In English and other Germanic languages, noun compounds are formed simply by “appending” the nouns in a certain order. For example, phrases like this are very common:

electricity price comparison service

In Portuguese, Spanish and probably other Romance languages, this kind of compounding has to be done with the aid of prepositions. The translation of the phrase above to Portuguese is:

serviço de comparação de preços de eletricidade

What determines how noun compounds are formed in a language? Is there something like a parameter (as in the principles and parameters theory) that explains this?

I'm focusing only on "open" compounds (not idiomatic ones) formed exclusively by nouns. As Molly mentioned, this might be related to the Adj/N word order in the language. This hypothesis is quite reasonable, since in the Germanic languages the head in these compounds comes at the end, whereas in Romance languages, they come at the beginning, matching the Adj/N word order.

But even if this correlation does hold cross-linguistically, how does it relate to the need for prepositions? Maybe prepositions help avoiding ambiguities when the adjectives are postnominal? Could it be related to processing preferences?

  • There are true compounds in Romance languages too, they're just far less common. Some nice examples from Spanish are quemarropa (point blank) and rascacielos (skyscraper). Commented Dec 6, 2011 at 18:50
  • Aren't these words formed by a verb and a noun? As far as I know, quemarropa comes from quemar (to burn) + ropa (clothes), and rascacielos comes from rascar (to scratch) + cielos (skies). I edited the question to clarify this point. Commented Dec 6, 2011 at 21:12
  • Yes they are compounds formed by compounding verbs and nouns, they are surely not the only kind of compounds in Spanish but the one with a verb plus a plural noun is a fairly large class in Spanish and I'm having trouble thinking of others off the top of my head. I thought bocapierna was one but that wasn't in the dictionary. There is no linguistic rule that says all parts of a compound must be nouns or that only words made up of nouns stuck together are called compounds. Commented Dec 6, 2011 at 21:15
  • I don't think you used a good English example. I'm not so sure that "electricity price comparison service" is actually a compound: coordination ellipsis is possible in this case (for example, gas and electricity price comparison services).
    – Alex B.
    Commented Dec 6, 2011 at 23:03
  • Thanks, @Alex. Do you have a better example in mind? Oh, and feel free to edit the question, if you prefer :-) Commented Dec 6, 2011 at 23:54

2 Answers 2


Compound words are really complex, even if we limit it to noun-noun compounds.

It looks like you're concerned with "open" compound nouns, not "closed" or hyphenated or idiomatic compound nouns like these (what a list! right? corpus alert!). So -

The syntactic definition of a compound noun is when the rule N -> N N can be recursively applied. Anything with more than two nouns is ambiguous in structure. ((coffee mug) holder) vs. (coffee (mug holder)) or whatever.

Most languages are more predictable than English when it comes to "headedness" or left/right branching of closed noun compounds; Downing's conclusion to "On the Creation and Use of English Compound Nouns" is basically that the relationship is unpredictable from the semantics of the two words.

My instinct (very scientific!) is that the concatenation vs. use of prepositions for noun-noun compounds is predictable by typology. My guess is that the form of compound nouns is correlated to the Adj/N word order tendency of a language. Of course, this isn't really an answer, but I think the answer you're looking for is too big. It's sort of like asking why a language is SVO or SOV.


Just thinking out aloud, but might it be related to the case systems in Germanic vs. Romance languages?

electricity price comparison service

...is a shortened form of:

service for the comparison of prices for electricity

This is the same as your Portuguese translation, where the prepositions act as case licensors. However, in Germanic languages, the case system is/was richer and the prepositions disappear (because they're redundant). In particular, "of" can be seen as a genitive licensor:

(service for the comparison) of (prices for electricity)

(prices for electricity)'s (service for the[ir] comparison)

Similarly, "for" is a dative licensor:

(electricity prices)'s (comparison service)

Dropping the genitive "'s" morpheme doesn't seem like a huge stretch of the imagination and serves brevity:

electricity price comparison service

  • Interesting! Where can I find more information about these "case licensors"? Commented Jul 10, 2012 at 16:27
  • Broadly: Nouns are thought to need case; even in languages where this has become morphologically moot (like English). As with a lot of syntactic constructions, the mechanism for this is a "giver and receiver"-type setup: the nouns are given case by licensors. Verbs do a lot of case licensing (e.g., nominative and accusative) as do, as we saw, prepositions; but in some languages, the licensor can also take the form of a clitic on another noun... Commented Jul 10, 2012 at 21:48
  • This is true of Germanic languages: However, the case systems in their modern descendents have (with some exceptions) withered away to nothing, but we still think of them as being there (I believe there's evidence to support this)... Interestingly, however, Portuguese is a Romance language: That is, derived from Latin, which also had a rich case morphology. It begs the question, therefore: Why did the Romance languages swap their nominal declensions for prepositions, whereas the Germanic ones just lost the morphemes? Commented Jul 10, 2012 at 21:54

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.