0
0
Linguistics and Humor
Linguistics is the study of language, its sounds, grammar and meaning. It is a multidisciplinary field that intersects with many other fields, including the humanities and social sciences. Its focus is on the role of language in human thought and behavior.
The linguistic humor theories discussed in this article are classified into three families. Each has its own specific characteristics. These include incongruity, hostility and release theories.
Incongruity theories
Some scholars have argued that humor involves a form of play with words. This theory is based on the notion that a joke has a set-up and a punch line. The set-up creates expectations and the punch line violates these expectations. The resulting dissonance can make the audience laugh.
These theories are not without their critics, but they do provide a useful framework for understanding the nature of humor. They also support recent neuroscience research, which suggests that areas of the brain involved in higher level cognitive thought are active during humor processing.
Linguists can use a variety of texts to examine humor, including Raskin’s primer of humor research and the Oxford Bibliographies Online. The latter provides a comprehensive collection of sources in multiple disciplines, such as psychology, anthropology, and communication. It also includes sections on linguistics, computational humor, and rhetoric. It also features a search function that allows users to look up specific terms and thinkers.
Hostility theories
When asked what traits they value in their spouses, many couples mention a sense of humor. Yet philosophers have said little about humor, and what they have said has been critical. This is surprising, since people everywhere seem to find humor amusing. Understanding the benefits of humor can help counter traditional objections to it, such as the Irrationality Objection. These theories focus on the cognitive aspects of humor and set aside social and emotional elements. They also assume that something malicious and potentially harmful must be involved in a joke for it to be humorous.
In contrast, hostility theories consider the effects of a joke on an individual’s emotional state. Jokes that ridicule members of a particular group are often seen as offensive, and may be appreciated for their incongruity or perceived as aggression. However, there is a “sweet spot” in which humor can be understood as playful and not aggressive. The challenge is to recognize that the jest is a joke before it causes an aggression response.
Release theories
Although humor is primarily studied by psychologists and social scientists, linguists also have an interest in it. One of the first studies of humor as an autonomous area of research was Victor Raskin’s Semantic Script Theory of Humor, published in 1985. His theory differentiates itself from incongruity theories and hostility theories by focusing on the linguistic characteristics of humor.
He argues that humor involves a twist of perspective. This shift allows us to see things differently than they really are. He argues that this occurs because of the use of semantic metaphors and metonymies in humorous texts.
Spencer’s hydraulic theory of laughter argues that nervous excitement and mental agitation generate energy that must be expended in some way. Laughter is an expressive outlet for this energy and relieves stress. This theory also explains why some people find certain things funny, even if they don’t share the same cultural context.
Computational models
Since the mid-1970s, humor research has been a recognized academic discipline with its own journals and conferences. While psychologists, sociologists, and literary scholars have long been concerned with the topic, linguistic approaches to humor are relatively new. Victor Raskin’s Semantic Mechanisms of Humor (1985) standardized the field by dividing humor theories into three families: incongruity, hostility, and release.
A computational approach to humor involves understanding the language and structure of humorous stimuli and generating humor accordingly. It has the advantage of being scalable to large corpora and allowing for more robust performance than manual methods. However, achieving a true computational sense of humor remains an AI-complete problem.
To improve humor detection, researchers have tried to include prosodic and multimodal features in their models. For example, the UR-FUNNY data set includes laughter markers and allows for the evaluation of humor in both spoken and written text. Moreover, a denoising autoencoding based pretraining has improved results compared to standard autoregressive training.
