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

Retaining Logical Connectives



Jim C writes:

>Just a brief note.  Old Loglan was intended to be a "logical language",
>and hence logical connectives were a major design feature, so that
>essentially you could "speak symbolic logic", believed in those days to
>be helpful for the thought processes.  Lojban is intended to carry on
>the Loglan tradition.  As a result there has come to be a plethora of
>logical connective grammar forms for just about every substructure in
>the grammar.
>
>In -gua!spi the logical connectives were one of the first features to
>go. In my Loglan writing I found that I used them surprisingly rarely. 
>I won't give examples, but -gua!spi emphasizes constructions like
>these, chosen because these are what I found I used often, all of which
>are handled via gismu and their arguments, not special grammar:
>
>1.  A set or list with explicitly stated members.  (Coffee, tea or milk:
>	choose one from the set.)  (An old Loglan puzzle; try translating
>	that in Lojban.)
>2.  The union or intersection of sets -- the result may be left as a set
>	or may be extended.
>3.  A logical sentence connective (also causal, also any other gismu with
>	two event arguments).
>4.  An anaphor for a previous sentence, with replacement arguments.  
>	(Example: Karen wants to go swimming.  Me too.  Meaning: I (speaker)
>	want to go swimming; "I" replaces "Karen".)

Sorry to repeat all that, but I want to deal briefly with it.  For my
own purposes (and I realize that they may be unique), I would greatly
regret removing the entire range of logical connectives from lojban.
To me, as a computing scientist with a radical formal bent, lojban is
one part of a continuum.  I am working toward more precision in the
specification of systems and in the construction of implementation of
those systems.  A lojban description is (to me) an extremely useful
part in this process.  A designer can refine and add rigor to his
description without leaving the spoken language by re-phrasing things
using more logical connectives in a more formal manner.  Eventually,
this description can be translated verbatim into predicate calculus
notations suitable for formal specification and proof of programs and
systems designs.

For this reason, I have become wary of using too much set notation and
set concepts in the specification process.  While I recognize, and
completely accept Jim's assertion that these are much more useful in
normal speech and expression (given the examples above), for my
purposes they may be overspecific.  In particular, recent research
(Dijkstra and Scholten, _Predicate_Calculus_and_Program_Semantics)
suggests that set notations (as opposed to formulae in predicate
calculus, which requires the full set of connectives) are more
difficult to manipulate.

So here is one vote for retaining the full set of connectives.  I
truly regret that my present knowledge of lojban does not allow me to
enter this discussion more fully; I have strongly held prejudices
concerning the use of logical connectives, particularly with regard to
the importance of the equivalence and avoiding overspecific grouping
of terms connected by associative operators (like the equivalence).
Soon, I hope.

Just one person's two cents worth......

						Dave Barton
						barton@i2wash.com