How the position of <em>at least</em> affects its interpretation: experimental data

Abstract

The Focus-particle at least is known to be ambiguous between two interpretations: an epistemic one conveying uncertainty/ignorance, and a concessive/evaluative one that conveys a desirability ranking. Prior literature has argued that the position of at least determines what interpretation is available: at least is concessive/evaluative adsententially and epistemic adnominally. We present three experiments that investigate how three properties which the two interpretations have been taken to differ on are restricted by the syntactic position of at least, namely entailment of the prejacent, truth of higher alternatives, and desirability. The results overall support the view that the interpretation of at least is restricted by its position as previously claimed by some accounts, but syntactic position seems to dissociate the three properties in ways incompatible with previous assumptions. We discuss the implications of these results for formal accounts of at least.