Deciding what we mean

Stipulation gives us a degree of control over meaning. By stipulating how I will use a term I am able to determine the meaning it will receive on future occasions of use. My stipulation will affect the truth conditional content of my future utterances. But the mechanisms of stipulation are mysterious. As Cappelen ([2018]. Fixing Language: An Essay on Conceptual Engineering .


Stipulation
In this paper, I will use the word 'fish' to mean creatures with fins that live in water.I have just stipulated a meaning for 'fish'.This stipulation will have a number of effects.Firstly, it will help you as a reader grasp what I intend to communicate when I say things like 'the only sea food I enjoy is fish'.You will be able to rule out the possibility that I enjoy shell fish.But you won't be able to rule out the possibility that I enjoy whale.My stipulation will also affect how my utterance can be reported.If I had not stipulated what I meant by 'fish' it would have been accurate to report me as saying that I don't like whale.However, given my stipulation, I cannot be reported as having said this.If you do report me as having said that I don't like whale then I will have grounds for complaint.If stipulation did not affect assertoric content I would have no grounds for complaint.Relatedly, my stipulation affects the truth-conditional content of my utterance.Suppose that I don't actually like whale, but I do like hag fish (a finless creature that falls within the usual extension of 'fish').In this case I'll have said something false.If I know that I like hag fish then I will have lied.And I'll also have violated the norm of assertion.But without my stipulation I'd have said something true (supposing that I genuinely do dislike all seafood that falls outside of the usual extension of 'fish').I won't have lied and, if I know that I dislike all seafood that falls outside of the typical extension of 'fish', I will have obeyed the norm of assertion.
So, there is an important sense in which my stipulation affects the meaning of 'fish' as it appears in my later utterances.My stipulation granted me a degree of control over the meaning of 'fish', at least within this limited setting.We regularly exercise such agency over meaning.And this agency plausibly extends beyond stipulation.For example, I can choose what object my use of 'that' refers to by pointing.And I can make it clear how other forms of linguistic context sensitivity are to be resolved by, for example, manipulating the question under discussion.But this agency is puzzling.
Firstly, it is not clear how stipulation enables us to manipulate meaning.As Cappelen (2018) emphasizes, linguistic meanings are determined by an incredibly complex myriad of external factors, the vast majority of which are beyond our control, perhaps even beyond our comprehension.If I had not stipulated what I meant by 'fish' then the meaning of 'fish' in 'the only seafood I enjoy is fish' would have been determined by, among other things, a long and complex pattern of use within my community, the opinions of marine biologists, and the initial historical conditions in which some word etymologically related to 'fish' was first introduced.These external factors would have determined the truth conditional content of my utterance.They would have affected whether or not I lied, and whether or not I violated the norm of assertion.But my stipulation somehow overruled all these factors.From whence does stipulation acquire this power?
Secondly, the power of stipulation is limited.For example, the effects of stipulation are typically (although not always) limited to the restricted context in which the stipulation takes place.If, on an unrelated occasion, I say 'the only sea food I enjoy is fish' then 'fish' will take on its normal meaning.My previous stipulation will have no effect.Moreover, stipulation can be unsuccessful even in the original context.Stipulation does not give us the power to mean whatever we want.Consider the following case: Slurring Stipulation A speaker is addressing a large crowd.The speaker says 'I will use '****' to mean people who are lazy and don't pay taxes', where '*****' is some powerful racial slur.The audience contains a group of people who fall in the usual extension of '****'.The speaker looks at this group as he says '*****'s don't belong in this country'.This stipulation does not get off the ground.The speaker can accurately be reported as having said that people of a certain ethnicity do not belong in his country.And his utterance is false precisely because ethnicity is not a legitimate basis for social exclusion.Why is stipulation limited this way?What sets the bounds on our powers of stipulation?
These are the questions I set out to answer in this paper.There are a number of related questions that I will not address: I don't aim to provide an account of the general speech act of stipulation.My focus here is meaning stipulation alone.Moreover, I am concerned with the metasemantic question of how stipulation affects meaning.I have no pretensions to settle the matter of whether stipulation is a kind of command, conversational exercitive, or a distinctive speech act of its own kind.
I will start by considering Pinder's (2021) speaker meaning approach.I will also consider the related claim that stipulations are proposals regarding meanings in micro-languages (Armstrong 2016;Ludlow 2014).I will suggest that, whatever their merits, these approaches are unable (without supplementation) to explain the distinctive metasemantic impact of stipulation.This discussion will lead us to a further question: I'll argue that stipulation can fail to have its characteristic metasemantic effects, and yet still function properly in facilitating communication.But if communication proceeds smoothly why should we care about any additional metasemantic effects of stipulation?I will argue that stipulation affects the word uses it is fitting to hold us to.That is, it affects the commitments we undertake when we communicate.I then combine this observation with the fittingness-based metasemantics recently defended by Peet (2022) to explain the effects of stipulation on meaning.This account is able to capture the mechanisms and limits of stipulation, whilst also explaining why we should care about full stipulative success.

Speaker meaning
The distinction between semantic meaning and speaker meaning is familiar.'Speaker meaning' denotes the way in which a speaker intends to use a term -what they mean by the term.'Semantic meaning' denotes what the word actually means, regardless of the speaker's intentions.Pinder (2021) argues that stipulation constitutes a proposal to use a term in a particular way: Suppose that I stipulate the meaning F for a term a.On the speaker meaning approach I have proposed that we use a to speaker-mean F. Initially, Pinder's account may appear to resolve our puzzles.It is no surprise that we can control what we speaker mean.It is also no surprise that stipulations are effective only within a limited context.If I stipulate the meaning F for a term a then my proposal will typically be limited to the current context.There are exceptions to this: Proposals in conceptual engineering (Pinder's primary concern) can plausibly be taken as stipulations with intended effects beyond the original context.But conceptual engineering is the exception rather than the rule.We will put conceptual engineering to one side for now (I will return to it in the conclusion).
Unfortunately, the speaker-meaning approach does not capture the metasemantic effects of stipulation.It does not explain why our public speaker's stipulation that '****' mean lazy people who don't pay taxes failed to get off the ground.Pinder acknowledges that speaker meaning has its limits.As he notes, one cannot mean that Paris is in France by 'it's another beautiful day' (153).He explains this in terms of the limits on what we can rationally intend.I cannot speaker mean Paris is in France by 'its another beautiful day' because I know that the audience will never recover Paris is in France from my utterance.Perhaps the same is true of our public speaker?Perhaps he cannot mean lazy people who don't pay taxes by '****' because he knows no rational audience would ever interpret him this way?
Well, if our public speaker is normal then they will no doubt realize this.So, they will not be able to speaker-mean lazy people who don't pay taxes by '****'.But if our speaker is irrational, deluded, or in possession of highly misleading evidence then there is nothing to stop them speaker-meaning lazy people who don't pay taxes by '****'.Yet, this makes no difference to the success of their stipulation.Even if they intended to say that lazy people who don't pay taxes don't belong in their country, this is not what they have asserted.
Pinder could perhaps respond that, for a stipulation to be successful, the audience must go along with the stipulation.That is, they must accept the speaker's proposal to use the relevant term in the stipulated way.But this doesn't help.Suppose that our public speaker, through an overwhelming stroke of luck, ends up addressing an audience that is happy to go along with his stipulation.They take him to be speaking out against lazy people who avoid taxes.They don't take him to be saying anything racist.In this case, the stipulation will, according to the speaker meaning approach, be successful.And there is clearly a sense in which this is right: the speaker luckily succeeds in making their intended meaning clear to the audience.
But the speaker's luck with their audience does nothing to change the fact that they have asserted something racist.They have asserted that a particular ethnic group has no place in their country.They can be fittingly reported as having said this.And their assertion is false precisely because ethnicity has nothing to do with who belongs in which country.To see this, suppose that a member of the original audience (who accepts the stipulation) later tells an outsider the exact words spoken by our public speaker.Even if they are aware of the original stipulation, the outsider will be able to report to others in an unqualified manner that 'The speaker said that ****s don't belong in this country' (or a neutral non-slurring translation referring to the relevant ethnic group).This stands in stark contrast to the case of successful stipulation with which we opened: If I stipulate that by 'fish' I mean 'creatures with fins that live in water', and I then state that 'the only seafood I like is fish', you will not be able, without qualification, to report my utterance to an outsider by saying 'X said the only seafood he likes is fish'.If you were to do so you would be making a false report.And, if this false report was of consequence (for example, if I was resultantly deprived of the opportunity to be served whale) I would be able to blame you.
So, even if they get incredibly lucky with their audience, our public speaker's stipulation still fails to affect the assertoric content of their utterance.Their stipulation fails to have the effects on meaning characteristic of successful stipulation.This cannot be captured on the speakermeaning approach. 1 related approach presented by Shields (2021) might appear to help here.Sheilds suggests that an act of stipulation is successful when it generates a shared inferential entitlement between the speaker and their audience.Shields's focus is on the speech act of stipulation in general, rather than meaning stipulation in particular.However, when a speaker stipulates that a term t mean F the relevant entitlement is naturally thought of as an entitlement to assume that, for the remainder of the conversation, t will be used to speaker mean F.
Shields's approach, at least as stated above, arguably secures the correct result with respect to the versions of Slurring Stipulation that we have considered so far.The audience in Slurring Stipulation accepted the speaker's stipulation, and thus accepted that he speaker meant (only) 'lazy tax avoiders' with his use of '****'.However, there is nothing to indicate that they were entitled to do so.Thus, there is nothing to indicate that the stipulation was successful. 2  This brings us to the broader problem with the speaker meaning approach: stipulation plausibly does clarify speaker meaning in the manner suggested by Pinder.But it does more than this -it also affects what a speaker asserts.And assertoric content can come apart from speaker meaning.This is illustrated by the following mundane case: Scandinavian Trip Theo has just been on a tour of Scandinavia.His favorite destination was Sweden.However, he often gets Denmark and Sweden mixed up.So, when he reports his trip to a friend he says 'my favorite country was Denmark', intending to communicate that his favorite country was Sweden.
Theo speaker means that his favorite country was Sweden.But this is not what he says or asserts.He asserts that Denmark was his favorite country.As a result, he has asserted something false.When all goes well, speaker-meaning and assertoric content coincide.But cases like Scandinavian Trip show that they can come apart.And successful stipulation affects assertoric content.So, the speaker-meaning account is at best incomplete.It may well be that the primary function of the speech act of help the speaker meaning approach.If our speaker sincerely specifies that they will only mean lazy people who don't pay taxes by '****' then the speaker meaning approach will be unable to explain why he is still reportable as having advocated for social exclusion on the basis of ethnicity. 2When presenting his view Shields seems to suggest that the relevant inferential entitlement is generated when the audience accepts the speaker's stipulation.This implies that the stipulation in Slurring Stipulation was a success, at least in the version of the case where the speaker gets lucky with his audience.So, as developed in Shields ( 2021), Shields's view does not secure the correct result about Slurring Stipulation after all.Still, this problem does not cut to the heart of the view as, intuitively, the audience in Slurring Stipulation is not entitled to accept the speaker's stipulation, and intuitively the stipulation does not generate the relevant inferential entitlements.
stipulation is to clarify speaker meaning, or to generate inferential entitlements.But there is still a mystery regarding the ability of stipulation to affect assertoric content.At the very least, speaker-meaning approaches will need to be supplemented to capture this. 3

Linguistic meaning
In Scandinavian Trip the speaker-meaning and assertoric content of Theo's utterance came apart.The assertoric content was determined by the linguistic meaning of 'Denmark'.That is, it was determined by the meaning of 'Denmark' in Theo's language.But if assertoric content is determined by linguistic meaning then it once again becomes mysterious how we are able to exercise agency over it.After all, as Cappelen (2018) and Deutsch (2020) argue, stipulation doesn't allow us to alter the meanings of words in English.Even authoritative bodies such as courts seemingly lack the ability to fix linguistic meanings.Cappelen argues as follows: Let's take a particular case, say the US courts deciding that corporations are in the extension of 'person' (this is simplifying the legal situation a bit, but the simplification is irrelevant for current purposes).That looks like a case of Control: they have made a decision about what that word should mean and now that's what it does mean.So Lack of Control is false, it would seem.But this objection misdescribes what has happened in such a case.What has happened isn't that they've decided on and created a new meaning for 'person'.What has happened is that they've made/forced certain people to misinterpret sentences containing 'person' in a particular way.The US Supreme Court or any other group can no more change the meaning of 'person' than I can.(Cappelen 2018, 76) There is a sense in which Cappelen is right, and a sense in which he is wrong.Clarifying the sense in which he is wrong will be suggestive of an alternative account of stipulation.Unfortunately, this alternative solution also fails to explain our data.
Talk of the meaning of a term in a language is really just a way of describing the norms, conventions, and patterns of use, deference, and 3 Such accounts will need supplementation even if, as it happens, whenever the audience acquires the relevant entitlements regarding speaker meaning, the speaker meaning corresponds to assertoric content.What we will need here is a metasemantic bridge principle explaining why this is the case -i.e.why whenever the audience is entitled to assume that the speaker means p the speaker thereby asserts p. Regardless, it is dubious whether this connection will always hold.Depending on their evidential background different audience members might acquire different entitlements on the basis of a speaker's stipulation.In that case, a strong connection between audience entitlements and assertoric content would require either a strong form of speech act pluralism, or a form of content relativism.
expectation within a particular linguistic community. 4At any given time a typical speaker will be a member of many different language communities.And each community will display slightly different patterns of use, deference, and expectation.For example, you the reader are a member of both the community of English speakers, and the community of English speakers minus me, the author.My own little linguistic idiosyncracies constitute one small part of the grand pattern of use, deference, and expectation that determine the meaning of terms in English: the language of the community of English speakers.But they do not play a role in determining the meaning of terms in the language spoken by the community of English speakers minus me.So, we may suppose, the language of each community will be slightly different. 5Thus, when we ask whether it is possible for stipulation to affect the meaning of a term in a language, we must ask 'which language?'.
When we speak of the meaning of a term in 'English' we have in mind the patterns of use in a large linguistic community -a community that contains over a billion speakers.Cappelen is right that a mere stipulation, even by an authoritative body, is powerless to change the meaning of a term in 'English'.It is fitting to describe a term as having a particular meaning in English only when this meaning fits the patterns of use, expectation, and deference (together with external factors such as referential eligibility) of a truly staggering population.A single stipulation doesn't have the power to affect the linguistic dispositions of a worldwide community.
However, an act of stipulation may have the power to affect the linguistic dispositions of a smaller community.For example, a stipulation may, for the duration of the relevant discourse, affect the linguistic dispositions of the speaker and their audience.This is a very small and short-lived community.But we will be able to identify dispositions of use, deference, and 4 This is not to say that there are no such things as languages -it could be that, for example the norms, conventions, and patterns of use, deference, and expectation prevalent in a community, in combination with facts about the world, ground facts about what words mean in the language of the community.However, it is also consistent with a more lightweight or fictionalist construal of languages according to which talk of the meaning of a term in a language is little more than a loose way of characterizing norms, conventions, and patterns of use, deference, and expectation.I remain neutral on the nature of languages here. 5Perhaps the removal of a single individual from a community will not be sufficient to trigger any differences in meaning.This is not of importance.1,348,000,000 people speak English.Suppose the removal of a million speakers is enough to make a difference to meaning.Then there will still be an unfathomable number of linguistic communities with different patterns of use and deference that the speaker is a member of at a given time.Although see Williams and Leckie (2019) for a rare attempt to provide an account of the linguistic community a speaker belongs to for the sake of determining which language they speak.
expectation within this community just as we can in with large linguistic communities.Thus, we can still think of such communities as speaking 'micro-languages' (Armstrong 2016;Ludlow 2014, see also Plunkett and Sundell 2013).So, perhaps we should say that stipulation aims to change the meaning of a term in a micro-language?Once again, this would seem to eliminate the mystery of stipulation.It is not especially mysterious how stipulation affects the expectations and dispositions of one's immediate audience.It is also obvious why these effects would be limited to the original context of stipulation: outside of this context the audience will revert to their default assignments of meaning (which will be more in line with those of the larger linguistic communities to which they belong).
Unfortunately, this approach also fails.There are two reasons for this.Firstly, an audience may stubbornly refuse to go along with a stipulation.As a result, the stipulation will not have its intended effect on the dispositions and expectations of the speaker's immediate interlocutors -it will not change the micro-language.But it may still affect the truth conditional content of their utterance.To see this, consider a paradigmatically reasonable stipulation (from Pinder (2021)): Luddites There are two types of people who do not use social media.There is the person who is opposed to social media, often having had (and closed) a Facebook account due to its negative effects on society.Call her the Luddite.In contrast, there is the person who has no specific objection to social media, but simply prefers to interact using more traditional means.Call her the Traditionalist.To fight against the spread of social media, we should not seek to increase the number of Luddites: the Luddite is reacting to the popularity of social media, and reactionaries stand little chance of changing mass movements.Rather, we should seek to increase the number of Traditionalists; we should make a movement of traditional means of social interaction.(Pinder 2021, 148-149) Suppose that Pinder's audience is either stubborn or inattentive.As a result, they fail to take on board his stipulated meaning for 'Luddite' and instead assume that he's discussing people who are opposed to the development of new technology.For example, when he says 'the Luddite is reacting to the popularity of social media' they take him to be suggesting that people who are opposed to the development of new technology are reacting to the popularity of social media.Perhaps this is what 'the Luddite is reacting to the popularity of social media' ends up meaning in the micro-language consisting of Pinder and his audience.Nonetheless, his audience is mistaken.Pinder has not said that people who are opposed to the development of new technology are reacting to the popularity of social media.He has not said anything false.Rather he has said that people who refuse to use social media because of its negative social effects are reacting against the popularity of social media.And this is true.So, in one sense at least, his stipulation was successful.Despite failing to change the meaning of 'Luddite' in the micro-language of his interlocutors, he has affected the meaning of 'Luddite' as it appears in his later utterances.
Secondly, the micro-languages story, like the speaker meaning approach, fails to account for the limits of stipulation.Consider Slurring Stipulation again.Our speaker stipulated that '****' (usually a powerful racial slur) was to mean lazy people who don't pay taxes.He then looked at a group of audience members who fall within the normal extension of the slur and said '****s have no place in this country'.As we noted above, even if our speaker does, by pure luck, end up addressing an audience that is willing to accept his stipulation, this will not affect the assertoric content of his utterance.His utterance will still be false precisely because race is not a legitimate basis for social exclusion.But the micro-language approach tells us that, if the speaker lucks out with their audience, then the stipulation was successful.After all, the speaker has succeeded in changing the meaning of '****' in the micro-language spoken by themselves and their immediate interlocutors.They have successfully set new patterns of use an expectation for the word '****' in this short-lived micro-community.
The proponent of micro-languages might respond by endorsing a form of content relativism.That is, they might suggest that what a speaker asserts must be relativized to a language.Relative to the micro-language of our public speaker and his immediate audience he asserts something true.However, in our micro-language -the language spoken by our fictional public speaker, myself, and you the reader, he says something false.After all, we would not be willing to go along with the stipulation.The same response can be given to the Luddite argument: Relative to our micro-language Pinder has said something about people who are opposed to social media.Relative to the micro-language of his imaginary stubborn or inattentive audience he has said something about people who are opposed to technological development. 6 don't find this response convincing.It seems clear that Pinder's stubborn audience would be making a mistake in tokening the thought 'Pinder said something false'.Likewise, our public speaker's audience is making a mistake when they token the thought 'he only said that lazy tax avoiders should be expelled'.They are mistaken in their judgment that he has not said anything racist.It is not clear how the content relativist can capture this.They could perhaps suggest that there is something defective about the micro-languages spoken by our misguided audiences and that their mistake lies in speaking the wrong micro-language.This is just to say that the audiences display the wrong patterns of use, deference, and expectation.Pinder's stipulation had an effect on what constitutes the right pattern of use, deference, and expectation.Our public speaker's stipulation had no such effect.But this just takes us back to square one.The question becomes, what constitutes the right meaning for a term?And how does stipulation affect this?My positive proposal will shed light on this.But before turning to the positive proposal it is worth mentioning a further question that is raised by the foregoing.

Why care about stipulative success?
I have considered two approaches to stipulation: the speaker-meaning approach and the micro-languages approach.The speaker-meaning approach holds that stipulation constitutes a proposal regarding what to speaker-mean with a term.The micro-language approach holds that stipulation alters meaning within the micro-language of the speaker and their interlocutors.These approaches were selected for discussion because they both give plausible accounts of the speech act of stipulation.However, it was found that whatever their merits these approaches were unable to capture the metasemantic impact of stipulation.
But this raises a question: the speaker meaning and micro-language approaches both make it obvious why we should care about stipulative success.After all, stipulation aids successful communication.But the cases I have presented suggest that stipulation can perform all of its communicative functions and still fall fail to bring about its characteristic metasemantic effects.That is, stipulation can affect the micro-language of the speaker and their interlocutors and support the recovery of speaker-meaning without affecting assertoric content.But if that is right, then why should we care about the metasemantic effects of stipulation?It would seem to be communicatively irrelevant, meaning that full stipulative success is of little importance.

Meaning and responsibility
In providing an account of the metasemantic effects of stipulation it will be instructive to begin with our final question: why should we care about full stipulative success?Stipulation, it seems, can perform all its communicative functions without having its characteristic effects on meaning.Why should we care about the effects of stipulation beyond those that contribute to communicative success?
The answer is that we do more than communicate when we speak.That is, speakers do more than simply cause their audiences to entertain particular propositions.When we speak we also undertake commitments.We become responsible for certain claims about the world.It becomes fitting to hold us to these claims to expect us to defend them, and to judge us negatively if they turn out to be false. 7The fact that speech generates such commitments plays a key role in explaining why testimony provides reasons for belief.It plays a key role in explaining why we generally speak truthfully.
Moreover, it is clear that speakers should care about what they commit themselves to.They should care about what they can be held to, and the reactive attitudes their speech acts will garner.This, I suggest, is why we should care about stipulative success: stipulation affects the claims we can be fittingly held to.When I stipulated that 'fish' mean creature with fins that lives under the water this affected the use of 'fish' it was fitting to hold me to.It was no longer fitting to hold me to the normal use of 'fish' when I uttered 'the only seafood I enjoy is fish'.Rather, because of my stipulation it became fitting to hold me to the claim that the only aquatic creatures I enjoy eating have fins.This is what my utterance committed me to.
The fact that a speaker unluckily ends up addressing a stubborn or inattentive audience does not affect the proposition or word use it is fitting to hold them to.Following Pinder's stipulation that 'Luddite' mean person who is opposed to social media it becomes fitting to hold him to this use of 'Luddite'.When he says 'the Luddite is reacting to the popularity of social media' he has committed to a claim about people who are opposed to social media.This is what it is fitting to hold him to.It is not fitting to hold him to a claim about people who are opposed to technological development.The fact that his audience was too stubborn or inattentive to go along with his stipulation does not change this.Likewise, the fact that our racist public speaker was lucky enough to address an audience willing to go along with their absurd stipulation does not affect the use of '****' it is fitting to hold him to.Any reasonable audience would interpret him as advocating for social exclusion on the basis of race.This is what he should be held to. 8o, we have identified an effect that stipulation has beyond its role in facilitating communication.And I have suggested that this is an effect that we should care about.But the question we are interested in concerns the metasemantic impact of stipulation -its impact on assertoric content.How do these observations help?Well, they give us a clue as to the type of metasemantic theory that will allow us to account for the metasemantic effects of stipulation.Stipulation determines the commitments a speaker undertakes with their utterance.So, metasemantic theories that draw a strong connection between assertoric content and commitment will have an advantage in explaining the metasemantic effects of assertion.
One such account, which is particularly useful for our purposes, has recently been developed by Peet (2022) who argues that the meaning a term receives on an occasion of use corresponds to the use it is fitting to hold the speaker to.That is, the value a term receives on an occasion of use corresponds to the use that there is the most objective reason to hold the speaker to.There will typically be a myriad of reasons for and against holding a speaker to a particular use of a term.Examples include the fact that they intended to use the term in a particular way, the fact that a typical audience member would interpret them a particular way, the fact that the term is used in a particular way in the speaker's linguistic community, and the fact that a particular referent is especially eligible.On a given occasion of use, according to Peet, the meaning of a term will be determined by weighing these (and many other) reasons.The assertoric content of an utterance is, Peet suggests, directly determined either compositionally or via bridge principles by the meanings of the words used on an occasion.So, an utterance's assertoric content is the proposition it is fitting to hold a speaker to directly in light of the specific word uses it is fitting to hold them to (for a similar proposal see Gauker (2007)).
We have seen that stipulation affects the word use it is fitting to hold a speaker to.So, if Peet (2022) is right, we are able to explain the impact of stipulation on meaning: the meaning a term receives on an occasion of use corresponds to the use it is fitting to hold the speaker to, and this is affected by stipulation.
But what about the questions with which we began?How does stipulation override the inscrutable external factors that normally determine the meanings of the words we use, and thus the propositions we assert?And, what explains the limits of stipulation?The fact that its effects are short-lived and local, and the fact that certain stipulations simply cannot get off the ground?I will address these questions in turn.
Assertoric content is typically determined by myriad of inscrutable factors that are beyond our control.These factors include, among other things, patterns of use within our community, the opinions and discoveries of experts, the eligibility of referents, and the lexical effects of our terms.The reason these factors determine assertoric content is that they play a role in determining the word use it is fitting to hold the speaker to.If experts in the speaker's community use a term in a particular way, this will often be a reason to hold the speaker to that use.If a particular referent is more eligible than others, this favors holding the speaker to the corresponding use.Metasemantics is inscrutable and intractable for a number of reasons: Firstly, the range of factors that could conceivably count for or against holding a speaker to a particular word use on a particular occasion is endless.Secondly, different factors will acquire different weights in different contexts.There is no clear way to systematize the weighing of these reasons.
However, the fact that a speaker has stipulated a particular meaning for a term strongly favors holding them to that use of the term.Their stipulation telegraphs the way in which they are using the term.Regardless of the way a word is used in the wider community, and regardless of which referents are more eligible, or the opinions of scientific experts, if a speaker makes it clear that they intend to use a word in a particular way, that will almost always be the use it is most fitting to hold them to.
But, of course, there are exceptions.There are cases in which stipulation does not get off the ground.These include cases in which the speaker is sincere (if somewhat deluded) in their stipulation, and their audience goes along with them.For example, in Slurring Stipulation our public speaker's stipulation that '****' mean lazy tax avoiders failed to get off the ground.If the stipulation had been successful they would have said that lazy tax avoiders have no place in their country.This is still probably false.But it is not as obviously false as what they did assert: that members of a particular ethnic group have no place in their country.
The reason this stipulation doesn't get off the ground is that, despite the stipulation, it is more fitting to hold the speaker to the standard use of '****'.No rational or reasonable audience would take the speaker's stipulation seriously.Any rational audience would take the stipulation to be a ham-fisted attempt to say something racist whilst retaining plausible deniability.Moreover, some words have such power that we cannot simply disassociate them from their usual meanings.Racial slurs are like this.The imagery and emotional responses they evoke are so powerful that they can be triggered even by etymologically unrelated but phonetically similar words.These associations count strongly in favor of holding users of slurs to their standard derogatory meanings.
We also saw that the effects of stipulation are temporary -typically restricted to the original context of utterance.This is because audiences outside the original context, even if they were present in the original context, will no longer expect the stipulated meaning to be in play.They may not be aware of, or may not remember the original stipulation.And, even if they do remember it, they will be unlikely to assume that the speaker expects them to remember it.So, reasonable and rational audiences outside of the original context will interpret the speaker in line with the normal meaning of the term rather than its stipulated meaning.This constitutes a strong reason to hold the speaker to the normal non-stipulated meaning in such contexts.
So, the fittingness approach to metasemantics is well-placed to explain the impact of stipulation on meaning.Indeed, it can do more than this.We are able to influence the meanings of our words through countless means other than stipulation.For example, we can fix demonstrative reference by raising a particular object to salience, and we can influence the resolution of context sensitivity by manipulating the question under discussion (see Schoubye and Stokke 2016;Stokke 2016).We are able to do this because there are countless ways of influencing the word uses it is fitting to hold us to.For example, the word use it is fitting to hold a speaker to will be typically be affected by the manner in which a reasonable hearer would interpret them.So, any way to prime a reasonable audience towards a particular interpretation could potentially affect the word use it is fitting to hold the speaker to.And there are innumerable ways to do this (a similar point is made in a different context by Buchanan and Schiller (2021)).

Conclusion
I started by outlining some puzzling features of stipulation.First, it is unclear why stipulation is able to affect meaning.Secondly, it is unclear what determines the limits of stipulation.I also went on to identify a third question: why should we care about stipulative success?I considered two related accounts of stipulation: the speaker-meaning approach and the micro-language approach.It was found that neither account was, without supplementation, able to account for the metasemantic impact of stipulation.I then argued that stipulation affects meaning by affecting the word uses it is fitting to hold the speaker to.Combined with Peet's (2022) fittingness-based metasemantics this explains the effect of stipulation on meaning.It also explains the limits of stipulation, and why we should care about stipulative success.
Finally, I pointed out that the fittingness approach explains our ability to manipulate meaning through means other than stipulation.There I was concerned with a subject's ability to control the meanings of terms in their own utterances.However, the point generalizes.There has been much recent interest in conceptual engineering: the idea that we can solve philosophical problems and fight for social justice by altering the meanings of our words. 9I take no stand on the viability of conceptual engineering as a means to promote social justice or resolve philosophical problems.But I do take seriously the idea that conceptual engineering is a widespread activity both within and outside of philosophy (i.e. in law and the sciences) and that, if possible, we should aim to make rational sense of this activity in general.However, the rationality of conceptual engineering is threatened by what is called the 'implementation problem' (Cappelen (2018), Deutsch (2020)): the question of how it is that conceptual engineers are able to affect the meaning of a term in a community given the fact that meaning is dependent in an inscrutable way on a myriad of factors beyond our control.
I am not going to attempt a full answer to the implementation problem.I agree that conceptual engineering will typically be very difficult, and it will often leave a great deal to chance.However, the 9 Not all conceptual engineers conceive of their task in this way.For example, Simion and Kelp (2020) and Flocke (2020) each, in different ways, present the conceptual engineer as being concerned with developing different ways of representing or thinking about the world.My own view is that there are various different activities that fall under the heading of 'conceptual engineering'.These include, as Simion and Kelp (2020) and Flocke (2020) would have it, developing different ways of representing the world.But they also include changing the meanings of words.And it is this latter activity with which I am concerned here.
fittingness view of metasemantics does give us some insight into how conceptual engineering proposals can get off the ground.They can do so by affecting the word uses it is reasonable to hold other speakers to.By introducing and advertising novel proposals for word use conceptual engineers can make these word uses available to speakers in a wide range of contexts.And, by arguing for the superiority of these uses over alternatives they introduce and raise to salience reasons to hold speakers to these uses.These arguments will have to be balanced against competing reasons to determine the word use it is fitting to hold a speaker to on a given occasion, and if the arguments for adopting a particular use are not widely known then they will have little weight in most cases.But this is just as it should be.Conceptual engineering proposals typically do not, by themselves, have a major impact on meaning.But the enterprise would be irrational if they had no impact.The fittingness proposal provides a mechanism for understanding the (typically minimal) impact conceptual engineering proposals have on meaning.