In the current literature on English compounding, the most common terms to designate those compounds whose internal semantic relation is coordinate (ex.: space-time, bittersweet, stop-go) are dvandva and copulative compound. 1 Several morphologists-e.g. ten Hacken, Plag, and Katamba and Stonham 2-consider that the two terms are synonymous, and they use them indiscriminately. Dvandva is however an ambiguous term as other linguists-e.g. Arnaud and Bauer 3-use it only in the case of heteroreferential coordinate compounding, i.e. for compounds composed of two nouns whose denotata are unfused (ex: dinner dance, tractor-trailer). They do so in accordance with the historical meaning of the term, which was originally used by Sanskrit grammarians to designate heteroreferential coordinate compounds. 4 This ambiguity sometimes leads to confusion, as the following remark by Plag 5 shows: "It is often stated that dvandva compounds are not very common in English (e.g. Bauer 1983:203), but in a more recent study by Olsen (2001) hundreds of attested forms are listed, which shows that such compounds are far from marginal." When using dvandva, Bauer refers to non-embedded heteroreferential compounds (ex.: Alsace-Lorraine), whereas Olsen's corpus contains almost exclusively either homoreferential or embedded heteroreferential compounds (ex.: dancer-singer; patient-doctor [partnership]). The extension of the concept DVANDVA sometimes includes compounds whose elements are not simply juxtaposed. Bauer 6 applies the term to syndetic coordinate compounds (ex.: bubble-and-squeak, milk-and-water), even though he stresses that these compounds differ from true dvandvas because a coordinator is inserted 1 I am grateful to Pierre Arnaud, Nicolas Ballier, Laurie Bauer, and Diana Lewis for helpful comments on earlier versions of this research. Errors are my own.

No abstract

This article deals with wordplay in word-formation and centers on lexical blending. It claims that, because of their very formation process, lexical blends are instances of wordplay. Drawing on examples from a variety of languages, it offers a categorization of the different features which may be argued to increase wordplayfulness into five classes: formal complexity, structural transgression, graphic play on words, semantic play on words, and functional ludicity.

Two sets of 97 French and 374 English lexical units identified as lexical blends are examined from a contrastive perspective. It appears that English displays a wider variety of patterns than French does – a larger number of marginal types of lexical input combination, of lexical shortening and of phonological splitting. Striking dissimilarities between the two languages also include an inclination for the pattern of double inner shortening in English and the pattern of left-hand-side inner shortening in French, as well as a preference for semantic and phonological right-headedness in English and the absence of a preferred lateral head position in French.