[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [lojban] Re: valfendi algorithm

At 08:19 AM 1/24/03 -0500, Pierre Abbat wrote:

On Friday 24 January 2003 07:56, Robert LeChevalier wrote:
> This does not sound like it is a proper defining algorithm for the Lojban
> morphology as you've described it, and as a first glance at the text
> indicates.  It may parse all well-formed Lojban words, but it also may
> successfully lex some not-well-formed Lojban (your algorithm seems to allow
> fu'ivla with embedded text strings that would invalidate the fu'ivla if it
> is a proper gismu or rafsi, but allows the fu'ivla if it is not).  This is
> merely another stage in our long running dispute as to whether type IV
> fu'ivla are to be constrained to specific forms positively defined, or can
> consist of anything lexable word that could be a brivla that isn't a gismu
> or lujvo.
I am planning further versions which will check all words for well-formedness. 
Currently it accepts anything ending in a consonant and not containing a
cmegadri as a cmene, including such unpronounceable messes as {mzantcesg}.
What do you mean by "fu'ivla with embedded text strings that would invalidate
the fu'ivla if it is a proper gismu or rafsi, but allows the fu'ivla if it is
not"? Can you give an example?


My understanding of:

A slinku'i, as far as word breaking is concerned, is anything that matches
the following regex:
^C[raf3]*([gim]?$|[raf4]?y)
where
C matches any consonant
[raf3] matches any 3-letter rafsi
[raf4] matches any 4-letter rafsi
[gim] matches any gismu.
A correct algorithm would use the structures CVC/CVV/CCV for raf3, CVCC/CCVC for raf4 and CVCCV/CCVCV for gim. It doesn't matter whether the values are in fact actually used. Post-freeze it seems logical that it would and should be easier to add and subtract from the gismu/rafsi lists than to change the entire morphology, so the morphology is defined at a higher level than the specific list of words.
(In addition "ala'um" is not an "option"; there should be no options in an official algorithm. It is either valid or invalid according to the rules.) 

lojbab

--
lojbab                                             lojbab@lojban.org
Bob LeChevalier, President, The Logical Language Group, Inc.
2904 Beau Lane, Fairfax VA 22031-1303 USA                    703-385-0273
Artificial language Loglan/Lojban:                 http://www.lojban.org