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

Some things about lojban

I received the level 0 mailings about half a year ago.  I read them
over, took some notes, and wanted to post some observations-
unfortunatly I couldn't find this lists address, and now that I have it,
I've lost my notes! (In fact, I don't even have the level 0 mailing in
front of me. gismu are the basic words, that fit inside the places
defined by the lujvo, right? This is from memroy)

The mailing mentioned that there was some trouble with representing
mathematical concepts, but there was no description of the current
method.  Could you mail me a copy of the math grammar?

One big problem that I remember- and it would take a lot of work to fix-
is lojban's use of the postfix grammar.  If I remember correctly, the
reason postfix was chosen was to make machine interpretation easier,
since most machines used a stack-based architechture.  However, today's
machines don't need to store data structures on a stack; stacks can be
used but they're no easier to represent than a binary tree, an array, or
a DAG.  (In fact, I suspect it would be better to represent a block of
lojban text as a list of trees, with the lujvo at the root of each
tree... just a hunch).  If it is necessary to store the text as a stack,
it would be simple to parse from prefix to postfix.

Using the arguments above, there's no clear reason to prefer either
prefix or postfix notation.  So why go through all the work of rewriting
textbooks, etc to go with prefix?

Human comprehension: with the current setup, a person has to keep track
of up to five gismu before hearing how they fit together (the lujvo). 
Since most people have a short-term memory of about seven items, people
won't notice this as a problem.  With more complicated structures, there
are times when people go near the limit of their short term memory. 
Putting the lujvo first gives the reader a framework in which to place
the subsequent gismu.  Since the reader is now able to chunk the
incoming words together, s/he can understand the information easier.

Furthermore, putting the lujvo first might help a parsers error
detection.  If a lujvo is left out from a postfix stream, the parser
will not be able to detect the error until it reaches the next (not left
out) lujvo.  With prefix notation, the parser could immediatly see that
it had at least one gismu too many and flag it.   I am not too sure of
the language and place structures, though, so this may be wrong.

It would take a bit of work to convert from postfix to prefix, but I
think the long-term comprehension benefits would be worth the effort.

What do you think? I am sorry that I don't have the grammar rules here,
with some concrete examples I'm sure I could make a more convincing