From - Thu Mar 14 11:54:04 1996 Reply-To: Logical Language Group Date: Thu Mar 14 11:54:04 1996 Sender: Lojban list From: Logical Language Group Subject: TECH: magnitudes X-To: lojban@cuvmb.cc.columbia.edu To: John Cowan X-Mozilla-Status: 0001 Content-Length: 5179 Message-ID: Veijo wrote: >mi pu cusku di'e > >>> Date: Tue, 12 Mar 1996 15:10:14 -0500 >>> From: John Cowan >>> Subject: TECH: magnitude >>> >>> I have been kicking around the idea of a new cmavo of selma'o KI, >>> "ki'i'i" temporarily. A sumti tcita involving this will be a specific >>> magnitude, thus: >>> >>> 1) mi dansu le bisli ve'a ki'i'i lo mitre be li pa >>> I dance on the ice over-interval one-meter. >> >> I'd prefer a separate selma'o NEhE which would allow a slightly more >> concise expression: >--More-- >> >> mi dansu le bisli ve'a ne'e mitre li pa >> >> This would require a one-line addition to 'modifier_84': >> >> | mod_head_490 NEhE_nnn selbri_130 tail_terms_71 > > Both of the above schemes have the shortcoming that a reference point > cannot be expressed. The following modification makes this possible: > > | mod_head_490 NEhE_nnn selbri_130 tail_terms_71 gap_450 > | mod_head_490 NEhE_nnn selbri_130 tail_terms_71 > TEhI_mmm sumti_90 > > {ne'e} and {te'i} would fit the already assigned cmavo quite well > as {ne'a} = 'next to', {ne'i} = 'within' and {te'e} = 'bordering' > > > > > co'o mi'e veion The problem with NEhE as you propose it, is that it will usually require a terminator except before CU or end of sentence. Using the selbri grammar rather than the bridi grammar as you have, thus has the same problem that SEI had. The verbosity of requiring the terminator is comparable to that of requiring the sumti and linkarg markers. Note that you can cut the latter by saying lo mitre pamei. As Cowan explained this proposal to me, you would not need the reference point grammar. If you wanted to specify the reference point, it would be: mi dansu le bisli ve'a le manri ki'i'i lo mitre be li pa I dance on the ice over-interval at the reference point of one-meter. If the ki'i'i is ambiguous as to what the quantity is quantifying you would repeat the ve'a OR attach the ki'i'i phrase using an afterthought termset. The attempt at afterthought termsets, which evolved from your own work on that grammar problem, stemmed from the desire to accomodate just this kind of linking. Of course this means that you need to be able to have afterthought termsets with only singular sumti, and no logical connective, as the grammar you orginally proposed for such offset termsets provided. Cowan seems to prefer the repeated tense to remove ambiguity when necessary, whereas I like the afterthought termset link (other tense elements having been set off with "ku"). So the combination of the new cmavo and the afterthought termsets allows a bit of playing around to find the optimal solution, while making sure that we HAVE a solution, as I promised Goran. As to the termset grammar requiring either two terms or NUhU before the connective, the needed contrast is with the sumti grammar connectives. Jorge: >> 1 shift/reduce conflict > >Is it possible to identify when this happens? When a {gi} is found, >the only possibility is that it is closing the latest open gek, >isn't it? In other words, under what circumstances would the first >nu'u not be elidable? The relevant grammar is:>term_set_83 : NUhI_587 GEK_807 terms_80 NUhU_gap_460 > GIK_816 terms_80 NUhU_gap_460 > | NUhI_587 terms_80 NUhU_gap_460 > JOIK_EK_421 terms_80 NUhU_gap_460 > >sumti_C_93 : sumti_D_94 > | GEK_807 sumti_90 GIK_816 sumti_C_93 > ; It can be seen that without the NUhU_gap, any single term in the first terms_80 will make the GEK term GIK look like sumti_93, i.e. a sumti rather than a complex structure - it will then slop up the 1st sumti after the GIk and the effective grammar is NUhI sumti_93 and the parser sits therew waiting for the rest of the termset, considering any other sumti that follow as part of the first half of the termset. With afterthought termsets not marked with NUhI, I think the NUhU is likely to be never elidable, because the "term JOIK_EK term is again a sumti_93, and added terms remain part of the incomplete termset. The bottom line is that NUhU was added when I created the termset grammar as an alternative to a full set of connective designed specifically for termsets - not a wise investment when termsets were seen as a rather uncommon construct in the language. NUhU may this be seen as parallelling the fronting consonants on the various connectives to keep them unambiguous, though taking an extra syllable, and often a pause: gi'V vs. nu'u.V vs. jV. though NUhU can also be used with JOIks as the connective. Given that we are trying in one fell swoop to allow afterthought termsets AND to allow termsets with no connective (to solve the ki'i'i problem and to allow added places to be expressed on BAI tags by an as yet undefined convention), we have a pretty limnited flexibility in how we play with the grammar. lojbab