This volume explores the extent to which LLM-generated language conforms to sociolinguistic expectations, positioning AI as a simulated language user and outlining new possibilities for AI-generated language as an object of linguistic inquiry.
Zarzycki makes the case for AI-generated texts as systematic linguistic artifacts which can be evaluated for lexical, syntactic, stylistic, and pragmatic features. The book combines theoretical discussion with controlled experimental research to examine whether changes in contextual parameters, such as register, dialect, sociolect, audience design, and communicative tone, result in predictable and replicable differences. The volume features a mixed-methods approach integrating controlled generative experiments, quantitative corpus-based analysis, computational stylometry, and natural language processing. This approach offers insights into how established sociolinguistic and pragmatic methods can be systematically applied to language produced by non-human agents, providing replicable research designs for expanding the study of AI-generated language in the future.
This book will be of interest to students and scholars in sociolinguistics, pragmatics, discourse analysis, and applied linguistics.