From Prompt Engineering to Prompt Communication

Something interesting is happening in the prompt literacy literature. Three papers — published between 2024 and 2025, collectively cited hundreds of times — are converging on the same insight without realizing it. They're all describing communication. They just don't know it yet.

The 589-Citation Taxonomy That's Actually Rhetoric

Walter (2024), in “Embracing the Future of AI in the Classroom,” built a seven-level taxonomy of prompt engineering techniques. With 589 citations, it's become the anchor paper for anyone studying prompt literacy in education. The taxonomy looks like this:

1. Input-Output Prompting (IOP) — direct commands
2. Chain-of-Thought (CoT) — requesting step-by-step reasoning
3. Expert Prompting (EP) — role assignment
4. Self-Consistency (SC) — generating multiple responses and comparing
5. Automatic Prompt Engineer (APE) — meta-prompting
6. Generated Knowledge (GKn) — asking the model to generate context before answering
7. Tree-of-Thought (ToT) — orchestrated branching reasoning

Walter presents this as an engineering taxonomy — a set of techniques to learn and apply. But read it again through a communicative lens:

• IOP is a basic directive speech act — “Do X”
• CoT is a metacognitive request — “Show your work”
• EP is identity co-construction — “Be an expert in Y”
• SC is dialogic verification — “Argue with yourself”
• ToT is orchestrated multi-voice debate

That's not engineering. That's rhetoric. Walter built a ladder of communicative sophistication and labeled it a technical taxonomy. The progression from simple directives to orchestrated multi-perspective reasoning is the same progression you'd find in any communication competence model — from basic transactional exchange to complex dialogic engagement.

The Four Resources Model Was Already There

Tour & Zadorozhnyy (2025) get closer to seeing it. In “Conceptualizing and Operationalizing Prompt Literacy for ELL,” they do something nobody else in the prompt literacy literature has done: they apply a literacy theory — specifically, Freebody & Luke's Four Resources Model — to prompt interaction.

The Four Resources Model describes four roles a literate person performs:

Tour & Zadorozhnyy map this directly onto prompt literacy, creating an iterative cycle: craft a prompt → engage with the output → review and refine → repeat. It's the most theoretically sophisticated paper in the prompt literacy cluster.

Here's what they don't say explicitly: the Four Resources Model is a communication model in disguise. Code-breaking is encoding. Text participation is meaning construction. Text use is pragmatic competence. Text analysis is critical awareness. These four dimensions map directly onto communicative competence as defined by communication theory going back to Hymes (1972).

Their iterative cycle — craft, engage, review — is dialogue. Not metaphorical dialogue. Actual turn-taking communicative exchange between a human and a system, mediated through the application layer.

The Control Group: What AI Literacy Looks Like Without Communication

Baskara (2025) provides the instructive contrast. In “Conceptualizing Digital Literacy for the AI Era,” the framework covers five dimensions: technical literacy, practical literacy, critical literacy, ethical literacy, and meta-learning literacy.

It's clean. It's comprehensive. And it's completely interchangeable with any other digital literacy framework from the last decade. Swap “AI” for “social media” or “the internet” and the framework still works — which means it's not actually capturing what's specific about AI interaction.

Baskara is what AI literacy looks like without the communicative dimension. It describes knowledge about systems. It never describes the communicative relationship with systems. The framework could produce someone who understands AI perfectly and still can't navigate a multi-turn conversation with one.

The Disciplinary Wall

Why does no one see this? The answer is disciplinary silos.

Walter is in education. Tour & Zadorozhnyy are in applied linguistics. Baskara is in digital literacy. All three are close to the communicative insight — Walter builds a rhetoric without naming it, Tour & Zadorozhnyy apply a proto-communication model, Baskara shows what happens without one — but none of them have communication theory in their toolkit.

Education researchers see “skills to teach.” Computer science researchers see “optimization problems.” Applied linguists see “language use patterns.” Nobody sees what a communication theorist would immediately recognize: a new communicative domain with its own registers, fluencies, and stratification patterns.

This is the gap that Application Layer Communication was built to fill.

From Prompt Literacy to ALC Fluency

Prompt literacy, as it's currently conceived, is a subset of a larger phenomenon. It focuses on one specific interaction modality — natural language prompting of generative AI — and treats it as either a technical skill (Walter) or a literacy practice (Tour & Zadorozhnyy). Both are true. Neither is complete.

ALC reframes prompt interaction as one instance of human communication through the application layer. The same communicative competencies that make someone effective at prompting — understanding system registers, reading feedback as dialogue, adapting communication strategies across contexts — are the same competencies that make someone effective at navigating any software interface.

The Four Resources Model mapping makes this concrete:

This isn't just renaming. It's a fundamental reorientation. When you treat prompt interaction as communication rather than engineering, different things become visible:

• Stratification — communicative competence is always unevenly distributed. Prompt literacy will stratify the same way writing literacy did, and for the same reasons.
• Transfer — communicative competence transfers across contexts. Someone fluent in application-layer communication doesn't just know one AI tool; they can navigate any system's communicative affordances.
• Power — who gets to communicate effectively through the application layer determines who captures value from AI systems. This is a political question, not a technical one.

The Publishable Insight

There's a paper waiting to be written here. Its title might be something like: “From Prompt Engineering to Prompt Communication: Why AI Literacy Needs Communication Theory.”

The argument writes itself: the prompt literacy literature is converging on communicative constructs without the theoretical vocabulary to name them. Walter's taxonomy is rhetoric. Tour & Zadorozhnyy's Four Resources Model is communicative competence. The gap between “prompt engineering” and “prompt communication” is exactly where the most important questions about AI equity, education, and access will be decided.

589 citations and counting. The field is ready for this. Someone just needs to say it.

Papers discussed:
Walter, Y. (2024). Embracing the future of AI in the classroom: Relevance of AI literacy, prompt engineering, and critical thinking in modern education. International Journal of Educational Technology in Higher Education, 21(15). 589 citations.

Tour, E., & Zadorozhnyy, A. (2025). Conceptualizing and Operationalizing Prompt Literacy for English Language Learners. TESOL Quarterly.

Baskara, F. R. (2025). Conceptualizing Digital Literacy for the AI Era: A Framework for Navigating an AI-Driven World.