| Before {ce'u} was invented or fully integrated into usage (i.e., up til now)
there has been a lot of usage of {ka} and considerable of {du'u}. We do not want to lose or have to revise this archive of material, so one goal of any decision on {ce'u} (and so on {ka}, {du'u} and perhaps {si'o}) is to keep the meaning of as much of this ({ce'u}-less) usage as possible. Part of this goal is summed in the claim that most of this earlier usage took {ka} as containing only one or at most two {ce'u} and that/those the first available place(s). So far as I can tell, this claim, while plausible, has never actually been tested. Nor would it be easy to test, since we only have the usage, not the explanations of what was intended. We can glork all we want (and try to remember as best we can) what was meant, but glorking has demonstrated a tendency to incompatible results with different glorkers (and memory is not much better). However, as I said, it does seem a plausible hypothesis and so can serve as a factor in defending any convention (subject to a later discovery that some other pattern is more common). The formalists (hardliners?) held that, just as all the empty places in {du'u} were filled by {zo'e}, so all the empty places in {ka} should be filled by {ce'u}, noting that that gave the nice pattern that the property expressed by {bridi} standing alone was {le ka bridi}. The other extreme point of view was that we rarely had occasion to talk of such things (we hadn't so far, as far as anyone could make of the cases available) and it was wrong to make the most efficient form do for the least used case. Besides, with the simple sumti forms and with questions, each place involved had to be marked separately (with {ke'a} or {ma} or whatever). At some point, someone worked out that a {ka} with all its places filled was essentially a {du'u} and then worked back to a {du'u} with a {ce'u} in some place was a {ka}. The pattern had been for empty places in {du'u} to be taken as {zo'e} and empty places in {ka} as {ce'u} or {zo'e} without any special rule. But now a rule was needed, since empty places in {du'u} might -- rather less regularly -- also be {ce'u}. The official position remains that any position in {ka} or {du'u} phrases that does not contain an overt sumti can be taken as filled by either {zo'e} or {ce'u}, which it is to be determined by goodwilled cooperatively intercommunicating glorking -- and asking for clarification if that noticeably fails. Neither end of the spectrum is very happy with that position and both want some conventions about what is which when. The official line is that these conventions are not abbreviations (always uniquely replaceable) but only guidelines to most likely patterns: gaps in {du'u} are most likely {zo'e}, gaps early in {ka} {ce'u}, for example. Against that, the following have been proposed as binding (up to "obvious exceptions" = cases glorked by both of goodwilled cooperatively intercommunicating conversants). 1. (generally agreed to, I think) all {ce'u} in {du'u} phrases are explicit, blanks are {zo'e} 2. the first blank in {ka} is {ce'u}, others are {zo'e} -- {ce'u} after the first must be explicit. 3. the first two blanks in {ka} are {ce'u}, others are {zo'e} -- later {ce'u} must be explicit. 4. all blanks in {ka} up to the first explicit {zo'e} are {ce'u}, all after are {zo'e} -- later {ce'u} must be explicit. 5. all blanks in {ka} are {ce'u} (The suggestion that the first place, even if filled, is where the {ce'u} goes is dropped for lack of a coherent explanation of what the sumti filling the space does.) In terms of the amount of potential abbreviation each of these offers, 2 (first free place is {ce'u} offers the most, followed by 4 and 3, with 1 and 5 tied for last -- about a third less effective. These figures ignore the case of no {ce'u} which 1 does unqualifiedly best (of course), followed by 4, the others being the same and requiring writing all the {zo'e} in. Further, 2 is most efficient in the case of a small number of {ce'u}, the assumed practice of existing non-{ce'u} writing, and so requires the least rewriting. But the most efficient abbreviation is actually a mixed strategy, using 2 for one or two {ce'u}, and 5 for three and four. this is about a third more efficient that any single line. Someone else suggested that {si'o} was also in the same cluster -- a concept or idea is like a property somehow -- and thus might be used to allow this strategy, taking the 5 abbreviation plan. The main objection to this is that {si'o} is not quite the same as {du'u} and {ka}, since it is explcitly tied to a person, whatever this may mean metaphysically. Thus, there may be many si'o brodi -- even one per person at each given time -- while there is only one ka brodi or du'u brodi -- assuming all the {zi'o} and {ce'u} are intended in the same way, at all times. But this line of solution, with {si'o} is still available. In its most effective places, scheme 2 uses 0, 1 or 2 (maybe 3) {ce'u} and no {zo'e}. In its most effective areas, schem 5 uses 0, 1 or 2 {zo'e} and no {ce'u}. There is, then almost no overlap between the two conventions, which cover different cases. The one problem issue is the case of 0 of each kind. On convention 2, this means that the first empty space is a {ce'u}, the only one in the phrase, the rest being {zo'e}. On convention 5, this means that all empty spaces are {ce'u} and no {zo'e} occur. But, on convention 2, {ce'u} never occurs in the first space not otherwise assigned and on convention 5, {ce'u} never occurs at all. Thus, a {ce'u} in the first not otherwise assigned space can be taken to mark that all empty spaces are {ce'u}. This does, admittedly, give this now rare case an inappropriately short form. But that does not affect the appriopriateness of the other forms and I suspect that this form will become more common when we get to doing Lojban semantics in Lojban. The best meeting of the various desiderata for {ka} then seems to be: all {zo'e} = {du'u} , 1 or 2 {ceu} use scheme 2 (first free space assumed {ce'u}), 3 or 4 {ce'u} use scheme 5 (show all {zo'e}), all {cu'e} : {cu'e} in exactly the first free space. |