On the Calculation of Local Implicatures

By contrast, the global implicature in (1c), though entailed by (1b), does not have a similar corresponding implicature of the embedded clause when it appears unembedded. For this reason, the local/global distinction is relevant only in complex sentences. In simplex sentence such as (2a) the local and global implicatures are one and the same. The fact that (1a) has the local implicature in (1b) is not under dispute. What is often debated is the source of (1b). According to some authors (e.g., Russell 2006), although (1b) is “local” in the sense that it is identical to the implicature of (2a), it is derived strictly in the pragmatic component, that is to say, after the meaning of the entire sentence in (1a) has been computed. Other authors (most notoriously, Chierchia 2004) argue that the local implicature in (1b) requires access to the meaning of the embedded clause in (1a) before the computation of the meaning of the entire clause is completed, and therefore its derivation is done alongside the compositional computation of the meaning of (1a). The primary goal of this paper is to argue for the latter approach. For ease of reference, we call the former approach to local implicatures ‘non-grammatical’ and the latter approach to local implicatures ‘grammatical’ (see Russell 2006). The paper also addresses some challenges for the grammatical approach, which predicts a correlation between NPI-licensing and cancellation of local implicatures. The most interesting of these challenges is perhaps the one posed by the fact that interrogative clauses produce local implicatures and, at the same time, license NPIs.