Meaning differences between the inputs to syntactic blends

Abstract

When speakers combine (or blend) expressions – that is, utter them in overlapping fashion – there is usually some degree of semantic overlap between the expressions. There is much agreement on this point. Some authors have gone further and suggested that the expressions involved are usually synonyms or near-synonyms. Others have gone further still, claiming, at least implicitly, that blends always involve synonyms or near-synonyms. In this paper I show that in many cases where expressions are combined they differ substantially in meaning, in some cases even expressing distinguishable events. I show that there is regularity in the nature of the differences in meaning, and that the differences often correspond to construal distinctions identified in the cognitive linguistics literature. Four degrees of semantic overlap between combined expressions are identified: synonymy; subsumption, where the meaning of one expression encompasses that of the other as well as contributing a bit extra; partial complementarity, where each expression conveys some element of meaning not conveyed by the other; and distinctness. Despite the parsing cost to the hearer of the speaker combining two expressions rather than choosing one over the other, the many cases of partial complementarity can be seen as serving a useful purpose. By combining semantic content that is not conventionally combined the speaker can cover more of his/her intended meaning. And by combining different construals of the same semantic content (e.g. interpretive and descriptive, different viewpoints, stative and dynamic, agentive and non-agentive, etc.) the speaker can equivocate in areas where conventionally a commitment is required. Twelve recurrent meaning differences are illustrated with examples from three different datasets.