A different approach to narrative analysis model, (over)simplified.
I promised to make a post about “my approach” to narrative analysis. When I say “mine” I mean it literally: a couple of years ago, a good friend of mine (who is a philosopher, with a knack for logical models and maths) and me (I’m sort of a CommSci major, with wide experience in cultural semiotics and directing narrative analysis thesis) were so frustrated by the finale of “Game of Thrones” that we started ranting, which was the natural way at the time.
What wasn’t natural was that we started veering towards the need of a different sort of analysis when talking about “serialized audiovisual narratives” (yeah, series). Because most of the tools you can find out there are:
Derived from literary analysis: where, in most cases, you have only one creator for the entire thing, while in series you have a lot of people involved in what you get on screen: writers, directors, showrunners, actors, editors… If you want to take the "autorial intent" route, you find that there are too many possible authors, each one with a different approach and ability to contribute. Also, there is the question of time, but I'll talk about this in a bit.
Influenced by psychology: which, I admit, is a lot of fun, but it is also speculative at best… because a character, by nature, has “no psychology” or “inner life”. We all love analyzing and overanalyzing this (it is sooooo much fun!) but in the end, in terms of achieving some sort of “objectivity”, it is really difficult to pinpoint and to get with some semblance of certainty. All we have to understand the characters are their actions, really.
Too abstract to generalize: I love narratology, really. The emphasis on characters being sort of “agents” that fulfill actions that are needed to propel the story forward is precious for many things and allows shifts in analysis that can be delicious (“What happens if you make your antagonist the actant?” might be my favorite). However, this tool set works better when analyzing written narratives, for specific scenes, and even when you could extrapolate it to compare or check out entire character trajectories… It is just not the right tool to do it.
Another factor that we considered was that, while books have an inner time that may or may not correspond to any external time logic, usually serialized audiovisual narratives are constrained by other sorts of time: episodes, seasons and such. Characters do not live in reality, but live in a “reality” that needs to create its own inner “sense”. As my coauthor would put it during one of our first discussions: “the main difference between being a human and being a character is that characters have a lot of ellipsis, while us humans have to go to the loo, pay taxes and do a lot of boring stuff”.
If you live in a “reality” where ellipsis are rules and not exceptions, every action counts. If an action is being shown, it is because it has relevance into building the plot. But how can we use this analyze the “probable trajectory” of a plot?
Our first proposal was centering the idea of actions over time. But… We needed something more. If we do not have psychological parameters, how can we observe the effects of actions over time in the narrative? We decided to center in relationships. Those are also “observable”: you can know if an action had an effect on a relationship between characters (usually, and we could also build a “valence” for those relationships: positive (strengthens the relationship) or “negative” (undermines the relationship).
We built this interaction scale, very modeled around the sort of actions you would find in a show like GoT, but that we felt can be easily extrapolated to other narratives:
Therefore, you could see how an action affected the relationship over time, and it also had to “build a trajectory” that had some internal logic in the internal world. A good narrative, we proposed, was not going to take a character from 0 to 100 in a blink, or that would break the verisimilitude of the whole world. Yup, we should remember that the last season of GoT inspired this.
What we were looking for was something as “neutral” as narratology, but that we could also transform into… TA-DA! A mathematical model. That was mostly the work of the other coauthors. They analyzed and programed all the interactions in the first three seasons of GoT… And well, it clicked, and allowed us to understand if a character arc was working in terms of verosimilitude, or not. You can find the whole paper here.
Even when I don’t use the math model for predictability, I usually go to the same basic principles when analyzing media: characters, actions, relationships and if that “makes sense” in the internal logic of the world. That is also why I tend to not invest myself deeply into “symbolic analysis”: it is fun, and it is great entertainment, but I’d bring up a couple of concepts borrowed from Umberto Eco: “obra aperta” (open work), as a finished and published work that becomes a playground for all the interpretations that the reader/spectator wants to superimpose over them (some of them “more accepted” than others, but none of them absolutely right or wrong, disquieting as it sounds) and “infinite semiosis”, the phenomenon through which we connect our references with one another and multiply their meanings. There is a real life complexity built into semiotic and symbolic analysis, in the purest sense of "complexity": too many moving parts, generating emerging qualities that have no clear causality. This is one of the most frequent criticisms received by these models in academia!
That's why, fellow tumblr citizens, I work the way I work... And I loved the idea of sharing this in here. It would probably be one of those posts with 2 notes, but I'm weirdly proud of this "toy" and I wanted to share with the class.
