Printer Friendly


1. Introduction

Although instances of predictive illocutions are not uncommonly referred to in empirical and formal studies of natural languages, systematic accounts of the very notion of prediction are rather scarce. Furthermore, there is a widely shared assumption in the literature that (nearly) all future assertions are predictive in nature. This is commonly supported by the epistemological claim that, given that the future is uncertain, any assertion that refers to a future event is predictive in its very essence. The assumption reaches far into compositional approaches to meaning as well, some of which tend to define the denotations of future morphemes by referring (informally, perhaps) to such a notion. (1)

In this paper, I would like to cast doubt on this assumption and propose an illocutionary distinction between predictions and what I will call "prospections." I do this in two stages: first, by showing that definitions in traditional accounts of predictive illocutions are in need of more restrictive amendments; and second, by showing that a dynamic approach to meaning can, in effect, account for the different elements that, in given conversational frame, map the (primitive) future meaning of a sentence into either predictive or prospective illocutionary force.

My critical approach brings to the fore one substantial caveat for speech act theories: that not every assertive speech act that refers to a future event is predictive in nature. More constructively, my proposal suggests that sentences which encode future meaning are understood as having predictive illocutionary force only under certain assumptions regarding the type of justificatory evidence and what is actually added and recorded in the conversational score at utterance time.

The paper is organized as follows. In section 2, I offer a critical review of one standard definition of predictive speech act, and suggest some amendments reflecting on a few counterexamples. Taking on these results, section 3 draws a distinction between prospective and predictive illocutions, and provides a contrastive dynamic analysis of prediction and other future oriented speech acts.

2. Predictive Illocutions: The Standard Definition

Although the literature on the subject of speech acts is considerably vast and well documented, formulations of a definition of predicting are rather uncommon. One particular attempt is the following:
To predict is to assert with the propositional content condition that
the propositional content is future with respect to the time of the
utterance and the additional preparatory condition that the speaker has
evidence in support of the proposition (Searle & Vanderveken, 1985:

I will refer to the above as the "Standard Definition" (for short: SD) and its background theory as the "Standard View" (for short: SV). (2) As it stands, SD reveals a basic tenet of SV: an illocutionary act is defined by a set of necessary and sufficient conditions for the performance of the act. (3) This preliminary setup provides a substantial starting point: predicting is not defined in terms of syntactic structure, prosodic patterns or social effects, but rather, by spelling out conditions that are primarily related to propositional content and the speaker's justificatory potential: (i) the propositional content condition that the content is future with respect to speech time, and (ii) the preparatory condition that the speaker possesses evidence in support of what is said. In the current section I will provide a critical examination of SD, reflecting on some counterexamples, and suggest some possible amendments. This preliminary approach will clear the ground for a more constructive approach in section 3 by putting in perspective the basic epistemic and semantic elements involved in predictive illocutions (and future oriented speech acts more generally).

An immediate problem with SD is that, as it stands, it is too strong, in the sense that it rules out illocutions that one would intuitively class as predictive. The problem is related to SD's first condition (namely, that the propositional content is future with respect to speech time). For one can certainly find instances of predictive illocutions that are not, strictly speaking, about the future (and, accordingly, that are not conveyed by future tensed propositions). One example would be "I predict that Mary won the election," uttered at a time at which the election has already been held, but participants remain ignorant of the results. (4)

A defender of SV can object to this critical point on methodological grounds: in theorizing about speech acts, it is recommendable to generalize in view of "paradigmatic" or "prototypical" cases, to only then accommodate the more exceptional ones. I will not assess this methodological issue here, but rather concede the point. For even though this first caveat hints at a critical point I will reflect on below (namely, that the assertion/prediction distinction is not reducible to purely temporal terms), my argumentative strategy will focus on the more characteristic futuretensed type of construction, to then argue that a large subset of them are not used for predictive purposes. This will make a stronger case for the overall critical aim of the article. As I hope it will be made clear, that not every future sentence is necessarily mapped into a predictive illocution is not inconsistent (and actually resonates) with the acknowledged fact that past tensed constructions can also yield, in relevant scenarios, a predictive effect.

Now, a second problem with SD is that it is also too weak, in the sense that it is satisfied by entities and events (namely, linguistic constructions and their utterances) that do not pertain to the class of things that are being defined. And this permissiveness is, of course, one major problem for any definition in any field of study. By way of illustration consider the following two utterances:

[1] Elections will be held in Wales on June 6th next year (as it is scheduled).

[2] Mr. Jones will win the Wales elections on June 6th next year.

The basic point here is that both [1] and [2] satisfy SD, but only [2] is meant as a prediction whereas [1] is not. In effect, the case with [1] is such that even though its propositional content is future oriented (i.e., it satisfies SD's propositional content condition), and even though the speaker may hold evidence in support of it (i.e., it satisfies SD's preparatory condition), speakers would not naturally report the utterance of [1] as making a prediction. And one chief reason why one would not consider the utterance of [1] as a clear-cut case of prediction is, arguably, that predictors do not characteristically target the holding of scheduled or planned events (that elections will be held in Wales on such-and-such day in the future), but rather, quite specific and uncertain aspects of these and other events (that either Mr. Jones or Mr. Davies will win the election). This is arguably why news reports often cover who predicts which candidate will win (and retrospectively, who got it right and wrong). It would be awkward, to say the least, a news report on who predicts what day the election would take place when a date has already been set. Thus, if we take the same speaker to utter [1] and [2] in the scenario where the date of the election has already been set, participants would naturally tend to report only [2] as a prediction. (5)

At this point, one may wonder whether a notion such as inertiality can be of any analytical use. The notion is, in effect, not new to semantic analysis of future oriented illocutions (see Copley, 2009). Informally, it refers to that aspect of an event in virtue of which the event is to be held in the future (either by means of convention, agreement, commitment, habit, regularity, nomic necessity, inter alia). The rationale for invoking such a notion in this particular regard would be that [1] constitutes a counterexample of SD in virtue of the fact that the speaker targets (an aspect of) an event that is inertial, whereas it is usually expected of predictors that they target those (aspects of) events that are non-inertial in nature, as the speaker who utters [2] exemplarily does. (6) Accordingly, what seems to be needed for a more restrictive definition, is a qualification to the effect that the event referred to by issuing a prediction is not of the inertial kind, but of the non-inertial kind.

The idea resonates with Searle's own constraints on other future oriented speech acts, such as promises:

[3] Preparatory Rule (promising): PRO is to be uttered only if it is not obvious to S and H that S will do A in the normal course of events (7) (slightly modified from Searle, 1969: 63, and 1996: 10-11, my italics).

Now, does [3] provides an indication of how a preparatory rule for prediction (PRE) shall be spelled out? Prima facie, what was needed to be brought into our definition was a qualification to the effect that the event referred to by issuing a prediction is not of the inertial kind, but of the non-inertial kind. However, notice that what is specified in Searle's rule as a condition of the illocutionary act is not exclusively related to the nature of the event, but to the knowledge and evidence that participants have of its coming into existence. For there is a subtle point in [3]: the condition spelled out is not imposed on the nature of the action/event itself, but on how that action/event is perceived by the participants. And this is a telling difference: what is at issue is not that the event shall not be, as a matter of objective fact, inertial, but rather, that it shall not be obvious, for both participants, that it is. This actually brings a significant effect when one considers its potential application to the case of predictions. For, a putative predictive analogue of [3] would leave one possibility open: that (objectively) inertial events, which are not known by participants to be so, satisfy the condition and consequently qualify as predictable.

And, in effect, such cases are not infrequent. A simple example to illustrate: a prediction about an event that has been scheduled to occur (say, a football game), but whose arrangement remains utterly unknown to participants. Under those circumstances, the speaker (perhaps on his way to the pub with the addressee) might speculate and make all sorts of conjectures about when the event shall occur, to eventually utter:

[4] The game will begin at 18:00.

In sharp contrast to [1] (and despite the fact that the event at issue is, by hypothesis, inertial), [4] does intuitively qualify as a prediction. And the main reason for that being the case is preparatory in nature: simply put, it is not obvious for the participants that the event will hold at the time that speaker predicts it will.

Let me illustrate this same point with a contrast. Consider two speakers, in diverting circumstances, issuing the same utterance and referring to the same event. The first speaker (let us call him "The Man in the Station") is sitting beside you in the train station and, while reading the timetable of departures on a screen, he utters:

[5] The train to Liverpool will depart at 11:00.

A second speaker (let us call him, "The Man in the Street") is on his way to the station, with no timetable at hand, nor any device to find out at what time the same train to Liverpool departs. Walking beside you, he mutters what seems to be a speculative ponderation regarding time patterns of arrivals and departures, perhaps remembers a few (rather inconclusive) facts, to eventually utter [5].

Now, it seems natural to report that while "The Man in the Station" is not issuing any prediction (but, perhaps, only conveying information), "The Man in the Street" does. And it is also apparent that the reason why this is the case does not relate to the inertiality of the event: by hypothesis, one and the same event has been scheduled, by competent authorities, to occur at 11:00. Rather, a telling difference between the two cases concerns the epistemicity of the same referred event: only to the man in the station (but not to the man in the street) is it obvious that the event will occur, as a matter of inertial course, at 11:00.

The above suggests that what is (and what is not) a qualifying subject matter of a predictive illocution is not determined by the ultimate objective nature of the referred event, but by the participants' (fallible) recognition of the relevant particular aspect of the event at speech time. (8) By the same token, what precludes SD from becoming a satisfactory definition of predictions is not the omission of a certain kind of future events (i.e., non-inertial future events), but of a certain kind (or amount) of (fallible) knowledge.

All of which leads us to the issue of evidence. And one thing that is worth noticing in this respect is that we seem to be facing a tension. And the tension is the following. A very essential claim in SD is that the predictor must have evidence in support of what he asserts. However, and considering our working (counter)examples, if one were to identify an asserter possessing robust evidence in support of what is said, that would be the man in the station, not the man in the street. Arguably, seeing the information on the screen is more reliable evidence than the conjectural, albeit successful, inferential process made by the man in the street. However, as it turned out, it is the man in the street who is issuing a prediction, not the man in the station. Evidence, thus, does not seem a discerning element after all. So either we delete from SD any constraint related to it, or we standardize the evidence possessed by the speaker so as to effectively rule out (cases like) the man in the station.

The first strategy does not seem very promising: dispensing with evidence as a necessary condition would leave room for blind guesses to qualify as predictive, and we certainly do not aim for such permissible definition of prediction. In our example, it is true that the man in the street does not possess compelling evidence (as the man in the station perhaps does), but he certainly has some evidence which allows him to do his ponderation. Otherwise, we would not consider him issuing a prediction.

The second strategy can attain a better effect in this same respect, although it may be a contentious issue how to standardize evidence. To clear the ground, I think a basic distinction is appropriate here, namely, that between constraints concerning the amount of evidence and those concerning the type of evidence. For it might be the case that, despite the fact that the man in the street is never in the position to provide conclusive evidence for what is said, evidence is nevertheless required, if only to a certain degree or of a certain kind, for us to ascribe him a predictive illocution.

Let us consider, first, quantity. The rationale of our putative constraint can be stated in the following lines: the predictor must have some supporting evidence for what he asserts (so as not to turn his future assertion into a blind guess), but not as much as to turn his utterance into a simple informative statement. In other words, the predictor must be in possession of some amount of evidence, yet only to the extent that he remains uncertain to a modest but relevant degree. That is why he predicts rather than assures, that is why he predicts rather than announces, that is why he predicts rather than reiterates what has been already taken for granted. Let me call this the Modest Evidence Condition. It seems obvious that, as a constraint, it could be easily added on SD. This would render SD*:


To predict is to assert with the propositional content condition that the propositional content is future with respect to the time of the utterance and the additional preparatory condition that the speaker has modest evidence in support of the proposition.

But, is this right? Modest evidence as a constraint for predictive illocutions? There is one serious problem with SD*: the definition rules out one particular and reputable subclass of predictive illocutions, namely, justified predictions. For, certainly, there is nothing contradictory in the idea of a prediction (scientific or ordinarily tailored), that is issued by some intelligent being in possession of an amount of evidence that turns his assertion into something very close to conclusive. In effect, robustness (rather than modesty) is the standard required for scientific predictive procedures. True: ordinary predictions, the one made by the man on the street, can tolerate modesty (as my examples above have shown), but modesty could not possibly be a general defining feature of predictive illocutions.

Let us, then, shift the focus from quantity to quality: that is to say, from the amount of evidence to the kind of evidence required. By this token, what determines what is predictable and what is not shall not rely on the degree of conclusiveness reached by the predictor, but on the type of evidence he possesses. And one reasonable qualification in this respect is the following: if one were to identify an aspect that is common ground for both scientific and ordinary predictions, it would be reasonable to point to some sort of personalized inferential process that justifies what is said. By stressing "personalized" I just intend to denote the idea of "taking part of," as opposed to merely acting as an "information-conveyor" that does not (at least not necessarily) participate in the inferential process that leads to the conclusive illocution. By these standards, cases such as the man in the station are easily ruled out, on the basis that the information about the future was merely conveyed by a deferential process (i.e., a process by which a speaker reasserts a reputable justified statement). In a more schematic way, a deferential process can be described as follows: when an asserter A utters p and p is justified as a conclusion, an asserter B can make use of his right to reassert p (or use p as a premise for further assertions) in virtue of A's reputation as an asserter. (9) To illustrate with another simple example: if you tell me that Cristina will go to Paris for the summer holidays (and openly let me know that she told you about her having already chosen that destination), it would be awkward, to say the least, that I describe you as making a prediction about her holidays. (10) But if none of us knew what her holiday plans were, and you start making some conjectures and speculations regarding her likes, past summer histories, perhaps flight offers and such, to finally conclude that she will choose Paris as her summer destination, it would come more than natural to ascribe you a predictive illocution. Needless to say, unless you are challenged (or, to the same effect, gently asked to justify your claim) you are not under the obligation to speak aloud the inferential chain that led you to your prediction. And in the case I am dubious whether you meant your statement as a prediction or simply as conveying information from an authorized source (namely, from Cristina herself), I can simply ask you to make explicit your evidence ("how do you know?" would suffice). Alternatively, I can rely on other evidence, such as asking Cristina if she told you. Notice, however, that this is not evidence about the facts (I am not asking Cristina if she will choose Paris as her summer destination), but evidence about evidence (I am asking her if she told you so, so I can get clear on how you meant your statement).

To summarize a bit, the distinction regarding evidence that I am introducing at this point is twofold. On the one hand, I distinguish between (i) first-order evidence a speaker may hold to justify his assertion (i.e., evidence about the facts referred to by the proposition), and (ii) second-order evidence a participant may hold about the kind of evidence that a speaker entertains in issuing his assertion (evidence about evidence). (11) On the other hand, I would like to draw a further distinction between (iii) inferential evidence (evidence that relates to some rational process) and (iv) deferential evidence (evidence that stems from a reputable original source) to justify a claim. Thus, with these distinctions in mind, the discussion above suggests that the first-order evidence that a predictor holds in issuing a prediction must be inferential in nature. Back to our original examples: the man in the station has reliable deferential first-order evidence to justify his utterance (evidence from an external authorized source and about the facts). The addressee, however, who is sitting beside him and sees him reading the timetable screen as he utters [5], handles a second-order type of evidence: evidence that the speaker's evidence is deferential. And because the addressee holds this type of evidence, she would not ascribe to the speaker's (honest, informative, justified) utterance a predictive force.

The intuitive picture suggested above is that of a predictor who, regardless the certitude reached by the evidence he possesses, entertains some sort of inferential procedure. Our paradigmatic counterexample, a man in the station looking at a timetable and simply reporting train departures to us on a deferential basis, brought some evidential elements to the fore. At this point, I would like to put some of these elements into a slightly different perspective, and reach a more conclusive diagnosis.

It is worth noticing that, although the deferential chain in our working example leads to an informational source stocked in the form of a timetable, the timetable itself is not what we characteristically conceive as an ordinary speaker. And here the thought goes that perhaps a better way to conceive timetables is by ascribing to them some sort of commissive aspect, as it seems that the core element conveyed by this informational source does not come in the form of an assertion about the future, but as an implicit commitment to the relevant course of action that leads to the arrivals/departures at the stipulated time. If this line of thought is on the right track, stretches of discourse such as timetables and fixtures are not entirely assertive in nature, insofar as they commit a background asserter (perhaps institutionally) to bring about the relevant events at the relevant times.

It is also worth noticing the role that the man in the station plays in this hypothetical conversational scheme. On the one hand, as a mere informant of the timetable contents, he is not engaging in any commitment (but implicitly reporting on one). On the other hand, as long as he only "re-assert[s] the original claim" (Brandom, 1983: 642), he is not issuing any prediction (let alone, if the original illocution was not meant as such).

The case raises some interest because it relates more generally to instances of speech acts in which a commitment or an obligation is in play. Consider the following sentence:

[6] John will have to sing boleros (tonight).

Now here are two scenarios where this sentence can be felicitously uttered. In the first scenario (henceforth, the predictive scenario), John's manager is deciding whether he needs a bolero repertoire for tonight's audience, and while he reaches a decision, participants in conversation speculate (and predict) about John's future obligations. In the second scenario (henceforth, the non-predictive scenario), John has already received the order from his manager to sing boleros, and informs the speaker so (who in turn takes John's words to be sincere and utters [6] to the addressee).

Now, it seems clear to me that the enforcing circumstances that trigger John's obligation (namely, the manager's order) are placed in two different temporal locations in each of these scenarios: while the relevant order is placed to the future of speech time in the predictive scenario, in the non-predictive case the order has, by hypothesis, already been issued. And this difference in the temporal placement of the relevant triggering fact of the denoted commissive/obligational state, has a significant effect on the very nature of both speech acts: while the speaker in the non-predictive scenario is reporting current enforcing circumstances that prospect a future event (John's singing boleros), those enforcing circumstances are not actualized yet, but only forecasted by the speaker of the predictive scenario. Hence, only in the latter case one would say that the speaker is, stricto sensu, predicting a future obligation. For the speaker in the non-predictive scenario is (perhaps implicitly) ascribing a current obligational state to John.

A crucial piece of evidence for supporting the diagnosis above is that only in the non-predictive scenario, but not in the predictive one, the present simple construction is pragmatically acceptable:

[7] Predictive scenario: John's manager is deciding tonight's repertoire but has not issued an order yet. Participants speculate and make their predictions.

a. 'John will have to sing boleros (tonight).' OK

b. ?? 'John has to sing boleros (tonight).'

[8] Non-predictive scenario: John's manager has reached a decision and ordered him to sing boleros tonight. Speaker informs the addressee about this.

a. 'John will have to sing boleros (tonight).' OK

b. 'John has to sing boleros (tonight).' OK

I will refer to the acceptability of [8b] in non-predictive scenarios as the "present simple alternation." Arguably, what such an alternation is pragmatically attaining is an implicit obligational ascription to John's current circumstances. And this ascription is clearly incompatible with a predictive reading of [6]. Hence, what the non-predictive scenario makes evident is that not all future talk is predictive. Or, simply put, that referring to future events does not necessarily entail predicting. (12)

How does the above connect with our central example, the predictive and nonpredictive utterances of [5] (to recall: 'The train to Liverpool will depart at 11:00')? Let me first point out that there is a crucial difference between the man-in-the street and the-man-in-the-station. For the former, it is not clear what the relevant commitments are, and by issuing a prediction about the train's departure the speaker is also making a prediction about that particular piece of (circumstantially inaccessible) information. In contrast, by uttering [5] (as he reads the information displayed on the screen), the man-in-the-station is ascribing the relevant commitments to the background institution (and insofar as he does that, he is not predicting what the commitment is).

Of course, these are just sample cases of what speakers can do. One can also build an example of a man-in-the-station who, skeptic about the claim in the timetable that the train to Liverpool will depart at 11, makes a prediction according to which the expectation will not be fulfilled:

[5'] It will depart at 11:15, you know how they are.

In a similar vein, one can also conceive the man-in-the-street reasoning in two inferential stages: first pondering what the commitment is, then considering a possible delay, to eventually utter [5']. But this does not affect our main idea, which points to an accountable difference in what is assumed and not assumed, pondered and not pondered, in the informational backgrounds of these distinctive speech acts. Crucially, insofar as the man-in-the-station does not call into question the commitment stipulated in the timetable's contents, he is not issuing a prediction, but only ascribing the commitments to an asserter (and perhaps adding them to our inertial expectations). Whereas for the man in the street those commitments are the very (terminal or partial) target of his prediction.

With these considerations in mind, I would like to suggest that, given that future tensed sentences are compatible with both predictive and non-predictive interpretations, the distinction to be drawn in the domain is not related to different temporal semantics, but to two distinct illocutionary forces. And the pragmatic ascriptional element referred above can be made one crucial distinctive factor between these two distinctive future-oriented illocutionary acts. For insofar as we are able to determine what is accepted (and what is not) in the informational backgrounds of the participants, an account of different speech acts is worth pursuing. To that effect, in the following section I give content to a dynamic analysis that points in that direction.

3. Assertions, Prospections and Predictions (and Bets)

In the effort of testing SD against a realistic set of counterexamples, we entertain the view that an utterance such as [5], repeated below in two distinctive scenarios ([9a] and [9b]), may not express a prediction, even though it satisfies the requirements of the definition.

[9] Utterance: The train to Liverpool will depart at 11:00.

a. Predictive scenario: the man-in-the-street.

b. Non-predictive scenario: the man-in-the station (reading the timetable).

This fact motivated not only the consideration of possible amendments for SD, so as to effectively rule out cases such as [9b], but also the methodological suggestion that an approach in terms of what is accepted into the informational background of the participants may well capture the illocutionary difference between [9a] and [9b].

Before we proceed with such an approach, a preliminary clarification about the taxonomy of these different types of illocutions is in order. More specifically, our query at this stage can be formulated in very simple terms: if not a prediction, what kind of speech act does [9b] represent? Until now, I have referred to it, somewhat informally, as a simple informative future statement. However, this simple nominal operation, as naive as it may seem, conflicts with both SD and SV, given certain tacit assumptions of the latter. One of such assumptions is that most, if not every, justified future assertion is a prediction. In effect, although it is true that SD is not formulated in a bi-conditional form, there seems to be too narrow conceptual space in the descriptive taxonomy of SV for a justified future assertion that is not meant as a prediction. The outcome, as I will show, is an unfortunate misconception of future oriented speech acts in general and predictions in particular. My proposal consists in replacing this misconception by incorporating into our taxonomy the subclass speech act of prospections.

One useful way of executing the task is to invoke Searle's notion of illocutionary point (i.e., the purpose of a speaker in making an utterance) and contrast [9b] with the case of promises. In effect, if the illocutionary point of making a promise is the "undertaking of an obligation by the speaker to do something" (1979: 2), an interesting point of contrast to the case of non-predictive [9b] is that only in the latter case, but not in the former, the referred course of events does not tend to be within the reach of speaker's agency (in the sense that the speaker is not to decide at which time the train at issue will depart). Simply put: by uttering [9b], the speaker is not committing to a course of action, he is only saying how things are going to turn out. I think this point is simply captured by the claim that the illocutionary point of uttering [9b] is to report a future event or state of affairs. However, this rather obvious claim is tackled as soon as we attempt to integrate it into SV's standard taxonomy, as there seems to be not the slightest suggestion in its categorization of an illocutionary point that reports a future event or state of affairs and is not a prediction. The effect of this rather arbitrary restriction is already clear: the traditional notion of prediction seems to be mistakenly broad as to include utterances such as [9b], and as soon as an utterance of this type is classed as a prediction, it tends not to be considered a report. As a matter of fact, Searle's claim strongly suggests that any future statement for which the speaker holds supporting evidence is a prediction and not a report:
The differences, for example, between a report and a prediction involve
the fact that a prediction must be about the future whereas a report
can be about the past or present (1979: 6).

Again, it is true that the above does not preclude future reports from constituting a sui generis speech act on their own, as Searle only claims that reports can be (rather than must be) about the past and the present. But the fact that there is no characterization of what a future report might be raises the question as to whether this omission leaves too much open ground for predictions to cover. In effect, as far as I can see, there is nothing contradictory in a report about the future that is not meant as a prediction. (13) Timetables in train stations all over the world are good examples. Annual fixtures of sport competitions are another. The information displayed in these prospective schemes is, in most cases, fully descriptive and accurate, and it can be quite misleading to describe the agents responsible for the delivery of this information as merely predicting. Intuitively, timetables are meant to inform, not to predict. And it is rather obscure how to account for their place in the natural taxonomy of future-oriented speech acts within SV's simplistic account.

Space prevents me to develop a full diagnosis of the source of SV's ungrounded assumption, but I suspect that a too heavily representational picture of assertion misreads the role of linguistic communication in general and predictions in particular. All in all, the only common aspect that one could trace as a continuum all along the wide range of speech acts called assertions, is representational in nature: to assert is, essentially, to represent the world's state of affairs. Asserting, in this view, establishes a distal relation to the world. And predicting, which is naturally an assertive subtype of speech act, establishes the same kind of relation, only with a non-actualized world that happens to be located in the future. My proposal in the following pages is that, instead of accounting for this rather large and uniform range of illocutionary acts, by means of an excessively broad and counterintuitive conception of prediction, one should better do two things: first, define prospection as a sui generis subtype of assertive speech act, and second, offer a more intuitive notion of prediction, by replacing SV's representational view with a framework that allows a more complete and dynamic account of linguistic communication and assertions in general.

The first task mentioned above can be easily carried out by stating the following (already familiar) definition:

[10] Prospection (Def.): An assertion with the propositional content condition that the propositional content is future with respect to the time of the utterance and the additional preparatory condition that the speaker has evidence in support of the proposition.

As expected, [10] is virtually identical to SD. This is explained by the fact that my proposal consists in replacing SV's two-fold distinction between assertions and predictions by a tripartite distinction between assertions, (what I refer to as) prospections, and (what are properly called) predictions. In this tripartite distinction, prospections are defined exactly as predictions (mistakenly) were, whereas predictions are tentatively redefined by adding the qualification on non-deferential evidence which I have introduced above.

Having reached a proper definition for prospections, then, our second task requires some justification. Why are we in the need of replacing SV's distal representational conception with another theoretical framework? A straight answer to this question is given by some authors in the field (Groenendijk et al., 1996; Kissine, 2013, inter alia) and points to the general fact that representing the world is not an essential feature of asserting. There are other elements, concerning participants' knowledge of both context and conversational rules, that are also necessary for a successful assertion to be performed. As one author put it, "I am reluctant to say that I have performed a successful assertion by telling an eightyear-old child that most formal systems are incomplete (although I have represented how the world is)" (Kissine, 2013: 63).

Now, one particular framework that has been developed in the last decades, with interesting results within and beyond pragmatics, is what has become known as "dynamic approaches to meaning." By this I will refer to a selected group of theories that advocate the view that the meaning of a proposition amounts not to what it represents, but to its capacity to produce changes and effects on the world (or, more specifically, on what these theories conceive as "context"). At first sight, this idea does not seem to express any theoretical novelty: that the meaning of a proposition has effects on the world or context was explicitly acknowledged by SV. In effect, the notion of perlocutionary effect captures that particular idea to a great extent. Yet, what becomes revealing in the dynamic approaches to meaning is, first, a more abstract and systematic notion of context, and second, a more integral model of its interactions with content. To take a prototypical example, consider the dynamic view advocated by Stalnaker (1973, 1999, 2002), whereby context is defined as the body of information that is taken for granted or presupposed by participants in conversation (cf. 1973: 448; 1999: 6, 53, 84, 98). What such a conception basically does is to equate context to a set of propositions or, in more heuristic terms, to the set of possible worlds in which those propositions are true--what Stalnaker calls the context set (1999: 6, 84-85, 99). This definitional move from propositions to possible worlds not only makes available a more abstract and operational representation of propositions, but also a much more dynamic picture of linguistic communication. In effect, if a proposition is conceived as a function from possible worlds into truth-values, and the context set as the set of possible worlds which contains all propositions that are taken to be true by participants, what the expression of a proposition in a particular context at a time attains is, essentially, to reduce the context set from a prior to a subsequent set of possible worlds compatible with the content of that proposition (cf. 1999: 86). Furthermore, to the extent that context is characterized as a platform for an everchanging process whereby participants distinguish "among alternative possible ways things may be" (1999: 85), the context-content interaction can be properly understood by a double order of influence: from context to content, and conversely, from content to context (cf. 1999: 4, 98). The first order of influence (context determines content) discloses context as a function from sentences to propositions, insofar as semantic rules for a natural language match sentences with propositions relative to a context (cf. Stalnaker, 1999: 34, 36). The second direction (content changes context) reveals how the resulting propositional content changes the context itself, by adding such content into what is mutually accepted (and eliminating what is incompatible). Thus, on Stalnaker's model of linguistic communication, assertion is described not in terms of how content "reflects" or "represents" context, but in terms of how context and content determine each other functionally. And that is arguably a much more alive picture than the distal view advocated by SV.

Of course, a note of concern at this point may be that some sort of "distal representation" is still in play here, given the centrality of the concept of information in Stalnaker's and many other dynamic accounts. For, to convey information about some state of affairs involving x is, arguably, to represent x as being a constitutive element of that state of affairs. And indeed, one natural motivation in many of these approaches, the contribution of which is specially appealing to what has been called "dynamic semantics" (cf. Nouwen et al., 2016; Groenendijk and Stokhof, 1991, inter alia), is to confer the concept of information some central role. Thus, according to most views within this framework, what an assertion essentially does is to update the background informational states of participants (or, in functional terms: map one informational state into another). Needless to say, the idea can be exploited to a great potential, but one particular query that has arisen, as authors approached more pragmatic arenas, is whether the concept of information is actually essential to a correct abstract account of asserting. Invoking an instructive analogy between language and games, Max Kolbel has suggested that it is not: conveying information is only an ultimate aim of assertion (as having fun is for games) and should be distinguished from what are its immediate constitutive objectives (as, for language and games, changing the score is):
[...] there is in language a distinction analogous to the distinction
between the immediate objective of a game--that is, changing the
score--and the ultimate aim of that game--for example, having fun,
entertaining, making money, impressing someone, and so on. [...] Many
people believe that the key to understanding language is that its
central aim is the exchange of information. In my view, however, the
exchange of information is only one among many ultimate purposes that
linguistic exchanges can have. When we converse in pursuit of the aim
of information exchange, we do so by pursuing the language-internal
objective of changing the conversational score, an objective that can
serve many other aims too. [...] What is characteristic of assertion is
the effect it has on the conversational score, and it is through this
language-internal effect that we, sometimes, effect the transmission of
information (Kolbel, 2011: 50-51).

According to Kolbel's view, then, what is characteristic of an assertion are the changes that it introduces into what he calls "conversational score." The notion, first introduced by Lewis' seminal paper (1979), is relatively equivalent to Stalnaker's "presupposition set" (1973: 450), "context set" (1978: 84), and "common ground" (2002: 701). In effect, inspired by Stalnaker's early motivations, (14) Kolbel characterizes the mentioned changes as comprising one single and abstract "essential effect": namely, the effect of adding the content of what is expressed by a proposition into the score. Interestingly, though, such an essential effect does not provide, as Stalnaker himself acknowledged (cf. 1999: 86-87), a definition: other things that are not assertions (e.g., suppositions) can have the same effect. The natural conclusion being that the essential effect can only constitute a necessary, but not a sufficient condition of the notion. Kolbel takes this as an incentive to fill the gap and provide the remaining definitional components. Building on Brandom's work (1983, 1994), Kolbel relates the additional elements to the normative consequences an assertion imposes on participants, "such as the obligation to provide a justification if asked to give one" (2011: 66). Even though Kolbel's intention is to present a simplified account of such normative effects, for the purpose of our survey it is important to emphasize that the "dual structure of authority and responsibility" that Brandom's original account ascribes to assertional practices (1983: 643), consist in: (i) the right of authorizing or licensing further assertion of one's claim, and (ii) the justificatory responsibility of vindicating one's claim if challenged. This double structure suits nicely the different status that an assertion can display within an inferential frame, both as a premise or as a conclusion of inferences (1983: 646). On the one hand, in licensing others to assert her claim, an asserter licenses an inferential chain in which her original claim works as a premise. On the other hand, in justifying her claims, the asserter makes explicit the inferential chain in which the assertion appears as a conclusion. By this token, the other participants can react in many different ways to an assertion, two of which are especially relevant: either they can challenge a speaker's assertion and evaluate her justification for bestowal or deprivation of its social force, or they can re-assert the claim and defer to the original speaker its justificatory responsibility. The latter move entails that an assertion can be made on the basis of personal deference and make use of that fact in justifying one's claim.

All in all, on Kolbel's view, a satisfactory account of assertion should not only point to something more "immediate" than conveying information (for conveying information is only an ultimate aim of assertion), but additionally to something more normative in nature (for changing the conversational score is characteristic of many linguistic acts, not only of assertion). The upshot is an integrated conception of linguistic communication, whereby the notion of assertion is explained by a dual standard: rules of score (e.g., the essential effect on the score) and normative principles (e.g., the rights and obligations that participants undertake by being participants in conversation).

Before considering how this picture can be applied to the case of predictions, let us first examine how Kolbel's rules and principles are spelled out. To simplify, I will only reflect on a basic subset of his formulations (for full details, see Kolbel, 2011: 60-72). Consider, then, the following two rules and two principles: one basic rule of score for assertion, one basic rule of score for supposition, a general normative conversational principle and a specific normative principle for assertion (correspondingly, Kolbel's SC2, SC5, NC1 and NC2).

[KRS1] If a participant asserts that p and the assertion is not rejected by any participant, then p becomes accepted. (If not-p was accepted at the time of assertion, then not-p is removed from the score together with propositions that obviously require not-p.)

[KRS2] If a participant proposes to suppose that p then, if no one objects, p is accepted temporarily. When the supposition is dropped again, all changes in the score that depend on the supposition that p will be reversed.

[KNP1] It is appropriate to make a linguistic move only if it is likely to change the score in ways that further the aim of the conversation.

[KNP2] If a participant asserts that p, then he or she thereby undertakes the obligation to justify* p upon request. (From Kolbel, 2011: 61-62, 68; my "*" in KNP2.)

Three brief comments are in place. First, the conditional qualification on both assertions and suppositions rules KRS1 and KRS2 (marked by expressions such as "if... not rejected" and "if no one objects") are of a different kind: while in the former rule the potential objection to p concerns, essentially, its truthfulness, the potential objection to p in the latter can only concern other related aspects, such as its utility, consistency or conceavibility. This is given by the fact that a supposition does not, in principle, aim at truth, and consequently, any objection that a participant can make by not accepting its content into the score cannot concern truth itself. If you propose to suppose that animal species had never evolved, I cannot object to your supposition in terms of its (presupposed) falsehood, but only in terms of the utility and perhaps conceavibility of such a supposition.

Second, one important addendum to the dynamic tenet formulated in KRS1 is that the score's changes can be described not only in terms of the propositional contents added to (or eliminated from) the score, but also in terms of the overt linguistic act that is being recorded. In other words, if p is asserted by a participant P, not only p is added to the score (if not challenged), but also the fact that P asserted that p (Kolbel calls this the conversational record: p. 64; cf. also Stalnaker, 1999: 101-102). In effect, as we shall later see, this information is not only conceptually distinct, but the source of a temporarily distinguishable event. This will prove relevant, and I would say, essential, to cases such as betting and predicting.

Third, my insertion of "*" in KNP2 aims to distinguish a sense of "justify" that is far more general than the one I will introduce in a later stage for the case of predicting. For now, justify* includes, broadly, a wide variety of warrants (observational, inferential, deferential, etc.).

Let us now take a step forward, before assessing the case of predicting, and consider an interesting intermediate case: placing bets. Arguably, this particular speech act must share with predictions some relevant aspects, insofar as both types of illocutions are (prototypically) future oriented and, moreover, truth-assessable: I can not only predict that Manchester City will defeat Liverpool (against your own convictions), but we can place a bet on it, making the relevant events an objective truth-assessable standard to determine who was right and who was wrong. And, in effect, as soon as one reflects on the rules of score that this might bring to the particular case of betting, a different variety of acceptance comes into play: as opposed to prototypical cases of assertion, it seems that participants in a bet can accept, in performing the very linguistic act of betting, each other's contradictory claims. To illustrate this simple point, let C stand for any conversation, P1 and P2 for participants, [t.sub.1]...[t.sub.2] for subsequent temporal intervals in a [t.sub.1]<[t.sub.2] order, and "..." some elapsing time (conversation omitted). In the following examples, C1 illustrates an odd and unacceptable assertional conversational frame, C2 a simple case of supposing, and C3 a simple case of betting.

#C1: asserting

[11] Paris is the capital of France (P1 at [t.sub.1]).

[12] Okay, Paris is not the capital of France (P2 at [t.sub.2]).

C2: supposing

[13] Let us suppose that Mary is a spy (P1, at [t.sub.1]).


[14] Let us now suppose that Mary is not a spy (P2, at [t.sub.2]).

C3: betting

[15] (I bet) Heads! (P1 at [t.sub.1]).

[16] Okay. (I bet) Tails! (P2 at [t.sub.2]).

As it is evident, C1 is at odds with KNP1 and, more specifically, with Stalnaker's claim that a participant "will not (...) assert things incompatible with the common background" (1999: 49; cf. also 1973: 450). The constraint thus expressed supports the idea that, by asserting, participants want "to reduce the context set, but not to eliminate it altogether" (1999: 89). Thus, the consistency of the score imposes a rational constraint not only on asserting, but perhaps at a more abstract level, on every linguistic move that Stalnaker classes as "acceptance." When one considers C2 and C3, though, this same constraint reveals a new guise, for both conversational frames are manifestly admissible despite its prima facie inconsistency: one participant proposes to suppose one thing, then the other proposes to suppose its contrary; one participant bets on one thing, then the other bets on the opposite. How can these admissible cases of linguistic exchange be accounted for within the framework of a consistent conversational score?

Conversations of the type C2 are easily explained away by the fact that a score is to be described and evaluated in terms of its synchronic states at a time. And in that sense, nothing like inconsistency is threatening C2: p is accepted at [t.sub.1], then dropped, and then not-p is accepted at [t.sub.2] (and then dropped). Insofar as "the state[s] of [the] context at any given moment" (1999: 86) are consistent, acceptance of the contents expressed by [13] and [14] comply with both KNP1 and KRS2. Intuitively, we can suppose that p (explore its consequences), and then drop it and suppose that not-p (and explore its consequences). But we cannot rationally explore the consequences of the supposition that 'p and not-p' at any single temporal stage.

When one considers C3, though, acceptance reveals yet another aspect, different from those exhibited in both assertional and suppositional frames. Let us reflect, first, on the putative rule of score. It seems clear that in betting that p, the alleged acceptance of [15] and [16]'s contents (in full: the coin will turn heads/tails), goes potentially beyond the limits of conversation (if, say, participants place a bet today for a flipping-the-coin kind of event taking place tomorrow). Moreover, the acceptance of the content will be assessed (or perhaps: retrospectively re-assessed) only relatively to the relevant facts once the coin is eventually flipped. In other words, there is a fact of the matter in respect to which the contents of [15] and [16] are to be truth-assessed. And this is clearly not the case of suppositions, as participants can indulge in counterfactual explorations (e.g., "Suppose that Napoleon never invaded Russia"). Thus, while the content of a supposition can in principle dispense with truth-assessment, the content of a bet must not, insofar as the relevant facts will determine correctness, and consequently, the rewarding effects that constitute the social and conceptual background of betting.

At this point, our current query can be reformulated in more technical terms: if the notion of betting is conceptually linked to truth-assessment, how can we make sense of the seeming inconsistency in conversational frames such as C3? The case raises some interest in that, as opposed to core cases of assertion (which, in turn, constitutes the paradigmatic case of truth-assessability), a conversational frame such as C3 displays no "conversational crisis" (15) when one participant expresses a content that is manifestly incompatible with what has already been accepted into the score.

One crucial point that is worth to reflect on, at this stage, is that the so called "essential effect" might have slightly changed its stage-setting. For it seems to me that what is essential in C3, and in cases of betting in general, is not so much to accept p into the score, but to record that this particular participant bets on p. We have already pointed out that this information (namely, the record) is incorporated into the score at a conceptually distinct stage from that specifically related to content. In effect, in an interesting passage (1999: 101-102), Stalnaker suggests that this overt incorporation can, in principle, occupy an interval of time prior to the content being added:
The prior context that is relevant to the interpretation of a speech
act is the context as it is changed by the fact that the speech act was
made, but prior to the acceptance or rejection of the speech act. A
successful statement will thus change the context in two different ways
that need to be distinguished. First, the fact that the statement was
made is information that is added to the context simply as a result of
the fact that it is a manifestly observable event that it was made.
Second (assuming the statement is not rejected), the content of
assertion will be added to the context. One might imagine a formal
language game in which the two kinds of change takes place successively
(1999: 101-102).

This seems to suggest that adding the record and adding the content are not merely two aspects of the same event but, potentially at least, two distinguishable events altogether. And my particular suggestion is that, in conversational frames that characteristically represent core cases of betting, it is the prior event that should be identified as the essential effect, not the latter.

It is important to stress that this claim (i.e., that content is not being added to the score in C3) does not rely on participant's P2 (reluctant) beliefs about what the first bettor envisages, but on both participants' overt public attitudes. This is not only aligned with Kolbel's motivation for treating the notions of acceptance and score non-reductively, in social or conventional terms, but also provides a better understanding of the actual linguistic moves in a betting conversational frame. For, independently of what her inner beliefs are, the second participant's public behavior is one of an entitled participant that treats the first participant's statement, not as true (nor as if she takes its truth for granted), but rather as one of a selected range of possibilities. And this is something that the different states of the score, all along the conversation, can account for. In effect, upon reflection, the presupposition set relevant for a flipping-the-coin type of bet must contain both modalised propositions "possibly heads" and "possibly tails," at any time during the conversation, and they remain to be in the set after the bets are placed. No reduction of the context set is effected by means of bettors statements (though perhaps only incidentally, as a collateral effect of specific internal regulations). (16) In pure pragmatic terms, the background presuppositions remain in the context set as true after the first bettor, and the second, have placed their bets, and their social behavior (despite what their true beliefs are) accounts for it. (17)

Now, does the above entail that acceptance plays no role in a betting conversational frame? Not entirely so. For first, the modalised presuppositions need not be part of the score prior the time of speech: they can be incorporated by the very act of betting (think of a bet on the lines of "I bet you that George will wear that yellow tie at his own funeral" to see more sharply the point). (18) Thus, it might be reasonable to recognize that in "accepting" someone's bet, even though one is not straightforwardly accepting the bettor's statement as if true, one accepts the set of possibilities that his statement brings along, to then bet on one element of that set ("I bet he will not," "I bet he will wear the red flowery one"), or refuse to participate in the bet for other reasons (cruelty, disrespect, inappropriateness, etc.).

Second, even if the contents expressed by both participants in C3 are inert with respect to any "reductive effect" on the score, this innocuousness is only temporary: by being "stocked" in the record, the content themselves remain pending for a subsequent point of re-assessment, whereby the facts prove retrospectively whose statement was right and whose wrong. Only then the conversational score can be modified by reducing the score accordingly. Of course, this later effect takes place independently, and in virtue of the facts. In other words: it is the world that speaks at this subsequent point. (19) However, in some interesting cases, albeit not all, the very linguistic act that give rise to a bet stage-setting, can reasonably be taken to be the responsible to trigger the chain of events that eventually determines truth (in the sense that if the first participant had never said "heads," the bet would have never taken place, and the coin would have never been flipped).

Having a clear picture of how the score is regulated in speech acts cases that are prototypical of betting, let me now turn briefly to its normative principles. The case makes immediately clear a difference between asserting, on one hand, and supposing and betting, on the other: while the asserter undertakes the obligation of justifying his claim (if challenged), no such constraint is imposed on supposing and betting. Certainly, there are other normative constraints of an even more social character (in betting, all the paying and rewarding commitments), but the crucial point is that none of these are of a justificatory kind: one cannot be obliged to justify why heads and not tails, for even a blind guess would be accepted as a sound bet. (20)

To sum up, then, betting is different from both asserting and supposing in that (i) it is the record, and not the content, what is essentially added to the score; and different from assertion (and similar to supposing) in that (ii) no justificatory constraints are imposed on participants.

With these sketchy notes on view, we are now in a position to address the case of predicting. In order to reach a clear understanding on its rules of score, let us first draw a conversational frame analogous with the ones we have been reflecting on:

C6: predicting

[17] (I predict that) Mr. Holmes will resign before Spring break (P1 at [t.sub.1]).

[18] (I predict that) Mr. Holmes will not resign before Spring break (P2 at [t.sub.2]).

The natural question arises as to whether: (i) participant P2, by uttering [18], is rejecting to accept [17]'s content into the score, or (ii) by analogy with the case of betting, none of the predictive contents are straightforwardly accepted (at least not until a subsequent point of truth-assessment settles the truth of the matter). To solve this query, I would like to suggest that, as "accepting" is conceptually linked to "agreeing" (in the trivial sense that if one accept p into the score, one agrees that p), a simple test initially suggests that the former picture, but not the later, is correct. In effect, this picture comes naturally when one reflects on the fact that a claim such as "I disagree" can naturally be placed ahead of [18] to render [20], whereas in (at least core) cases of betting the result is rather odd:

C7: predicting

[19] (I predict that) Mr. Holmes will resign before Easter break (P1 at [t.sub.1]).


[20] I disagree. (I predict that) Mr. Holmes will not resign before Easter break (P2 at [t.sub.2]).

C8: betting

[21] (I bet) Heads (P1 at [t.sub.1]).


# [22] I disagree. (I bet) Tails (P2 at [t.sub.2]).

It seems that a natural description of C7 would be to say that P2 is refusing to accept [19]'s content into the score by issuing his own prediction. In contrast, the addition of "I disagree" in C8 triggers an odd effect in the conversation. Moreover, it is reasonable to think that (judgment of taste and such notwithstanding), if disagreement is in place, so is the possibility of acceptance (in the sense that if there is a fact of the matter where participants objectively can disagree about, there is also a correct judgment to eventually accept into the score). This suggests that, prima facie, in predictive conversational frames acceptance has a regulatory role to play.

It is worth noticing, though, that we are looking for regulatory constraints, not merely descriptive generalizations. This means that we are interested not in what predictors usually do, but in what they can and cannot do. And it is a telling fact that predictive conversational frames such as C7 are, if anything, only one shape the conversation can take, among many. In effect, more often than not, participants can and do remain neutral regarding the other participant's predictions, as the following conversation illustrates:

C9: predicting

[23] (I predict that) Mr. Holmes will resign before Spring break (P1 at [t.sub.1]).

[24] Is that so? / Interesting... (P2, at [t.sub.2]).

Leaving aside possible cases of irony, what I think P2 attains by such a neutral stance is neither to reject nor accept P1's expressed proposition (perhaps in the same way as a passive participant witnesses a bet). (21) In effect, such a neutrality can even be depicted in conversational frames in which the second participant makes his own predictions but expresses no intention whatsoever to "prove" his point or "discard" his opponents' claim (call this the "time will tell" attitude). What is crucial here is that despite the fact that disagreement is sometimes expressed, participants can in principle remain neutral and adopt a "time will tell" attitude. And, in a reasonable analogy to the case of betting, the effect of this potential neutrality relies not on the addition of the propositional content into the score, but on leaving trace of the record. (22)

Naturally, this neutrality will not last forever: time will tell. Specifically, the future events that are the very subject matter of the prediction's propositional content, will eventually supply an objective standard for truth-assessment, and not only prove the truth or falsehood of the recorded statements, but change the score accordingly. Of course, as in the case of betting, this procrastinated change in the score constitutes the effect of facts that are independent of participants' claims. Notice, however, that truth-assessment not only creates a normative channel between the relevant facts and the score (Brandom's world asserter), but also between the relevant facts and participants: in tracking the record and truth-assessing the proposition stocked, a linguistic community not only proves that the asserted contents expressed by the predictor were true or false--it also proves him right or wrong.

All of which supplies substantial matter for a simple formulation of a rule of score for predicting:

[PRS] If a participant predicts that p, the record that he predicted p is added to the score, and a subsequent point of truth-assessment will prove him right or wrong.

There are varying conceptual modes to describe the ways of a linguistic community to an individual (cf. Brandom, 1983; Kripke, 2004, for some suggestive ideas). Space prevents me from developing a full-blown picture. Instead, I would like to focus on one normative corollary of PRS. For, if keeping a record serves for the retrospective personal valuation of the predictor, the justificatory responsibilities that he is committed to, as a predictor by issuing his prediction, must not defer the content of his statement to another asserter's assertion. And, in effect, as soon as one reflects on a normative principle for predicting, the need to stipulate a contrasting point to the case of standard assertion immediately arises: as it has been suggested all along in this paper, it seems evident that predicting (unlike suppositions and bets) does require justification (so as not to turn a statement into a blind guess), but not any type of justification. Unlike standard assertions and prospections (to recall: the man in the station), the predictor cannot defer his assertion to another asserter. It is his own, genuine and personal commitment what is being essentially accepted into the record. By contrast, insofar as the man in the station justifies his claim by pointing to a timetable, he would be justified in issuing a common assertion, but we would not consider him as issuing a prediction. This could be easily captured in the following normative constraint:

[PNP] If a participant predicts that p, then he or she thereby undertakes the obligation to justify** p upon request (where justify** means "justify on the base of a restricted type of warrants, non-deferential in kind").

Interestingly, applied to our main example, the principle suggests that justification unveils illocutionary force. And it seems reasonable, indeed, that in justifying her claims a speaker clarifies how she meant what she said. The whole point of our example was that a statement like "The train to Liverpool will depart at 11:00" can be meant alternatively as a prospection or as a prediction, perhaps even as a promise, depending on the speaker's circumstances. In case the context does not make it clear what her intentions were (how she meant what she said), justification can be a telling factor. In effect, one way, among others, to clarify the illocutionary point of an utterance (and consequently, its force) is to be responsive to a question such as "Why do you say so?" issued by the challenger (who might not be interested in revealing how true the statement is likely to be, but simply in how it was meant). By pointing to the timetable, the man on the station would make clear that he is not making any prediction, but just conveying information from one reputable source: that he is committed to the content by being licensed by the timetable. But if the man on the street were similarly asked to justify his prediction, he would have to reformulate his pondering, perhaps the whole inferential chain that led him to his personal conclusion. And that, of course, is a prototypical example of what a prediction is, what a prediction demand and what a predictor does.

The approach proposed in the above has also a clarifying effect on our second set of examples--the ones related to obligations, repeated here:

[25] Predictive scenario: John's manager is deciding tonight's repertoire but has not issued an order yet. Participants speculate and make their predictions.

a. John will have to sing boleros (tonight). OK

b. ?? John has to sing boleros (tonight).

[26] Prospective scenario: John's manager has reached a decision and ordered him to sing boleros tonight. Speaker informs the addressee about this.

a. John will have to sing boleros (tonight). OK

b. John has to sing boleros (tonight). OK

To recall, only [25a] is fully predicting an obligation, as the triggering fact of the obligation (the manager's order) is asserted to occur to the future of utterance time. In contrast, the prospective scenario is by hypothesis one in which the order has already been issued, and the speaker is therefore reporting a current obligational state of John (which is confirmed by the present simple alternation).

Now, let's add a further assumption. Let's suppose that in none of these scenarios John ends up singing boleros, but rancheras instead. Let's further assume that in the predictive scenario, he receives a different order than the predicted one, whereas in the prospective one, John simply disobeyed the reported order (that, recall, was already issued at speech time). Now, it is interesting how the different factors can be accounted for in our proposal. Let's first examine the predictive case. According to PRS, by adding the record to the score, participants leave an illocutionary trace that is retrospectively assessed once the relevant facts become a standard of truth assessment. Thus, in our example, future facts about John prove the predictor wrong: it turned out that John's manager ordered him to sing rancheras, so the predictor's dictum about John's future obligations was mistaken. If asked to justify his initial prediction, the speaker can spell out all the inferential chain that led him to conclude that John's manager was going to order him to sing boleros, and acknowledge that he was wrong. Depending on his justification (inter alia), his reputation as a predictor is more or less affected.

What about the speaker in the prospective scenario? Was he wrong? I think the correct answer to this question is that he wasn't: John had to sing boleros at some temporal point (and that is why we say that he eventually disobeyed). This simple fact supports the consistency of our illocutionary distinction: what was essentially added into the prospective score (but wasn't for the predictive one) was an obligational ascription (which, recall, could have also been expressed with the present simple alternation "John has to sing boleros"). This is why the prospector's justification is deferential: he was told from a reputable source (John himself) that there was a deontic/commissive state of affairs in force, and what the speaker did by uttering [26a] was simply to report about it. Thus, what superficially seemed to be (by SD standards) a pure predictive assertion, was not: the primitive future semantics of [26a] triggered other illocutionary force, different from prediction. Crucially, in virtue of an analysis sensitive to what is (and what is not) added into a conversational score, we can account for these illocutionary differences.

4. Conclusion

The following table summarizes the results of my examination. For clarity, nonessential effects appear in square brackets (CS stands for conversational score): (23)
Force       Essential Effect

Assert p    Add p to CS [and add record to CS]
Prospect p  Add p to CS [and add record to CS]
Suppose p   Add p temporarily to C [and add
            record to CS]
Bet p       Add record to CS
Predict p   Add record to CS
Force       Normative Constraint

Assert p    Justify* p
Prospect p  Justify* p
Suppose p   No justificatory responsibility
Bet p       No justificatory responsibility
Predict p   Justify** p

One final query, if only to suggest a discussion: if adding the record (rather than accepting the content) is a prediction's essential effect, to what extent are predictions a subtype of assertions? One possibility is to assume that predictions are not assertions simply on the basis of Stalnaker and Kolbel's claim that adding content to the score is essential to assertion. However, this would be highly counterintuitive, given the two other components of predicting: its truth-assessability and the justificatory responsibilities. A second alternative is to claim that predictions are a subtype of assertions only in virtue of their truth-assessability. This would allow, among other things, that bets qualify as assertions (something not entirely intuitive, as the disagreement test shows). I would like to suggest, then, a third possibility: what is an essential aspect of assertions in general, and predictions, in particular, relies in its justificatory responsibilities. This entails that the essential effect of assertions, in general, relies not on adding content to the score (although, of course, this might be the case), but on imposing normative (and specifically: justificatory) constraints on the asserters. This would allow prospections and predictions (but not bets) as assertion subtypes, an arguably desirable result. It seems to me that this simple move is akin to a basic intuition: to assert (in a broad sense, including predictions and prospections) that p, is to be committed to justify p, independently if the content is added or meant to be added to the score. (24)

Author Contributions

The author confirms being the sole contributor of this work and approved it for publication.

Conflict of Interest Statement

The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.


(1.) See for example Enc (1996) and, more recently, Giannakidou and Mari (2017).

(2.) SV will refer, then, to classical Speech Act Theory, which is epitomized by Searle's seminal 1969, reviewed in his 1975a and 1975b, and formally refurnished in his collaborative work with Vanderveken (Searle and Vanderveken, 1985). As the principles of SV are widely known, I will only develop a sketchy outline, as I go along with the main discussion. This will bring into focus only those aspects of the theory that have a substantial bearing on our subject matter and that are particularly relevant for the critical review of SD in this section.

(3.) In the strict sense, Searle's analysis of illocutionary acts first spells out a set of necessary and sufficient conditions, and subsequently extracts from those conditions a set of semantic rules. While the former are conditions for the performance of the defined speech acts, the latter are rules for the use of the linguistic device responsible for making a particular utterance a special kind of speech act (cf. 1969: 22).

(4.) I thank two anonymous reviewers for pointing out to me this particular problem. I am inclined to think that some subclass of scientific predictions

can also be referred to to illustrate the point. In effect, a linguistic theory can be said to predict some feature of an extinct language in more or less the same way as a natural science theory can be said to predict past events about dinosaurs. Arguably, past tensed sentences would be easily construable to report on the predictions being made. For the sake of concreteness, though, scientific predictions (and non-future tensed constructions more generally), will be set aside in this paper for the reason I explain above.

(5.) It is, of course, a totally different scenario when people predict whether (and if so, for when) a sitting government will call a snap election, for the simple reason that the prediction would be made before the election has actually been scheduled.

(6.) For the confined purpose of this paper, I assume that events are particulars (rather than universals, cf. Montague, 1969: 149-150; and Chisholm, 1970: 20; and Pianesi & Varzi, 2000, for an instructive synthesis), but remain neutral on other metaphysical issues, such as individuation criteria and ontological thickness. I am sympathetic to the idea of inertiality and non-inertiality as instantiated aspects of one and the same event token. As far as I can see, there is nothing contradictory in the idea that these opposite properties emerge during the event denoument ("an event may come to acquire new properties as the result of later happenings," Pianesi & Varzi, 2000: 16). All in all, the fine-grained metaphysics of events (whether we think of inertiality and non-inertiality either as discerning properties of distinct events or as emerging aspects of one and the same event), does not affect the central point of my insight.

(7.) As already mentioned, a rule like this is meant to guide the use of what Searle calls "illocutionary force indicating device" (sometimes called "a function indicating device," cf. 1965: 6; 1969: 30; 1985: 2), referred here as PRO, where PRO (p) represents the illocutionary force indicating device (for promising) having scope over a proposition. A proposition is taken to be, in an uncontroversial fashion, what is expressed by the speaker in a context when he utters a sentence.

(8.) It is worth noticing that this point raises an interesting temporal constraint regarding the ascribability of an undetermined range of speech acts, presumably future oriented ones. Take a preparatory condition that defines an illocutionary force F in terms of how an aspect A of a referred event E is recognized by participants at speech time T. The constraint relates to cases in which the ultimate objective nature of E reveals to be non-A at T' (where T < T'), and stipulates that this shall not force a reassessment of the original ascription (and "backtrack" from F to non-F). To put a concrete example, consider threats. Arguably, one preparatory condition on such speech act is that the propositional content of the threat is undesirable for the addressee at T. Now, it might turn out that what was undesirable at T becomes strongly wished for by the same addressee at T'. But it seems clear that this does not impose any reassessment of the ascription of the original speech act, which was meant and recorded as a threat.

(9.) Cf. Brandom (1983) for a detailed account of the authorizing processes in assertional practices.

(10.) I thank an anonymous reviewer for pointing out to me that the clarification in italics was essential: if what Cristina reported to the speaker were that she predicts that she will ultimately choose Paris as a summer destination (not that she has already decided that and, say, scheduled a flight), then perhaps we should qualify the asserter claim as a prediction after all.

(11.) I used the second-order terminology instead of the higher-order one that has come to the fore in recent epistemological discussions (Christensen, 2010; Kelly, 2010), only because the evidence that is relevant for determining illocutionary force does not necessarily reveals the defeasibility of first-order evidence, but more precisely its origin or "source-route".

(12.) Interesting semantic questions arise at this juncture: what is precisely the semantic contribution of the future marker will in the non-predictive scenario? How can we account for the fact that the deontic semi-modal have to is under the scope of the future morpheme in the non-predictive reading of [6]? Prima facie, the future marker seems not only innocuous to the temporal location of the enforcing circumstances in the relevant scenario: it also leaves completely underspecified the temporal range of the subject's obligational state. I offer a full semantically-driven analysis of this and other technical issues regarding the semantics of future obligations in Fuentes (in prep.).

(13.) As this might be a controversial remark, and it is likely to be rejected on the base that one cannot report an inexistent state of affairs, I would like to point out that it may be worth reflecting on how tense shapes our judgments on the matter. For certainly I concede that one cannot possibly report how the future is (unless, of course, one holds a deterministic view of the universe). However, notice that there is nothing in the notion of report that precludes the idea of reporting how, in view of the inertial course of events, the future will be. If it is further argued that this cannot be a report on the more epistemological ground that the alluded inertial course of events is fallible (and hence, that future is uncertain), it should be born in mind that many reports about past and current state of affairs may also be subject of uncertainty. Arguably, there is nothing unconceivable in the idea that a reporter on a civil war delivers more uncertain data than a reliable timetable, from a reputable railway company, does. One further point to reflect on is the pragmatic acceptability of orders such as "Jones, I need a report of what possible scenarios Chinese economy will confront next year," in comparison to the also acceptable "Jones, I need a report of what possible scenarios Chinese economy would have confronted today, had Mr. Chung been elected last year." As far as I can see, there is nothing contradictory in the idea that reporting about a (possible, unrealized) future parallels reporting about a (possible, unrealized) past. Further research is needed at this point. I thank two anonymous reviewers for useful comments and discussion on this contentious issue.

(14.) Kolbel interpretation of Stalnaker's work compellingly suggest that, despite the fact that his early work made use of propositional attitudes terminology, his attempts of reaching a non-reductive definition of speaker presupposition point to a conceptual link between this notion and the rules of score that govern linguistic exchange. This is what motivates Kolbel to present a simplified account of linguistic communication and assertion in particular. One advantage of the non-reductive account that he recommends is, thus, that it raises no need of referring to iterated beliefs, an issue about which Stalnaker himself has been somewhat ambivalent. In this regard, it is of considerable theoretical interest (although beyond the specific purpose of this article) to determine the extent to which Kolbel's nonreductive approach copes with the difficulties ascribed to cognitive conceptions of context, and especially with the demand of a more objective notion of it (see Sbisa, 2002). I thank an anonymous reviewer for bringing up this point.

(15.) The term is taken from Farkas & Bruce (2010).

(16.) In the case of a flipping-the-coin kind of bet, the first participant's bet determines a reduced set of available choices for the second participant, in the sense that if the first bettor bets on heads, the second participant must bet on tails. This, however, seems to be an idiosyncratic rule, not a linguistic one, and let alone a constitutive constraint on the speech act of betting. One reviewer pointed out that this might suggest that a more productive way to characterize bets would be to class them as games, and not as sui generis speech act. I leave the point for further research.

(17.) A simple example to illustrate: a mini-roulette that only contains three possibilities, say, Black, White and Red. One participant bets Black. A second participant bets White. No participant bets on Red. Turns out red. Now, participants wouldn't be surprised. They wouldn't say: "we didn't expect that one coming," "how queer," or "something went awry, let's do it again." This can only be explained by the fact that at the time their statements were made, the three possibilities were part of background assumptions and they continued to be after the bets were placed. This innocuousness shows that what was essentially added to the score, by the linguistic act of both bettors, was the record, not the content.

(18.) To clarify: a bet that points to a set of possibilities that were largely ignored, or not considered by participants, prior to the bettor's statement. Interestingly, and commenting on Stalnaker's idea of reductive effects on the set, Kolbel manifest the belief that the objective of a non-defective conversation is, less specifically, merely changing the set, which may involve widening it, "i.e. adding possibilities that had previously been discarded" (p. 50, note 2, my italics).

(19.) Cf. Brandom (1983: 649, note 12) on the idea of a world assertor.

(20.) Of course, one can supply a justification, or even adopt the practical principle of placing bets only when justified, but this is not a necessary condition for the communicative act of betting. Related to this same point, a reviewer has pointed out that one can bet on what one thinks (=predicts) about the future. I do not argue against this possibility. I am only making the point that this is not essential for the act of betting: I can bet on something I don't have the slightest idea how it will turn out, or even on something that I think it will not happen (say, for the sake of gaining a worthy profit, in case events turns as nobody thinks will turn, including me). More generally, there seems to be a lack of justificatory mechanism in bets which seems constitutive of predictions, as the following contrasting conversations show:

C: ??

A: I predict that Huddersfield will defeat Chelsea.

B: Do you really think so?

A: No. Not really. But I predict so.

C: OK.

A: I bet that Huddersfield will defeat Chelsea.

B: Do you really think so?

A: No. Not really. But I am betting on it just in case I get a worthy award.

(21.) It is of great empirical interest which linguistic constructions can express, in virtue of their meaning, a neutral stance such as the one I aim to illustrate with this simple example. Needless to say: as there is relevant linguistic data to infer what types of expression count as assertions and what as challenges, there is every reason to think that concrete languages allow constructions that can count as expression of a neutral stance, and perhaps more specifically, as a neutral stance regarding future contingencies.

(22.) Note that, at least in some common cases, these replies would not have the same attitudinal effect for present and past assertions: "is that so?" commonly manifests doubt (or at least resistance to the incorporation of the content into the score); while "interesting..." concedes credit to the speaker (and accepts to incorporate the content). These observations are not meant as strict generalizations, though, and are subject to contextual factors. Further research is necessary to determine whether a dynamic approach is suitable for the analysis of linguistic constructions that (potentially) express acceptance, rejection and the neutral stance alluded in the previous note.

(23.) The reader may notice that, according to the table above, prospections come out as no different from assertions. This underspecification indicates that the defining element for taking the former as a subclass of the latter can only be captured in terms of tensed propositional content, as in the definition on p. 21. What is shown in the table, instead, are the distinctive elements between assertions and predictions, none of which involve temporal specifications. This is the driving factor to allow future assertions (=prospection) that are not predictions.

(24.) I am grateful to Martina Faller, Delia Bentley, Graham Stevens, and three anonymous reviewers for useful comments and discussion. This research has been supported by a CONICYT Chile grant.


Brandom, R. (1994). Making it Explicit. Cambridge, MA: Harvard University Press.

Brandom, R. (1983). "Asserting," Nous 17: 637-50.

Chisholm, R. M. (1970). "Events and Propositions," Nous 4: 15-24.

Copley, B. (2014). "Causal Chains in Futurates," in P. De Brabanter, M. Kissine, & S. Sharifzadeh (eds.), Future Times, Future Tenses. Oxford: Oxford University Press, 72-86.

Copley, B. (2009). The Semantics of the Future. New York: Routledge.

Christensen, D. (2010). "High-Order Evidence," Philosophy and Phenomenological Research 81(1): 185-215.

Enc, M. (1996). "Tense and Modality," in S. Lappin (ed.), The Handbook of Contemporary Semantic Theory. Oxford: Blackwell, 345-358.

Farkas, D., & K. Bruce (2010). "On Reacting to Assertions and Polar Questions," Journal of Semantics 27(1): 81-118.

Fuentes, P. (in prep.). "Future Obligations." [Unpublished manuscript].

Giannakidou, A., and A. Mari (2017). "A Unified Account of the Future as Epistemic Modality. The View from Greek and Italian," Natural Language and Linguistic Theory (AOL). doi:10.1007/s11049-017-9366-z

Groenendijk, J., M. Stokhof, & F. Veltman (1996). "Coreference and Modality," in S. Lappin (ed.), The Handbook of Contemporary Semantic Theory. Oxford: Blackwell, 179-216.

Groenendijk, J., & M. Stokhof (1991). "Dynamic Predicate Logic," Linguistics and Philosophy 14(1): 39-100.

Kelly, T. (2010). "Peer Disagreement and Higher-Order Evidence," in R. Feldman, & T. Warfields (eds.), Disagreement. Oxford: Oxford University Press, 111-174.

Kissine, M. (2013). From Utterances to Speech Acts. Cambridge: Cambridge University Press.

Kolbel, M. (2011). "Conversational Score, Assertion and Testimony," in J. Brown & H. Cappelen (eds.), Assertion: New Philosophical Essays. Oxford: Oxford University Press, 49-78.

Kripke, S. (1982/2004). Wittgenstein on Rules and Private Language: An Elementary Exposition. Oxford: Blackwell.

Lewis, D. (1979). "Scorekeeping in a Language Game," Journal of Philosophical Logic 8(3): 339-359.

Montague, R. (1969). "On the Nature of Certain Philosophical Entities," The Monist 353: 159-194. Reprinted in R. H. Thomason (ed.), (1974), 148-187.

Nouwen, R., A. Brasoveanu, J. van Eijck, & A. Visser (2016). "Dynamic Semantics," The Stanford Encyclopedia of Philosophy (Winter edn.), Edward N. Zalta (ed.),>.

Pianesi, F., & A. Varzi (2000). "Event and Event Talk: An Introduction," in J. Higginbotham, F. Pianesi, & A. Varzi (eds.), Speaking of Events. Oxford: Oxford University Press, 3-47.

Sbisa, M. (2002). "Speech Acts in Context," Language and Communication 22(4): 421-436.

Searle, J. R. (1965). "What is a Speech Act?," in M. Black (ed.), Philosophy in America. Ithaca, NY: Cornell University Press, 39-53.

Searle, J. R. (1969). Speech Acts. An Essay in the Philosophy of Language. Cambridge: Cambridge University Press.

Searle, J. R. (1975a). "A Taxonomy of Illocutionary Acts," in K. Gunderson (ed.), Language, Mind and Knowledge, Minnesota Studies of Science. Minnesota, MN: University of Minnesota Press, 344--369.

Searle, J. R. (1975b). "Indirect Speech Acts," in P. Cole and J. L. Morgan (eds.), Syntax and Semantics, Vol. 3, Speech Acts. New York: Academic Press, 59-82.

Searle, J. R. (1979). Expression and Meaning. Studies in the Theory of Speech Acts. Cambridge: Cambridge University Press.

Searle, J. R., & D. Vanderveken (1985). Foundations of Illocutionary Logic. Cambridge: Cambridge University Press.

Stalnaker, R. (2002). "Common Ground," Linguistics and Philosophy 25: 701-721.

Stalnaker, R. (1999). Context and Content: Essays on Intentionality in Speech and Thought. Oxford: Oxford University Press.

Stalnaker, R. (1998). "On the Representation of Context," Journal of Logic, Language, and Information 7(1): 3-19. Reprinted in R. Stalnaker, (1999), 96-114.

Stalnaker, R. (1978). "Assertion," Syntax and Semantics 9: 315-32. Reprinted in R. Stalnaker, (1999), 78-95.

Stalnaker, R. (1973). "Presuppositions," Journal of Philosophical Logic 2(4): 447-457.

Thomason, R. H. (ed.) (1974). Formal Philosophy. Selected Papers of Richard Montague. New Haven, CT: Yale University Press.

PABLO FUENTES, School of Arts, Languages and Cultures, The University of Manchester, England

How to cite: Fuentes, Pablo (2019). "Predictive Illocutions and Conversational Scores," Linguistic and Philosophical Investigations 18: 7-36.

Received 10 October 2017 * Received in revised form 8 January 2018

Accepted 9 January 2018 * Available online 25 January 2018

COPYRIGHT 2019 Addleton Academic Publishers
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2019 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Fuentes, Pablo
Publication:Linguistic and Philosophical Investigations
Geographic Code:4EUFR
Date:Jan 1, 2019

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters