Even: The conventional implicature approach reconsidered

Conclusion

Like Bennett's account of ‘even’, my analysis incorporates the following plausible and widespread intuitions. (a) The word ‘even’ does not make a truth-functional difference; it makes a difference only in conventional implicature. In particular, ‘even’ functions neither as a universal quantifier, nor amost ormany quantifier. The only quantified statement that ‘EvenA isF’ implies is the existential claim ‘There is anx (namely,A) that isF’, but this implication is nothing more than what the Equivalence Thesis already demands. (b) ‘Even’ is epistemic in character, implying some type of unexpectedness, surprise, or unlikelihood. Moreover, despite Kay's arguments to the contrary, this implication is part of the meaning of ‘even’. (c) ‘Even’ is a scalar term, since unexpectedness comes in degrees. And, finally, (d) the felicity of an ‘even’-sentenceS requires thatS* be sufficiently surprising in comparison to its true neighbors. However,pace Bennett, being more surprising than just one true neighbor will not suffice. At the same time, being more surprising than all true neighbors is unnecessary. Suffice it thatS* is more surprising than most true neighbors.