From cooperation to collaboration via conjecture

Discoursediscourse: A communicative exchange among two or more people. can take a wide variety of forms, but academic study has been primarily concerned with the subclass of discourses that fall under the heading of cooperative exchanges of information. These discourses involve two or more individuals engaged in a project of achieving the goal of increasing their mutual knowledge, using each other’s stock of information to augment their own. In this form, discourse implements a jointly undertaken inquiryinquiry: The enterprise of forming, testing, and revising beliefs.. An inquiry is, in the words of Robert Stalnaker, “the enterprise of forming, testing, and revising beliefs” (1987, p. ix).

Discourse that implements a joint inquiry is cooperativecooperation: Alignment of the contributions of two or more individuals engaged in a joint activity. Each person ensures that their contribution accords with the contributions that have come previously and anticipates contributions to come.: A speaker presents some content in a way that makes room for responses that she anticipates her interloctorsinterlocutor: Any participant in a discourse. are able to provide. A hearer considers the contribution, adds it to her personal model of the conversation, and adjusts her response to accord with the expanded model. And the process iterates with the discourse participants alternating roles, always trying to make their contribution fit with what has predeeded it. Paul Grice took his cooperative principle, “make your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged,” to be the one general principle by which participants in conversations derive the import of their interlocutorsinterlocutor: Any participant in a discourse. utterances (Grice 1989). But discourse isn’t just cooperative in this sense – it is collaborativecollaboration: Integrated action of two or more individuals engaged in a joint activity. Each person provides an essential element to the contribution..

You and I are set to build a stone wall. We have before us a pile of stones of various sizes, as well as mortar and wheel-barrows and other tools necessary for the job. Some of the stones are small enough for a single person to move into position. To productively add these stones to the wall, one of us need only recognize the positioning of previously laid stones and place the new stone next to them in a neat fashion. This requries that we cooperate with each other in our placement of smaller stones. But some of the stones are quite large; they require us both to lift in unison and carefully place into position on the wall. To get these stones in place, we must the willing to collaborate in our efforts. While placing these larger stones adds complication through the increased demand on coordination, the value of having such a stone in place is potentially great in minimizing the overall effort required to bring our wall building task to a close.

In many ways, inquiry is a similar endeavor to wall-building. Informationinformation: Any bit of data about the world (an event or state-of-affairs) that is communicable and available to cognition. Also, the content of a contribution in discourse. takes the place of stones as the building blocks participants collectively heft, and utterances take the place of hands and wheelbarrows as the conveyances of the building materials. If all goes well, our analogy maps the finished wall to resolution of whatever issuesissue: An uncertainty within the common ground. A bit of information that is lacking. A set of alternatives none of which is settled upon as correct. spurred our entering into the inquiry.

I think that conversation commonly realizes a collaborative inquiry, in the sense that successful contribution to the discourse requires the efforts of multiple parties working in concert, not just serially. It is the goal of this project to both motivate this idea and provide a model for how it is implemented.

The principle departure point for the discussion is the speech act of conjectureconjecture: A conversational move that initiates an uncertain but potentially viable proposal for further investigation in the inquiry.. In the course of conversation, individuals will ocassionally put forward information for which they do not possess conclusive evidence. Sometimes, the basis for such a maneuver is laziness, deceit, or obstinance. But often, this kind of move is just what the conversation needs to move forward. For instance, in brainstorming sessions, the scientific context of discovery, or tip-of-the-tongue style paralysis, discurrsants set aside the restriction that all contributions be considered to be accurate in favor of entering into the conversational recordWord not found: something that the parties can use as a jumping off point. In these situations, speculative informational profferments form an important part in advancing the inquiry.

At its base, this common conversational move, a conjectureconjecture: A conversational move that initiates an uncertain but potentially viable proposal for further investigation in the inquiry., is a distinctive kind of speech actspeech act: Any action performed by way of uttering meaningful words in discourse. in which the speaker offers up her content with a world-directedWord not found: , but less than assertoricassertion: A conversational contribution that adds information to the common ground (either provisionally or automatically). An interlocutor undertakes certain commitments related to the truth of the content asserted in offering this contribution., force. But a conjecture does not advance the inquiry on its own. Putting forward information one does not stand behind as accurate risks muddling the inquiry unless it is taken as merely a jumping off point. A conjecture calls out for a response, and the contribution it makes is not complete until that response is made.

A conjecture needs to be vetted, and the canonical response to a conjecture is a move that provides this – a correctioncorrection: An attempted contribution to inquiry that involves a denial of some element of the preceding attempted contribution to discourse and a positive contribution that is related in a salient way to the preceding attempted contribution.. Corrections can take many forms, but they normally involve two parts, a denial of some portion of the information previously put forward and a substitution for the denied information. The conjecture lays a rough foundation, and the correction builds upon it.

The dual conversational moves of conjecture and correction are what I call a discourse pairdiscourse: A communicative exchange among two or more people.. The key precept of collaborative discourse dynamics is that discourse pairs are the basic units by which contributionscontrastive topic: An information structural element located within the topic of a sentence that sets its scope in contrast to a (commonly contextually determined) set of alternatives. to the inquiry are made. That is, we do not represent the conjecturer and corrector as each providing indvidually complete and jointly cooperative contributions to the discourse. Instead, the conjecture/correction pair together contribute to the inquiry; the parts are individually inert.

The project of using the notion of discourse pairs to develop and motivate a theory of collaborative discourse dynamicsdiscourse dynamics: has 3 parts:

Conceptual: Deriving collaborative discourse from inquiry by way of knowledge. Inquiry is deeply collaborative, and via a mutual, conceptual connection to knowledge, so too is discourse.

Theoretical: Development of a deeply collaborativecollaboration: Integrated action of two or more individuals engaged in a joint activity. Each person provides an essential element to the contribution. account of discourse dynamics and a formal model of conjecturesconjecture: A conversational move that initiates an uncertain but potentially viable proposal for further investigation in the inquiry. and correctionscorrection: An attempted contribution to inquiry that involves a denial of some element of the preceding attempted contribution to discourse and a positive contribution that is related in a salient way to the preceding attempted contribution. within the update semanticsupdate semantics: Any of a family of theories of formal semantics that treat contributions in discourse as ordered sets of additions to and operations on the discourse platform. tradition.

Practical: Application of the collaborative framework to the phenomenon of contrastive topiccontrastive topic: An information structural element located within the topic of a sentence that sets its scope in contrast to a (commonly contextually determined) set of alternatives., implications of the collaborative model for the concept of linguistic disagreementdisagreement: , and the contextualismcontextualism: The thesis that semantic value is, at least occasionally, determined in some way by the context of the speaker of an utterance. May encompass the broader thesis that semantic value is determined by features of context generally. Usually contrasted to relativism and absolutism. v. relativismrelativism: The thesis that semantic value is, at least occasionally, determined in some way by the context of the interpreter of an utterance. Usually contrasted to contextualism and absolutism. debate in semantics.

Knowledge in the image of inquiry

It is tempting to maintain that the end of inquiryend of inquiry: The goal state of the process of inquiry, generally resolution of the issue(s) for which the inquiry was undertaken, as well as any sub-issues raised during course of the inquiry. is truth. While possession of the truth may be what the inquirer is after, she wishes to possess it in such a way that she both recognizes and accepts it as the truth. The goal of engaging in inquiry is to achieve a cognitive state with true propositions as its content.

In contrast, C.S. Peirce and others following in his wake (Levi 1980), have suggested that truth is too lofty a goal; that the most we can genuinely hope for at the end of inquiry is the settlement of opinion, for:

“it is clear that nothing out of the sphere of our knowlege can be our object, for nothing which does not affect the mind can be the motive for mental effort. The most that can be maintained is, that we seek for a belief that we shall think to be true. But we think each one of our beliefs to be true, and indeed, it is mere tautology to say so.” (Peirce 1877/1955, p. 11)

In response, we could take the pragmatist bait and insist that truth is the end of inquiry simply because whatever is output from the process of inquiry just is truth. But we needn’t take this line. We can cleave to a correspondence account of truth while maintaining that final aim of inquiry involves wide contentwide content: Mental content that essentially involves a part of the mind-external world. by delimiting a natural, restricted category of proper inquiry that can be reliably expected to produce true propositions.

Principally, proper inquiry is a rule governed process. Rule-following has the practical advantage of improving the reliability of the inquiry, but the connection to rule following is a conceptual one. The cognitive state that the inquirer seeks is praiseworthy, and one does not garner praise simply by stumbling onto true propositions. Undertaking the process of inquiry is important to earning the accolades that come with possessing the truth.

The end of inquiry, then is a valuable, cognitive state with true propositions as its content. Specifically, proper inquiry is a process that reliably produces justified, true belief. Modulo Gettier-type complications, the end of inquiry is knowledge.

If we had a solid understanding of the nature of knowledge, we could use the end of connection to extend our understanding to encompass inquiry as well. But the nature of knowledge is highly contested, and connections go both ways. I propose to investigate the basic structure of proper inquiry by looking to specific historical accounts of its importance, to take at face value the idea that the outputs of the process of inquiry are knowledge, and thereby to provide insight into the structure of the knowledge relation.

The key insight that arises from this investigation into proper inquiry is that it is a deeply collaborative process. By way of the end of connection, we have reason to represent the knowledge relation as similarly collaborative. In one sense, this idea is quite radical, in that it seems to eliminate the potential for solipsitic knowledge. But we can perhaps buff a few of the burrs off this conceptual edge, and the proposal garners support from the puzzles it can help to solve.

Knowledge and the process of justification

One such puzzle derives from the fact that knowledge is more valuable than mere true belief, but as far as all one wants is to get to Larissa, a merely true belief will do as well as a justified, true one. The swamping problem is that there seems to be no value over and above the truth of the justified belief provided by its being so justified.

Reliabilist accounts of knowledge offer a reasonable story about the justification process, but they go wrong in suggesting that the value inherent in the process transfers over to its product. Linda Zagzebski (2003) maintains that this problem suffices to undermine the reliabilist approach. But this is true only if we commit to treating knowledge as merely the output of a justifiying process. If we build the process into the knowledge relation, there is no need for the value to transfer at all.

The reliability of beliefs formed via inquiry is incidental. The central virtue of inquiry is that it provides one with an account. It allows an individual to stand up to challenges of their belief. And possessing this is more valuable than merely possessing the true belief. Socrates speculates with Meno that what separates knowledge from true belief is that knowledge is tied down by an account. The account doesn’t get you to Larissa any more surely, but it can help you respond to detractors you find along the route. And this ability is worth posessing.

The inquiry-first account of knowledge is compatible with a view in which knowledge is the constitutive norm of assertionassertion: A conversational contribution that adds information to the common ground (either provisionally or automatically). An interlocutor undertakes certain commitments related to the truth of the content asserted in offering this contribution.. But it is ecumenical, in that this does not mean giving up on the virutes of a commitment-based account of the speech act (MacFarlane 2005). Since knowledge contains the commitment fulfilling process of inquiry as a part, it can serve as both the norm of assertion and ground the constitutivity of other commitments engendered by assertion.

Dispute and collaborate

Disagreement has been a bit of a star figure in recent debates in the philosophy of language. Primarily, it has been used as a tool for assessing the adequacy of different proposals for the semantic contribution of certain expressions. Genuine disagreements, as opposed to spurious or merely apparent ones, intuitively require in the linguistic context or the minds of the disputants, the presence of some kind of content toward which the participants have incompatible commitments.

The contestants in battle disagreement are absolutism, contextualism, and relativism, and the standard scorecard has absolutism and relativism coming out on top in virtue of their ability to secure the requisite constant content across uses of the expression. Contextualism, it is charged, assigns contents to the utterances and thoughts of different individuals that have them incurring different, compatible commitments as a result of their utterances or thoughts.

Whatever stance we take on the outcome of this debate, disagreement does seem to be a valuable tool to have in the linguist’s toolkit. Language users have intuitions about the compatibility of sentences used in discourse, and linguistic theories ought to respect those intutions. But simply marking the presence of incompatible contents in a discourse is a relatively peripheral role for disagreement to play in lingusitic theorizing. As they stand, the semantic theories considered above give no insight into disputative discourse itself; they merely accord to a greater or lesser extent with one interesting consequence of such discourse.

The previous discussion has shown that dispute plays a significant role in the very structure of knowledge. Given broadly accepted connections between knowledge and discourse maneuvers, it also has implications for understanding the nature of discourse dynamics.

Collaborative update semantics

Conjectures are best understood as one half of a pair of moves in a discourse. A contribution is incomplete without both parts being incorporated. In general, discourse pairs are completed by someone other than the one who initiates them (Clark and Wilkes-Gibbs 1986). Thus, discourse pairs represent genuinely collaborative contributions to the inquiry, which can be fruitfully modeled within a natural extension to the update semanticsupdate semantics: Any of a family of theories of formal semantics that treat contributions in discourse as ordered sets of additions to and operations on the discourse platform. framework.

Standard update semantics is build upon a system of sequential additions to and operations on a shared discourse spacediscourse: A communicative exchange among two or more people.. By supplementing this framework with a middle ground working space, a discourse sandboxWord not found: , we can represent interlocutors’ joint contributions, decomposed into initiationinitiation: An utterance that constitutes the opening phase of a contribution to inquiry. An initiation is paired with a completion, and the resulting discourse pair provides the contribution to the inquiry. and completioncompletion: An utterance that constitutes the closing phase of a contribution to inquiry. A completion is paired with an initiation, and the resulting discourse pair provides the contribution to the inquiry. phases.

I consider accounting for conjecture to be an adequacy condition on semantic theory in that conjectures form a well-defined class of linguistic tools that do important interpretive work that cannot be done better by some other (set of) tool(s).

Genuine contribution types

A conjecture is a kind of conversational move. It is defined in terms of the role it plays in advancing the inquiry in which it occurs. While speech acts are finely individuated, there is a small set of contributioncontribution: Any linguistic performance, or the semantico-pragmatic structure associated with it, that advances the inquiry realized by the discourse of which it is a part. types derivable from the goals of inqury.¹

The primary goal of inquiry is to come to know something one did not know at its outset.² Thus, an individual can advance us toward that goal by submitting (proposing) information to be taken on as mutually accepted. We call this kind of contribution an assertionassertion: A conversational contribution that adds information to the common ground (either provisionally or automatically). An interlocutor undertakes certain commitments related to the truth of the content asserted in offering this contribution..

The end point of an inquiry depends in part on its starting point, which is the issue that is to be resolved by the process of inquiry. An individual can submit a new issue to the inquiry to be taken on as providing a new goal for the inquiry. We call this contribution a questionquestion: A conversational contribution that adds an issue to the common ground with the implication that the issue is to be resolved in the course of the inquiry. Acceptance of the question commits interlocutors to providing contributions that offer at least partial answers to it..³⁴

These commonly cited examples do not provide an exhaustive list of conceptually possible contribution types. I don’t intend to provide identity conditions for types of contribution, but the examples do share certain things in common, and their common traits can provide us with criteria for recognizing novel contribution types.

• Genuine contribution types operate on informationinformation: Any bit of data about the world (an event or state-of-affairs) that is communicable and available to cognition. Also, the content of a contribution in discourse. in a distinctive way. Assertions expand the context set, and questions partition it. The currency of inquiry is information, and its value is tied to the ways in which information can be put to use to advance the inquiry toward its goal.
• Conversational moves that contribute to inquiry anticipate an appropriate response, and place constraints on the form of the response. Assertion anticipates either acceptance or rejection of the asserted content, and (modulo repair tactics for failed presuppositions) restricts responses to either acceptance or rejection. Questions anticipate an answer, and restrict responses to those assertions that at least partially resolve the issue raised.⁵
• Genuine contribution types are performed by way of a conventionalized linguistic tool. If a contribution belongs to a specific type, language will have found a way to implement that type. Assertions and questions are stronly tied to linguistic mood, although the connection between these contribution types and mood is a tricky one (Portner 1997) (starr2010).

I maintain that conjecture, when paired with a corresponding correction, constitutes a distinct, genuine contribution type. Conjecture is similar to assertion in that it seems to provide a proposal of added information. And similarly to questions, conjectures seem to present an issue that is up for further discussion.⁶ But conjectures are distinct from each of these, and call for a unique treatment. That they earn such status is what I hope to show throughout the course of this project.

Levels of dynamism

It is folly, Frege (1884/1997) warned, to ask for the meaning of a word in isolation, prescribing instead that the sentence as a whole be the proper unit of semantic investigation. The dynamic turn in semantics did Frege’s context principle one better. On the basis of anomolous linguistic data involving anaphora resolution, attention was shifted from single sentences to the broader linguistic context in which their specific instantiations appear. Thus it has been suggested that the proper units of linguistic evaluation are extended texts or conversations, what I refer to here as discoursesdiscourse: A communicative exchange among two or more people..

The many varieties of dynamic semantic accounts share in common a model of discourses in which there is exists a theoretical platform upon which information is built in a stepwise manner throughout the discourse. Each subsequent contribution to the discourse is analyzed in terms of what it constructs upon the platform or what previous construction it wipes away. The materials of which discourses are built include possible worlds, situations, events, discourse referents, and set-theoretic constructs thereof. This platform and its associated construction go variously by names such as discourse representation, conversational scoreboard, common ground, and information state. I couch my discussion within this broad tradition of theories and refer to the semantic representation variously as the information state or state of information of the discourse.

These semantic theories are dynamic in that they describe how new contributions to a discourse respond to and modify what has preceded, and discourse is very much dynamic in this sense. But by and large, they fail to capture another way in which discourse is dynamic. Contributions to discourse also anticipate what is to follow in the discussion, and this feature of how discourses proceed ought to be brought under the umbrella of semantic theory.

Initiations, defaults, and sandboxes

To capture this anticipatory level of dynamics, I propose that we treat individual utterances as together forming a single evaluable contribution. We can model this idea by incorporating another Fregean insight into update semantics, treating the initiation phase of discourse contributions as incomplete, with the completion phase saturating it with the requisite informational material. It is only once saturated that the discourse content of an utterance is brought to light. However, initiations will ocasionally be treated on their own for purposes of evaluation. While I contend that instances where this is done are limited to non-canonical contexts such as presuppostion repair, it isn’t strictly true that initations are inevaluable without a complementary completion. For this reason, we must also incorporate a notion of defaultWord not found: saturationWord not found: , in which context supplies the material needed for evaluation of the initiation. Defaults turn an initiation into a complete contribution, but they are also subject to override by a standard completion presented by a discourse participant.

The resulting picture is one of an extended and evolving, collaborative process surrounding each contribution to discourse. Initiations undergo a vetting procedure before they are entered into the official conversational record. To avoid contamination of the common ground with the untreated mess that many initiations carry, interlocutors must isolate the collaborative process from the conversational space. For this purpose, I draw the notion of a sandbox from the field of computer programming. A sandbox is platform cordoned off within a system where untested code can be executed and debugged under real conditions but without infecting the system as a whole if things go wrong. If the executed code is found to be functional, the restrictions can be lifted and it can be incorporated into the system. By introducing a sandbox-style working space within the conversational space as a whole, we can incorporate collaboration into standard update semantics.

The system of initiations, defaults, and completions, encompassed within a sandbox operating theater, provides our update semantics with a new type of dynamism. Standard dynamic accounts have it that utterance meaning depends on the surrounding linguistic context. But in collaborative update semantics, utterance meaning can actually change as the conversation progresses, and the status of an utterance evolves from initiation saturated with a default, through the sandbox vetting, and into the conversational record as a completed contribution.

Collaborative disagreement

Disputes can be frivolous, fractured, and fraudulent. They can derail conversations, and generate so much animosity that perpetuating the dispute itself comes to mean more than the prospect of resolving it. These disputes may have very little to do with actual conflict over the informational potential of the disputative language used to express them, instead being driven by ad hominem and ad ideologiam tendencies.

But dispute can also be a fruitful tool for people involved in an honest, cooperative attempt to further mutual knowledge. There is the ancillary fact that interlocutors sometimes lose track of the cooperative effort, and it is valuable to have a means of correcting them to bring them back on to the task at hand. But dispute can play also more central role in the project of inquiry. As we have seen, the speech act of correction, in which an individual denies a proposal initiated by an interlocutor is a crucial element in properly vetting conjectures in discourse. As such, some element of dispute is central to the enterprise of inquiry. I contend that the conjecture/correction discourse pair provides a genuine contribution to inquiry. In the previous chapter, I motivated this status by providing a model of discourse dynamics that carves out a distinctive operation on information that this contribution makes. In this chapter I explore another criterion for genuine contributions: that it be represented in some conventionalized linguistic tool. I think that this tool is focusfocus: An information structuring device of natural language. It serves to divide the content conveyed into parts based on the role they play in the broader dicourse. A principle such division is between given and new information. In English, focus is usually realized as intonational stress., and that the virtues of collaborative discourse dynamics come to prominence in explaining focus interpretation when we examine contrastive topiccontrastive topic: An information structural element located within the topic of a sentence that sets its scope in contrast to a (commonly contextually determined) set of alternatives. specifically.

Consider the following little discourse:

Pretheoretically, the response utterance seems to do two things:

• It denies an element of the preceding contribution to discourse.
• It provides a positive contribution that is related in a salient way to the previous contribution.

I think that it pulls off this dual task because it serves as a completion of the initiation provided by the first utterance. To motivate this claim, I look at two alternative accounts of this sort of disputative discourse and find them lacking.

Questions under discussion and plans of inquiry

A salient feature of the Bears dialogue is the presence of focusfocus: An information structuring device of natural language. It serves to divide the content conveyed into parts based on the role they play in the broader dicourse. A principle such division is between given and new information. In English, focus is usually realized as intonational stress. on grizzly in the response utterance. Focus is a lingusitic tool whose use serves primarily to package information as opposed to providing its own contribution. It distinguishes the elements of a sentence that are new to the discussion from those that have already been introduced, thus showing to the other participants in the discourse how the contribution is intended to fit in with what has preceeded. In this sense, focus has a distinctly backward looking function. It depends on, and responds to, previous moves in the conversation. So it makes sense that the most common analyses of focus interpretation assign it a presuppositional pragmatic function. Focus adds a felicity condition to the overall import of a sentence; the sentence is interpretable in situ only if material matching the focus semantic value of the sentence can be found in the previously constructed conversation.

The Question Under Discussion model of discourse dynamics, developed principally by Craige Roberts (1998), expounds on the alternative semantics account by treating discourse as governed by a stack of questions. The governing question at any point in a discourse may have been explicitly asked or it may be implicit, presumed by the discourse participants and revealed in the information packaging of their utterances.

Daniel Büring (2003) (,buering2008) extends the QUD model to capture second-occurrence focus, also called contrastive topic. But there is an issue in this extension when applied to cases such as the Bears dialogue. It is not so much an issue of empirical coverage as it is a conceptual issue having to do with the role of corrections in discourse evolution. The QUD model of contrastive topic makes its discourse contribution entirely backward looking. Whatever role the focus plays is just a matter of checking prior discourse for an antecedent to the focus-generated alternative set.

The Bears dialogue is an example of felicitous use of contrastive focus in which there is no expicit question on the table at all. It seems to me that the role of the focus in Bears is more than simply to mark the way the response fits into previous discourse; it genuinely advances the inquiry along by correcting what predeeded. But to capture this added feature, we needn’t dispose of Büring’s useful plan of inquiryplan of inquiry: The structure of issues and proposals governing the inquiry at hand. The POI places constraints on appropriate future moves; in order to be relevant, they must fit into the plan in some way. entirely. What we need is an expansion of the structure involved in such plans to include conjecture/correction congruence in addition to question/answer congruence. Focus, fundamentally, serves to indicate information structure, but information structure includes much more than just what is given and new. It incorporates a record of where the inquiry has been as well as where it is going.

Correction as downdate

Equally salient in the Bears dialogue is the disputative nature of the response statement. It seems to simultaneously deny the original statement and substitute an alternative to it. Let’s call utterances with this dual function corrections. Jennifer Spenader and Emar Maier (2009), expanding on work of Maier and Bart Geurts (1998) (,geurtsa), attempt to treat corrections as downdating the information state. On this proposal, the function of corrections is to remove from the discourse representation something that was previous added to it. A wrinkle is that corrections do not eliminate information wholesale. Consider, fo example:

Crucially, the correction above does not eliminate the information that the movie was funny. In fact, the replacement is perfectly compatible with this information. What the correction denies is the presupposition of the initial statement that the moview was merely funny. To account for the nuanced nature of corrections, Spendader and Maier develop their proposal within the framework of Layered Discourse Representation Theory (LDRT), in which information states are segmented and different information contributed to discourse is tagged and placed in different segments depending on its import. Corrections can eliminate information from one segment without affecting the information in any other segment.

The challenge is to track information across utterance boundaries. LDRT allows for a single bit of information to be added to multiple segments, but when corrections take place at a remove from the utterance they correct, the proposal runs into a binding problem. This problem is, I think, undermining, and it reveals an inherent limitation of the LDRT account. LDRT treats utterances as providing independent contributions to the discourse either in the form of updates or downdates. But the Bears dialogue is not just a pair of independent utterances. They cohere in an important way that should be accounted for in any representation of the discourse.

Collaborative update semantics

If we treat the Bears dialogue as a minimal example of collaborative discourse, we can capture the forward-looking import of corrections while representing the strong coherence of the utterances in the dilogue.

According to collaborative update semantics, the first utterance in the dialogue initiates a contribution that is completed by the second, correcting utterance. Focus is treated as a sandboxing marker and denial serves to replace the default saturation of the propositional radical with an alternative one.