Not really sure who to quote at this point, so I'll ju= st talk in general.

My opinion on the business of empty places and z= o'e is that an empty place is not strictly equivalent to an overt {zo&#= 39;e}. First of all, the idea that an empty place always equates to an over= t {zo'e} (minus place structure advancement) falls apart in relative cl= auses and {ka}-abstractions, where implicit {ke'a} and {ce'u} are c= rucial to brevity. I much prefer the idea that if a place is left empty, th= en it could very well be {zi'o} due to the underlying idea that I subsc= ribe to such that {ko'a broda} implies {zi'o broda}.

So an empty place can give rise to {zo'e} and {zi'o} ({zu'i= } is just a special kind of {zo'e} IMO) and in certain clauses, the emp= ty place can also give rise to {ce'u} and {ke'a}. Furthermore, I be= lieve that it is in some sense an axiom of Lojban that implicit information= cannot produce a contradiction, va'i that an empty place can give rise= to {no da}, since that would result in a contradictory negation.

As for the problem of the place structure of a tanru whose tertau is a = JA-connected selbri, I think that the most sensible solution is the toss ou= r hands into the air and let pragmatics decide. If one says {mi jipci je su= tra lo ka tavla}, then it would be silly for one to assume that {lo ka tavl= a} is equally filling jipci2. That's just nonsense, and I think that we= should try to limit nonsensical interpretations as much as possible. If yo= u want to explicitly make all the places line up, then do so using giheks; = they can do that quite well, and no one really argues about how to interpre= t {broda gi'e brode vau ko'a}. Yes, it's longer, but that's= the price you pay for concision. (I think of that as the length-complexity= axiom: precision and length should be proportional as best as possible.) S= o I think that in the general case of {broda je brode ko'a}, the most s= ensible thing to say is that we don't know the structure of the = trailing sumti in such a case, in much the same way that we don't kn= ow the distributivity of a bare lo-sumti. We are thus complying with th= e length-complexity axiom by making a very short construct such as a je-con= nective in a tertau have a very vague meaning. The listener is expected to = infer the most sensible structure based on context. On the other hand, it i= s pretty useful to require the x1 of the left and right operands of the con= nective to line up, and is a very weak requirement, much less weak than tha= t all the places line up, which, I agree, is just plain silly, bordering on= totally useless in practice.

.i mi'e la tsani mu'o