5 Card Nancy and Panel Transitions

One of Scott McCloud’s more wacky inventions is the game Five Card Nancy which is based on the old comic strip Nancy. The basic premise of the game is that you can create lots of different (and fun) novel strips by combining random panels together. Scott recently posted an old collage he did that led to the game.

Of immediate note in his collage is that the sequence doesn’t exactly make much sense, despite some cohesion between the panels. I’d say that it may have a narrative structure (i.e. visual grammar), but no meaning (semantics).

In some cases though, the juxtaposed panels do make sense, but the global meaning does not. In linguistics (borrowed from math), we’d call this a “first-order Markov chain“, since only the units right next to each other have a connection. If a panel had a connection to two panels next to it, it’d become a “second-order chain”, etc…

Markov chains were the primary way that people thought about language’s grammar up until the 1950s, when Noam Chomsky then showed that grammar needed to account for connections farther than just countable individual word relationships (an approach I then applied to comics’ sequences).

Essentially, McCloud’s theory of panel relationships is a first-order Markov chain theory. It only looks at juxtaposed relationships. Interestingly, his Five Card Nancy game follows the same characteristic. Since players put down one panel at a time, it appears as though they are just making choices linearly. However, I’m guessing that the higher scoring combos are all ones that gel on a global scale, not just a local connection.

Also, the limitation of the panel transition viewpoint is really highlighted by McCloud’s Nancy collage. How can panel transitions be correct if only local connections make sense but ones further down the sequence do not? Though we may draw and read comics one panel at time, it doesn’t mean we don’t build or project a bigger structure in our minds beyond the linear relations.


  • on the contrary, I would say that McCloud's collage has a valid higher-order structure that only breaks down when you try to parse it semantically — very similar to Chomsky's "colorless green ideas sleep furiously," which is grammatical but semantically senseless.

    The collage looks and acts like a full-page comic story, with a very traditional structure (intro scene, introduction of problem, dialogue, journey to solve the problem, return home, denoument, punchline). It even holds up visually: McCloud was careful to choose panels that comprised tiers of equal width, and even included one panel with a date and one panel with the syndicate copyright tag.

    You claim that in this story "only local connections make sense." On the contrary, the individual panel-to-panel transitions are often very weak connections, but each panel does make a good deal of sense with respect to its function in the overall story (e.g. the wide panel of Nancy jumping rope, which encapsulates a whole journey, and the cole slaw patch, which implies a destination). The final panel looks exactly like a final panel should — indeed, the Nth panel generally looks like an Nth panel should. It's only the semantic connections that fail (which you indeed pointed out).

    So, I guess, I agree with you that Chomsky's theory of grammar is better than McCloud-the-theorist's. I just think that this example by McCloud-the-artist feels grammatical, which actually advances the argument for Chomsky, because only Chomsky's theory explains why it (almost) works.

  • Leigh, thanks for the comment. On further inspection of the strips, I actually wouldn't disagree with you. I'm not sure that the whole thing holds up with a narrative structure, but each tier seems to.

    Actually, "grammatical but not semantic" sequences are the basis for my last two experiments.

    Nice observations!

  • Write a Reply or Comment