This book is an ambitious attempt to outline and justify a theory of preferred interpretations and explore its implications for pragmatics, semantics, and grammatical theory. Levinson assumes that both the content and the metalinguistic properties of an utterance carry the message. He argues that preferred interpretations exist to overcome a bottleneck in transmission, providing an economical means to maximize inference and amplify the communicational load of speech. Levinson revives the Gricean notion of generalized conversational implicature (GCI), extending it to include implicatures based on metalinguistic knowledge. He proposes three heuristics closely linked to Gricean maxims, which induce preferred interpretations of three general kinds. Levinson argues that GCI theory is at odds with Sperber and Wilson's Relevance Theory (RT), which treats all implicatures as 'nonce' or 'one-off'. Levinson partitions the inferred information into two categories: utterance-token meaning and utterance-type meaning, adding a third level for generalizations. He argues that RT fails to provide criteria distinguishing explicatures from implicatures. About half the book discusses the wider implications of GCI theory, arguing that implicatures play a role in the assignment of truth-conditional content, and that semantics and pragmatics are interdependent. Levinson's views challenge traditional grammatical theories, suggesting that many areas of grammatical theory may be reanalyzed within the GCI framework. The book is clear, well-argued, and provides a compelling case for GCIs, challenging traditional views and offering a new perspective on language. It is a significant contribution to linguistics and pragmatics.This book is an ambitious attempt to outline and justify a theory of preferred interpretations and explore its implications for pragmatics, semantics, and grammatical theory. Levinson assumes that both the content and the metalinguistic properties of an utterance carry the message. He argues that preferred interpretations exist to overcome a bottleneck in transmission, providing an economical means to maximize inference and amplify the communicational load of speech. Levinson revives the Gricean notion of generalized conversational implicature (GCI), extending it to include implicatures based on metalinguistic knowledge. He proposes three heuristics closely linked to Gricean maxims, which induce preferred interpretations of three general kinds. Levinson argues that GCI theory is at odds with Sperber and Wilson's Relevance Theory (RT), which treats all implicatures as 'nonce' or 'one-off'. Levinson partitions the inferred information into two categories: utterance-token meaning and utterance-type meaning, adding a third level for generalizations. He argues that RT fails to provide criteria distinguishing explicatures from implicatures. About half the book discusses the wider implications of GCI theory, arguing that implicatures play a role in the assignment of truth-conditional content, and that semantics and pragmatics are interdependent. Levinson's views challenge traditional grammatical theories, suggesting that many areas of grammatical theory may be reanalyzed within the GCI framework. The book is clear, well-argued, and provides a compelling case for GCIs, challenging traditional views and offering a new perspective on language. It is a significant contribution to linguistics and pragmatics.