From the Lojban web site FAQ (

fu'ivla: borrowed word (literally, "copy word")

Formerly called le'avla, a more literal calque of 'loan word'.

The shift from le'avla to fu'ivla may have come from Colin Fine; see, where he announces that he wants to start using fu'ivla instead of le'avla. (At that time, there appears to have been a little discussion as to whether fu'ivla was a better word for the concept; see the couple of messages starting at --pne

There are four ways to borrow a word into Lojban, with increasing degree of integration:

Type 1 fu'ivla: me la'o gy. + word in original spelling + gy.; e.g. me la'o gy. Phascolarctos gy. *How is it gy.? It's not an English word, shouldn't it be ly. for Linnean, since that's the kind of name it is?Officially, the choice of word is quite arbitrary. People get used to gy, and forget it's gy for glico. Is that explained in any of the current learning materials?
Type 2 fu'ivla: me la + Lojbanised cmene; e.g. me la faskolerktos.OK for noun-like brivla, but less fitting for verb-like brivla, which might have other places. Granted, noun-like ones are much more common. Still, even something like cmacrnintegrali ("integral" in the calculus sense) is likely to have a place structure like "x1 is the integral of x2 with respect to variable x3" or some such. How these other places are to be intuited for Type 3&4 fu'ivla I don't know, but since me only has one place (does it?) it wouldn't work for these Types 1&2. mi'e mark Oops. And already said this below.
Type 3 fu'ivla
gismu + buffer consonant(s) + Lojbanised word; e.g. mabrnfaskolarkto
Type 4 fu'ivla: Lojbanised word, with clusters to guarantee it will not fall apart morphologically; e.g. fasxo larto

The Book indicates that the canonical form of fu'ivla morphologically is Type 3, with four-letter rafsi prefixes.

  • Four-letter prefixes are recommended only because if you use a three-letter prefix and don't check for a consonant cluster, you might end up making a non-fu'ivla. With tools like vlatai at our disposal, we shouldn't be afraid of 3-letter rafsi prefixes, which can sometimes give nicer, less "crunchy" words (djarspageti, for example, is much nicer than cidjrspageti). --rab.spir
  • Generic ideological objections: (1) the 4-letter version is completely predictable, the 3-letter isn't; (2) I won't be running vlatai in face-to-face interaction; (3) of course fu'ivla should be crunchy — how else will I realise immediately they aren't lujvo? — nitcion.
  • 3-letter rafsi fu'ivla are valid type 3 only if the rafsi is of CVC form. At least that's what vlatai thinks. --phma
    • My copy of vlatai parses djarspageti just fine. --rab.spir
    • Do make sure you're using the absolutely latest vlatai. It was a target of many bug fixes in 0.37

No convention exists for what vowel to choose as the final vowel of a loan word, if it ends in a consonant in the source language, and that language's morphology does not suggest a suitable final vowel (unlike the case for Latin.) Nick Nicholas proposes in the lessons that the final vowel simply be repeated; e.g. Mamluke - mamluk - prenrmamluku .

Another (previous?) proposal was that the final vowel of the gismu used as prefix be used: pren-r-mamluk-u (in this case happily both conventions agree)

Counterexample: zgike + rok -> zgikrnroko vs. zgikrnroke .

  • I think you've added an extra hyphen to zgiknroko and zgiknroke. I've never heard of nrock music.

The advantage the former proposal has is that it is also usable for Type 4 fu'ivla, whereas the latter is only usable for Type 3.

Another variant on Method 1 is (me) la'e zoi gy. Phascolarctos gy.

Methods 1 & 2 are also less integrated syntactically since (I (=And) think) me has only an x1. Hence they won't work for fuhivla that need to be polyadic. The ma'oste I have says "convert sumti to selbri/tanru element; x1 is specific to [sumti] in aspect x2", which looks like two places to me. On the other hand, that's still not a lot. --pne