEIn. "King Hussein" There is actually a further wrinkle you can put in this one: in the NAhA construction (not in the LE construction above) you could allow a selbri, and interpret this as an omitted "poi" (you could actually go to a sumti-tail or a full bridi-tail, but then you would always need to terminate it, so I don't think that's worth it. ) This allows a rather neat construction for titles that can be expressed by selbri: na'a nolraitru la xusEIn. "The Hussein who kings" na'a kicmu la uotsn. na'a misno la ringos. star. (restrictive: as opposed to all the non-famous ones) na'a nolraitru bero natmrbelgi'e la boduan "King of the Belgians Baudouin" Any takers? Colin c.j.fine@bradford.ac.uk