Thorstein Fretheim and Ildikó Vaskó

It may be in order to say a few words initially about the linguistic term 'concessive' and its relationship to the (object-language) English noun concession and the verb concede (i.e. "admit", "acknowledge", "allow", "accept", "grant", etc.). Unlike the corresponding noun and verb, the adjective concessive and the adverb concessively are seldom found in discourse where they are not used metalinguistically, as a linguist's technical terms, and the noun concessive is a strictly linguistic term denoting any function word that has a meaning roughly similar to anything properly defined as a concessive subordinator (although, albeit, (adversative) while, whereas), concessive coordinator ((al)though, but), concessive adverbial item (nevertheless, nonetheless, all the same, even so, still, yet, after all), concessive preposition (in spite of, despite, notwithstanding), or concessive parenthetical adverbial expression ((it's) true, true enough, sure enough, to be sure, admittedly). The abstract noun concessivity is also, of course, strictly confined to metalinguistic discourse. Two propositions p and q are in a concessive relation to one another, if q is true in a context where p is also true, although the truth of p would (stereo)typically cause the communicator, the addressee, some specific third person, or people in general to believe that q would be false. Because p and q would not both be true in the most commonplace of contexts, the speaker/writer may find it pertinent to metarepresent the thought that a true p does not affect the truth conditions of q. We will henceforth refer to p as the 'concession' in a concessive relation, and we will refer to q as the 'counter-expectation'—for convenience, as our usage does not imply that the assertion of the truth of q in a concessive relation will always contradict the speaker's expectation, or contradict a metarepresented expectation that the communicator attributes to the interlocutor or to a third person. The concession segment in a concessive relation between p and q is functionally subordinate to the complementary counter-expectation segment. Pragmatically, the truth of q overrides, or plays down, the relevance of a true p, which is generally to be construed as the immediate context in which the utterance expressing q should be processed. This subordination is also reflected structurally in many, and maybe most languages. The counter-expectation q is expressed in a main clause, and the concession is frequently in an embedded clause, a so-called concessive clause, which is an adverbial adjunct at the matrix clause level of syntactic analysis. But it is also quite common for both concession and counter-expectation to be expressed in two juxtaposed matrix clauses, in a conjunction of clauses, or in what may be analysed as contrastive coordinate constituents of a single matrix clause. The latter option is illustrated in (1), where though connects two predicates, both complements of the copula. There is only a subtle difference in pragmatically derived meaning, if any at all, between (1a) and the conjunction of (1b), where though is replaced by but, the most unmarked of the concessive coordinating connectives in English.