miscellaneous responses to many postings

[Logical Language Group <cbmvax!uunet!GREBYN.COM!pucc.PRINCETON.EDU!lojbab>]

- On "conversion" (Rosta/Cowan) - this is folk linguistics from the "I
think I know JCB well enough school":  I think it is derived from
"converse", just as JCB.

"lexeme" was chosen because the assignment of selma'o by the parser is
done by a "lexical analyzer", by analogy with "phonemes" identified by
phonetic analysis.

These have nothing to do with anti-linguist sentiments, but much to do
with JCB's analogic approach to understanding how language works (one
must see some of his convoluted linguistic arguments in the trademark
case to understand how much stock he places in his personal semantic
interpretation of a word)

- On assimilation across morpheme boundaries
(Rosta/Morneau/Shoulson/etc.)  - I think syllabification is a factor in
all this, too.  I think non-homorganic consonant phonemes tend to be
split into separate syllables where possible and homorganic ones tend to
be joined, and assimilation is less likely across a syllable boundary.
This is intuition based on looking at Rick's examples.  (This doesn't
explain Gilson's Japanese example, of course - but then the Japanese
have so few clusters they will almost always use a heavily buffered
Lojban dialect) So one perhaps should talk of balanced and unbalanced
syllables.  Lojban's morphology I believe allows one to always split
non-homorganic clusters into separate syllables.  The only clusters that
one can't make such guarantees on are the 'permissible initials'.  But
our selection of these sets, I admit, was not based on this type of
analysis.  We had a Russophile professional translator/ amateur linguist
arguing for as much variety of consonant clustering as possible to suit
Slavic aesthetics to balance the rather polynesian tendencies of much of
the rest of the language.  Lojban's predicate words must have consonant
clusters - why not make these as versatile as possible?  But we did
therefore have someone skilled in the problems of phonemic assimilation
in a heavily consonantal language when we redid the Lojban phoneme set
from old Loglan (and Russian DOES have much such assimilation).

Mark writes:
>Note also that the rules permit some clusters which (sorry, Lojbab) are
>not so easy to work with.  One which I found just recently and marveled
>at was "pm".  This is a real toughy.  Homorganic stop/continuant
>clusters are usually fairly touch, as opposed to homorganic
>continuant/stop clusters, which tend to be easy.  Something like "pm"
>would likely get swallowed up into the nose; not what we'd like.  Ditto
>"tn" or "dn" or "bm".  "dl" and "tl" have their own problems.

Actually, I have more problems with stops articulated in different
places like "mapku", but again, these are separate syllables, and the
worst risk I notice in actual speech is that a listener might hear the
space between them as a pause if we really talked fast enough.  I know
of no problems in pronunciation with 'dl' and 'tl', but then my "l"s and
"r" probably are very un English when I speak Lojban (I've been told
that my rapid Lojban speech sounds like a south Slavic language; e.g.
Montenegro - I've been tempted to send Ivan a tape to see what he
thinks.)  In English we are used to aspirating rather than swallowing
syllable final stops - Malayan I believe does otherwise, as will 'good
Lojban' for a "pm" cluster.

- On necessary places (Gilson, after others):

Bruce writes:
>Now I think I finally see what disturbs me.  Whenever a gismu has its
>place structures baselined, the decision has been made for all time what
>are the essential components.  If the place structures for "xunre" and
>"blanu" do not include "under illumination xn" (n = whatever) then the
>consensus of lojbanists who accepted it do not -- to _my_ eyes -- accept
>the reality of the fact that perceived colors depend on the light
>source.  An apple may be black under the blue-violet lights used to
>cause fluorescence, yellow under a sodium-vapor lamp, and red under
>sunlight.  On the other hand, since "klama" has the 5 places that were
>given previously, I am _forced_ to imply that those places are important
>enough to think about, _even_ if (by using "fa" type particles or
>"zo'e") I omit those places.  Sorry, but to me this is a _serious_
>weakness.  It still allows Lojban as a SWH test -- in fact may very well
>make it even _more_ useful as a SWH test, but it makes Lojban very much
>less useful as a means of communication.

Funny.  I can talk about whether an apple is red in English without
specifying the illumination.  To the average person in normal usage,
scientific fact notwithstanding, "red" does not involve illumination,
and the average person can judge truth or falsity without caring about
illumination.

There are several members of BAI that are used to convey what might be
called the relativist metaphysics school of meaning.  These include
comparative more than ..., by standard ..., to observer ..., under
conditions ...  (which would encompass the illumination question).
Whether these are metaphysically necessary to meaning is dependent on
your metaphysics, and modern metaphysics is predominantly relativist, so
these things SEEM obviously necessary.  But in effect EACH of these
places could be determined mandatory for every gismu; i.e, they are part
of our means of knowing truth, not part of the nature of truth itself.

Lojban tries to minimize metaphysical assumptions, hence leaves these
places optional.  Put them in every sentence, though, if you need to.
Only in places where it truly is impossible to evaluate truth of a
useful concept without knowing these are they included in the place
structure.  For example, "left" and "right" have no meaning without an
observer and a frame of reference.

Yes, the 5 places of klama are all important - to thinking about
"klama".  If you are expressing "klama", you are expressing an
at-least-5-place relationship.  If you have a relationship in mind that
doesn't involve one of these, it would be a different relationship,
hence a different word in Lojban.

Lojban has room for words of similar place structure.  If one were to
advocate a version of Lojban with two different place structures for
each 'concept', I would say - make them different words.  Using the same
words would cause dialectization, but Bruce is correct - I would still
call both dialects "Lojban" and/or "Loglan".  Indeed we have that
nowadays since some people are using the new gismu list in writing and
some use the draft lessons lists, and some use the public domain list.
Lojban is communicative even if the exact place structures aren't known
or firmly defined (though a machine might have more trouble).  This is
why I tell people not to place heavy importance on memorizing place
structures:  1) they will change eventually because language changes 2)
you don't need to memorize them to communicate - you will learn the ones
you need to know as you need to use them, just as you learn the
idiosyncratic meaning of the English preposition "to" for each verb that
it is found with only as you need to do so.

- on "kill" vs.  "cause to die", "klama" and "litru" and various
other similar gismu (And and Bruce (&&Bruce %^)?

As above, if it expresses a different relation, it is a different
predicate, hence a different word in Lojban.  The only real question
that should be involved here is whether the words should both be gismu.
But as I said in my glossary posting, gismu are not 'primitive' in
Lojban - the set is selected for pragmatic reasons including historical
compatibility and usefulness in making compounds (lujvo).

- on "remi" (Cortesi >> and Cowan >):
>> That's good; how about "le zdani be remi"?  (I am very taken with "remi"
>> "cimi" etc. in part because they have no English equivalent.)
>
>Actually, they do:  "remi" means "the two of us, we two".  "le zdani be
>remi" is perfect for "the house of the two of us".

I don't think this is right.  "remi" means "two of us" with the total
number of "us" unspecified (though context may tell you that there are
only two in "mi" to choose from), not "the two of us"; "mi poi remei"
would be "we who are two", for "the two of us".

I would use "roremi" "all-two of us/ Both of us" to get the latter
across.  It relies on assigning a semantic meaning to the number "rore"
which would otherwise be meaningless.  I will note that not everyone
likes this convention, but I personally favor coming up with useful
semantic interpretations for apparently meaningless strings, since I do
not like the idea of calling them "bad usage and hence illegal" like JCB
does.  The latter is a cheating way of saying that the 'unambiguous
syntax' of the language describes a superset of the language, not the
language itself, and we rejected it in developing the Lojban grammar.
Thus there are perfectly grammatical sentences that have no clear
'meaning', but which I will not call 'wrong' unless they fail to
communicate.  The development of meaningful semantics for such constructs
is a potentially creative area for the language that will take years
to reach its glory, but rthe English has its "curious green ideas sleep
furiously".

lojbab

Footnote on remi: after reading cowan's posting on the subject, I am reminded
that "le re mi" is also valid for "the two of us".