The sources for this web site live at GitHub ModularWriting. I invite thinkers, teachers, and practitioners to review the arguments here and consider joining the conversation. Email me at stan@modularwriting.com to join the discussion and to contribute content.
Some of the questions that shape that discussion are:
Conventional thinking about modular information development suggests that it is quite recent (1961+) and confined in scope to technology-driven communications facilitated by computers and web servers. Superficially, the rise of AI (artificial intelligence) systems as content authors, consumers, and assemblers suggests that the prospects for modular information are bright.
I can't dispute that this conventional thinking is inaccurate, but I do wonder whether it is complete. The more I have worked with content strategy, DITA, microdocuments, chatbots, AI-driven content delivery, conversational AI, and self-assembling content, the more points of resonance I find with seemingly unrelated fields such as oral composition theory, transformational linguistics, and structuralism. I keep getting pulled back into discussions of Homer, Hesiod, Beowulf, schools of Roman and Renaissance rhetoric, fairy tales, the Old Irish Book of Leinster, the Roman breviary, Medieval chant, and American Southern Baptist homiletics. Successful practitioners of modular information development and delivery have been with us for a long time and have much to share. I wish to give voice to the ones with which I am familiar and invite others to contribute in all the areas where I am clueless -- notably the African, Asian, and South American traditions.
Do DITA and Beowulf share the same tradition? are they consistent structurally?
No - fundamentally DITA is expository whereas Beowulf is narrative. Beowulf relies upon the poet to shape traditional themes, formulae, and metrics whereas DITA relies upon XML parsers and DTDs to govern how and where modules can be referenced. DITA currently produces static deliverables whereas the poet delivered many variant renditions of the Beowulf saga, each session being equally "original" and responsive to the interactions with the audience.
Are DITA and Beowulf congruent structurally? Do they operate within the same modular, transformational framework, but differ by in the mechanisms and traditions that they employ?
Yes - I would argue that the Odyssey, Iliad, Beowulf, the Prudentius Breviary, OASIS DITA spec, and the AIMLtechnology Mitsuku chatbot share some common assumptions about what is required to build a successful modular information development framework.
Shared frameworks that talk about "deep structures" and "abstract components" are, by definition, rubbish if you want to understand how a particular implementation works -- say Southern Baptist homiletics. Structural frameworks in general and the following framework in particular are valuable only if you want to test connections between implementations and want to ask questions that span conventional boundaries. What skills do Homer, the Beowulf poet, Quintilian, Don Day, Joanne Hackos, Michael Priestley, and IBM Watson have in common? If, for example, the ability to imagine how instances of a "type" come together requires a developed imagination, what does that tell us about the need to assess and cultivate imagination in the communicators that we hire? How do program conversational AI systems accordingly?
Here goes. When I look at modular information development from Homer through to IBM Watson, I see a "stack" of sorts. I see some common ingredients for success.
From the bottom of the stack:
"Types"/"Themes": Successful modular information systems have a crisp sense of recurring "types" or "categories" of units/modules.
The Iliad and Odyssey have multiple instances of the narrative theme "messenger arriving at the war council with important news." The poet inserts differentiating details such as names, locations, and titles -- but the shape of the narrative theme behind each instance is very stable.
Similarly, behind any Information Mapping or DITA publication, we recognize Robert Horn's basic information types: procedure, process, principle, concept, structure, and fact.
Instances/variations of "types"/"themes": Successful modular information systems need to adapt the context-free types of themes that it keeps in its repository (its structuralist langue) to the immediate context (its structuralist parole).
My go-to scholars on the Oral Composition theory, Albert Lord and Vladimir Propp, developed methods to identify and map out variations between narrative themes. The fairy tale poet, for example, never repeats the same language in describing the abduction of the princess by the dragon. The poet inserts names, locations, descriptions, embellishments, and local color to contextualize each instance of the narrative theme.
Look at any 20 DITA reference topics on "Pre-requisites for Installation" and you'll find similar, recurring variations and contextual substitutions. Whereas the substitutions made by Homer cannot violate the dactyllic hexameter, the XML writer cannot insert a @conkeyref to a product name that violates the XML content model or reaches outside the scope of the root map.
It takes years of training and practice to produce the poem or the WebHelp collection that simultaneously leverages and skillfully disguises its underlying themes (be they narrative or expository).
Assembly logic for theme instances/variations: Successful modular information systems have rules (explicit or implicit) for assembling many instances of contextualized themes/types into a coherent deliverable.
The Roman breviary is a printed collection of readings, psalms, prayers, and songs from which Catholic clergy and laity assembled daily cycles of prayers ("the hours"). The exact order of readings, songs, psalms, and prayers for any morning "hour" (matins) was not included in the breviary, but was maintained by the abbot or Master of Prayer as a separate list of references to the various contents of the printed breviary. Over time, these ordo lists contained tips on how to further customize songs and prayers to the season or locale. Similarly, the story of the return of Odysseus from Troy in the Odyssey was one of many "return-from-war" epics (nostoi) familiar to Homer's audience and contained no plot surprises for his contemporary audience. The "delight" for listeners was in the way Homer does innovative things with the assembly of those themes. For example, in creating parallel journeys for the searching son (Telemachus) and the searched-for father (Odysseus), Homer arranges the themes to foreshadow or echo one another. Instantiating the same theme multiple times is secondary to the effect in the overall assembly of themes. The theme of "recognizing-the-returning-veteran-by-his-childhood-scar" appears in three variant instances in the final books of the Odyssey -- each instance illuminating a different aspect of Odysseus' connections with his family and home.
We should not measure the future potential for AIs to develop and deliver dynamic content by today's mediocre chatbots. The metadata associated with underlying customer support themes is pretty weak -- but it will get better. As will the ability of Machine Learning to augment the metadata and inferences used by AIs to assemble content into useful, interactive conversations. The content assembled dynamically by future AIs to help me book a trip or to rebalance my portfolio will not be wooden or mechanical. They will have a more rich context for delivering useful information to me because they would have access to every previous session with me. It's not unlike the poet returning to a hall for the fifth time. The poet remembers which themes seem to have evoked the best response and can re-assemble the themes on the fly for maximum effect.
Delivery/performance logic for assembled instances/variations: Successful modular information systems support delivery methods optimized for the type of assembled content.
As best we can tell, the earliest forms of Christian plainchant optimized the singing of psalms and devotions for small-ish communities of non-professional singers. There were probably a limited number of monophonic melodies to which the community could sing various psalms or prayers in the same metric phrasings. Adding, removing, or customizing specific lines to reflect some particular topic or seasonal theme would have been common. The lines of plainchant inserted dynamically into monastic chant had to conform to the delivery method -- prescriptive phrasings and monophonic melodies. The Beowulf poet assembled the final performance to conform to half-line, alliterative segments. The JSON-driven chatbot "composes" its interactive messages to a customer according to a combination of grammatical and logical rules. It attempts no dactyllic hexameter.
The rules by which the poet, writer, singer, or AI make final adjustments to the assembled modular content differ by delivery medium, but they reveal some common denominators as well -- repetition, transposition, inversion, rule transformation, substitution, and so on.
The following sections summarize the work of various thinkers who have been wrestling with these issues.
The oral poet makes no conscious effort to break the traditional phrases and incidents; he is forced by the rapidity of composition in performance to use these traditional elements. . . . His art consists not so much in learning through repetition the time-worn formulas as in the ability to compose and recompose the phrases for the idea of the moment.
- Pg 5.
We have been given a bare statement of the council theme, with which the song opens. The theme ends with the suggestion by the wise councilor that the Sultan send to Bosnia for Derdeliz Alija and the Bosnian armies, a suggestion which is accepted, and a messenger is sent with letter to the hero. This is the framework on which the singer will build. Although he thinks of the theme as a unit, it can be broken down into smaller parts; the receipt of the letter, the summoning of the council, and so forth. Yet these are subsidiary to the larger theme. They will be useful perhaps in other contexts later on, but the singer learns them first for use in the specific council of the specific song, with the appropriate names of people and places and their characteristics. - page 74f
Our concept of the "original", of the "song" simply makes no sense in oral tradition. To us it seems so basic, so logical, since we are brought up in a society in which writing has fixed the norm of a stable first creation in art, we feel that there must be an "original" for everything. . . . It follows, then, that we cannot correctly speak of a "variant," since there is no "original" to be varied! Yet songs are related to one another in varying degrees; not, however, in relationship of variant to original. - p 107
The act that the same song occurs attached to different heroes would seem to indicate that the story is more important than the historical hero to which it is attached. There is a close relationship between the hero and the tale, but with some tales at least the type of hero is more significant than the specific hero. It is convenient to group songs according to their try content, or thematic configurations, because songs seem to continue in spite of the particular historical hero; they are not connected irrevocably to any single hero. If we classify songs by their content, we find a number of well-defined categories: weddings, rescues, returns, and captures of cities. -p 128
A theme is a "group of ideas regularly used in telling a tale in the formulaic style of traditional song."
The word "morphology" means the study of forms. In botany, the term "morphology" means the study of the component parts of a plant, of their relationship to each other and to the whole -- in other words, the study of the plant's structure. . . . It is possible to make an examination of the forms of the tale which will be as exact as the morphology of organic formations.
- p xxiii
We are undertaking a comparison of the themes of these tales. For the sake of comparison we shall separate the component parts of fairy tales y special methods; and then, we shall make a comparison of the tales according to their components. The result will be a morphology (i.e. a description of the tale according to its component parts and relationship of these components to each other and to the whole.
= p 19
Propp developed a notation to categorize each recurring narrative "form" (A B C D) and variations of that "form" occurring in different versions of a fairy tale. Here is a diagram capturing variations in two "struggle-victory" tales. The first contains "forms" narrating the killing of an enemy without a fight, the second with a fight.
Expressed as a single "schema" for both "struggle-victory" tales, we see the central sequence of "forms" branch and merge.
Examining princess-abduction tales, Propp concludes, "the entire store of fairy tales ought to be examined as a chain of variants. Were we able to unfold the picture of transformations, it would be possible to satisfy ourselves that all of the tales given can be morpohologically deduced from the tales about the kidnapping of a princess by a dragon -- from that form which we are inclined to consider basic. - p114