Live Discourse Analysis

Repost if your connections would find this interesting:

This semester, my encounters in fieldwork work shift the interpretive sands around me. I now conduct discourse analysis from the point of encounter rather than representation. It is a methodological shift that leads to profoundly different writing. But why? Why do I find encounter so valuable in a field of scholaly writing so committed to representation?

In an of AI, where GenAI tools can not only write for us, but search out, “read,” and summarize text for us, I see many people outsourcing the encounter of reading – of stumbling on paragraphs, sentences, ideas that move our thinking in unexpected ways. It is so easy, so frictionless, for scholars to use AI-assisted search programs to solve the enduring problem of find literature gaps. But the problem with outsourcing such tasks is it restricts the questions asked to the gaps AI returns. Yet some of the most innovative or exciting scholarship does not pinpoint gaps in the literature, it asks the field questions the field never thought to ask. Gap-filling is a logical extension of accepted premises in the field. Innovative scholarship points conversation at the premises themselves.

For that reason, I have fallen out of love with the 8,000 – 10,000 word empirical research article. It is too repetitive. Too verbose. Too much effort to tread back over well-tread ground. I now prefer shorter articles: essays, commentaries, conceptual pieces. These papers assume the intelligence of their reader by declining to define every term using the same definitions and citations used by all. They do not state the obvious. Or, better, they state the obvious and tease out their less-than-obvious implications.

As I think through this accelerating crisis of academic writing, peer review, and the editors caught in-between, I think about a LinkedIn post that to me exemplifies this problem. I will share the post piece-by-piece in bold, followed up by my own Live Discourse Analytic reading.

What is happening to the journal peer-review process??? 🤯🤯🤯

The first thing that jumps out is an emoji triplet indicating that a mind is blown coupled with a triplet of question marks. We see the voice of someone with a lot of energy, somebody surprised or perplexed by something, or perhaps even apoplectic about something.

In this headline, we see an approximation of Labov and Waletsky’s abstract — prefacing work that sets the stakes and tellability of a forthcoming story (or, as Alexandra Georgakopoulou or Michele Koven often calls it on social media, “small stories”). Put differently, the author situates herself within a broader discourse that presupposes (1) her experience is new or unexpected, (2) she is upset about this change of affairs, and (3) the length and comments on the contact [?] hint that something is not right. that if this was a person who wrote an honesty [honest?] review with their attention, then it would be shocking if it looked like this. More importantly she opens her small story with a broader discursive question which suggests at least that face value familiarity with a discourse around a rapidly changing conditions within the academic peer review process

We just received the reviews for a paper submitted to a high-impact Elsevier journal, and I am shocked: 36 pages of reviewer comments from only 2 reviewers!!!

I like the way that she frames this shock by inviting a reader to calculate: 36 pages of reviewer comments divided by two reviewers equals 18 pages per reviewer.  We don’t know the author’s field, and we do not know an ideal or expected page count for a review. But we can guess, from the telling, that it is much less than 18 pages. It also foreshadows who the author wants to hold accountable to this aberrant review – Elsevier and its entailing impact factor ideology. In other words, it presupposes that such reviewer behavior may exist, but not within the standards of a high-impact journal.

Much of it is filled with jargon, repetition, and irrelevant critiques.

Gricean framework of pragmatics, we can see big time flouting of at least three conversational maxims — the idea that one’s contribution should say what is necessary and no more (maxim of quantity), that it should be true (maxim of quality), that it should be relevant (maxim of relation), and should communicate clearly in a way that avoids confusion (maxim of manner). 

Reading this from my own experience and the author’s language, it is the overabundance of these three substances rather than their presence that most contributes to the author’s shock. In other words, it is within the genre of the review to, at times, include jargon, repetition, or even some irrelevant critiques. Yet when we look at the grammar of “much of it is filled with,” we can also speculate a review document where these three rhetorical substances are not only excessive, but evenly distributed — not total, but also omnipresent to the point that the author cannot articulate its frequence through more precise spatial or volumetric formulations.

I fully appreciate rigorous peer review, and detailed feedback can absolutely improve a paper. But when 80–90% of the comments are difficult to connect to the actual manuscript or simply do not make sense, something is clearly going wrong.

Going back to the abstract of the post, we can read this first sentence and confirm (or at least suspect) the author’s participation within a community of authors and peer reviewers. Or, at the very least, they are socialized past a threshold to hold a belief about the community and its expected or idealized practices.

The unstated information, but a piece of information consonant with my own experience of peer review small stories is that the author suspects uncritical AI usage in the generation of the peer review. This is a big problem for Grice’s maxims. In Gricean terms, the flouting or violation of maxims is not a failure of communication, but communicative in its defiance. To deviate from expected pragmatics of a given situation is to produce a stance on that situation. By generating a peer review report, AI flouts maxims without rhyme or reason attached to a speaker. Or, if we assume AI does not carry communicative intent, then AI does not flout maxims, rather it simulates the document we call a “peer review report” while remaining ignorant of the pragmatic maxims that govern interactions between human peer reviewers and human authors. 

One maxim would be of quantity. The author already asserts an 18-page review to flout this maxim via its excess. Were the reviewer a human, we can envision someone (let’s call them Reviewer #2), who flouts the maxim of quantity to express their distaste for a manuscript and its plethora of perceived conceptual, methodological, or stylistic shortcomings. However, to accord to a maxim of quantity as a human reviewer would involve judgment and restraint – to focus on the 20% of problems that would make a manuscript 80% stronger and leave the rest alone. This is not just because a human reviewer wants to be cooperative, but because lengthy reviews carry costs of attention and time – valuable resources many academics would prefer to apply elsewhere. An AI has no such limitations nor compunctions. To an AI reviewer, all errors are easy to point out and itemize, with no triage necessary. The problem? Such triage is essential to make a peer review into a coherent story. When the author suggests that “80-90%” of the comments are hard to connect to the manuscript, I sense more than mere irrelevance. I sense a flat structure of comments with no cohesive review narrative.

And while Claude pushed me by saying we cannot confirm AI usage, or a human could also produce an itemized list of comments without a review story, or prompt AI to make a more triaged review, then the point still stands. If a human prompted AI in a way that supported commentary triage and cohesion, the author would not be complaining on LinkedIn at this time. In that way, such review griping is less about AI and more about the idle human that does not critically engage with AI output and therefore takes AI’s default information-driven textual structuration. But AI’s default is problematic. It does not take human limitations of time, attention, or incentives for parsimony into account.

Peer review is already a demanding process for authors, reviewers, and editors. But when reviews are incoherent, irrelevant, or AI-generated, they should not simply be forwarded to the authors without scrutiny!

This final sentence provides warrant for our prior speculation on suspected AI-generation. What I find striking is the passive voice that frames responsibility — “saying [long unhelpful reviews] should not simply be forwarded to the authors without scrutiny!” A poetically minded reader can duplicate the hidden subject as both a reviewer and an editor, yet we cannot know with certainty who the author accuses of neglecting scrutiny. What is more interesting is how the specter of AI, and its co-constructed human users asleep at the wheel, manifest in an agentless sentence. This is entwined with the author’s struggle to interpret a text that, through her eyes, lacks traces of human engagement — echoes behind a representation that echolocate back to a thinking, breathing, feeling being behind the screen and LLM algorithm. Back when we counted on human-led peer review, we took those echoes for granted — the unevenness that signals a language user who defies probabilistic next-token guesses to surprise us with typos, inclusions, or omissions, or misunderstandings that trace the contours of an only human sense-maker.

Yet I linger on the word scrutiny. In a sociolinguistic sense, scrutiny stands in a paradigmatic relation with comparable words that can slot in this opening of discourse: words like “without care,” “without attention,” “without checking,” “without verifying.” The first two indicate an ethic of human connection. The latter two indicate an ethic of quality control. In fact, as I re-read the author’s post now, I wonder beyond her shock or confusion at the state of peer review and instead wonder if her gripes rest in quality control, or something deeper. Would the author be satisfied with a different AI-generated review as long as the human quality control nodes in the AI-informational circuit did their jobs? Or are they mourning the loss something else fundamental, less namable, something beyond language as a technology that conveys information. Does their complaint rest with the leeching of feeling or traces of human triage? Or with a failure to provide coherent information?

I read this post and her shock as this realization of a lack of human touch. But what if it’s not? That is the one thing I cannot read into this text and cannot know from the author. It is an ambiguity left hanging in the air as the post wraps up.

But it remains an important question. In this postdigital AI turn in scholarship and schooling, what future do we want? A human-as-verification-device future prioritizes information over expression, and positions humans as quality control nodes in an AI-directed informational assembly line. Language in this future selects for transparency, one-to-one form-meaning mappings, and the eradication of poetic ambiguity. I hate this future. I hate the idea of a future where productivity becomes not only a God, but an operating system – where creating more output supersedes anything I might feel about that output.

So I choose encounter. Encounters cannot be sped up, cannot be automated, cannot be simulated. Encounter is a space I find to be, if not human, then life-affirming. AI does not encounter. AI gorges itself on data, reads and produces patterns. Encounter sits in a space of uncertainty, partiality, otherwise, sense-making what is becoming. In that space, patterns form, break apart, and re-form into something unexpected, unfixed, unfinished. And crucially, writing from an encounter’s-eye view resists the information economy of language – writing from a space of present-tense becoming rather than past-tense authority – a lamination of time and space where surprise engulfs certainty. The problem with AI is not its hallucinations or misinformation – it is its certainty in saying things that may or may not be true. Encounter suspends disbelief in the certain and true to welcome the uncertain and possible. And that is the space I want to live in.