Scalar implicature and non-alternatives

Scalar implicatures (SIs) have been studied using priming. Previous studies on implicature priming observe that SIs arise more often after strong primes (priming trials that force the computation of an SI) than after weak primes (priming trials that force literal interpretations without SIs). To explain these results, Bott & Chemla (2016) claim that what is primed in implicature priming is certain aspects of the computational mechanism for SI. Specifically, computing an SI is considered to involve two steps: (i) referencing an alternative sentence and (ii) negating it. For example, participants refer to ‘all’ when they compute SI in a sentence such as ‘some squares are here,' and then they negate ‘all’ and generate SI as ‘not all squares are here.’ However, the results from Skordos & Papafragou (2016) seem to put an objection to this computation mechanism. In their experiments, presenting a non-alternative ‘no’ in the initial priming stage also helped children derive more SIs. If participants only refer to an alternative and negate it, we would not see the increased priming results when ‘no’ is in place. To account for this, we propose that context adaption and inverse preference will provide a better explanation.