[Corpora-List] Primitives, semantic and otherwise

[amsler at cs.utexas.edu]

Primitives have a curious habit of being split into smaller units  
every few decades. Chemical elements became primitives with the  
creation of the periodic table of atoms, but physicists split the atom  
and have continued subdividing those results. To me, the useful  
properties of a set of primitives derives from their completeness  
(covering all the known combinations of the next larger units) and  
their organizational ability (revealing useful properties about which  
combinations are possible, predicting what combinations will  
occur)--rather than their indivisibility.

Theories of semantic primitives could serve many purposes (and  
fields). Charles Osgood et al. created the semantic differential to  
analyze connotative meaning and used factor analysis to extract  
several independent dimensions of 'semantic space'. However,  
connotative meaning doesn't explain denotative meaning. Primitives  
there seem to call for an approach more like that of the periodic  
table. Alas, our current state of knowledge there seems more akin to  
alchemists trying to figure out what things are 'elemental'.

A true set of semantic primitives for denotative meaning should do  
something useful, such as predict the meaning of compound nouns or  
predict which compound nouns would be more likely to be used. I  
suspect that isn't uniquely possible since compound nouns seem to  
sometmes acquire their meanings by hapstance. I.e., some day we need a  
new term for something and grab a couple of nouns and stick them  
together and people accept the new term, since you have to call  
something new by some new name. We could call this the discover's or  
inventor's principle. You discover/create it, you get to name it.

Now, sometimes interesting things happen when the new term is  
translated between languages or even within regional variations of a  
language. For example, in the United States, we created 'shopping  
carts' which people use in stores to load up their purchases and to  
carry their purchases into the parking lot. But in British English the  
word 'cart' was wrong, they use 'trolley' to mean what American  
English means by a 'cart' (and likewise, American English couldn't use  
'trolley' since that is reserved for, among other thing, large  
bus-like vehicles, which typically run on tracks or use overhead  
electrical lines for power) so for British English, 'shopping trolly'  
was the correct compound.

That's revealing of underlying primitives at work. When you have to  
take a perfectly reasonable compound apart because it clashes with  
your society's semantics and change its words you could be showing  
that the component words have different primitives in those two  
variants of English. It also shows that those component words are  
'alive' in the sense that they actively affect the creation/acceptance  
of new compounds rather than just being historical accidents of the  
past.

But it all feels like alchemy to me right now.

_______________________________________________
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora