Abstract

Structurally, Swedish, a Germanic language, follows the general Germanic pattern in which compounds prototypically have their categorial and semantic head on the right-most lexical component. The structural analysis is here complemented with textual frequency calculations of compounds in a morphologically tagged newspaper corpus of 24.2 million word-tokens. The frequency analysis shows that the major types of Swedish two-component nominal, adjectival and verbal compounds are the following: N+N, A+N and V+N in nouns; N+A and A+A in adjectives; and N+V, A+V, ADV+V and PREP+V in verbs. The textual frequency analysis of nominal, adjectival and verbal compounds reveals that Swedish compounds are not extremely productive: (i) with the exception of one compound (with a referentially non-fixed, i.e., numeral modifi er), the corpus does not contain a single compound hapax; (ii) moreover, the frequency ranges of the compounds start at relative high textual frequency values, and the compounds of the lowest frequency values are intuitively common lexical items in the language. Thus, if a compound is used in Swedish, it is used relatively frequently. This leads us to claim that Swedish heavily relies on compounds which are conventionalized, established and lexicalized.