FAQ: LOGDATA.RAW format - what are rafsi, keyword, clue word

[Logical Language Group <lojbab@GREBYN.COM>]

This is a repost of a good [frequently asked] question from Ian Gordon.
My recent posting of the new rafsi baseline contains much of this info,
but maybe not as clearly for beginning Lojbanists.

*** Ian G.  Please note addendum at the end of my response.

>From: gordon@baslg1.enet.dec.com (Ian Gordon, Computer Systems Administrator,
 DTN 7 826 2251  25-May-1993 0852)
>
>This one is mainly a query which (ISTM) is a little elementary for the list
>(compared with the level of discussion I have seen so far).
>
>On the gismu list (I have grabbed logdata.raw) I am a little puzzled by the
>layout - it seems to be of the form
>
>gismu   rafsi1   rafsi2   rafsi3   meaning   'loose meaning'   exact meaning
>
>Where _any_ (one or more) of the rafsi may be omitted, even to the extent of
>only having number 2 or number 3 present.
>
>I have (just about) figured that a rafsi is a shortened form of the gismu for
>use in building more complex words.  My question is - when do you use each of
>the possibilities?  Is one of them only used at the start, one only in the
>middle and one only at the end, or what?

AdvTHANKSance

Ian

& r
To: gordon@baslg1.enet.dec.com
Subject: Re: self-segregating morphemes

Actually, this is the type of question I, and a lot of other people,
would like to see more of on the list.  Most of the people on the list
are at your level of skill and question, but they tend to be lurkers.
The people who write 'higher' stuff are perfectly happy to talk on these
subjects, but don't have the questions.

The PLS file newrafsi.unf will contain the more detailed answer to your
question.  Briefly, a given gismu has up to 5 combining forms.  I will
use zmadu as an example, since it has all of them

-zmadu is the long form found at the end of a word zmady is the long
form used anywhere else

Thus blanyzmadu is the lujvo equivalent for blanu zmadu (blue-more) =
(bluer)

When any of the other three forms exist, they may be used in substitute
for the longer forms, subject only to the morphology restrictions on
compounds.

These are - a lujvo must end with a vowel (hence you cannot substitute a
CVC-short form in final position, but you may substitute either of the
others:  blanymau blanyzma

and certain phonology and morphology rules dictate modifications when
certain combinations of letters come together.  If they are the wrong
ones, then you have to stick a 'y' in between, pronounced as an
unstressed schwa - the final vowel of 'sofa'.

You may also have to insert either a 'y' or an r/n in two rarer
instances where missing those letters would cause the compouynd to fall
apart and be heard as two or more words.

All of the forms, short, or long, are considered to be the identical
word, with identical place strucxture and semantics - you use a long
word when talking to someone who doesn't know the rafsi short--forms,
since you can easily reconstruct what it was built from then.  You might
also use it in 'noisy environements' like talking on a radio to an
airplane or ham radio, talking at a noisy party, etc.  Otherwise, you
probably use a word that sounds good to you, mostlikley the shortest
possible one.  But it is still the same word.

Feel free to post this question on the List, and let someone else
answer it (or I'll repost this answer).  I know there are others who
would ask the question or want to know the answer.

*****
Addendum on reposting:

BTW the fields that you call "meaning" and "loose meaning" have a
considerably different intent than your titles suggest.  The first we
call 'keyword', the second 'clue word'.

The keyword field is a short word or phrase which is intended to serve
as a unique one word (or very short phrase) English reminder of the
meaning - it must be unique because the field is used in flash card
drilling 9such as with the LogFlash program), wherein you must give the
Lojban word for the English.  As it must be unique, it therefore
sometimes is not really a "meaning"; e.g. sfani 'fly' (the insect) and
vofli 'flight' (but really 'fly' the verb - "flight" would probably be
"nu vofli")

The 'clue word' has an similar purpose for people studying the words.
Most (but not all) of the Lojban words have some English portion in
their etymology, according to the classic Loglan algorithm for grinding
together the sounds of the 6 (8 in old Loglan) source languages.
Usually the keyword is the same word that was used in the Lojban
word-making etymology (or is very similar as per fly/flight).  If the
etymology word is NOT similar to the keyword, it is given in the 'clue
word' field.

Thus, Lojbanists like Frank S. who may be having trouble learning the
gismu, and who haven't found memory hooks, may get some from the clue
word, if it exists, and/or the keyword.  To do this, Lojbanize the
American pronunciation of the English word, and look for matches with
the Lojban - they will usually be evident and similar (though people
have noted that we probably shouldn't have matched later/'LeItr' with
'baLvI' and may give you either a good solid hook on the word as a
whole, or at least on some letters of the word.

(We didn't want to try to do British pronunciation, which we who built
the Lojban vocabulary don't know as well and in any case seems to reduce
more vowels and consonants to non-contrasting values than American
pronunciations.)


lojbab