This chapter explores models and measures that describe language users, focusing on talkers and listeners rather than the language itself. It emphasizes the distinction between a person's knowledge and their actual or potential behavior, highlighting that formal characterizations of language do not simultaneously serve as models of language users. The chapter discusses the limitations of bounded memory devices in producing and understanding natural languages and the importance of aligning these limitations with human capabilities. The chapter introduces stochastic models, which assume that message elements can be represented by probability distributions and that communication processes transform these distributions according to known transitional probabilities. Markov sources, a type of stochastic model, are discussed in detail, including their properties and how they can be generalized to higher-order Markov sources. These models are used to approximate the statistical structure of natural languages, such as English, and to understand the limitations of these approximations in capturing the complexity of human language. The chapter also covers the concept of redundancy, which measures the efficiency of coding in natural languages. It discusses methods for estimating redundancy and the challenges of directly estimating the probabilities required for stochastic models. Finally, it introduces the measure of selective information, which quantifies the amount of uncertainty reduced by a message, and its applications in psychology and communication research.This chapter explores models and measures that describe language users, focusing on talkers and listeners rather than the language itself. It emphasizes the distinction between a person's knowledge and their actual or potential behavior, highlighting that formal characterizations of language do not simultaneously serve as models of language users. The chapter discusses the limitations of bounded memory devices in producing and understanding natural languages and the importance of aligning these limitations with human capabilities. The chapter introduces stochastic models, which assume that message elements can be represented by probability distributions and that communication processes transform these distributions according to known transitional probabilities. Markov sources, a type of stochastic model, are discussed in detail, including their properties and how they can be generalized to higher-order Markov sources. These models are used to approximate the statistical structure of natural languages, such as English, and to understand the limitations of these approximations in capturing the complexity of human language. The chapter also covers the concept of redundancy, which measures the efficiency of coding in natural languages. It discusses methods for estimating redundancy and the challenges of directly estimating the probabilities required for stochastic models. Finally, it introduces the measure of selective information, which quantifies the amount of uncertainty reduced by a message, and its applications in psychology and communication research.