This is the abstract of a TULCON14 presentation.
Haley Hsu, Elsi Kaiser (University of Southern California)
We report a corpus analysis on the linguistic complexity and humor types in television sitcoms and cartoons. Animated sitcoms for adults (e.g. The Simpsons, Rick and Morty) are very popular, having surpassed the lifespans of live-action sitcoms (e.g. Friends, Modern Family). To shed light
on this, we investigated whether the humor and linguistic complexity of animated sitcoms more closely resemble children’s cartoons (e.g. Arthur, Kim Possible) or live-action sitcoms for adults.
We test two broad hypotheses: An audience-oriented account of humor and linguistic complexity leads us to expect that animated sitcoms will resemble live-action sitcoms (both are for adults) in terms of humor types and text complexity (predicted to be higher than in children’s cartoons). In contrast, a simplicity-based account predicts that the humor and text complexity in
animated sitcoms resembles children’s animated cartoons and that this may underlie the popularity of animated sitcoms. Given findings that children’s cartoons use simple language (Poštič 2015, Malhotra 2019), and adults experience an increased processing load with complex sentences (e.g.
Gibson 1998), the popular genre of animated sitcoms is predicted to resemble children’s cartoons more than live-action sitcoms, in terms of humor and text complexity. We build on work on humor taxonomies (e.g. Dynel 2009, Long & Graesser 1988, Attardo 2017), acquisition of humor (e.g.
Airenti 2016, Glenwright & Pexman 2010), discourse coherence (Hobbs 1979, Prasad & Webber 2009), and sentence and text complexity (e.g. Gibson 1998, McNamara et al. 2010, Frantz 2015).
Corpus. We compared 3 genres: (i) animated sitcoms for adults (e.g. The Simpsons), (ii) live-action sitcoms for adults (e.g. Friends), and (iii) animated cartoons for children (e.g. Arthur). Our corpus consisted of 29 episodes (97710 words) from 8 different live-action sitcoms, 8 animated
sitcoms, 13 children’s cartoon shows (we included more to compensate for short duration). We only analyzed linguistically-expressed jokes, not purely visual ‘gags’.
Humor type and joke structure: For each joke, we annotated (a) its humor type, building on categories in prior work (e.g. sarcasm, puns, absurd humor, jokes based on lying) and (b) its linguistic structure (did it have a set-up in addition to the punchline, did it include humor after the
Results. Frequency of different types of humor: Animated sitcoms resemble children’s cartoons more than live-action sitcoms. E.g., jokes based on lying are twice as frequent in live-action sitcoms (28 jokes) than animated sitcoms (13) or children’s cartoons (12) (p’s < .02, chi-squared). Similar differences occur with other types e.g. absurd jokes (more common in animated sitcoms, children’s cartoons). As for jokes’ linguistic structure, we find this ranking (most to least complex): live-action>animated>children’s cartoons (differences shown by chi-squared).
For text complexity analyses, we used Coh-Metrix (http://cohmetrix.com/), a computational tool that provides text complexity and readability measures (e.g. McNamara et al. 2010). We used Coh-Metrix to quantify the Flesch readability score, lexical diversity, syntactic complexity of each
joke, as well as the age of acquisition of the words in the joke. Results: Overall, multiple text complexity measures suggest that animated sitcoms have linguistic properties that more closely resemble children’s cartoons than live-action sitcoms (e.g. Flesch readability: animated sitcoms
87.12, children’s cartoons 87.42, live-action sitcoms 90.75. Interestingly, animated sitcoms and cartoons both show higher lexical diversity, on multiple metrics, than live-action sitcoms.)
Conclusions: Our analyses indicate that animated sitcoms, though intended for an adult audience, more closely resemble children’s cartoons than live-action sitcoms (also intended for adults). The resemblances between the linguistic properties of jokes in animated sitcoms and cartoons surfaced in multiple analyses, including humor type, joke structure and measures of text complexity. Our research suggests that the linguistic properties of humor in different genres deserve more research, especially in an age where television is a dominant form of entertainment.