On another board, Lojbab said that there are approximately 1350 morphemes in Lojban, and evidently that's pretty much enough. It seems very few to me. Anyhow, as most here know, I made a stab at reforming Loglan many years ago, called 'Ceqli,' and it continues, but it's developed into a non-Loglan, and I'm still interested in the idea of actually reforming Loglan so as to get rid of the allomorphs. My original idea still holds — invariant morphemes that automatically form compounds unless a separator particle intervenes. I went on with Ceqli to change the wordshape rules, which you can read about on my Ceqli site at http://ceqli.pbwiki.com/ But now I have something less revolutionary in mind. Define Neologlan wordshape as Cn (nVC). Given that we have 5 vowels and 21 consonants, the simplest form of a Neologlan morpheme, CVC, alone yields about 2000 morphemes. Evidently an adequate number all by itself. But when you add in the other possibilities allowed, you end up with an effectively infinite possible vocabulary. Continuing the Loglan system, we need a set of little words — I propose they be restricted to the shape XVX, where X is the SAME consonant. That is, lw's would be bab, beb, bib, etc., or a set of approximately 100 words. If that's not enough, it could be stretched to include XVVX. This would, of course, result in a less pleasant-sounding language than Loglan, Lojban, or Ceqli. OTOH, it would probably sound a lot like German or English, which are pleasant enough. A loglan without allomorphs. Sounds like a worthwhile project.