Matthew Dicks is backstage at the Brooklyn Historical Society. He is about to give a talk to a substantial audience. He has prepared the talk over weeks. He has the slides; he has the structure; he has the rehearsed phrasings. Fifteen minutes before he is supposed to go on, he realizes — with the kind of clarity that arrives when there is no time to fix it — that the talk he has prepared is being delivered at the audience, not spoken to them. The phrasings are formal. The transitions are stiff. The cadence is presentation-mode, not conversation-mode. He has built a speech, not a talk.
He scraps it.
In the fifteen minutes before he goes on, Dicks rewrites the entire approach in his head. He throws out the rehearsed phrasings. He throws out the formal transitions. He commits to going on stage and talking to the audience the way he would talk to friends at a dinner party. He keeps the content — the points he wants to make, the stories he wants to tell — but changes the register from at-mode to to-mode. He goes on stage. The talk lands at a level his original prepared version would not have approached.1
The fifteen-minute rewrite is one of Dicks's most-told personal cases for the principle: every form of public speaking, including storytelling, should be spoken to the audience as if they were the storyteller's friends in conversation, not delivered at the audience as if they were a wall the storyteller was firing words at. The principle distinguishes the registers and operationalizes the shift.
The principle has three components.
1. There are two registers — speaking-to and speaking-at. They differ in vocabulary, sentence length, cadence, eye contact, body language, and the storyteller's relationship to the prepared material. Speaking-at is the default register most public speakers adopt when given a stage — formal, prepared, distant. Speaking-to is the conversational register — informal, present, immediate. The two are not on a continuum; they are discrete states. The storyteller is in one or the other, not partway between.
2. The shift between registers is largely a matter of imagined audience. Speakers in at-mode imagine the audience as an undifferentiated mass — a crowd, the room, the public. Speakers in to-mode imagine the audience as specific people they are talking with — even if those people are strangers. The mental image of the audience determines the register; the register determines the delivery. The shift is therefore an upstream cognitive operation, not a downstream delivery technique.
3. Audiences receive the registers as fundamentally different kinds of experience. An audience receiving speaking-at experiences the speaker as performing — they are observing the speaker; they are evaluating the performance. An audience receiving speaking-to experiences the speaker as talking with them — they are participants in a conversation; they are included. The two experiences produce different downstream effects: at-mode produces respectful applause; to-mode produces felt connection.
The cognitive mechanism: audiences run social-status calculations on speakers continuously. The calculations are unconscious but consequential — is this speaker above me / below me / equal to me; am I being addressed as a peer or as a subordinate; is the speaker inviting me in or holding me at distance.
Speaking-at register signals status-asymmetric. The speaker is on stage; the audience is in seats; the speaker is performing; the audience is watching. The speaker's prepared phrasings and formal cadence reinforce the asymmetry. The audience's response is calibrated to the asymmetry — they show respect, they applaud at the appropriate moments, they do not feel personally addressed.
Speaking-to register signals status-symmetric or invited-into-peer-relationship. The speaker is talking to the audience as if to friends; the audience is being treated as conversation-partners. The cadence is conversational; the phrasings sound spontaneous; the speaker's relationship to the prepared material is I'm telling you this rather than I have prepared this for you. The audience's response is calibrated to the symmetry — they lean in, they feel addressed personally, they form felt connection with the speaker.
The downstream effect is enormous. Status-asymmetric receiving produces respect but not transformation; status-symmetric receiving produces both connection and internalization. Audiences who receive a speech in at-mode walk out remembering having attended; audiences who receive a story in to-mode walk out remembering having been spoken to. The two memory-states have different durabilities and different effects on subsequent behavior.
A second mechanism: humans have far more practice listening to conversations than to speeches. The to-mode register engages the listener's conversation-listening neural circuitry, which is more refined, more automatic, and more emotionally engaged than the speech-listening circuitry most adults use rarely. The to-mode speaker is therefore exploiting better-developed audience-side machinery; the at-mode speaker is forcing the audience to use less-developed machinery for the same operation.
A third mechanism: at-mode produces cognitive distance between speaker and audience because the audience perceives the speaker as performing for them. The performance frame triggers evaluation — the audience-member becomes a critic. To-mode collapses the cognitive distance because the speaker is with the audience rather than for them. The conversational frame triggers participation — the audience-member becomes a partner.
The registers can be diagnosed from specific vocabulary, sentence structure, and delivery features.
Speaking-at markers:
Speaking-to markers:
The markers cluster. A speaker who is in at-mode will display most of the at-markers; a speaker who is in to-mode will display most of the to-markers. Mixed-register speakers are rare — once one or two markers shift to one register, the rest tend to follow.
The principle is the delivery-mode complement to the no-one-wants-to-hear-anything mindset. The mindset is the attentional discipline (assume the audience is uninterested); to-mode is the delivery register that converts an uninterested audience into an interested one. At-mode amplifies the audience's default uninterest because at-mode treats the audience as a passive receiver; to-mode dissolves the default uninterest because to-mode treats the audience as a conversation partner.
It pairs with imperfection is the beauty (Catherine Burns / Moth): the Moth tradition's institutional preference for imperfection — the doubling-back, the visible memory-recovery, the occasional stammer — is the institutional codification of to-mode delivery. Burns's maxim is the aesthetic framing of what speaking-to-vs-at frames operationally.
It connects to nervousness as connector: disclosed nervousness is a to-mode marker (at-mode speakers do not disclose nervousness; they perform composure). The willingness to disclose nervousness is one of the highest-leverage moves a speaker can make to shift register, because it instantly signals I am here with you as a fellow human rather than I am a polished performer producing a speech.
It contrasts with the game-played-with-audience principle of remembering rather than memorizing — the remember-not-memorize discipline is part of what enables to-mode delivery. Memorized lines tend to come out in at-mode cadence; remembered material tends to come out in to-mode cadence. The two principles work together — remember-not-memorize at the preparation level; speak-to-not-at at the delivery level.
Dicks's account of the fifteen-minute rewrite is one of the clearest worked cases of register-shift in real time.2
The original prepared talk:
The fifteen-minute realization: Dicks realizes the talk is in at-mode. Specifically: the prepared phrasings are formal, the transitions are stiff, the audience is being addressed as a mass. The talk will land, in the sense that audience-members will hear the content and applaud appropriately, but the talk will not connect — audience-members will leave remembering attended-a-talk, not remembering having-been-spoken-to.
The shift operation: In fifteen minutes Dicks cannot rewrite the slides or the content. What he can do is shift the imagined audience in his head — from the audience at the Brooklyn Historical Society to the specific friends I would be telling this to at a dinner party. The shift in imagined audience changes the upstream cognitive state, which changes the downstream delivery.
The resulting on-stage delivery:
The audience response: The audience receives the talk as being talked to rather than being talked at. They lean in. Their evaluation shifts from is this talk good to I am in conversation with this speaker. The post-talk response is qualitatively different from what the original prepared version would have produced — felt connection, follow-up conversations, audience-members staying afterward to talk with Dicks personally.
The instructive part of the case: The shift was upstream cognitive, not downstream tactical. Dicks did not change the content; he changed the imagined audience. Everything downstream — the vocabulary, the cadence, the eye contact — followed automatically from the upstream shift. This is why the shift could be executed in fifteen minutes without rewriting any of the prepared material: the upstream operation is fast; the downstream tactical adjustments fall out of it.
You are about to give a talk. You have prepared it. You are backstage or in the wings. You suspect you might be in at-mode and want to check.
It is twelve minutes before your scheduled start. You step into a quiet corner. You take the first paragraph of your prepared opening and you read it aloud — quietly, not for anyone else to hear. You listen to the cadence. Does it sound like something you would say to a friend at a dinner party, or does it sound like something a speaker would say at a podium?
If the second — you are in at-mode. You have eleven minutes to shift.
You do not rewrite the words. You shift the imagined audience. You think of one specific friend — not in abstract but with their face, their name, their specific way of listening. You think I am going to tell this to Sara. You read the first paragraph again, this time imagining Sara sitting across from you at a coffee shop.
The reading changes. The vocabulary you would use to Sara replaces some of the formal vocabulary automatically. The sentence structure shortens. Hesitations appear. The cadence becomes conversational. You are not consciously editing; you are downstream-following the upstream shift.
You repeat with the second paragraph, then the third. By the time the door opens and you walk on stage, the imagined audience in your head is Sara (extended to Sara plus a hundred and fifty other Saras). You walk to the stage. You open your mouth. The first sentence comes out in to-mode because that is where you have been mentally for the past eight minutes.
The audience leans in. The talk is doing what at-mode talks cannot do.
The pattern is: diagnose register from a quiet pre-talk read-aloud test, identify a specific individual as the imagined audience, hold that individual in your head during the run-up to delivery, allow the downstream tactical features to fall out of the upstream cognitive state, walk on stage already in the register you want to perform in.
Your applause is loud at the end but no one approaches you afterward. At-mode delivery produces audible appreciation but not personal connection; audience-members do not feel they were spoken to and therefore do not feel invited into post-talk interaction. To-mode delivery produces both audible appreciation and follow-up — audience-members come up to talk, exchange contact information, ask follow-up questions.
Your video review reveals you behind the lectern the whole time. Physical position correlates with register. Speakers who stay behind lectures default to at-mode; speakers who move out from behind the lectern shift toward to-mode by physical position alone. If your video shows you stationary at the lectern for the full talk, the register was almost certainly at-mode.
Your transcript reads like a written essay. When you transcribe your talk and read it as text, does it read like conversation transcribed (with fragments, hesitations, restarts) or like essay delivered aloud (with full sentences and prepared transitions)? Essay-prose transcripts are at-mode evidence; conversation-prose transcripts are to-mode evidence.
Audience members compliment your presentation rather than your story. You're a great presenter and that was a polished talk are at-mode compliments. I felt like you were talking to me and I loved the way you tell stories are to-mode compliments. The compliment-vocabulary the audience uses diagnoses the register they received.
You feel performed-out rather than connected-out after the talk. Speakers who delivered in at-mode tend to feel performed-out — tired in a way analogous to having executed a long monologue. Speakers who delivered in to-mode tend to feel connected-out — tired in a way analogous to having had a long deep conversation with many people. The post-talk subjective state diagnoses the register the speaker was operating in.
Dicks across both interviews holds the speaking-to-vs-at principle as one of the central delivery disciplines.3 He emphasizes the cognitive upstream nature of the shift — changing the imagined audience changes everything downstream, but the change has to start with the cognitive operation, not with surface tactical adjustments. Speakers who try to shift register by speaking more casually without changing the imagined audience produce a hybrid (informal at-mode) that is in some ways worse than either pure register.
Lewis treats the principle in his synthesis as a Dicks-tradition emphasis on delivery as relational, not performative.4 Lewis's framing extends slightly beyond Dicks's specific articulation to include the relational ethics of the to-mode shift — speaking-to treats the audience as humans deserving of relational address, whereas speaking-at treats the audience as units to be impressed.
Jasper Claus does not articulate the at-vs-to distinction in this specific vocabulary but operates an analogous principle in his corporate filmmaking — the protagonist in a Claus film typically delivers their key moments in to-mode (speaking conversationally as if to the viewer) rather than at-mode (delivering scripted lines as if for the viewer's evaluation). The convergence suggests the principle generalizes across speaking forms (oral storytelling, corporate filmmaking, perhaps written-essay voice) and is not specific to live oral performance.
Picture the two delivery-modes side by side: performance-mode address (the speaker performs at the audience) versus conversation-mode address (the speaker talks with the audience). The two produce wildly different downstream effects on audience reception. Multiple other domains formalize equivalent distinctions.
Pastoral / clergy traditions: Effective preaching traditions distinguish between sermon-as-performance (delivered at the congregation) and sermon-as-pastoral-address (spoken to the congregation as known individuals). The pastoral-tradition vocabulary names the to-mode shift as meeting the congregation where they are — a phrase whose precise operational meaning maps onto Dicks's principle. Effective preachers cultivate the to-mode register over years of practice; ineffective preachers stay in at-mode and produce respectful but transformation-free attendance. Look at how old the distinction actually is: at-vs-to is centuries old as a craft discipline in religious traditions, and storytelling-craft is rediscovering vocabulary the pastoral tradition has refined since at least the medieval era. Neither tradition alone gives you the full account: storytelling-craft supplies the secular operational framework; pastoral tradition supplies the deep theological and relational vocabulary for the to-mode register's ethics. Pairing them suggests storytellers might productively study sermon-tradition theology of meeting the listener where they are as a deepening of the register-shift practice.
Therapy / clinical communication: Effective therapists distinguish between clinical-mode communication (delivered at the patient from professional distance) and therapeutic-presence communication (spoken to the patient as a peer in the room). The two registers produce wildly different therapeutic outcomes — clinical-mode produces received instruction; therapeutic-presence produces internalized change. Carl Rogers's unconditional positive regard and the broader humanistic therapy tradition formalize the to-mode register as the central clinical discipline. Put therapy and storytelling together: the at-vs-to distinction turns out to be fundamental to any communicative discipline aiming at internalized change in the listener, not just to storytelling. Storytellers might import therapeutic-presence practices (Rogerian reflection, deliberate slowing of pace, eye-contact discipline) as register-shift training; therapists might import storytelling's specific tactical markers (vocabulary shift, sentence-length shift, conversational hesitations) as faster-deploying versions of presence-establishment.
Teaching / classroom presence: The same distinction structures classroom pedagogy. Lecture-mode teachers operate in at-mode (delivering content at the students from podium-distance); discussion-mode teachers operate in to-mode (talking with the students as conversation partners). The two produce different learning outcomes — lecture-mode produces information-retention with low engagement; discussion-mode produces lower information-density per minute but higher engagement and durability. Excellent classroom teachers cultivate the discussion-mode register as their default and reserve lecture-mode for specific information-dense situations. Pairing with the platonic ideal: discussion-mode classroom teachers continuously approach the platonic ideal across class periods; lecture-mode teachers operate continuously below the ideal. The two pedagogical practices converge on the same operational target storytelling craft articulates.
The Sharpest Implication The at-vs-to distinction implies that most of the perceived difficulty of public speaking is an artifact of speakers defaulting to at-mode. When speakers shift to to-mode, the cognitive load drops dramatically — they are having a conversation, which is a near-automatic cognitive operation, instead of executing a performance, which is a high-effort cognitive operation. The terror many speakers feel before stages is largely terror of executing the at-mode performance; the to-mode register transforms the stage from a performance-arena into an enlarged conversation-room. Speakers who make the shift report dramatic reductions in pre-talk anxiety. The implication: speech-anxiety treatment might productively focus on register-shift rather than on confidence-building, since the register-shift addresses the upstream cause while confidence-building addresses the downstream symptom.
Generative Questions