Re: building from primitives (was Re: Langauge Constets)

While I agree with Jörg about the essential dangers of making people
think like computers I can sympathize with the desire for a limited
vocabulary. However any classificatory scheme will get dated, mostly
sooner or later - witness Dalgarno and Wilkins. The best option is to
use word frequency statistics, preferably from several languages and
bearing in mind that the boundary between phrases and words is not a
clear one. I've put some material to this end on FrathWiki. The
Longman list is very good for vocab building. Although I don't know
how they arrived at it it's actually a mini-conlang for dictionary
definitions.

2007/11/26, Jörg Rhiemeier <joerg_rhiemeier@...>:
> Hallo!
>
> On Thu, 22 Nov 2007 15:12:28 -0000, Michael Poxon wrote:
>
> > Bravo!
> > Someone (I forget who) once said that "the danger is not that computers
> will
> > think like people, but people will think like computers".
>
> AMEN.  People "thinking" like computers are indeed a great danger
> - because they no longer really think but merely obey orders.
> It happened here in Germany (as well as in other places, but the
> German case was the worst of all) - the gory results are well-known.
>
> If you read web sites of transhumanists, you find this kind of
> computerish hyperrational thinking all the time.  I find that
> very unsettling.  Those people appear to be bored nerds who wish
> to play Hitler once they get the money to do that :(
>
> One of the strengths of human beings over computers is that they
> can question what others tell them - if they fail to do so, great
> evil usually is the result.
>
> >       There are so many
> > wonderful differences between computer "languages" and human ones.
> > Human languages (or a decent conlang) have cultures and histories behind
> > them to which they refer and are inextricably part of, for one thing.
>
> Yep.  I feel that all that "philosophical"/"semantic primitive based"/
> oligosynthetic engelanging is thoroughly misguided.  Those languages are
> cumbersome, lifeless and ugly.  I have experimented with such beasties
> in the past, and was utterly unsatisfied with the meager results.
> I found it doesn't really work: reality is far too complex to pigeonhole
> it into a set of "semantic primitives".  I found my naturalistic conlang
> projects, all of which have a history behind them, to work out much better.
>
> Loglangs and their ilk are not even useful as auxlangs - they are way
> too difficult to learn because they are so unlike human languages.
> You have to be familiar with first order predicate calculus in order
> to understand the grammar of Lojban (let alone learn the language)
> - can we really treat tourists and businesspeople to undergo that
> effort?  Even I, who has a university degree in computer science and
> certainly an above-average predilection for languages, found the grammar
> of Lojban to be a book of seven seals.
>
> All that "testing the Sapir-Whorf hypothesis", "eradicating semantic
> sloppiness", "speeding up thought" and whatever things some people
> want to achieve with engineered languages appears to be completely
> misguided to me.
>
> ... brought to you by the Weeping Elf
>

--
/ BP