Book review The Linguistic Analysis of Jokes

Graeme Ritchie’s book is definitely a learning experience. Reader-friendly, written in a lucid style, and yet rigorous in argumentation, it is a pleasure to read. Ritchie takes his readers on a tour of selected sights of the complex world of humor research and shows them his perception of things: this is the perspective of a formal linguist, who spent most of his professional life dealing with artificial intelligence (AI) and computational linguistics. And from that vantage point, the state of the art of humor research looks rather disheartening, even though Ritchie focuses on the analysis of simple jokes: we are still very far away from being able successfully to formalize joke generation, not to mention joke understanding. From my vantage point, however, that of a researcher working on the interface of linguistics and literature and trying to handle texts longer than jokes (such as humorous short stories) in terms of humor analysis, this state of affairs has a blessing: successful, complete formalization achieved by AI alone could be a setback for the interdisciplinarity of humor research. Since the time of the heyday of generative linguistics, which as Ritchie admits is his ‘‘conceptual frame of reference’’ (p. 11), literary scholars and other social scientists have found it difficult to find a common language with linguists, precisely due to the key assumption of the latter: ‘‘The descriptions that a linguist builds up should be sufficiently well defined that it would be possible for a machine to follow the grammatical rules’’ (p. 11). A machine could not possibly follow the paths of human creativity as expressed, e.g. in literary works, the literati argued. Luckily, in recent decades the world of linguistics has opened up to non-linguists thanks to the rise of the cognitive strand of research. As is well known, cognitivists (such as Langacker) no longer focus on morphology and syntax; instead they talk of the primacy of semantics, the role of metaphor, iconicity, and prototypical, non-discrete categorization. Admittedly, they have not gone much further than sentence-level research, but what counts most in their approach is ‘‘studying specific linguistic facts in detail’’ (p. 11), which is a perspective they apparently share with the Chomskyan tradition. What also counts is the reaching out towards non-linguists, and dropping of the notorious ‘‘linguistic imperialism’’, which has poisoned the interdisciplinary relations for decades. Graeme Ritchie’s book is not part of this new trend, and not surprisingly so, as adopting the cognitive assumptions could put the AI goals even further away. Indeed, he is mainly interested in structural properties of propositional and linguistic jokes (his generativewww.elsevier.com/locate/pragma Journal of Pragmatics 37 (2005) 961–965