An approach to abductive inference developed in the TACITUS project has resulted in a dramatic simplification of how the problem of interpreting texts is conceptualized. Its use in solving the local pragmatics problems of reference, compound nominals, syntactic ambiguity, and metonymy is described and illustrated. It also suggests an elegant and thorough integration of syntax, semantics, and pragmatics. 1

Many researchers believe that certain aspects of natural language processing, such as word sense disambiguation and plan recognition in stories, constitute abductive inferences. We have been working with a specific model of abduction, called parsimonious covering, applied in diagnostic problem solving, word sense disambiguation and logical form generation in some restricted settings. Diagnostic parsimonious covering has been extended into a dualroute model to account for syntactic and semantic aspects of natural language. The two routes of covering are integrated by defining "open class " linguistic concepts, aiding each other. The diagnostic model has dealt with sets, while the extended version, where syntactic considerations dictate word order, deals with sequences of linguistic concepts. Here we briefly describe the original model and the extended version, and briefly characterize the notions of covering and different criteria of parsimony. Finally we examine the question of whether parsimonious covering can serve as a general framework for parsing. 1

have developed an approach to abductive inference, called "weighted abduction", that has resulted in a significant simplification of how the problem of interpreting texts is conceptualized. The interpretation of a text is the minimal explanation of why the text would be true. More precisely, to interpret a text, one must prove the logical form of the text from what is already mutually known, allowing for coercions, merging redundancies where possible, and making assumptions where necessary. It is shown how such "local pragmatics " problems as reference resolution, the interpretation of compound nominals, the resolution of syntactic ambiguity and metonymy, and schema recognition can be solved in this manner. Moreover, this approach of "interpretation as abduction " can be combined with the older view of "parsing as deduction " to produce an elegant and thorough integration of syntax, semantics, and pragmatics, one that spans the range of linguistic phenomena from phonology to discourse structure. Finally, we discuss means for making the abduction process efficient, possibilities for extending the approach to other pragmatics phenomena, and the semantics of the weights and costs in the abduction scheme. 1.

Abduction is inference to the best explanation. In the TACITUS project at SFLI we have developed an approach to abductive inference, called "weighted abduction", that has resulted in a significant simplification of how the problem of interpreting texts is conceptualized. The interpretation of a text is the minimal explanation of why the text would be true. More precisely, to interpret a text, one must prove the logical form of the text from what is already mutually known, allowing for coercions, merging redundancies where possible, and making assumptions where necessary. It is shown how such "local pragmatics" problems as reference resolution, the interpretation of compound nominals, the resolution of syntactic ambiguity and metonymy, and schema recognition can be solved in this manner. Moreover, this approach of "interpretation as abduction" can be combined with the older view of "parsing as deduction" to produce an elegant and thorough integration of syntax, semantics, and pragmatics, one that spans the range of linguistic phenomena from phonology to discourse structure.

We are able to understand language so well because we know so much. When we read the sentence John drove down the street in a car. we know immediately that the driving and hence John are in the car and that the street isn’t. We attach the prepositional phrase to the verb “drove” rather than to the noun “street”. This is not syntactic knowledge, because in the syntactically similar sentence John drove down a street in Chicago. it is the street that is in Chicago. Therefore, a large part of the study of language should be an investigation of the question of how we use our knowledge of the world to understand discourse. This question has been examined primarily by researchers in the field of artificial intelligence (AI), in part because they have been interested in linking language with actual behavior in specific situations, which has led them to an attempt to represent and reason about fairly complex world

We are able to understand language so well because we know so much. When we read the sentence (1) John drove down the street in a car. we know immediately that the driving and hence John are in the car and that the street isn't. We attach the prepositional phrase to the verb \drove " rather than to the noun \street". This is not syntactic knowledge, because in the syntactically similar sentence (2) John drove down a street in Chicago. it is the street that is in Chicago. Therefore, a large part of the study of language should be an investigation of the question of how we use our knowledge of the world to understand discourse. This question has been examined primarily by researchers in the eld of articial intelligence (AI), in part because they have been interested in linking language with actual behavior in specic situations, which has led them to an attempt to represent and reason about fairly complex world knowledge. In this chapter I describe how a particular kind of reasoning, called abduction, pro-