Two-Tracked Conversations�
Sam Berstler (MIT)
Northwestern Colloquium Series
21 November 2025
slides available: www.samberstler.com
To understand a conversation as such is in part to understand its organizing structure.
Our talk exchanges do not normally consist of a succession of disconnected remarks, and would not be rational if they did.
(Grice 1989: 26)
To understand a conversation as such is in part to understand its organizing structure.
Q. What organizes conversation?
To understand a conversation as such is in part to understand its organizing structure.
Grammar
Q. What organizes conversation?
To understand a conversation as such is in part to understand its organizing structure.
Grammar
Participants’ goals & participants’ rationality
Q. What organizes conversation?
To understand a conversation as such is in part to understand its organizing structure.
Grammar
Participants’ goals & participants’ rationality
Constitutive norms of conversation & of speech acts
Q. What organizes conversation?
To understand a conversation as such is in part to understand its organizing structure.
Grammar
Participants’ goals & participants’ rationality
Constitutive norms of conversation & of speech acts
Social norms
Q. What organizes conversation?
To understand a conversation as such is in part to understand its organizing structure.
Grammar
Participants’ goals & participants’ rationality
Constitutive norms of conversation & of speech acts
Social norms
Philosophers tend to assume that social norms cannot and do not provide deep explanations of conversation’s structure.
Q. What organizes conversation?
To understand a conversation as such is in part to understand its organizing structure.
Grammar
Participants’ goals & participants’ rationality
Constitutive norms of conversation & of speech acts
Social norms
Philosophers tend to assume that social norms cannot and do not provide deep explanations of conversation’s structure.
Their content is arbitrary.
Their content is culturally variable.
Q. What organizes conversation?
To understand a conversation as such is in part to understand its organizing structure.
Grammar
Participants’ goals & participants’ rationality
Constitutive norms of conversation & of speech acts
Social norms
Philosophers tend to assume that social norms cannot and do not provide deep explanations of conversation’s structure.
Their content is arbitrary.
Their content is culturally variable.
Their scope is not distinctly conversational.
Q. What organizes conversation?
To understand a conversation as such is in part to understand its organizing structure.
Q. What organizes conversation?
Grammar
Participants’ goals & participants’ rationality
Constitutive norms of conversation & of speech acts
Social norms
Orthodoxy:
The philosophy of conversation should generate deep and universal explanations for conversational structure. So the philosophy of conversation should avoid appealing to social norms.
To understand a conversation as such is in part to understand its organizing structure.
Q. What organizes conversation?
Grammar
Participants’ goals & participants’ rationality
Constitutive norms of conversation & of speech acts
Social norms
This talk is a case study.
Appeals to social norms can provide deep and default-universal* explanations for conversational structure.
This is because not all social norms are arbitrary.
(*default universal: we should expect but are not guaranteed to find this form of explanation for all societies)
Case study question.
Speakers sometimes seem to conduct two conversations simultaneously. Why?
interlude
movie time
Two-tracking
When we two-track, we seem to conduct an official and unofficial conversations simultaneously. The unofficial conversation is “unofficial” in a strange sense: it’s what we’re really talking about.
Two-tracking
When we two-track, we seem to conduct an official and unofficial conversations simultaneously. The unofficial conversation is “unofficial” in a strange sense: it’s what we’re really talking about.
We might call it…
Double-talk
Passive-aggressiveness
Conversation and subtext
Conversation and shadow conversation
Two-tracking
When we two-track, we seem to conduct an official and unofficial conversations simultaneously. The unofficial conversation is “unofficial” in a strange sense: it’s what we’re really talking about.
We often do it when…
Flirting
Propositioning others for sex
Asking for awkward or illegal favors
Talking “corporate”
Threatening others
Negotiating bribes
Case study question.
Speakers sometimes seem to conduct two conversations simultaneously. Why?
Case study question.
Speakers sometimes seem to conduct two conversations simultaneously. Why?
It seems to be cognitively costly.
It seems to (massively!) increase the risk of miscommunication.
Some characteristics of two-tracking
Two-tracking involves…
…double-talk
…non-contamination of tracks
…pretense
Two-tracking involves…
…double-talk
At least one speakers makes an utterance U and, in doing so, makes two simultaneous (real/pretended) speech acts G and G*.
…non-contamination of tracks
…pretense
Two-tracking involves…
Our speech reporting practices seem to confirm that there is indeed a secondary (inexplicit) speech act:
.
…double-talk
At least one speakers makes an utterance U and, in doing so, makes two simultaneous (real/pretended) speech acts G and G*.
…non-contamination of tracks
…pretense
Two-tracking involves…
The secondary speech act is available for anaphoric reference and other context-sensitive variables.
Ethan: Yes, let’s hook up.
…double-talk
At least one speakers makes an utterance U and, in doing so, makes two simultaneous (real/pretended) speech acts G and G*.
…non-contamination of first track
…pretense
Two-tracking involves…
Pretenses and ritual speech that do not involve double-talk do not constitute two-tracking:
…double-talk
At least one speakers makes an utterance U and, in doing so, makes two simultaneous (real/pretended) speech acts G and G*.
…non-contamination of first track
…pretense
Two-tracking involves…
…double-talk
…non-contamination of first track
Don’t openly acknowledge the shadow conversation. Or: if something happens in the second track, don’t acknowledge (presuppose) it in the first track.
…pretense
Two-tracking involves…
Ordinary implicatures are usually not two-tracked.
Peter: How’s MIT?
Sam: Well…I haven’t been fired yet!
Implicature: It’s going badly.
Peter: That bad, huh?
…double-talk
…non-contamination of first track
Don’t openly acknowledge the shadow conversation. Or: if something happens in the second track, don’t acknowledge (presuppose) it in the first track.
…pretense
Two-tracking involves…
…double-talk
…non-contamination of first track
…pretense
The speaker and addressee collaboratively maintain a pretense that they are not having the shadow conversation.
Two-tracking involves…
…double-talk
…non-contamination of first track
…pretense
The speaker and addressee collaboratively maintain a pretense that they are not having the shadow conversation.
Sequential, deniable speech acts do not constitute-two tracking:
Sam: Hmmm…unrelatedly, I have been wanting to go to
Hawaii.
Implicature: I’ll give you an A if you pay me a bribe equivalent to a
Hawaii vacation.
Two-tracking involves…
…double-talk
…non-contamination + pretense
We can use layered bodies of information (common grounds) in order to represent this.
To formally represent the two-tracks, we can use what I call common ground layering:
Interlocutors in a conversation c layer common ground cg2 over common ground cg1 just in case:
(2) if cg1 entails p, then the interlocutors aim to ensure that cg2 entails [ that cg1 entails p ], and
(3) interlocutors aim to ensure that cg1 entails that cg2 does not exist.
Two-tracking involves…
…double-talk
…pretense
…non-contamination of tracks
The first should entail that activity within the second track is not occurring. The speakers have two sets of presuppositions, about what’s happening in each track.
Unfortunately (?), we need take tracks as primitively individuating, and interlocutors must accept p for the sake of track track (layer) 1, 2, 3, and so on….
To formally represent the two-tracks, we can use what I call common ground layering:
Interlocutors in a conversation c layer common ground cg2 over common ground cg1 just in case:
simultaneously rely on cg1 and cg2, and
(2) if cg1 entails p, then the interlocutors aim to ensure that cg2 entails [ that cg1 entails p ], and
(3) interlocutors aim to ensure that cg1 entails that cg2 does not exist.
Are the two tracks just the pretense-y and belief-y tracks?
(cf. Yalcin 2007 “conversational tone”)
Are the two tracks just the pretense-y and belief-y tracks?
(cf. Yalcin 2007 “conversational tone”)
No. Much of what occurs in the first track is “belief-y.” We presuppose it because we believe it. We assert it because we want others to come to believe it.
Huffington: I just talked to Travis.
Are the two tracks just the direct and indirect speech act tracks?
Are the two tracks just the direct and indirect speech act tracks?
No. We don’t always update the second track via implicature:
You need to be careful with what you’re doing.
First track: You need to be careful managing this crisis.
Second track: You need to be careful in how you threaten me.
Are the two tracks just the explicit and inexplicit track?
Two-tracking involves…
No. Sometimes the “explicit” speech act is the speech act that updates the second track:
If you ever do that again, I’ll kill you! Hahaha!
First track: (hyperbole) If you ever do that again, I’ll do
something really bad (but not kill you).
Second track: (threat) If you ever do that again, I will
literally kill you.
Are the two tracks just the explicit and inexplicit track?
Two-tracking involves…
…double-talk + pretense + non-contamination of tracks
Two-tracking constraint
If you utter u within a two-tracked conversations, then the first track entails that you uttered u.
Two-tracking intelligibility constraint
If you utter u within a two-tracked conversation, then make sure the first track entails that in making u, you made some speech act G compatible with the first track presuppositions.
Double-talk is optional
With your turn at talk, you can but needn’t update the secondary track. But you must update the first.
Case study question.
Speakers sometimes seem to conduct two conversations simultaneously. Why?
Case study question.
Speakers sometimes seem to conduct two conversations simultaneously. Why?
It seems to be cognitively costly.
It seems to (massively!) increase the risk of miscommunication.
A natural thought
Two-trackers want deniability!
A sanctioning orthodoxy
Two-trackers aim to avoid some kind of sanction for what they are doing in the second-track. Characteristically, in two-tracking, they aim to modulate what someone does / could / will know about what they’re doing.
Walton (1996); Pinker (2007); Pinker et al (2008); Lee and Pinker (2010); Fricker (2012);
Peet (2015); Camp (2018); Davies (2019); Mazzarella (2021); Dinges and Zakkou (2023); Berstler (forth)
Daphne/Ethan
A sanctioning orthodoxy
Two-trackers aim to avoid some kind of sanction for what they are doing in the second-track. Characteristically, in two-tracking, they aim to modulate what someone does / could / will know about what they’re doing.
Daphne/Ethan
Huffington/Gurley
A sanctioning orthodoxy
Two-trackers aim to avoid some kind of sanction for what they are doing in the second-track. Characteristically, in two-tracking, they aim to modulate what someone does / could / will know about what they’re doing.
Commitment?
Two-trackers aim to avoid undertaking commitments to what they put in the second track.
But the commitment they’re avoiding doesn’t seem to be epistemic:
1. If you don’t fix this, I will.
It doesn’t seem to be practical either:
2. If you don’t pay me the money, I’ll kill you!
Hahaha!
Commitment?
Two-trackers aim to avoid undertaking commitments to what they put in the second track.
Showing off?
Two-trackers aim to show off their rhetorical cleverness and ability to read each other’s minds.
Showing off?
Two-trackers aim to show off their rhetorical cleverness and ability to read each other’s minds.
But there are many ways to show off in conversation. For example, they could try to avoid words with the letter “b.” Why show off this way?
An invisible audience?
Two-trackers are playing to an invisible or imaginary audience.
(Goffman [1956] 1959;
Camp 2018)
By this communication technique [of double-talk] individuals may convey information to one another in a manner or on a matter that is inconsistent with their official relationship. Double-talk involves the kind of innuendo that can be conveyed by both sides and carried on for a sustained period of time. It is a kind of collusive communication different from other types of collusion in that the characters against whom the collusion is sustained are projected by the very persons who enter into the collusion… double-talk regularly occurs…as a safe means of making and refusing requests and commands that could not be openly made or refused without altering the relationship. (Goffman [1956] 1959: 194-195)
Two-tracking involves…
…double-talk + pretense + non-contamination of tracks
Two-tracking constraint
If you utter u within a two-tracked conversations, then the first track entails that you uttered u.
Two-tracking intelligibility constraint
If you utter u within a two-tracked conversation, then make sure the first track entails that in making u, you made some speech act G compatible with the first track presuppositions.
Double-talk is optional
With your turn at talk, you can but needn’t update the secondary track. But you must update the first.
The hypothesis that we are trying to deceive an onlooker—so that she believes that we are not having a second shadow conversation—explains the odd structure of these constraints.
An invisible audience?
How does and why would would this work?
In the interpersonal realm, an audience to an interaction generates additional perspective relative to which we must manage our reputational and (recognized) status.
In addition, audiences can interfere with our conversation in way that can escalate the social consequences of them.
We use strategies to manage audiences and thereby manage these risks.
Costly signaling theory: Spence 1973; Zahavi
1975; Zahavi and Zahavi 1998; Grafen 1990;
Gambetta 2011
Covertness is an essential tool for engaging in limited war. Tacitly cooperating to hide the most extreme forms of rivalry allows adversaries to operate within a kind of backstage and preserve the appearance of limited cooperation. Sequestering activity in the covert sphere reduces mobilization of external audiences, the reputational and domestic stakes involved in an incident, and hard-to-control escalation pressures…To develop this argument I draw on insights about secrecy from the sociology of Erving Goffman…Rivals may tacitly cooperate to steer dangerous encounters to the backstage as a way to safeguard to external impression of their encounter as a limited conflict. (Carson 2016: 103-104)
At an MIT faculty meeting, we are debating whether to institute a new logic requirement. During negotiations, Sam and Justin lobby passive-aggressive insults at each other. Sam and Justin intend for their other colleagues not to recognize what they’re doing. Why?
At an MIT faculty meeting, we are debating whether to institute a new logic requirement. During negotiations, Sam and Justin lobby passive-aggressive insults at each other. Sam and Justin intend for their other colleagues not to recognize what they’re doing. Why?
Containing reputational effects: If Justin knows that Brad realizes that Sam has insulted Justin, Justin has additional reason to retaliate/ retaliate more harshly against Sam. He doesn’t want Brad to think he’s a wuss!
At an MIT faculty meeting, we are debating whether to institute a new logic requirement. During negotiations, Sam and Justin lobby passive-aggressive insults at each other. Sam and Justin intend for their other colleagues not to recognize what they’re doing. Why?
Containing allies: Suppose that Alex and Sam belong to faction A and Justin belongs to faction B. If Alex knows that Sam and Justin are sniping, Alex has reason to join in on the snipefest against Justin. Rinse, repeat for all members of faction A and B.
At an MIT faculty meeting, we are debating whether to institute a new logic requirement. During negotiations, Sam and Justin lobby passive-aggressive insults at each other. Sam and Justin intend for their other colleagues not to recognize what they’re doing. Why?
Containing pro-escalation forces: Suppose that Matthias likes drama. Anytime he witnesses a conflict, he always wants to make it worse. If he knows that Sam and Justin are sniping at each other, he’ll try to find a way to “stir the pot.”
At an MIT faculty meeting, we are debating whether to institute a new logic requirement. During negotiations, Sam and Justin lobby passive-aggressive insults at each other. Sam and Justin intend for their other colleagues not to recognize what they’re doing. Why?
While Sam and Justin want to deceive third parties, their primary aim is not (or not necessarily) to escape social sanctioning.
At an MIT faculty meeting, we are debating whether to institute a new logic requirement. During negotiations, Sam and Justin lobby passive-aggressive insults at each other. Sam and Justin intend for their other colleagues not to recognize what they’re doing. Why?
While Sam and Justin want to deceive third parties, their primary aim is not (or not necessarily) to escape social sanctioning.
They are cooperating to contain risk. And in cooperating to contain risk in each other’s presence, they holistically communicate with each other.
Practical display
In performing some act G, an agent A practically displays that she has some disposition, ability, or skill X (relative to some audience B) iff:
(i) the fact that A has X explains why or how A G-ed,
(ii) in virtue of (i), B acquires evidence that A has X.
Feigning
In performing some act G, an agent A feigns (relative to some audience B) that she has some disposition, ability or skill iff:
(i) A believes that she does not have X,
(ii) in G-ing, A intends for B to infer that A G-ed because she has X.
Signal strength
(i) The less likely it is (relative to B’s information state) that A could have G-ed without having X, the stronger the signal.
(ii) The more likely it is (relative to B’s information state) that A G-ed in virtue of X, the stronger the signal.
When we two-track in front of an audience, we thereby:
…reduce the risk of relationship /status change
When we two-track in front of an audience, we thereby:
…reduce the risk of relationship /status change
…and thereby actually commit to reducing the risk of
relationship / status change
When we two-track in front of an audience, we thereby:
…reduce the risk of relationship /status change
…and thereby actually commit to reducing the risk of
relationship / status change
…and thereby credibly signal that we want and intend to reduce this risk
Practical display
When we two-track in front of an audience, we thereby:
…reduce the risk of relationship /status change
…and thereby actually commit to reducing the risk of
relationship / status change
…and thereby credibly signal that we want and intend to reduce this risk
…and thereby credibly signal that we do not want to deeply change our relationship just in virtue of our communication
When we two-track in front of an audience, we thereby:
…reduce the risk of relationship /status change
…and thereby actually commit to reducing the risk of
relationship / status change
…and thereby credibly signal that we want and intend to reduce this risk
…and thereby credibly signal that we do not want to deeply change our relationship just in virtue of our communication
…and thereby credibly signal that we can reduce the risk
Practical display
When we two-track in front of an audience, we thereby:
…reduce the risk of relationship /status change
…and thereby actually commit to reducing the risk of
relationship / status change
…and thereby credibly signal that we want and intend to reduce this risk
…and thereby credibly signal that we do not want to deeply change our relationship just in virtue of our communication
…and thereby credibly signal that we can reduce the risk
….and thereby make it rational for us each to pursue this strategy
When we two-track in front of an audience, we thereby:
…reduce the risk of relationship /status change
…and thereby actually commit to reducing the risk of
relationship / status change
…and thereby credibly signal that we want and intend to reduce this risk
…and thereby credibly signal that we do not want to deeply change our relationship just in virtue of our communication
…and thereby credibly signal that we can reduce the risk
….and thereby make it rational for us each to pursue this strategy
….and sometimes thereby make it the case that our relationship doesn’t change as much as it otherwise would have
Often the risk we reduce is status and reputation management risk.
A gangster two-tracks, when threatening a civilian to pay protection, in order to signal that the gangster does not want the civilian’s official social status to change in virtue of giving into a threat.
Two-tracking
Two-tracking is a ritualization of the signaling two-tracking strategy. What enables two-tracking are a set of mutually understood social norms, which arise in virtue of the signaling strategy and which we understand in virtue of (tacitly) understanding the signaling structure.
Two-tracking
Two-tracking is a ritualization of the signaling two-tracking strategy. What enables two-tracking are a set of mutually understood social norms, which arise in virtue of the signaling strategy and which we understand in virtue of (tacitly) understanding the signaling structure.
Genre
This is a norm-constituted but deep conversational genre.
(pace Harris and Unnsteinsson MS 2025)
Two-tracking
Two-tracking is a ritualization of the signaling two-tracking strategy. What enables two-tracking are a set of mutually understood social norms, which arise in virtue of the signaling strategy and which we understand in virtue of (tacitly) understanding the signaling structure.
Signaling
Two-tracking continues to work expressively, but the expressive engine is transformed.
Two-tracking
Two-tracking is a ritualization of the signaling two-tracking strategy. What enables two-tracking are a set of mutually understood social norms, which arise in virtue of the signaling strategy and which we understand in virtue of (tacitly) understanding the signaling structure.
Signaling
Two-tracking continues to work expressively, but the expressive engine is transformed.
The content, “I want to minimize changes in our relationship that my second track communications normally trigger” is now conventionally expressed in virtue of our selection of the genre of two-tracking.
We still credibly express and credibly commit to this content. But what generates the credibility and commitment is our use of a genre that itself imposes high risks and costs on our communication.
Dramatic communication
Two-tracking is an example of what I have, elsewhere, called dramatic communication:
a form of communication (information flow) that occurs in virtue of “what scene and persona we holistically dramatically enact and how we enact it”
the communication does not properly attach to any single speech act within the scene although it can depend upon speech act’s features
Sociolinguistic variation
Dramatic communication
Two-tracking is an example of what I have, elsewhere, called dramatic communication:
a form of communication (information flow) that occurs in virtue of “what scene and persona we holistically dramatically enact and how we enact it”
the communication does not properly attach to any single speech act within the scene although it can depend upon speech act’s features
Sociolinguistic variation
Social distance and rank variable management
Dramatic communication
Two-tracking is an example of what I have, elsewhere, called dramatic communication:
a form of communication (information flow) that occurs in virtue of what scene and persona we holistically dramatically enact and how we enact it
the communication does not properly attach to any single speech act within the scene although it can depend upon speech act’s features
Sociolinguistic variation
Social distance and rank variable management
Small talk performance
Dramatic communication
Two-tracking is an example of what I have, elsewhere, called dramatic communication:
a form of communication (information flow) that occurs in virtue of “what scene and persona we holistically dramatically enact and how we enact it”
the communication does not properly attach to any single speech act within the scene although it can depend upon speech act’s features
Two-tracking
Two-tracking is a ritualization of the signaling two-tracking strategy. What enables two-tracking are a set of mutually understood social norms, which arise in virtue of the signaling strategy and which we understand in virtue of (tacitly) understanding the signaling structure.
Function and use may differ
Once the genre becomes available, we may use the genre for different dramatic reasons. Two-tracking is fun and enables us to show off in strategically useful ways.
But what we dramatically communicate in virtue of selecting a genre is always context-sensitive.
I am explaining the emergence and persistence of the genre.
Two-tracking
Two-tracking is a ritualization of the signaling two-tracking strategy. What enables two-tracking are a set of mutually understood social norms, which arise in virtue of the signaling strategy and which we understand in virtue of (tacitly) understanding the signaling structure.
Two-tracking is expected to be universal
According to a vast philosophical and sociological tradition, there is no socially neutral way to exchange information. (Simmel 1906; Goffman 1967, 72; Rachels 1975; Brown and Levinson 1978/87)
Two-tracking
Two-tracking is a ritualization of the signaling two-tracking strategy. What enables two-tracking are a set of mutually understood social norms, which arise in virtue of the signaling strategy and which we understand in virtue of (tacitly) understanding the signaling structure.
Two-tracking is expected to be universal
According to a vast philosophical and sociological tradition, there is no socially neutral way to exchange information. (Simmel 1906; Goffman 1967, 72; Rachels 1975; Brown and Levinson 1978/87)
What information we exchange and how we exchange it always carries information about how we view our shared relationship.
So we should expect all language communities to create strategies to enable communication while mitigating relationship-change risk.
Two-tracking
Two-tracking is a ritualization of the signaling two-tracking strategy. What enables two-tracking are a set of mutually understood social norms, which arise in virtue of the signaling strategy and which we understand in virtue of (tacitly) understanding the signaling structure.
Two-tracking is expected to be universal
This genre should naturally arise everywhere because it ritualizes a natural (non-conventional) rational strategy.
To understand a conversation as such is in part to understand its organizing structure.
Q. What organizes conversation?
Grammar
Participants’ goals & participants’ rationality
Constitutive norms of conversation & of speech acts
Social norms
Social norms are not always arbitrary ways to solve coordination problems.
Social norms sometimes have proper functions: functions that explain why they persist. These functions promote the good of the group in which the norm is embedded. Sometimes, there are only a few (or one) good solution to a problem.
To understand a conversation as such is in part to understand its organizing structure.
Q. What organizes conversation?
Grammar
Participants’ goals & participants’ rationality
Constitutive norms of conversation & of speech acts
Social norms
Some social norms constitute genres that function to mitigate extremely deep tensions between our social needs and our communication needs.
Two-tracking is one such genre. It deeply and cross-culturally organizes conversation.
Thank you
Sam Berstler (MIT)
Northwestern Colloquium Series
21 November 2025
slides available: www.samberstler.com