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

[lojban] Re: morphology paper announced



On 2/20/07, Cyril Slobin <slobin@ice.ru> wrote:

http://www.lojban.org/tiki/tiki-index.php?page=Morphology+analysis+programs+comparasion&bl

Comments are solicited.

coi kir

Have you looked at
<http://www.lojban.org/tiki/tiki-index.php?page=BPFK%20Section%3A%20PEG%20Morphology%20Algorithm>
?


<<
2.2.2 Leading cmavo

 There is no common agreement about breaking a potential brivla into
 leading cmavo and the rest. Published word breaking algorithm gives
 a set of patterns for breaking words, but it is unclear whether the
 rest of the word after cutting off a leading cmavo must be a valid
 word by itself.


Yes, the rest must be one or more words, otherwise you cannot separate
a cmavo.

<<
Brkwords program treats this as the fact of being
 a valid word for the resting part is irrelevant: for example, the word
 "iglu" breaks into cmavo "i" plus resting "glu" and therefore is not
 a valid fu'ivla (the fact that "glu" is not a valid word by itself is
 irrelevant). On the other hand, vlatai insists that "iglu" is valid
 word, *because* "glu" is not a valid word and therefore "iglu" is not
 breakable. The Vim syntax plugin follows the first approach (brkwords
 compatible) by default, but can be coerced into vlatai-compatible mode
 by setting a flag variable.


{.iglu} is no different from {ciblu}. If you break it into {.i} + {glu}
then you would also break {ciblu} into {ci} + {blu}.

<<
2.2.3 Obscure case

 Vlatai does not recognize as brivla some words that I failed to find
 any reason not to be a valid brivla. The shortest possible example is
 "adjdga". If someone knows why this is not a brivla, mail me please!
 For the Vim syntax plugin this word is a valid brivla.


The PEG morphology rejects it because "jdg" is not a valid initial cluster.
It only accepts non-initial clusters that consist of one consonant plus
a valid initial cluster.

mu'o mi'e xorxes


To unsubscribe from this list, send mail to lojban-list-request@lojban.org
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
you're really stuck, send mail to secretary@lojban.org for help.