The TACITUS project has developed an abductive inference approach that simplifies the interpretation of texts by integrating syntax, semantics, and pragmatics. This method addresses local pragmatics problems such as reference resolution, compound nominal interpretation, syntactic ambiguity, and metonymy. The approach involves deriving the logical form of sentences and the constraints imposed by predicates, allowing for coercions and merging redundancies to achieve a minimal, best interpretation. Abduction is used to infer the best explanation for observed phenomena. In the context of text interpretation, this involves deriving logical forms and constraints from background knowledge or previous text. The process includes assigning costs to atoms based on their likelihood of being new information, with definite noun phrases having higher costs than indefinite ones. The method also incorporates a type hierarchy to manage the specificity of assumptions and avoid inconsistencies. This hierarchy helps in controlling the abduction process by specifying the types that predicate-argument positions can take and their likely disjointness relations. This approach allows for efficient and accurate interpretation of texts, including handling of redundancy and ensuring consistency. The integration of syntax, semantics, and pragmatics through abductive inference enables the system to handle complex interpretations, including resolving ambiguities and identifying references. The approach has been implemented in the TACITUS system and has been used to interpret various texts, including casualty reports. Future directions include extending the interpretation processes to handle lexical ambiguity, quantifier scope ambiguity, and metaphor interpretation, as well as integrating discourse structure and utterance relations with the hearer's interests.The TACITUS project has developed an abductive inference approach that simplifies the interpretation of texts by integrating syntax, semantics, and pragmatics. This method addresses local pragmatics problems such as reference resolution, compound nominal interpretation, syntactic ambiguity, and metonymy. The approach involves deriving the logical form of sentences and the constraints imposed by predicates, allowing for coercions and merging redundancies to achieve a minimal, best interpretation. Abduction is used to infer the best explanation for observed phenomena. In the context of text interpretation, this involves deriving logical forms and constraints from background knowledge or previous text. The process includes assigning costs to atoms based on their likelihood of being new information, with definite noun phrases having higher costs than indefinite ones. The method also incorporates a type hierarchy to manage the specificity of assumptions and avoid inconsistencies. This hierarchy helps in controlling the abduction process by specifying the types that predicate-argument positions can take and their likely disjointness relations. This approach allows for efficient and accurate interpretation of texts, including handling of redundancy and ensuring consistency. The integration of syntax, semantics, and pragmatics through abductive inference enables the system to handle complex interpretations, including resolving ambiguities and identifying references. The approach has been implemented in the TACITUS system and has been used to interpret various texts, including casualty reports. Future directions include extending the interpretation processes to handle lexical ambiguity, quantifier scope ambiguity, and metaphor interpretation, as well as integrating discourse structure and utterance relations with the hearer's interests.