Message-Id: <9211251435.AA11839@getafix.oasis.icl.co.uk> Date: Wed, 25 Nov 92 14:35:08 GMT From: I.Alexander.bra0122@oasis.icl.co.uk Subject: TECH RE: Pretty Little Girls' School To: cowan Cc: lojbab@grebyn.com Cc: nsn@munagin.ee.mu.oz.au Cc: c.j.fine@bradford.ac.uk Cc: vilva@viiki21.helsinki.fi Cc: shoulson@ctr.columbia.edu X-Mozilla-Status: 0001 I've got some detailed comments to come, but there's one major issue to be resolved here. What does cmalu je nixli ckule mean? Example 15.3 suggests that it means A: school for small things and for girls and example 15.19 suggests B: school for things which are small and are girls and example 15.39 can't seem to make up it's mind. I think that since cmalu je nixli means [something which is] small and [is a] girl then example 15.19 must have the right of it. We _can't_ let tanru modification distribute over logical connectives, since there is a perfectly straightforward interpretation which means something different. Or is it just ambiguous? I doesn't matter much in your examples in section 6, where the modifiers are essentially adjectival, but it does matter in general. Forget that last statement. There is already an ambiguity between citno ckule young ("new") school youth (young children['s]) school so cmalu je citno ckule could mean school for things which are small and young or school which is small and young or small school for young things or new school for small things but some of these work better than others in the distributed interpretation ke cmalu ckule ke'e jeke citno ckule [ke'e] I'm not convinced I like all this rampant ambiguity. There _are_ technical solutions for expressing "A" above unambiguously. I would borrow some programming language technology from e.g. John Backus' 1977 Turing Award Lecture ("Can Programming Be Liberated from the von Neumann Style? A Functional Style and Its Algebra of Programs", CACM Aug 1978 Vol 21 #8 pp. 613-641). Use {ce'o} to construct a sequence of brivla. This is an essentially formal construction, whose main purpose is to facilitate further manipulation. Define tanru modification such that it distributes over sequences, so that cmalu ce'o nixli ckule is the same as ke cmalu ckule ke'e ce'o ke nixli ckule [ke'e] Then define a "reduction" operator {xe'o} which "inserts" it's logical-connective left operand between each element of the right operand sequence's members, so that je xe'o cmalu ce'o nixli ckule expands to ke cmalu ckule ke'e je ke nixli ckule [ke'e] Of course this would be another grammar change. Some more detailed comments. Line after example 3.9: "Example 2.5" -> "Example 2.8" Examples 6.10 and 6.11: "crinu" -> "crino" Example 7.3: Are you sure that it isn't {la nu,iark.}? :) Example 8.6: To my mind, a _right_-grouping rule would make it ta ckule co (nixli co cmalu) ta ckule co cmalu nixli ta cmalu nixli ckule Section 13: pinsi kilri'a: I was going to suggest {kiltci}, but Nick's {kilbra} is good. jipci pimla: {pimlu} ractu mapke: {mapku} cutci sudsau: (and elsewhere) - {sudysau} nanmu bakni: should at least be {ninmu} cidja barja: The description's gone wrong here. It's the _tertanru_ which is describing the place where the _seltanru_ is sold. And then it's not a specific example of the foregoing. And in the final group in this section, I would classify as follows: smacu terkavbu: zdani turni: zerle'a [nun]terpa: cevni zekri: Object of (potential) action {cevni zekri} is the most dubiously thus classified, but it _does_ correspond to the x2 of {zekri}. kanla djacu: Source of product cag[y]cecmu zdani: Place where object (typically) is OK, I give up on {ladru denci}. Section 14: sonci toe'rdarsi: {to'erdarsi} - and I might have chosen {to'ervirnu} Oh, and for completeness, you might as well mention {jai} in section 10. mi'e .i,n.