Message-Id: <9208051430.AA12306@getafix.oasis.icl.co.uk> From: I.Alexander.bra0122@oasis.icl.co.uk Date: Wed, 5 Aug 92 15:30:54 BST Reply-To: I.Alexander.bra0122@oasis.icl.co.uk Subject: Grammar changes zo'u Content-Length: 4037 Lines: 104 I lost track of which of your messages were to the list, and which only to the baseline review committee, so I've just replied to the latter. This is just a quick first pass through, to get the easy ones out the way, and mention some of the issues arising. 1. YESBUT. Looks superficially OK, although your description in the "rationale" appears to have got confused between high and low precedence. Although it makes sense as it stands, there may be other considerations which make it obsolete - see later. 2-4. YES. They look useful, and no apparent complications. I second Mark's comments on 4. 5. SHRUG. I can't get excited about this. I don't really understand the LALR(1) problem, although I haven't looked deeply into it, and my parsing theory's very rusty. (But I guess it's not surprising if LALR(1) imposes _some_ restrictions.) 6. YES. A borderline "yes" on the grounds of compatibility with {ve'o}. (What's this "separate cmavo for the subscript"?) 7. SHRUG. Mark's objection has some merit. 8-12. YES. 13. NO. If it ain't broke, don't fix it. 14-19. YES. Like Mark, I didn't understand 15 until I checked with the grammar. Re 17, I never :) understood why these three were separated in the first place. 20-21. ABSTAIN. I need more time on these, and it may depend on sorting out some problems in the grammar:- BNF (I haven't got round to checking most of these in the YACC yet.) sumti-5<95>: I notice that in getting rid of multiple quantifiers on an indefinite description, you've ended up with multiple quantifiers on a sumti-6 :-) (I now see that Veijo's already pointed this out.) sumti-tail<111>: There's a missing ']' in "[sumti-6". I take it these [nested-relative-clauses] refer to the sumti-6. sumti-tail<113>: This relic of the previous grammar appears between relative-clause<122> and selbri<130>. (I now see that Veijo's already pointed this out.) selbri-1<131>: It seems odd putting after 'CO' - this means you can have a tag after the {co}, as long as you start with a {na}. If you can think of a use for the tag here, use , otherwise move all the tag/NA stuff up into selbri<130>, (which'll probably cost you another YACC rule). Rewind sumti<90>: My first thought was why didn't you make "KE sumti /KEhE/" a sumti-6, analogously to what is now tanru-unit-2. Then I remembered two reasons - the ambiguity with the use of {ke} in tanru grouping, and the desire to use "ek stag ke". Pity, it would have been a potential solution to a lot of grouping problems (see e.g. sumti-6 below) - the position of 'NU' and 'NAhE' in the old selbri grammar was awkward, but could always have been circumvented by using {ke ... ke'e}. I suppose you can still do the same sort of thing with sumti in an emergency using which ever LUhI fits, but this isn't very attractive. sumti-6<96>: I wrote a whole tirade on the "gek sumti gik sumti-4" here, but eventually I checked the YACC, where it is absent. It did however help to draw my attention to the grouping problem. You _do_ want some way of attaching relative clauses unambiguously to a whole "gek ... gik" or "sumti ek sumti" (which I don't think the old grammar could do, except using the LUhI trick mentioned above), and I think the only way to do that is some kind of terminator or grouping mechanism. Similarly we _do_ need to say things like "Three of the people who voted", or "Three of the men who voted". But you can either use some sort of inside quantifier or use "ci lu'a ... lu'u", so we're covered. If you come back and say LUhI is the answer, then I'll accept that. My main concern is that we can say what we want to say, and that we document what the various constructions mean, so that we can figure out how to say it. Fine tuning of elidable terminators is a new problem to me, so I'll rely on your obvious experience in this area. I think this means my vote on 1 becomes a YES.