This began as a comment on MattK’s post of restlessness about textons and scriptons. Since my comment sort of … expanded … I thought a post/trackback was in order, lest I crash the comment box. The relevant contextual posts are:
MattK – Restless about Textons and Scriptons
Matt Bowen – CodedAndRecoded
MattK – Wakey Wakey

My thoughts follow Francois’ comment.

Skimming through the Hayles piece – the stipple engraving example struck me [see paragraph 9]

Stipple engraving, although it is normally perceived by the reader as a continuous image, operates through the binary digital distinction of ink dot/no ink dot; here the scripton is the image and the ink dots are the textons.[4]

Does this really work as an example for texton to scripton? Aren’t we talking about two levels of scriptons here – micro and macro (collection of dots v. big picture)? 13 ways of looking at a blackbird…? There’s no configuration here, which seems to be necessary to the texton–>scripton dynamic?

The remainder of that paragraph (#9) certainly provides a lengthier explanation of texton and scripton than we see in Aarseth’s Cybertext, with various nods to the idea that whether something is texton or scripton depends on the reader (the browser, a person, a compiler, etc). Oddly, no mention whatsoever of the traversal function, which seems to be necessary third leg on this particular stool.

Her earlier “Point One: Electronic Hypertexts Are Dynamic Images” (paragraph 5) seems to be an attempt to recoup the “flickering signifier” and bring it into, perhaps, a larger discourse that appears more recently informed by the textual studies kind of work done by MattK, in line with McGann, Drucker, etc. She writes:

In the computer the signifier exists not as a durably inscribed flat mark but as a screenic image produced by layers of code precisely correlated through correspondence rules. Even when electronic hypertexts simulate the appearance of durably inscribed marks, they are transitory images that need to be constantly refreshed to give the illusion of stable endurance through time.

So, we do have the very real assertion of the screen here; the refresh rate reminding us that on the screened surface, image does abound (Francois’ nod towards a possible occularcentrism noted). But as I think I recall from Matt’s essay (so recently – and wrongly – reviewed as some sort of Humanities Computing fallacy), we also need to talk *around* the screen – and focus on the actual, functional computational boundaries and differences in text and image that still operate with or without a screen. Matt’s more recent work (hex editors and so forth) also emphasizes these various layers that are, in fact, marks – magnetic or otherwise – with real consequence.

So, I’m going to talk around the screen, and wonder aloud, without having read any of this recently (that’s my caveat), if Aarseth’s true purpose wasn’t so much in determining the multiple layers that comprise the computing experience (a materiality approach), but really perhaps an attempt to answer a terribly difficult conundrum when confronting any sort of interactive narrative – the relationship between author and reader. In other words, maybe we’re asking his model to answer a question it wasn’t intended to answer.

Given that we all know ludology’s secret passion for narratology (this is the dirty secret of our study – and for any intending to complain, I’m poking fun at myself), let’s suppose that the real uncertainty is that when examining the traditional models of readership relationships, such as Chatman, the easy split between author and reader is rather more complicated when the discourse of the story is not predetermined by the author. That is to say, in a hypertext/IF/game: though the story (the plot over time) may be established, and though the author – by providing the linking and configuration mechanisms (the game engine, as it were) – enforces a certain level of discourse (how the story is told), there remains the difficult middle-ground of choice and configuration in the eventual reception of the overall narrative (the scripton). How, in other words, does one account for the fact that a functional text output can vary reading to reading, beyond the normal expectations of interpretation, reader response, and so on (or, as Aarseth states: “Scriptons are not necessarily identical to what readers actually read, which is yet another entity”).

If we read the relationship of author to reader as established partly through the author’s shaping of the story through discourse, then how does one account for a readerly return to the discourse in the instance of configurable/interactive narrative. What the rather vague “traversal function” provides is a term that describes the interaction between the reader and the discourse-engine provided by the hypertext writer/creator that eventually leads to the final output text – the scripton. The traversal function may not be so much a material consideration, but a relationship consideration – the engine of possible configurations that aid the reader in the creation of a scripton.

I don’t think it’s meant to account for the multiple layers of textual construction process (read: materiality), as Hayles seems to suggest in her interpretation of it (though maybe it should), and as MattK was hoping it would, but rather provide a gloss on a rather thorny issue – a question of who has more control, author or reader. This is not a new problem in the theory of hypertext, which is covered quite well by Marie-Laure Ryan when she points out that early proponents of hypertext might have better focused on “controlled freedom” rather than “unbridled license.”

When we’re talking about a print edition of, to take a random example, The Great Gatsby, we can talk about textual variants, we can talk about textual history, but ultimately, we talk about an edition of a text. You and I can both point to the same page and read a quotation together, by virtue of the fact that the role of narrative participants is (at least to a reasonable degree) established and maintained via a completed text.


Chatman’s model, in the image above, shows a singular flow: F. Scott Fitzgerald -> Nick -> …story of Jay G… –> Implied Reader (Us, idealized). I don’t remember an explicit narratee in the story. The “story” begins with Jay Gatsby’s early life, continues through this rise to success, his quest for Daisy, his death, and finishes with Nick’s observations. The “discourse” arranges the story so that Nick frames the tale; we don’t discover Gatsby’s “secret” until the end, nor do we understand Daisy’s selfishness until the end either. The establishment of this order shapes our sympathy for the characters, our understanding of character motives, and so forth. By providing a complete narrative text, with no room for configuration (as we would understand it in a computation sense), the flow of narrative follows Chatman’s model, from Implied Author to Implied Reader.

Since, by virtue of the nature of IF/hypertext, the discourse is not set completely (though it is set in part), that means that the reader – in controlled and limited ways – participates in the establishment of the discourse, either through asking questions (such as in Aarseth’s example of Deadline), or by clicking (think Joyce’s Afternoon: a story), or by piecing together various footnotes (think, House of Leaves). To varying degrees (always dependent on the mechanisms in place), the reader has some limited flexibility in choosing the discourse, based on how the broader delivery engine is built. Once you introduce a measure of configuration in a text, like we see in Aarseth’s cybertextual examples, some sort of function has to be described to account for textual variants developed in the process of reading/playing a work.

All this to say, then, that the texton –> traversal function –> scripton question may well be one of narrative participation (author/reader), rather than one of textual materiality. Not that, of course, those are mutually exclusive entities, but I suspect that Aarseth might think that they are. I realize I’m playing my own sort of intentionality game here, but this line from Aarseth leads me to believe that the material considerations are to the side (although not forgotten):

(1) a text cannot operate independently of some material medium, and this influences its behavior,

I don’t think that his model is set up to account for materiality, but rather that this line establishes a nod that materiality is, indeed, a factor. I’ll need to revisit the pages in question so I consider the full context on this one…

This may be an aside, or maybe not, but I’m trying to work in another consideration for the role of the author. As Montfort has argued (Twisty Little Passages) in regards to IF, the function of the riddle establishes a dynamic that provides an agreement between riddler and riddlee. While there’s certainly some sort of agreement in place in any literary exchange, the riddle does address another particularly thorny issue in literary studies: intentional fallacy. In IF and many sorts of games, solving the puzzle/game is a rather obvious intention on the part of the creator.

James Phelan gave a talk at UMD a little while ago on the Implied Author that in some respects sought to address the question of intentional fallacy. Note to self: dig up notes.


11 Responses to The Traversal Function

  1. Matt K. says:


    Recentering the terms of the debate from material layers to author/reader dynamics is a brilliant intervention, and exactly what I was looking for when I wrote (in my original post) “I’m not confident that my understanding of them is as rigorous and nuanced as Espen would want to insist.” This is very helpful, and I certainly hope it finds its way into your own work in some form.

    I hope to have some work in progress to send around by the end of next week. You have an attachment in your future.

  2. Jason says:

    Thanks, Matt, for posing the original question, which helped me work on this thorny issue of narrative relationships from a different angle than I had been (somewhat less successfully) pursuing.

    Looking forward to seeing the attachment.

  3. Matt Bowen says:

    I agree with MGK that the recentering is useful. However, I’m still not entirely soothed. My problem is that I cannot nail down what /is/ a texton when discussing a piece of interesting hypertext. Using a blog post for example: I markup my post in a simplified markup using [tags] that don’t resemble HTML. This is saved in a flatfile and then served as HTML which is then rendered. In a sense, both the flatfile and the HTML that the reader eventually sees are textons for the same eventual scripton. However, Aarseth doesn’t really give me a way to talk about this problem, and I really think the problem exists. Maybe I’m missing the point?

  4. Jason says:

    MattB, I share your discomfort. I don’t think Aarseth’s model covers it, and I don’t think it can. While useful in the contexts of discussing the problem of shared creation of a narrative’s discourse, I’m not sure it is built to handle the materiality of a text. Though, it’s clear that Aarseth views this as a problem, which you can see when he voices his concerns about ‘surface only’ reading in his discussion of a semiotic approach to games. I think the relevant discussion is on page 48? I came across it when I was writing this weekend…

    Point being, the problem still exists, but I think the solution is elsewhere. Which means that your wiki post remains quite useful in detailing the various levels so that it *can* be talked about.

  5. Matt Bowen says:

    I guess I just was hoping against hope that with enough subscripts and caveats one could weld the necessary changes onto Aarseth’s framework to do the talking necessary. I guess my next question is: where would one look for the necessary tools to do the kind of talking MGK was talking about?

  6. It is a bit strange to enter this discussion, because it feels more like I am one of the boys trying to figure out what this Aarseth guy could have meant. “I” wrote that stuff about 15 years ago, and haven’t given much thought to it since 1995 or so.

    There is actually more discussion about materiality, authorship etc in my essay in Landdow’s Hyper/Text/Theory, and hopefully also useful elaborations on the scripton/texton model. Back then, I made the big mistake of not repeating most of the points from the article in my book. Seriously, a big, big mistake. Big. Very big. I wish I could sue myself.

    Anyway, to add my two øre, Here goes:

    Yes, I try to deal with it, by constructing a model that is broad enough to account for texts in any medium. By taking a step back, and describe what is common for texts independent of material medium, we can better see what role materiality does play. This is not ignoring materiality, as some critics have claimed, but rather an attempt to avoid privileging any one particular material medium, and setting up an instrument that will allow us to describe the significance of a specific material consequence in a more nuanced way.

    Hypertext as scripton/texton:
    Hypertext could be defined as a text where the textons are identical to the scriptons. (The internal fragments equal the external fragments.)
    In the case of an HTML file, or the text I am writing here, the markup would be part of the traversal function. With regard to plain link-chunk hypertext, the S/T/T model is clearly overkill, but not (to me at least) problematic.

    Hope this helps some, and if not, it is high time for a better model!

  7. marc says:

    I think a good place to start refocusing this discussion, then, would be to situate it in a context similar to that proposed by Marie-Laure Ryan, whose recent collection of essays, Narrative Across Media, is essential reading.

    In describing the need for transmedial narratology (like Aarseth), Ryan dismisses both the “exclusive language-centered approach” (narrative is an act of storytelling passed from narrator to narratee) and the “inclusive language centered approach” to narratology (i.e. narrator –> narratee exchange mapped onto aural/visual means of communication– film, painting, etc.; adherence to the author – implied author – narrator – naratee – implied spectator – spectator dynamic is correlative to status as a “narrative”), claiming that each , in their own way, limit the scope of the discussion of both the centralities of narrative and, obviously, the role that a medium has (if any) in shaping the narrative.

    But she is also quick to dismiss what she calls the “doctrine of radical medial relativism”, which views media as “self-contained systems of signs, and their resources as incommensurable with the resources of other media”. “Just as two languages cannot convey the same semantic values under the doctrine of linguistic relativism”, she states, “two different media cannot convey similar meanings or use similar devices under the doctrine of medial relativism”. True indeed, even if it’s nothing more than a pointer to rampant remediation and the need under this model to have different narratological grammars for different media.

    Ryan’s proposal for transmedial narratology rests, then, entirely on the notion that even though narrative requires signifier and signified, any theory of transmedial narratology rests entirely on the signified. In other words, narrative is a mental image “which can be isolated from the stimuli that trigger its construction”. She gives the example of our innate human ability to form narratives around, say, a string of events in our life. Whatever stimuli causes this construction, be it the loss of a set of car keys, the reading of a paragraph or a live performance of the Tchaikovsky’s 1812 overture, often results for Ryan in the construction of a “narrative script”, which is more or less the evoked product of the coming into contact with something that, at the very least, “has” narrativity. To put this a bit differently, Ryan makes the distinction between “narratives” and things that “have narrativity”. A picture, in her view, can have narrativity (the trigger) which might subsequently produce a narrative (the mental construct).

    This isolation between signifier and signified is hugely important here because it, at least momentarily, places the construction of narrative outside of any mediation. In other words, perhaps we’re somewhat off in attempting to categorize the scripton/texton/ traversal dynamic in the terms of any sort of “power” relation (i.e. reader/author, medium/narrative, etc.). For Ryan, I think it could be argued that the property of “having” narrativity is most similar to the texton, the trigger, while the scripton, it’s manifestation, is entirely a mental construct. Note that the traversal function is not entirely necessary in this relationship– unless, of course, you consider human cognition as a traversal (which, in many ways, I think it is). Now obviously, talking about blog posts or IF in this regard does somewhat require a reframing of the traversal, but it’s not beyond comprehension that someone might look at the code for such a work or even a simple HTML page and explain it as a narrative (Geoffrey Rockwell did some nice work on whether code can be narrative a while back).

    I guess what I’m arguing is that perhaps the scripton-as-product model doesn’t really fit in with what I feel is the central truth to Ryan’s assertions– that, yes, the properties of a medium can have different impacts on the level of narrative script produced but that, as a whole, we construct narratives independent of medium. Words and images constitute highly separate evocational mechanisms, but they ultimately trigger the same process.

    And that’s precisely where I think we need to start– with process. The category of scripton is not simply a “product” of fucntional antecedents– it’s a process, continually defined by the mind of the user, and continually redefining the [Scripton = texton + traversal] model we already outlined. Process, not product, in both the hardware and the recipient.

  8. Jason says:

    Thanks Espen for the reference in H/T/T – I’ll have to go back and reread it (it’s been quite some time since I’ve dipped into that volume). Maybe you can include it in the 2nd edition of Cybertext 😉 As I mention above, I think that you do deal with materiality in some ways throughout, and I certainly grant that texton–>tf–>scripton provides a broad way to consider these issues.

    At the same time – and without the broader vision from your H/T/T essay in my mind, this might be an unfair response – I’m not sure I see that your texton–>tf–>scripton is overkill even with regards to simple html. I’m also a bit unclear how the texton and the scripton would both be the same in general hypertext (as you say: “The internal fragments equal the external fragments”), since (for example) the use of a different browser (Opera v. IE) could introduce differences. Or, I could use commenting in HTML to obscure aspects of a texton from the scripton.

    I suppose the question in any discussion of materiality is, at least for me, the same one I generally put forward in this earlier conversation over on MattK’s blog:

    “what kinds of theoretical questions would we want to answer that would require a certain levels of depth in the, as [MGK] put it so well, receding city lights? E.g. I know that in some pursuits, knowing what kind of ink Jane Austen’s first edition of Sense and Sensibility is scribed in, or the watermark of her paper, is absolutely crucial for a theoretical point. In other arguments, not so much …”

    In considering the material conditions of online games, for example, I might think that knowing how Asheron’s Call sends and receives data, and how that data generates images of land-masses and the items that populate that land-mass, is generally useful when considering how the use of player plug-ins exploit that data in ways that specifically and radically alter gameplay.

  9. Jason says:

    So now I know who has Ryan’s Narrative Across Media checked out of McKeldin library. 😉

    Sorry, Marc, that your comment only now just appeared. It was lost somehow in the approval system for MT, which I only looked at today (my past week being otherwise occupied with juggling two feverish loved ones battling the flu).

    Do the keys only “have narrativity” once they are lost?

    Seriously, I’ve written about four different responses, but I’m unsatisfied with all of them. Perhaps reading the Ryan might help (so maybe I can borrow it?). First of all, I’m a bit troubled by the notion that all things have narrativity. Does lyric poetry? Sure, I can make a narrative out of my reading experience, but that’s not the same thing. And that reading experience (reader response?) appears to be the driving force here, which is another thing that sort of concerns me, because it’s unclear from your summary (and it might be quite clear in her text) the degree to which she articulates how the signifier -and- the referent, and all their material attachments, influence the process of our mental-image signifieds.

    Because if we are still talking about process in that fashion, then I think I stand by what I articulated before. Because, as you argue “it’s a process, continually defined by the mind of the user,” but it’s also a process, structured in some ways by the medium, and defined by the mind of the (implied) author. I’m not sure how we get away from that…

  10. marc says:

    I agree, Jason, that “it’s also a process, structured in some ways by the medium”, but there is a real need for centering this debate outside of the medium, otherwise we restrict ourselves to insulated views of literature, film, etc, etc. I think Ryan’s real contribution is that she views narrative not as some binary (either wholly structured by the medium or not at all) but rather, like David Herman, as a gradient, where there are analogous constraints and freedoms present in all narrative domains. What I was trying to argue is not for a system where everything is narrative, nor one where reader-response dominates. Ryan is clear to set up these distinctions. It’s my fault that this wasn’t apparent (I wrote the comment in the midst of an excrutiatingly dull meeting using old notes). Ryan’s distinctions between 1) things that “have” narrativity and 2) things that “are” narratives is crucial as it’s only in finding the similarities between these two that any transmedial theory of narrative can occur. But your point about the lost keys is well-taken– what qualities does something need to have in order to have this narrativity?

    I don’t have the time to get into that, nor should I. Ryan spells it out far better than I could. Really, this whole thing is far from what I wanted to argue originally, namely that, if we are to look at, say, Queneau’s Cent mille milliards de poemes, I don’t feel quite right with saying that the scripton occurs in the final combination of the texton parts. I would argue that, instead, we should view the process of the combination as the scripton, not the “completed” text of one reading. But seriously, we’re so far from the probably original use of Aarseth’s terms that a lot of this feels somewhat forced.

    And btw– it’s not me that has Ryan’s book out. But you’re welcome to grab my copy whenever you like. Maybe next time we get together? Like, say, next year? 😉

  11. With all due respect to marc, some of us have never outlined the relation of between scription, texton and traversal function as an equation with scription on one side. See in particular the discussion at mgk (Matt Kirschenbaum’s blog) where Text is on one side of the equation. However I must admit that marc’s equation helps clarify for me that Text can be viewed as tf(scripton:texton). This is different from the representation put forth by jason where the traversal function mediates between the textron and the scription spaces. The text is a traversal of the space of relation between the textron to the scripton. One other way of approaching this model of textual relations is to consider the work Ryan on possible worlds alongside that of Don Idhe on the phenomenology of perception (see Existential Technics and his discussion of Necker Cubes).

    Narrative, like any mental image, is but one variation in a series. The series of possible variations is a textron. One possibility is a scripton. A traversal function would than map the paths of transforms.

    Jason’s introduction of the problematic of the materialtiy and intentionality underscores the dynamic. Consider a instance of a couple rounds of tic-tac-toe played on paper and played on screen. Both can keep present to the players previous rounds of games played (scriptons). In a human-computer interface context there are implementation decisions to be made about whether each round wipes the screen just as on paper players can decide on whether to use a fresh sheet of paper for every round or let the grids accumulate on a page. What I am trying to get at here is a simple assertion that affordance have an effect on how one will access the textron from any given set of scriptons. The textron in this example would be the sample space of all possible moves in the game. It now may become obvious that a text is not simply a set of scriptons figured as a subset of textons.

    Try mapping Narration to Scripton; Narrative to Texton; and Narrativity to Traversal Function. Somewhere somewhen somone wrote that human beings were “prone to narrativity” and that a “human self will project its self-making onto the world in order to
    generate stories from sequences and to break stories into recombinant sequences.”

    There can be no intentionality without an encounter with materiality. There can be no abstraction from materiality without multiple encounters with materiality. To think combinations requires the experience of variations.

Leave a Reply

Your email address will not be published. Required fields are marked *

Set your Twitter account name in your settings to use the TwitterBar Section.