Friday, January 04, 2019

Six Plots, or, Everything is Up or Down

Some ideas just don't seem to go away.

I wrote earlier about the idea that all stories seem to fall into one of a small number of basic plot types. Depending on your source, the number is anywhere from half a dozen to a dozen. The best list I could find gave seven.

Lately, some researchers have been trying to classify plots quantitatively, in fiction and in movies. The fiction project was carried out at Washington State U. and later at U. of Vermont, while the film project is out of the Turing Institute. However, they used a similar, and in my opinion similarly misleading, methodology.

The technique is called sentiment analysis. A unit of analysis is chosen: the fiction project used individual words, while the film project used sentences. Each word or sentence is scored as emotionally "positive" or "negative." An emotional score can then be calculated for each page, chapter, scene, or X-minute interval, and these can be used to generate a graph or trajectory for the whole work. The fiction article shows several example graphs based on classic novels or stories. The movie article also attempts to measure "success" of the different story types, financially (box office returns) and artistically (Oscars won).

Speaking as a professional quantitative scientist, this is one of the most ridiculous pieces of reductionism I have ever seen. Critique from the bottom up...

Context-free scoring. Words may be positive, negative, or neutral, in different people's opinions. I remember being dumbfounded that apparently the word "moist" is considered disgusting by a majority of English-speakers. I find it mostly neutral, and in some contexts positive: a "moist, fluffy cake" beats a "dry, crumbly cake" every time.

I have a very hard time evaluating words or phrases out of context. Consider: "The test was positive." For what? Cancer? Pregnancy? A planned pregnancy, or an unwanted one? Or were we testing for alien life on an unexplored planet?

Words aren't everything. In movies particularly, there can be a lot of emotional input that features no words at all. Jurassic Park, the first and best of the series: The male biologist almost fainting at his first glimpse of living dinosaurs. Or watch the last couple minutes of the first Hobbit movie.

It's particularly weird to apply this assumption to a play such as Romeo and Juliet (featured in the fiction article linked above). Play scripts as we know them from Shakespeare are made up almost entirely of dialogue, since his stage directions were minimal and optional. But a real stage production, or film adaptation for that matter, is much, much more than just words. Hamlet's famous monologue took on a whole different life in Branagh's film production: Hamlet knew or suspected that he was being watched, which made the whole thing a performance for the benefit of his enemies and not the piece of morbid introspection it's usually seen as.

Words can conflict. Let's say we (characters in a novel or film) did just discover alien, sentient life for the first time ever. One of us is delighted: I get to use my hitherto completely theoretical xenolinguistic skills! Another is horrified: My humano-centric universe view is shattered! Another responds with a calm assessment of possible threat.

Any way you look at it, this discovery represents a major turning point in the story, one that creates excitement and anticipation in the reader. But as the different characters each use words or sentences reflecting their own response, the overall score may end up being emotionally close to neutral. In fact, much of the interest for the reader/viewer may be in the conflict or contrast among the characters as much as the excitement of the discovery itself. Back to Jurassic Park: Jeff Goldblum's character was using a lot of downer words about the whole idea, yet the two biologists were ecstatic. Wonder how those scenes would score?

Emotions don't just go up or down. What if we applied this kind of analysis to more than one dimension? For example, if the graph for Shelley's Frankenstein had a second emotional axis representing Frankenstein's scientific curiosity-- which intensifies and diminishes throughout the novel, but not necessarily in step with his fears and griefs around his relationship with the Creature. Wouldn't that create a much more complex and interesting view of the story?

Emotion is an emergent property. A story or scene induces an emotional response in the reader or viewer. (Or not. But let's assume there's some effect, otherwise the whole exercise becomes pretty pointless.) This response may or may not be the same as the emotion or emotions portrayed in the scene, or experienced by the characters in the scene. For example, the Jewish characters in Schindler's List express hope and confidence at points that can only strike a viewer as horrifying: as they're being shipped off to the camp, one woman remarks happily "The war is over for us, we are workers now!"

I respect the attempt to impose some sort of quantitative rigor on this subject. I particularly liked that the movie article thought it was important to evaluate success of the different plot types-- although, in the capitalistic and cultural framework of modern movie-making, I don't find the measures they used at all convincing. (And no, I don't have a better suggestion.)

But I just can't take this analysis seriously. Part of the problem is that it really is content-neutral. Referring to the seven-plot list from Wikipedia: at least four out of the seven fall into the sentiment-analysis category of "rags to riches," i.e. generally an upward movement. Yet, "Overcoming the Monster" stories feel definitely different to us than "Quest" stories. The Monster and the Journey are culturally important symbols, or symbol-packages, that influence how we perceive and respond to a story, beyond the emotional rise and fall.

Available! High-Voltage Lines, Knocking from Inside

No comments: