¹ Such integration may also help resolve some well-known difficulties with the analysis of implicature. Grice's insufficiently nuanced account of what it is to ‘say’, as distinct from to ‘implicate’, something (see, for example, Travis 1991) can be greatly enriched by use of Ross's account of linguistic inertia anchored by benchmarks. This account also provides a notion of initial context rather more plausible than the assumption of initially undifferentiated data which underlies Sperber and Wilson's ‘Principle of Relevance’, in terms of which they attempt to unify Grice's Principles of Conversation and to grade the strength of implicatures (Sperber and Wilson 1986: 46–50, 123–132; see also Cooper 1991: 6).