Typomora: Neuroscience, Naming and Framing for Paper 3 (the Typed Delay and the Return of Voice)

Summary

The neuroscience: Literacy measurably changes the brain. Dehaene et al. (2010, Science) found increased responses to letters in the visual word form area (VWFA), a small displacement of face responses, and enhanced early-visual and phonological responses. Skeide et al. (2017, Science Advances) found that 6 months of reading training changed thalamic–brainstem coupling in illiterate adults. But the "loss" side of the trade-off is contested: Hervais-Adelman et al. (2019) and van Paridon et al. (2021) found recycling "without destruction". Typing, handwriting and speech share central lexical and semantic systems but use different peripheral and feedback circuits. Speech comprehension and reading converge on almost identical semantic maps (Deniz et al. 2019).
What the voice shift actually changes: It removes a transcription bottleneck in production, not in comprehension. Dictation-versus-writing studies run from Gould (1978) through Ruan et al. (2016), where speech was 2.93× faster than phone typing with fewer errors. Working-memory studies (Bourdin & Fayol 1994, 2002) show that writing costs more cognitively than speaking, even in adults. Two lines of research give mechanism-level support for your observation that speaking to an AI improved your speech: self-explanation (Chi et al. 1994) and psychological treatment of speaking anxiety (Ebrahimi et al. 2019: "0.74 (Hedges g; 95% CI: 0.61-0.87)" across 62 interventions). This support is analogical: no study tests that exact situation.
The coinage and frame: Call the typed era the typomora: a delay, counted in spoken time units, that bandwidth and compute imposed on the McLuhanite return to acoustic space. Use epanaphony for its ending. The defensible thesis is that the keyboard era was a production-side bottleneck that temporarily locked a typographic configuration (orthographic working memory, visual feedback, serial composition) onto an electric medium.

Key figures

Dictation-versus-writing studies run from Gould (1978) through Ruan et al. (2016), where speech was 2.93× faster than phone typing with fewer errors.Summary →

Gould's IBM studies found dictated letters equal in quality to written ones after a few hours of practice; in Gould (1981), "text-edited letters required 50% more time to compose than did written letters."Section 1 →

Research document · Canopticon · received 2026-10-04 · McLuhan & media theory · condensed from chat; summary, findings, verdicts and caveats verbatim, long tables shortened

The neuroscience supports a narrower version of your thesis than the strong one. Literacy is the best peer-reviewed evidence that a communication medium measurably reconfigures perceptual and production systems in the brain, which fits your claim in Paper 1 that a "typographic" configuration is physiological. But the best evidence also shows that comprehension converges on a largely modality-invariant semantic system. So the shift back to voice should be argued as a change in the production, timing, feedback and social loop of language, not as a change in what is understood. For the name, I recommend typomora (Greek typos, "impression, type" + Latin mora, "delay", which is also the phonological unit of spoken timing). It appears unclaimed, it does the structural double work that syncrasis did, and it pairs well with a second term, epanaphony ("the return of voice"), for the closing phase.

1Key Findings

**Literacy is the cleanest demonstration that a medium changes the brain, but what changes is mostly addition and retuning, not destruction.** Responses to letter strings grow in left ventral occipito-temporal cortex. Early visual cortex (down to V1) becomes more responsive. Activation of the spoken-language network increases in planum temporale. Face responses shift rightward. Mirror-image generalisation for letters is suppressed. Whether anything is lost is an open dispute, and the 2019–2021 evidence leans toward "no net loss".
Writing and speaking are partly separate production systems. Purcell, Turkeltaub, Eden & Rapp (2011) separate central spelling substrates from peripheral motor ones. Speech production relies on auditory–motor feedback loops (Hickok & Poeppel; Guenther's DIVA model). The modalities share lexical and semantic stages and split at the buffer and feedback stages.
Writing costs more cognitively than speaking, and dictation lets mature composers keep quality while gaining speed. Bourdin & Fayol (1994); "Even in adults, written production is still more costly than oral production" (2002). Gould's IBM studies found dictated letters equal in quality to written ones after a few hours of practice; in Gould (1981), "text-edited letters required 50% more time to compose than did written letters."
Semantic comprehension is largely modality-invariant. Deniz et al. (2019, J Neurosci) found that semantic representations evoked by listening and by reading are "almost identical". Rogowsky, Calhoun & Tallal (2016) found no comprehension or retention difference across audiobook, e-text and dual modality (n = 91).
Spoken practice reduces speaking anxiety with large effects, and verbal elaboration improves understanding. g = 0.74 across 62 interventions (Ebrahimi et al. 2019); in vivo and VR exposure g ≈ 1.4–1.6 (Reeves et al. 2022); self-explanation d ≈ 1.14 (Chi et al. 1994). None tested voice-to-AI interaction directly.
Voice adoption is real but survey-dependent. DataReportal/GWI Q4 2025: 22.5% of internet users worldwide use voice assistants weekly. Research on cognition with conversational AI is preliminary: Kosmyna et al. 2025 is an un-peer-reviewed preprint with a published methodological critique.
Naming: typomora, phonomora, clavistasis, typolatency, phonolatency, nostophony and epanaphony returned no prior uses. Typographic hysteresis collides with Theophanidis & Thibault's "media hysteresis". Akroatic collides with Lipari's "akroatic thinking" (2014). Typostasis collides with Schindewolf's paleontology.

2Assessments (verbatim)

Literacy is the cleanest case of a medium altering "sense ratios" in a literal sense. It is a culturally imposed, trainable, dose-dependent change in the weighting of visual-orthographic versus other inputs, and it can be induced in adults within months.
I found no peer-reviewed study showing that heavy typing degrades speech production, or that heavy speaking improves typed composition. The cross-modal transfer question is open.
The "bandwidth artifact" thesis gets direct historical support from Gould, Conti & Hovanyecz (1983): a decade before the web, IBM showed that dictation was cognitively viable and that recognition technology was the binding constraint.
I found no controlled study of "explaining to an LLM by voice" versus by typing. Treat the rubber-duck and AI analogy as original synthesis.
I found no peer-reviewed study of voice-first computer interfaces producing gains in spontaneous speech fluency. That is a genuine gap your paper can name.
At the semantic level of reception, very little [changes]. The paper should argue that the typed era configured the production and interaction loop, not the semantic core.
I found no peer-reviewed study of how voice conversation with AI affects memory, composition or cognition.

3Strongest defensible thesis

The typographic character of networked media from about 1970 to 2015 was a production-side bottleneck, not a reception-side destiny. Electric media's acoustic bias, as McLuhan described it, was held in abeyance because the only economical input channel was the keyboard. As recognition error, bandwidth and compute constraints fell (with dictation reaching about 3× typing speed with fewer errors by 2016), the interface began to return to voice. The return changes the production and interaction loop of language (timing, feedback modality, persistence, prosody, transcription cost), not its semantic core.

4Five testable predictions

Transcription-cost release. 2. Production-loop retuning. 3. Asymmetric atrophy (the typomoric deficit). 4. Persistence re-weighting. 5. Semantic invariance holds.

5Caveats

Citations not verified in this research pass: Pegado et al. 2014; Kolinsky et al. 2011 details; Morais et al. 1979 page numbers; Dehaene-Lambertz et al. 2018 DOI; Reece & Cumming 1996; Houde & Jordan 1998; dystypia case reports; Recanzone; Chen et al. 2024 venue.
Chi et al. (1994) d ≈ 1.14 comes from a secondary review table. Adoption figures come partly from aggregators; cite primary reports.
Kosmyna et al. (2025) and the Stanković et al. critique are on arXiv and are not peer-reviewed.
Original synthesis (flagged): the typomora thesis itself, the mapping onto the syncrasis dimensions, the "production-not-reception" relocation, the voice-to-AI-as-exposure mechanism, and all five predictions. No source frames the typed era this way.
Collision checks are limited to general-web search; confirm with exact-phrase searches in Scholar, JSTOR and Google Books.

Cite this

Telford, C. (2026). Typomora: Neuroscience, Naming and Framing for Paper 3 (the Typed Delay and the Return of Voice). Canopticon research document. canopticon.com/p/typomora-neuroscience-naming-and-framing-for-paper-3-the-typed-delay-a