Raymond Brown writes:
 > At 2:08 am +0100 20/4/02, Tim May wrote:
 > >Nik Taylor writes:
 > [snip]
 > > > Those are just some of the problems that make it impractical.  :-)
 > > > There's also the fact that 800 words CANNOT be used to describe any
 > > > concept.
 > >
 > >Well, it's largely a matter of how you define "word" - you can use
 > >idiomatic phrases, but they'll need to be seperately learned, you
 > >can't derive the meaning solely from the constituents.  This is what
 > >Basic English does, I believe - things like "make good" for succeed.
 > Or does it mean "repair" ?
 > That's one of the problems with idioms - they are (potentially) ambiguous
 > and, as you rightly say, have to be learnt as separate "words".  IMO they
 > are a source of confusion and a real, separate word rather than a spurious
 > compound is preferable.
Oh, I agree absolutely.  There's definitely a sense in which such a
compound is really a seperate word, which is one of the flaws in the
idea of Basic English.  I was just being pedantic.

 > >In this sense, you can express everything with just 2 words - 1 and 0.
 > You can, but unless your brain is equipped with a processor dealing with a
 > billion 'words' or so a second, communication is going to be a long,
 > tedious and error-ridden process  :)
 > [snip]
 > >
 > >On the other hand, if you're not concerned with how long it takes to
 > >say something...  the Longman Defining Vocabulary is only 2197 words
 > >and some affixes, and can be used to define any word in the Longman
 > >dictionary.  In theory, you can replace any word with its definition
 > Yes, but to freeze the language at 2197 words might make the expression of
 > new concepts in the future very awkward - the set of words, it seems to me,
 > has to be open-ended.
My point was that it's possible, not that it's a good idea.

 > [snip]
 > >tool for dictionaries, not a theoretical exercise.  I wonder what size
 > >the minimal set of concepts is?  I'm not tempted to try to create it
 > Some have  :)
 > Reginal Dutton thought it was 491 - that's the number of "root words" in
 > final version of Speedwords (1951).
 > More recently, Jeffrey Henning has experimented with the minimal set in a
 > project originally called 'minimalex' but later renamed 'Dublex'.   There's
 > a link to it from his Langmaker web-site.

Yeah, but these are attempts to find the minimum set of roots that can
be used practically, not the theoretical minimum possible.  I'm pretty
sure they make use of compounds which are not solely defined by the
sum of their parts, too (what in lojban would be lujvo rather than
tanru, I think).  What I was imagining was the absolute minimum set of
roots which would allow complete discription without such things, and
without any regard for practicality.

I've thought about it a little more, and it's even more hideous than I
imagined.  Many dictionary definitions _don't_ absolutely describe the
meaning of the word.  Instead, they do things like identify a set
which contains all things designated by the word, and imply that the
word designates a subset, such that a human can work out the subset
and apply the word to it through experience.

For example, dog

From WordNet (r) 1.7 [wn]:

       n 1: a member of the genus Canis (probably descended from the
            common wolf) that has been domesticated by man since
            prehistoric times; occurs in many breeds; "the dog
            barked all night" [syn: {domestic dog}, {Canis

(ignoring the other meanings, which would have to be designated
differently in this hypothetical language of minimal words)

Firstly we have some information on the relationship of the species to
others, and its history.  I'm not sure that "common wolf" is any
easier to define than "dog", so unless you include the entire
evolutionary history of the species back to the dawn of life, it
does't help.  Neither does the fact that dogs have been domesticated
by man since prehistoric times - firstly, I don't think that this is
fundamental to the idea of a dog (it would still be a dog even if this
weren't true).  More importantly, you can take all these points (apart
from the ancestry) and imagine another animal that meets this
description but is not a dog.  As a result, it's almost impossible to
define "dog" precisely, at least without writing pages of text (maybe
a complete listing of the canine genome? (with each atom seperately
described in terms of fundamental particles and forces)).  But this
language must adopt some such measure whenever it wants to refer to
such a thing.  Whenever _any_ word in the language can be paraphrased
in terms of more fundamental elements, it must be.

Such a minimal lexicon must theoretically exist, but obviously no-one
will ever speak it.