View source for desirability of rafsi

* mark:
** I don't completely remember the details of Lojban-poesis, so this may have been addressed, but I've been thinking about the relative utility of [[rafsi|rafsi]]. Were they just given out based on what was more available and which [[gismu|gismu]] would be likely to need more rafsi, or was some thought given to the relative desirability of the different rafsi forms? I seem to recall some discussion of better rafsi (and not just more of them) for oft-compounded words, but I still find myself wondering.  This may be painfully obvious, or not...
*** [[User:Bob LeChevalier|Bob LeChevalier:]]
**** Yes, I considered what gismu would need more rafsi, and relative desirability of rafsi forms.
*** For example, is there really any reason why a gismu that has a CCV rafsi would need <u>any</u> other rafsi?  (I don't count the 4-rafsi and 5-rafsi, of course, which are special cases). A CCV rafsi can be an initial element in a [[lujvo|lujvo]] with no danger of [[tosmabru|tosmabru]], or a final element, or a middle element with no more worry of [[hyphen|hyphenation]] than any other rafsi. That's really the only possible advantage of having another rafsi when you already have a CCV: if the other one starts with another consonant that can mesh hyphen-lessly with consonants that the CCV would have to hyphenate. But '''mel''' and '''mle''' are thus redundant for '''melbi''' (though a casual look shows that in many cases, at least, it is a different initial consonant). It is also true that fairly few gismu have additional rafsi beyond CCV, but some do.
**** Ooh!  Thought just occurred to me: what would you think about permitting CCV rafsi with the CC <u>not</u> being a permitted initial, though still a permitted medial? Could still be useful, no?
***** It would cause morphological problems. How would you break up /tosmApku/?
**** [[User:Bob LeChevalier|Bob LeChevalier:]]
***** For the most part, post-1993, if a gismu has a CCV rafsi, it does <u>not</u> have any other rafsi.  If it does, then there was an explicit reason.  Among these reasons are
*****# an alternate CVC rafsi starts with a different letter allowing less hyphenation;
*****# other rafsi were considered "sacred" because they had been used in significant texts or examples ('''gic''', and the triplets for '''condi''' and '''bloti''' are major examples of this)
*****# the CCV is not 3 consecutive letters of the gismu and there was still some doubt that people would learn and prefer such rafsi (we still were allowing some room for aesthetics);
*****# we retained a couple of CCV/CVC pairs simply because they had been there before the revision and there was no good reason to eliminate them (reducing the number of rafsi to be learned was not considered a significant reason since we were talking about only a couple rafsi in a list of a thousand).
***** There may be a couple other reasons I've forgotten, in case there is one or two not covered by the above, but a quick skim of the gismu/rafsi list seemed to cover most cases with those four reasons.
** Plainly, CCV are the best rafsi to have. (aesthetically, CCV rafsi may be considered "ugly" after a CVC rafsi because of the consonant heaviness; in such cases, having a CVC or CVV alternative that could be used would be nice - lojbab) What then?  Well, CVV rafsi are obviously better than CV'V, as they're monosyllabic.  I think that counts as a ''big'' advantage.  Saving syllables is a big plus in lujvo: if you don't save them, the lujvo can be as cumbersome as the veljvo tanru.  True, there is the semantic difference that gives them their advantage, but there isn't a savings in ease of use.  CVV and CV'V may need to be hyphenated, but the r or n hyphen rarely results in another syllable (my most favorite/despised example is '''voirli'u'''/''fly'', which in my mouth anyway develops an extra syllable [more or less] in the r-hyphen. If only I had the freedom to use '''*voinli'u''', which doesn't have that problem).
** CVC can be initial but never final, but at least they're monosyllabic.  They may, however, incur an extra syllable in a y-hyphen not just before (like all the others) but also after, and not just as a result of consonant-clashes: tosmabru can cause it too.  One might argue that because of the extra syllable, CV'V are close to useless, since they never save a syllable over the unreduced rafsi.  There's something to be said for this viewpoint (though all the y-hyphens in an unreduced lujvo are annoying, even when it's same syllabic length as its reduced version).
** OK, so where am I going with all this glaringly obvious stuff?  Well, I'm wondering if the rafsi are, in fact, a resource that should be used sparingly.  You could (and probably will) answer the question of "why does '''melbi''' have both '''mle''' and '''mel'''?" with "Well, nothing else can use '''mel''', why not?"  And maybe that's a good answer.  But is it?  We're finding a lot of uses these days for those unallocated [[cmavo|cmavo]], and the [[type 4 fu'ivla]] that I never thought would see daylight are getting serious consideration.  Might we not want to consider non-obvious rafsi for some of these someday?  Or keep rafsi for future use?
** What do you think, is this page a waste of electrons (probably)?  Maybe there's something to consider here.  Or perhaps a broader question: how would you reallocate rafsi now if you could?  (i.e., any favorite gismu of yours that had its perfect rafsi given to something you think is useless?)
** [[User:Nick Nicholas|nitcion]]:
# Unallocated cmavo and type 4 fu'ivla - well, you know what I think of them.
# Noone's learned all the rafsi, and noone will.
# Of course they were flooded onto anything that could have them. You remember [[the Great Rafsi Reallocation|the Great Rafsi Reallocation]], right?
# rafsi should indeed be used sparingly, precisely because they give rise to even more minimal distinctions between words, and are not widely known. There is a <u>lot</u> to be said for sticking to 4-letter rafsi for nonce lujvo, and only going with 3-letter if you're sure this will be a commonly-used concept.
** [[Adam|Adam]]:
*** Too bad it's baselined. :-)