The demise of the college essay is overdetermined. One need only spend 5 minutes on Substack to find dozens of pieces lamenting the death of the research paper, and traditional approaches to education more generally.
The knee-jerk conclusion of most humanists is that Large Language Models (LLM) will replace human invention. Since “Invention” is one of the five canons of rhetoric as practiced by the Greeks and Romans, I as a speech teacher remain haunted by this prospect. At the same time, I know that the fact of students relying on Gen-AI is a fait accompli.
So rather than add to the obnoxious chorus that proclaims, “the sky is falling,” I wanted to speculate about what may be an opportunity for communication pedagogy to change course. If we assume that human invention will henceforth be inextricably linked to digital technologies, then we as teachers can focus our efforts on other elements of the speechmaking process that are largely neglected.
For years, I stressed to my speech students that they should focus their attention on invention and arrangement, the first two of the rhetorical canons.1 I did this to minimize the stress they felt about elocution.2 By emphasizing that the content of the speech was– at least with regard to their grade– more important than the smoothness of their delivery, I hoped students would focus instead on the development of sound arguments. This strategy was at least partially successful, but it was a band aid that covered over the fact that students received surprisingly little training in how to actually deliver speeches.
Now Gen-AI can construct an argument and arrange it in whichever patterns we like in a few clicks. However, to present this information verbally still requires human memory and iterative practice. It still needs the canons of memory and delivery. Put differently, even once a student has a strong, well-structured argument, most of their work as a speaker is still ahead of them.
What follows is the strongest version of an argument I can surmise for why Gen-AI may actually be good for the public speaking course. But before I proceed to accentuate the positive, a few caveats are in order. First, I do not believe that the rhetorical canons can exist in isolation. I concur absolutely with the editorial in the journal Nature last month, which made the obvious connection between thinking and writing, warning that trying to separate the two is impossible.
Second, human memory is not merely the ability to recall information and spit it out. Rather, it is to know broadly and deeply enough to present confidently, and to defend, clarify, and expound on the fly. This ability to draw further connections as part of an hermeneutic invention3 is what distinguishes human memory from machine intelligence.
How we got here
Since many students’ first real instruction in public speaking does not occur until college, our curriculum tries to do an entire course of rhetorical instruction (which in the ancient world took 4-6 years to complete) into a single semester. This less than ideal scenario is what happens at the universities that still require speech. Many majors do not require it. Much of the time, English teachers, many of whom lack formal training in speech, are directed to integrate “oral communication” into their composition courses. Then administrators check the box for that learning objective— without bothering to look at what is actually being taught.
Rhetoric students in antiquity had a much more in-depth introductory course. It was called the Progymnasmata, a set of elementary exercises that students undertook progressively throughout their rhetorical education.4 It started with declamation (recitation) of speeches written by others across prominent speech genres of the day. Before teachers even asked their students to invent original arguments, those students had already mastered the art of memory and delivery and habituated themselves to important generic forms.
The program had many benefits, chief among them being that students had learned to manage their communication apprehension (CA) (i.e., speech anxiety) by the time they began the controversia stage of their training. (It was in this phase that students would start composing their own original arguments.) The condensed one-semester training plan expects students to overcome CA, while also inventing compelling arguments, and mastering delivery. It is completely unreasonable to expect this will all be done well.
So what we do is focus on invention and arrangement within narrow genres (organizational patterns) and downplay the role of style, memory, and delivery in the name of mitigating anxiety. But really, we do it because it is the path of least resistance.
What if we instead focused on two different canons (setting “style” aside for the moment): memory and delivery? What might this look like? We could take a hands off approach to Gen-AI, assuming students will use it to “invent” content regardless of what we say. At the same time, we would expect students to practice their delivery and become fluent enough with their speech content to deliver it from memory, or at least predominantly extemporaneously (with very limited notes— and certainly no slides).
I know, it sounds crazy. Keep in mind I am mostly just trying to play the role of provocateur.
If we Taught Memory
A curriculum that took memory seriously would militate against the rot at the heart of higher education today: the idea that finding information and copying it is the same as understanding and generating it. Why bother memorizing a sonnet or the details of a historical event when I can look it up whenever I want? This sentiment is widespread and continues to grow with the power and usability of our technology.
If we expected students to memorize their speeches, they would engage parts of their brain that aren’t often activated. More importantly, in doing so, they would innately draw new connections between ideas and develop a better understanding of their topics. Debate is such a valuable tool for learning because in having to defend an idea, you have to really know it. A debater, then, needs to memorize enough about their arguments to present confidently without having to refresh their understanding.
All the time we spend focusing on subjects like organizational patterns could be reallocated to mnemonic strategies, such as Cicero’s memory palace.5 In so doing, we would necessarily be encouraging imagination and creativity, two God terms in education. What most people don’t know is that these terms are relatively new. In the Middle Ages, a learned person would not be called “imaginative,” they would, as the Medieval scholar Mary Carruthers put it, “be said to have richly retentive memories, which they expressed in intricate reasoning and original discovery.”6
To “discover” is to invent. So what I’m suggesting is actually radical— albeit hardly new. To teach memory seriously, and in the spirit of the classical and Medieval traditions, would be a way to teaching invention! This is because of that hermeneutic circle I mentioned earlier.
If Delivery mattered like we say it does
Nearly all speech textbooks cite the psychologist Albert Mehrabian’s work to illustrate the importance of non-verbal communication. Mehrabian’s experiments looked at the channels that are most influential for receivers of communication. The famous 7-38-55 rule posits that the impression of a speaker comes primarily from their use of voice (38%); their bodily communication, to include facial expressions, eye behavior, and movement (55%); and semantics of the words chosen (7%).7
At the same time that we expect our students to treat this as a fact, we assess their speeches as if it weren’t true. Sure, delivery matters a bit but typically we reward those students who make substantial or “creative” arguments as long as they reach a minimum level of verbal fluency. So we’re not measuring their performance against what we are teaching them is really important, or at least not in proportion to how important we are telling them it is.8
We are setting students up to for failure by pretending that delivery doesn’t matter. Not only has the principle Mehrabian discovered largely held up, human voice and embodied non-verbal channels bring a humanity to communication; they enable us to share emotion in ways that a disembodied machine cannot. Especially if we cultivate an analog classroom environment on presentation days, this focus on delivery will reveal to students what is unique about face-to-face interactions, something they experience infrequently thanks to the ubiquity of smart phones and what we call “social” media.
So, if there is a silver lining to Gen AI for the speech course, it is this. The constraints of the contemporary course make it impossible for students to become proficient at all of the rhetorical canons in 1 semester. So if we must choose, perhaps the ubiquity of Gen AI will prompt us to re-assess where our limited time might be put to the best use.
Uncomfortable as I am with giving up on invention in the traditional sense, there is something to be said for tried and true practices like the progymnasmata. We should at least have the courage to ask hard questions about the efficacy of our curriculum. I believe there is a reasonable argument that going back to the basics may be the best way forward in these uncertain times.
The canons of rhetoric are most fully described in Book I of the Rhetorica ad Herennium, an instructional treatise from the Roman Republican period that is sometimes attributed to Cicero. Its true author cannot be known with any certainty.
The majority of fears among beginning speakers center around failure to execute either delivery (saying “um” incessantly) or memory (freezing up, forgetting everything I was going to say). See Lefebvre, Lefebvre, & Allen, “Training the Butterflies to Fly in Formation: Cataloguing Student Fears about Public Speaking,” Communication Education 67, no. 3 (2018): 348-362. doi:10.1080/03634523.2018.1468915
This is the generation of ideas that occurs via the process of interpretation. It is an iterative process that tacks back and forth between text(s) and context(s). Each new encounter of a text or idea becomes colored by previous interpretations and experiences. Right now machines aren’t capable of this, but they may get here sooner than we know.
The most well-known surviving set of these exercises dates to the 4th century CE and is attributed to a man called Aphthonius of Antioch. His treatise survives nearly in full.
For a fascinating history of memory studies in antiquity, see the seminal book on the subject: Frances Yates, The Art of Memory (Chicago: University of Chicago Press, 1966).
Mary Carruthers, The Book of Memory: A study of memory in medieval culture 2nd ed. (Cambridge, UK: Cambridge University Press, 2008), 4.
Albert Mehrabian, “Decoding of inconsistent communications,” Journal of Personality and Social Psychology 6, no. 1 (1967): 109-114. doi: 10.1037/h0024532
We have moved away from stressing delivery out of a concern for anxiety and depression among college students and because traditional measures of delivery are reinforce ableism. When mitigating CA became one of the primary objectives for the basic communication course and critical theory took over the humanities and social sciences, programs shifted their focus to make the course a coddled, “feel good” one rather than one that challenges students and forces them to adapt to real-world presentational constraints.