Monday, August 14, 2017

Reactions to Schank

In my previous post,  I reviewed Roger Schank's Tell me a Story, New York (Scribener) 1990 (= NWU Press, 1995) as to the problem of how to construct narratives, what constituent pieces are, and especially how to challenge or critique them. Though in the previous post, I mentioned three examples of analysis by Schank, I mostly focused on Tawana Brawley and on Iran Air Flight 655 getting shot down by the USS Vincennes, leaving Canadian Olympic gold-medalist runner Ben Johnson aside.

In musing over my dialogue with Schank, I wanted to offer some preliminary stabilization of my thoughts in terms of some theses to historiography---especially in the aftermath of the political display of revisionist Southern supremacy history in Charlottesville. Some of these are issues that Schank is concerned about and some of them I do not take him to care about.

  1. Schank is right that story skeletons and their selection form an initial stance on the problem that even leads to the filtering of information or adaptation of memories. In fact, if Schank had the benefit of hindsight that we have, he could have pointed to the US Navy's explanation of scenario confirmation offered to the BBC in 2000 for the precise behavior that his story skeletons had predicted in terms of filtering incoming information.
  2. Schank correctly distinguishes between story, skeleton and gist. The story is the actual production of the human talker, targeting the listener. The skeleton is the summary of the basic telos and gives explanation patterns at hand (more of that probably in Schank's prior book on Explanation Patterns). The gist is the cake mix, to use an analogy from cooking, that can be turned into a variety of cakes that all share a distinct family resemblance. Knowledge about what my current guests like and what I have quickly at hand inform the actual execution.
  3. Schank's skeletons provide an easy ingress into his theory for content analysis. The arguments made and the supports advanced and the information elided (if we can reconstruct it) come from the telos of the skeletons that is truly most like a political stance.

  1. Because Schank does not do much with the evidence/story or fact/narrative distinction (possibly because his biases against expert systems and automated theorem proving), he misses out on the ability to postulate that facts are stories for a subculture that no longer can or wishes to treat these stories narratively. They are as if baked, and questioning them can be interpreted as a violation of the norms of the subculture. Even low level sensor readings have these stories behind them, and this holds true even in particle physics (cf Knorr-Czetina), where the individual detectors in a setup such as the CERN super-collider are attributed 'personality' (in a anaphoric sense) because of their non-interchangeable behaviors. Unlike edible cookies however, facts can be unbaked back into stories if there are anomalies, provided the research data is available. Thus, bad footnotes or page references ghost through the literature until some brave grad-student hunts them down and slays them. For many historical documents, that is not possible, and here the community finds the line drawn for it. 
  2. Schank has no locus for the social role of power in the success of narratives, because he does not distinguish the UN Security Council adequately from a US district court or the admiralty of the US Navy. Schank tries to rope this in while looking at the story expectations, but that is really a small aspect of the problem only. His use of made-up stories or divorcee self-reports is equally ill-suited to discover this, as in the first case everyone knows the story is not real (thus there are no real validity requirement or possibility) while in the second case our culture considers it flat-out rude to question people's divorce narratives. 
  3. Schank is handicapped by his focus on the genre newspaper articles and transcript of psychological experiment. He looks a bit at screenplays as well, but never clarifies how the fictional status of these works interacts with the realistic status of the newspaper article or the emotional stance of the transcribed individual.

No comments:

Post a Comment