A Short Description of -gua!spi

I have never been satisfied with the mess which English makes of
grammar and semantics, and I was intrigued by Jim Brown's 1960 article
in Scientific American about Loglan.  However, Loglan doesn't go far
enough.  I implemented a number of suggestions from the Loglan
community to produce -gua!spi, which is based on Loglan but (from their
perspective) is radically different.

Morphology: C = (bcdfgjkpqstvxz:), V = (aeilmnoruwy), word = C^n V^n
with official assignments in CV, CCV, CVV, CCVV, CVVV.  : is a glottal
stop, used for transforming vowel-initial words into a CV^n form.  This
morphology is trivially resolvable into words.  Stolen foreign words
(like :au stralo pi te ku) tend to break up into syllables but the
compound word rules keep them together.  Each word has a Chinese-style
tone, represented by the symbols -/|!^=

Grammar: Words are divided into prefixes and predicates (plus sentence
start prefixes and two words to force phrase endings).  A phrase
consists of optional prefixes, a predicate (which may be several words
as a compound) and subphrases among the prefixes or after (not among)
the predicate.  The tones cue the start and end of (most) phrases. 
Words can be grouped in phrases without any reference to their meaning.

Organization (again nearly meaning-independent): 
   A.  A pronoun represents words, not the referent of words, and
is interpreted by copying the referent in place of the pronoun.  This
greatly simplifies semantic analysis.  By the time copying occurs, 
though, showers of context have joined the original, so the copy will
be interpreted the same.  The effect is almost the same as if a pointer
to the original's referent had been copied.
   B.  One form of subphrase is a "modal phrase" for tenses, speaker
ID, listener, etc.  A stack is provided whereby these can be supplied
to phrases automatically, and can be changed and restored -- e.g. for
story dialog even with nested quotes, so you don't have to say "Joe
said" over and over.
   C.  Compound words are eventually split up so each predicate word has
its own subphrase; arguments are replicated as needed.  

Vocabulary:  The Loglan - Lojban lexicon is the starting point for
-gua!spi, though I have added and removed a few words and have
extensively adjusted the word definitions to work well in compound
words.  There are about 1400 content words; any compound word can be
interpreted as a combination of these basis words.

Semantics: The predicates are interpreted by predicate calculus: each
predicate word represents a boolean (or fuzzy logic) function which is
true for thus-related arguments.  Argument sub-phrases fill cases in
order, up to five cases, except that certain optional prefixes act like
Latin case endings to override the natural order.  While the labels X1,
X2 etc. are used to designate the cases, X1 and X2 frequently act like
a traditional nominative and accusative case.  In arguments a (normally
unspoken, normally nominative) placeholder gives a "that which is..."
kind of interpretation; variants on this interpretation are available
through prefixes.  Related words have similar definitions, e.g. -ber =
X1 carries X2 to X3 from X4 via X5, and all transitive motion words
have a similar definition.  Any predicate can appear in any kind of
phrase (main, argument, or modal).

Further documentation available at this time:

guaspi.txt		A longer discussion of -gua!spi (65 Kbytes)
guaspi.sty		LaTeX style file to print it out
xankua.part1, part2, part3   Dictionary of -gua!spi, tab separated database