Sapir's language classification

The last month I've been reading about Sapir's
language classification, and I'd like to know if
you know it, and what you think of it. It's complex
and it's almost 80 years old now, so I may be trying
to understand something that has proven wrong by
now.

For those who don't know what I'm talking about,
the classification takes into account that there are
four types of concepts in languages:

I. Basic concepts (root elements)
II. Derivative concepts (concrete non-radical elements)
III. Concrete relationship concepts
IV. Pure relationship concepts

and it examines what languages do with these groups.
For example, in Chinese you only have types I and IV,
roots and pure relationships expressed by adpositions
and word order. In English, the plural ending -s
belongs to group III, and the diminutive suffix -let
(as in eaglet, piglet) to group II, but in Tibetan
the plural is expressed by a separate root "several",
"many" (group I), and in Bantu langs the diminutive is
not a derivative concept but a relationship concepts;
it's a gender and belongs to group III.


What do you think of this classification? Have you
seen it used anywhere? It's certainly more useful than
the old agglutinative/flexional/isolating system, at
least to me, when it comes to imagine how a language works.


--Pablo Flores

* * * * * * * * * * * * * * * * * * * * * * * * * * *
 And the Lord said unto Job, "There's no
 reason for it.  It's just policy."
                                       Kelvin Throop