My stance on this issue, which I believe is a decent compromise between the different issues raised in this thread (which I will not take the time to respond to individually) is that glides should be permitted, but be non-clustering.
What I mean by "non-clustering" is that clusters of the form /Cg/ should be forbidden. In other words, the occurrance of Cg means a syllable break must occur between the C and g.
This breaks words such as "guaspi" and "tuitsku", and even cmevla such as "tuitr" and "nitcion".
For cmevla, the problem is not so dramatic, as we can just break up the gV into ghV, getting "tu'itr" and "nitci'on", or we can displace the glide by inserting another vowel, yielding "tuuitr" and "nitciion" (I address those specific forms below).
For zi'evla/fu'ivla, the issue is trickier as "gu'aspi" is no longer a fu'ivla. I suspect that some/many such fu'ivla will become so-called tcizbaga-form words: they will become fu'ivla masquerading as lujvo. That is another debate to be had, however. I propose correcting these words, which begin with with Cg, more robustly so that the result is still in fu'ivla form.
Furthermore, I propose restricting valid pronunciations of the 'u' in 'uV' to only be the approximant /w/. This solves the problem with the string "tuuitsku": it would therefore need to be interpreted "tu uitsku". Disyllabic realizations of the diphthongs would be forbidden.
As for the case as hand which sparked this discussion, namely the experimental cmavo "jie'e'e", it would be forbidden as it would contain a cluster involving a glide.
This solution also acts to increase consistency in the morphology between zi'evla, cmevla, and lujvo.
Also, since this proposal effectively considers /j/ and /w/ as consonants, I think I should clarify that I would be opposed to considering words such as "uitki" to be gismu, simply on the basis that expanding the gismu space should be done with more care. We may wish to consider them gismu at a later time, but that would require changing words such as "uitki" that do not deserve to be gismu. (I'm not specifically saying uitki is undeserving of a gismu. That's a separate debate, unrelated to the one at hand.) For now they should still be considered zi'evla/fu'ivla.
Besides breaking certain fu'ivla and cmevla, neither of which is core to the language like the gismu or cmavo, both of which can be fixed a posteriori, what issues does this solution raise?
On Sunday, 19 October 2014 20:19:33 UTC-4, guskant wrote:
coi ro me byfy
peg morphology allows cmavo beginning with "consonant glide":
http://www.lojban.org/tiki/BPFK+Section%3A+PEG+Morphology+Algorithmcmavo-form <- !h !cluster onset (nucleus h)* (!stressed nucleus / nucleus !cluster) / y+ / digit
onset <- h / consonant? glide / initial
while CLL3.4 disallows it:
"the ten following ones [diphtongs with on-glide, i.e. beginning with i or u] are used only as stand-alone words and in Lojbanized names and borrowings."
Which is official now? La jbovlaste allows it based on peg morphology, and la gleki and I don't agree to the new rule. Concretely, an experimental cmavo "jie'e'e" is now discussed:
http://jbovlaste.lojban.org/comments.html?valsi=27527;natlangword=0;commentid=0;definition=0If this new rule is official, CLL3.4 must be modified.
mi'e la guskant mu'o