this thread con fused my brain. I went to this site. Might not be what you're looking for, but it made senses to me. It sorts compounds out into several main varieties.
Is the word "compound" a Latin compound of the two words "put" and "together"?

webster.commnet.edu/grammar/...ounds.htm

Sati, well first of all, I took a deep breath and went and got some books and read up on what generativists (i.e., post-Chomskyan linguistics) think of compounding. I started by looking at Anderson 1992, which pretty much convinced that the view I have been espousing in this thread is that of traditional, structuralist/classical linguistics, (pre-Chomsky). There morphology is divided into inflectional, derivational, and compounding. Infelctional and derivational morphology both have to do with roots and affixes. So, boxes (= box + pl. marker, i.e., inflectional), but understand (= under- preverb + stand vb). The traditional wisdom about the differences between inflectional and derivational is that inflectional tends to be quite productive and derivational not so. Also, inflectional morphology keeps words in the same part of speech (syntactic category), wherreas derivational tends to change things, e.g., deadjectival nouns by suffixation in English redness (red adj. + -ness). All these parts are called morphemes, but the inflectional / derivational affixes are called bound morphemes because unlike words they do not occur in isolation.

One of the major places I was having trouble is that I basically interpreted "salt and pepper" as a common enough phrase denoting the spices salt and pepper. Not as a modifying term "salt-and-pepper" as in "he has salt-and-pepper hair." So, in my obstinant way it was difficult for me to see how I could interpret "salt and pepper" as a "compound". I guess I can see "salt-and-pepper" as a composite (a term that Anderson uses) modifier. (I would still like to reserve compound for things that look like words rather than hyphenated phrases, but that's my personal quirk and is probably not shared by many other folks these days.

Anyway, I'd take a look at these books for a little more in-depth discussion of compounding:

Stephen A. Anderson. 1992. A-Morphous Morphology. CUP.
Mark Aronoff. 1976. Word Formation in Generative Grammar. MIT.
Leonard Bloomfield. 1933. Language.
Peter H. Matthews. 1974. Morphology. CUP.
Anna Maria Di Sciullo and Edwin Williams. 1987. On the Definition of Word. MIT.
Elizabeth Selkirk. 1982. The Syntax of Words. MIT.

Hope this helps.

Usually people, before they make their mind, do some search. You are just repeating what has been said, that`s not a very difficult task, is it?
I do advise you to look at the site WIL gave (I was there some time ago), and there some more:

www.epri.com/corporate/d.../style3.html
www.answers.com/topic/english-compound

"I believe compounds are always defined as being composed of independent words"

That's how I feel, and most traditionalists, but generativists don't necessarily see the three-way distinction to be all that important. Take a look at some of the later books I cited above. The feeling there (especially Anderson) seems to be that sentences have structure and so do words. Is there some possible way for sentences and words to share a grammar. That collapses morphology and syntax. This is seen as a good thing.

Sorry about my forms, I had included a boxes (box n. + -s pl) , too, but it seemed to get erased during editing of the post.