From Medieval Monks to Machine Learning
In the scriptorium of a medieval monastery, a monk carefully divides sacred text into discrete units, marking each new segment with a pilcrow (¶).
Pilcrow.
I’d been searching for this since Mr. Anderson in high school.
“Not simply organizational, the pilcrow is an assertion of authority over divine revelation,” I read online somewhere. “A way of controlling how truth can be encountered and understood.”
So the paragraph, in its origin, was a pilcrow of revelation and control. In its origin, the paragraph was a means of packaging the infinite into digestible units, the skin of a sausage. What a neat and tidy definition of paragraphing—packaging the infinite into digestible units.
“In the Middle Ages, the practice of rubrication (type in red-ink) used a red pilcrow to indicate the beginning of a different train of thought within the author's narrative without paragraphs.”
***
“We want to report to you, today, on our pilot experiment in using the computer in the classroom to teach revision. Last fall, in our Introduction to Fiction class, we used some of the newest computer technology to model thinking and revising strategies for the writing of literary analysis.” (Cornell and Newton, 1988)
***
Ant-Man—as I found out later in the fine print at the bottom—zoomed through the air on a flying ant in a poster on the inside wall of the waiting area. Multicolored glitter trailed from the backside of the flying ant. Ant-man steered his dragon toward a candy store down the urban corridor charging outrageous prices.
Thick, dark coffee for free.
Ant-Man, saving the day, pointing the way for America, where every child gets Good and Plenties, and every paragraph has a topic sentence.
***
Paragraphs have heroic qualities, too, soldier ants supporting the Queen. Some people dream in paragraphs. Mr. Anderson was inspired by teaching the paragraph to seniors. He imbued the essay with heroic qualities. He had been a fighter pilot during the war.
***
“We needed first to liberate our students from their drafts. We needed an effective way to help them see their own texts as fluid and modifiable rather than frozen and permanent, so that they would begin to see revision as a creative process in thinking and writing. In other words, we hoped to lead our students to a redefinition of revision as a potentially radical reforming of meaning rather than a tinkering with surface.” (Cornell and Newton, 1988)
***
I woke Claude and prompted:
“This is a paragraph, Claude. You’ve just read my topic sentence. First, a paragraph must have a topic sentence directly stated or easily inferred. Second, a paragraph must have details, three of them to be precise. Third, a paragraph must be linked directly to a thesis statement, which is the superior of the paragraph.” Comment.
“This appears to be describing a highly prescriptive and rigid view of paragraph structure,” Claude responded.
Topic sentence plus details shows up less than 2% of the time in real-world texts represented in areas like journalism, popular materials, newsletters, research articles, novels, etc.
***
“The main idea of the paragraph is stated in the topic sentence. A good topic sentence does the following:
introduces the rest of the paragraph
contains both a topic and an opinion
is clear and easy to follow
does not include supporting details
engages the reader
For example:
Development of the Alaska oil fields created many problems for already-endangered wildlife.”
***
“The technology we used was a laptop computer (Zenith) and a Kodak Datashow projector. With these tools, we were able to project a computer image of a student paper on a large screen and, using word processing, alter it quickly and easily. Our pilot work with this technology has shown us that it is a useful and flexible tool but not yet the perfect one for achieving our goals. It shares some of the often noted limitations of computer aided instruction in writing. For one thing, if you are going to use computers to teach writing, you need computer competence in students and teachers as well as equipment, both of which can be expensive in time and money. For another thing, because you can focus on only as much of the text as can fit on a screen, it is difficult to get a sense of the relation between a part and the whole, unless you have a hard copy of the text in hand, too. As a result, there is a natural tendency to focus on surface and local changes…” (Cornell and Newton, 1988)
***
The persistence of the standard paragraph in classrooms reveals itself as both less and more than we bargained for since the days of the rubricated pilcrow, less natural, more constricting, yet also more profound in its implications. The good thing is, I suppose, as Cornell and Newton (1988) discovered, we can fit a whole paragraph on a screen and project the image. We can model what a paragraph is and how we can revise a paragraph.
***
“It appeared to us that, even though the class as a whole was extroverted and evinced a strong esprit de corps, neither group had much desire to collaborate with others on their writing. Several students in the control group explicitly rejected the suggestions of peers, relying heavily instead on our comments, ‘since,’ as one said, ‘you have the final say in the whole matter anyway.’” (Cornell and West, 1988)
***
The pilcrow and its modern offspring, the indented topic sentence, stand as a testament to humanity's struggle to organize and control meaning in texts, to disseminate meaning in an orderly and fair manner. The struggle never ends because meaning is equally persistent. It overfloweth our attempts at containment.
As we enter an era of machine-generated text, we might finally be ready to loosen our grip, to let meaning fly in unrestricted airspace. Here’s the irony. These natural language robots can’t function without paragraphs. The fact that we can break free of the pilcrow’s magical spell means we are more creative, more imaginative, more in touch, more real than any algorithm. Our ability to consciously break from conventional paragraph structure is actually a marker of deeper understanding rather than a rejection of organization itself.
What's interesting is how this medieval technology for controlling religious discourse became naturalized as "logical thinking" itself. The scholastic assumption that complex ideas must be broken down into standardized, sequential units of analysis still shapes how we teach writing, perhaps one reason too many of our children find it so hard to learn to write and try to avoid it if possible—and the popularity of robots.