German is well-known for its propensity for nominal compounding. This claim is put on a firmer empirical footing by means of a bidirectional translation study between German and English. The difference between the two languages crystallizes in the competition between compounds and phrases. Two complementary asymmetries emerge: first, German compounds are more frequently translated by English phrases than English compounds by German phrases; second, English phrases are more frequently translated by German compounds than German phrases by English compounds. An extension to other word classes shows that the compounding bias in German is not restricted to nouns. It is tentatively argued that the token frequency of word classes plays a role in the emergence of compound propensity. The heavier use of nouns and adjectives in German than in English might be partly responsible for the higher rate of nominal and adjectival compounding in the former than the latter language.