This paper presents a framework for modeling the local coherence of discourse, focusing on the relationship between focus of attention, choice of referring expressions, and perceived coherence of utterances within a discourse segment. The framework, called "centering," models the local component of attentional state and examines how local coherence interacts with the choice of referring expressions. It argues that differences in coherence correspond to the inference demands made by different types of referring expressions, given a particular attentional state. The paper discusses the development of the centering framework, which was initially proposed in 1983 and later expanded. It defines various centering constructs and proposes two centering rules. The framework is based on the idea that the coherence of discourse is influenced by the compatibility between centering properties of an utterance and the choice of referring expression. The paper also explores the relationship between coherence and inference load, and how both interact with attentional state and choices in linguistic expression. The paper introduces the concept of "centers" of an utterance, which are entities that link the utterance to other utterances in the discourse segment. Each utterance is assigned forward-looking and backward-looking centers. The backward-looking center of an utterance connects with one of the forward-looking centers of the previous utterance. The connection between these centers may be of several types, and the realization relation combines syntactic, semantic, discourse, and intentional factors. The paper discusses the constraints on centering transitions, including the preference for continuation over retention and retention over shifting. It also explores the factors that govern centering, such as grammatical role and pronominalization, and how these affect the ranking of centers. The paper provides examples illustrating how different choices of referring expressions can affect the coherence of a discourse. The paper concludes by discussing the implications of the centering framework for discourse interpretation and the need for a semantic theory that supports partial interpretations. It also references subsequent research that expands on the original work and examines the cross-linguistic applicability and empirical validity of centering.This paper presents a framework for modeling the local coherence of discourse, focusing on the relationship between focus of attention, choice of referring expressions, and perceived coherence of utterances within a discourse segment. The framework, called "centering," models the local component of attentional state and examines how local coherence interacts with the choice of referring expressions. It argues that differences in coherence correspond to the inference demands made by different types of referring expressions, given a particular attentional state. The paper discusses the development of the centering framework, which was initially proposed in 1983 and later expanded. It defines various centering constructs and proposes two centering rules. The framework is based on the idea that the coherence of discourse is influenced by the compatibility between centering properties of an utterance and the choice of referring expression. The paper also explores the relationship between coherence and inference load, and how both interact with attentional state and choices in linguistic expression. The paper introduces the concept of "centers" of an utterance, which are entities that link the utterance to other utterances in the discourse segment. Each utterance is assigned forward-looking and backward-looking centers. The backward-looking center of an utterance connects with one of the forward-looking centers of the previous utterance. The connection between these centers may be of several types, and the realization relation combines syntactic, semantic, discourse, and intentional factors. The paper discusses the constraints on centering transitions, including the preference for continuation over retention and retention over shifting. It also explores the factors that govern centering, such as grammatical role and pronominalization, and how these affect the ranking of centers. The paper provides examples illustrating how different choices of referring expressions can affect the coherence of a discourse. The paper concludes by discussing the implications of the centering framework for discourse interpretation and the need for a semantic theory that supports partial interpretations. It also references subsequent research that expands on the original work and examines the cross-linguistic applicability and empirical validity of centering.