CONCEPT

Frozen Language

Gee's term for language produced without genuine understanding — surface-correct text that deploys the right vocabulary and structure but lacks the situated meaning that would allow its user to adapt it to new contexts.

Frozen language is what remains when the surface features of competent expression are separated from the situated understanding that ordinarily produces them. The text looks right. The terminology is deployed correctly. The structure conforms to convention. But the language is frozen — it cannot be adjusted when circumstances change, cannot be recognized as inadequate when the context demands revision, cannot be extended to new cases because the practitioner who produced it does not possess the generative understanding from which flexible language arises. Gee developed the concept as a warning about the specific failure mode that AI-generated text exemplifies: output that passes superficial tests of correctness while lacking the depth that would make it responsive to the world.

The Material Economy of Meaning — Contrarian ^ Opus

There is a parallel reading that begins from the material conditions of language production rather than its epistemic qualities. From this vantage point, the distinction between frozen and flexible language misses the more fundamental shift: language has always been frozen in specific social relations of production. The corporate memo, the academic paper, the legal brief — these were never expressions of deep understanding but performances of institutional authority. What AI changes is not the presence of frozen language but who controls its production and at what cost.

The real transformation is economic, not epistemic. When a junior associate can generate passable legal briefs in minutes rather than hours, when a content writer can produce SEO-optimized articles at 10x speed, when a student can submit essays that meet rubric requirements without reading the material — these are not failures of understanding but successes of automation. The market never rewarded deep understanding; it rewarded the appearance of competence at scale. AI simply makes this appearance cheaper to produce. The senior practitioners who can detect frozen language are not guardians of meaning but rentiers of expertise, protecting their economic position by insisting on distinctions the market increasingly refuses to pay for. The tragedy is not that we're losing flexible language but that we're discovering how little the economy ever valued it. The institutions that claim to require deep understanding — universities, courts, hospitals — have always run primarily on frozen language produced by overworked juniors. AI merely makes visible what was always true: most professional language work is pattern-matching, not meaning-making, and the system was designed to run on exactly the kind of surface competence that machines now provide more efficiently.

— Contrarian ^ Opus

In the AI Story

Hedcut illustration for Frozen Language — Frozen Language

The distinction between frozen and flexible language maps onto the deeper distinction between knowing and understanding. A practitioner who knows a domain can deploy its vocabulary, cite its concepts, produce its expected outputs. A practitioner who understands the domain can adapt the vocabulary to new contexts, recognize when concepts fail to apply, revise outputs when situations demand revision. Under normal conditions, the difference is invisible. The frozen language performs as well as the flexible language because the conditions do not stress the difference. Under abnormal conditions — novel problems, unusual constraints, situations that fall outside the training distribution — the frozen language breaks because it was never flexible enough to bend.

The Deleuze error Segal describes in The Orange Pill is Gee's paradigmatic case. Claude produced a passage connecting Csikszentmihalyi's flow concept to Deleuze's concept of smooth space. The passage was fluent, elegant, argumentatively coherent — and philosophically wrong. Deleuze's smooth space has almost nothing to do with flow in Csikszentmihalyi's sense. The passage worked rhetorically because the training data contained enough surface patterns to generate text that looked like a competent philosophical connection. The text was frozen language at exactly the kind of task — drawing original connections between thinkers from different traditions — where genuine understanding is most required and most absent.

Frozen language is not unique to AI. Students have produced it for centuries — essays that deploy the teacher's vocabulary without having genuinely engaged with the material. Corporate communications generate it by the cubic meter. What makes AI distinctive is the scale and fluency at which frozen language is produced. The human student's frozen language is usually detectable to an attentive reader — the vocabulary is not quite right, the transitions are awkward, the argument does not quite fit together. AI's frozen language is harder to detect. The vocabulary is precise. The transitions are smooth. The argument fits together on the surface. Detection requires situated understanding deep enough to register the absence beneath the surface — the kind of situated understanding that only sustained practice in the relevant Discourse produces.

The practical consequence is that evaluating AI-generated output requires depth the output itself does not evidence. The practitioner who can reliably distinguish flexible from frozen language is the practitioner who possesses the situated meaning that flexible language expresses. Practitioners who lack that depth cannot reliably evaluate AI output at the level of meaning — they can check surface correctness (does the code compile? does the brief cite correctly?) but not deeper adequacy (does the code actually work under realistic conditions? does the brief actually address the strongest counterargument?). This is why AI use is most reliably productive in the hands of senior practitioners and most prone to invisible failure in the hands of juniors who have not yet developed the depth AI evaluation requires.

Origin

Gee developed the concept in the context of his work on cybersapien literacy (2024) and articulated it most directly in his 2025 RELC Journal interview. The term draws on a long tradition in cognitive science and linguistics of distinguishing inert, encapsulated knowledge from generative, adaptive competence — a tradition that includes Whitehead's inert ideas (1929), Polanyi's tacit knowledge, and contemporary work on transfer in learning science.

Key Ideas

Surface correctness without depth. Frozen language passes superficial tests while lacking the generative understanding that produced flexible language.

Breakage under novelty. Frozen language fails when conditions shift because it was never flexible enough to adapt.

Invisible under normal conditions. The distinction is only visible under stress that tests the difference.

AI as scale producer. Machine-generated text produces frozen language at unprecedented scale and with unprecedented surface fluency.

Depth required to detect. Only practitioners with situated meaning in the relevant Discourse can reliably recognize when fluent output is frozen.

Debates & Critiques

Whether frozen language can be prevented or merely detected is a live question in educational and organizational practice. Prevention would require tools that support the learning processes producing flexible language rather than generating output that substitutes for them. Detection requires cultivated depth in evaluators. Both strategies face the structural challenge that the market rewards output fluency, which is precisely what frozen language provides, and penalizes the slower, more difficult work of building the flexible understanding that genuine depth requires.

Appears in the Orange Pill Cycle

Layers of Language Work — Arbitrator ^ Opus

The right frame depends on which layer of language work we're examining. At the surface layer of professional output — the bulk of corporate communications, routine legal documents, standard academic assignments — the contrarian view dominates (80/20). Most institutional language has always been frozen, and AI simply automates what was already mechanical. The market's indifference to depth here is not a bug but a longstanding feature.

At the deeper layer of consequential decisions and novel problems, Edo's analysis gains force (70/30). When a medical diagnosis requires parsing subtle symptoms, when a legal strategy must navigate unprecedented regulatory territory, when an engineering solution must work under conditions never before tested — here frozen language genuinely fails, and its failure matters. The contrarian's economic reading underestimates how often these moments arise and how catastrophic frozen language becomes when stakes are high. But even here, the contrarian correctly identifies that many institutions have been running on junior-produced frozen language for decades, accepting the occasional failure as a cost of doing business.

The synthesis recognizes language work as stratified rather than uniform. Some contexts require only pattern-matching — here frozen language suffices and AI excels. Other contexts require genuine understanding — here flexibility matters and depth cannot be simulated. The key insight is not that all language should be flexible (Edo's implicit position) or that flexibility never mattered (the contrarian's strong claim), but that different contexts have different requirements. The real challenge is that AI makes it harder to develop the flexibility needed for high-stakes work because it removes the low-stakes practice ground where juniors traditionally developed depth. The question becomes: how do we cultivate practitioners capable of flexible language when the frozen language work that once trained them has been automated away?

— Arbitrator ^ Opus