Earth Minimal

EarthMinimal - or E-minimal for short - is a complete language with only 300 words.

http://ebtx.com/lang/eminfrm.htm

Outline "tutorial" here:

http://ebtx.com/lang/emintutr.htm

300 Words is rather misleading though. These 300 elements are rather meaning carrying elements which can be combined in clever linguistic ways to give every meaning required.

Quite easy to learn and very elegant.

Some parts of the lexicon:

suffix-transitive prefix-intransitive EM Past EN Present EL Future ER- a question?

- N That which Acts - L That Acted on - M Adjective - J Adverb - Z Plural - EV Possessive (ehv)

pro- = preposition di- = conjunction te- = mated principle pi- = the study of -po = the opposite of -to = an increment of

Some words:

VI : I
SU : You
TI : He
FE : She
KI : It
DE : This
DO : That
HO : Building
BA : Collision
VU : Family
HU : Person
KO : Companion
VE : Possession
MU : Easy
FAA : Father
NAA : Not
ZAA : Vehicle
PWA : Furniture
SLU : Contents
GLA : Cover
SLE : Sleep
KWO : Help
SPRO : Rope
SPRAA : Seat

Seems to work on the basis of a root nouns with orthogonal/independent "conversions" to get other components of language. Then it smurfs it together and uses the idea that people generally extract meaning, even if what you actually said doesn't make sense.

It may remind you of NewSpeak because it prefers positive base terms and derives negatives with -PO, e.g. very bad = HAAPOTOM (HAA=the good, -PO=opposite, -TO=increment, -M=adjective).

I think EarthMinimal would be a very interesting language for automated language learning, recognition and production because it has so simple and few base terms. No need to build complex rules to deal with a big vocabulary. EarthMinimal is a Lisp for natural languages. And you cannot disappoint too many people by sub optimum results because there are few native speakers probably. And because the language is intended for on-the-fly construction of composite words (because it lacks any complex fixed terms) there is no need to capture the meaning of terms exactly.

I think it would be extremely bad for automated language learning, recognition and production because it has so simple and few base terms. It relies almost exclusively on the apparent build-in ability of humans to extract meaning from strings of sounds. It says on the site:

Don't underestimate the ability of others to "piece together" what you mean. They have an enormous understanding of language in general (same as you do) which no one has even begun to codify.

That is exactly why I meant that it would be good study language. The structure of real language is fairly well understood by now, but the meaning isn't. Having a language that focuses on the latter without the overhead of the former would be a GoodThing, or?

So you don't mean "it's a good thing" because you think it's a good place to start from to build a working system, you mean "it's a good thing" in the sense that "This encompasses all the things we haven't got a clue how to do." Is that right? If so, perhaps you could delete all of this section, and replace it with a clearer statement.

I also question your assertion that the structure of real language is fairly well understood. I've asked a few linguists and they've simply snorted. However, I agree that focussing on the content and avoiding the question of the structure gives us a place to start where we have no idea at all where we are.

From my point of view as a non-linuist with an interest in computer linguistics and cognition it looked different. We may not really know the structure of a particular language, but we can infer the structure of any language automatically. See the refs at the end of CategoryNaturalLanguage. So for me this means "well understood" enough. Sure there remains much to be done to use the knowledge on real language and to build useful software on it. But the leading edge seems to move from structure to semantics and this is where I'd guess EarthMinimal to be a useful subject for study.

CategoryLanguage / ConLang