According to the Wikipedia article, Compound (linguistic), compound words that occur in natural languages can be semantically grouped into four categories. Witness this quote from the article:

  1. “endocentric: A+B denotes a special kind of B: examples: darkroom, smalltalk”

  2. “exocentric: A+B denotes a special kind of an unexpressed semantic head: examples: skinhead, paleface (head: ‘person’)”

  3. “copulative: A+B denotes ‘the sum’ of what A and B denote: examples: bittersweet, sleepwalk”

  4. “appositional: A and B provide different descriptions for the same referent: examples: actor-director, maidservant”

Generally speaking, how common is each of these types of compounds across languages? For example, are appositional compounds rare, not that common, common, or extremely common across the documented natural languages?

  • 3
    As usual, wikipedia is redefining reality. You ought to add the sanskrit terms for the compounds, as that's what people that work with compounds use. 3 is a dvandva, 2 is a bahuvrihi etc.
    – kaleissin
    Commented Apr 12, 2012 at 10:53

1 Answer 1


At first, it seems that the (semantic) classification of compounds you found in Wikipedia was proposed by Geert Booij, with some modifications (he uses the term appositive). The thing is that it's not the only one - I know at least a dozen of other proposals (by Laurie Bauer, Martin Haspelmath, Rochelle Lieber, Sergio Scalise, among many others). Although some linguists would say that "bittersweet" is appositional because "bitter" and "sweet" describe the same referent. See English co-compounds? Is bittersweet a co-compound? for more details

Of course, such a terminological muddle would pose a serious problem for any cross-linguistic research. It is quite possible to do such research though - you just have to keep in mind who means what by this or that term. And, to make things worse, there are languages where it's hard to determine whether there is compounding in them or not.

Guevara and Scalise 2009 used another classification in their cross-linguistic research: they divided all compounds into subordinate, attributive, and coordinate. All of those can be both exo- and endocentric. Here's what they found, on the basis of sixteen languages in their sample:


Laurie Bauer also did a cross-linguistic comparison of compounding on the basis of thirty-six languages. Here are some of his findings:

  1. it is not clear whether compounding can be regarded as a universal;

  2. the majority of languages that have compounding have endocentric (tatpurusa) nominal compounds;

  3. the use of dvandva compounds seems to be an areal feature.

See Bauer 2001 for more details.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.