This is a post by Dario Mortini (University of Barcelona).
Simple ‘outright’ assertions serve a wide range of familiar communicative purposes in our daily lives: from giving directions (‘The Sagrada Familia is up that street to the right’) to sharing novel and interesting facts about the world (‘The new bakery in the neighbourhood sells apple cake’). However, we are not always well-positioned to make outright assertions. When uncertainty looms, we must settle for something weaker.
Perhaps you only recently moved to Barcelona, and you don’t feel too confident about your knowledge of the city. In this situation, it seems more appropriate for you to say something like ‘The Sagrada Familia is up that street to the right, I think’. Or perhaps you casually read about the new bakery on an Instagram food page whose recommendations have both hit and missed in the past. In this case, it is more prudent make a weaker assertion: ‘The new bakery in the neighbourhood sells apple cake, or so they say’. A bit more precisely, these weaker speech acts are called ‘hedged assertions.’ Recently, philosophers have appealed to hedged assertions to reach a familiar conclusion – that outright assertion obeys to a knowledge norm. But does this new argument succeed?
First, it helps to give a general characterisation of the philosophical debate on norms of assertion. Philosophers have long been interested in which norm, if any, regulates the speech act of assertion. Suppose that you want to make an unqualified claim about something or someone. In which position would you have to be for this claim to be in epistemically good standing? An influential answer to this question appeals to knowledge. The basic idea is that you have to know what you say. Some philosophers have defended a weaker requirement: you don’t have to know what you say, but you have to believe it with a good reason – or perhaps even just believe it. Others have been attracted to a stronger equirement: you have to be certain of it. These are not the only options. The literature on norms of assertion is rich (and, at times, admittedly unwieldy).
Philosophers are in the business of giving arguments, so you may wonder about which arguments have been offered in support of these options. Overall, the argumentation has proceeded mostly on the basis of linguistic data. Philosophers have used more or less realistic examples of assertions and checked which of the proposed norms best explains these examples. The question, then, becomes the following: which of these norms best explains our ordinary practice of asserting? The answer matters greatly, as the norm that provides the best explanation will be the right one. In philosophy, we call this type of argument abductive. The conclusion of an abductive argument is not guaranteed to be true, but we have good reasons to take it to be true because it explains a wide range of different phenomena.
Now, the literature on norms of assertion has focused on multiple sets of linguistic data, which I will not review here. One of the options mentioned above – the knowledge norm – does particularly well in explaining these sets of data. Interestingly, and back to the topic of this post, hedged assertions have been recently added to the mix.
It has been noted that hedged assertions sound fine with denials of knowledge but come off as very odd when paired with knowledge claims. To see this clearly, let’s revisit our previous examples by adding knowledge denials and knowledge claims:
(1) The Sagrada Familia is up that street, I think, but I don’t know if it is.
(2) #The Sagrada Familia is up that street, I think. Indeed, I know it is.
If a speaker utters (1), we wouldn’t bat an eye. But if a speaker utters (2) without follow-ups or clarifications, we would immediately perceive that something is off. The second part of their speech act – the knowledge claim – seems out of place after hedging. Why? What explains this contrast?
If outright assertion obeys to a knowledge norm, this contrast is easily explained. Take (1): the speaker successfully distances themselves from what they are saying because they do not know. That’s precisely what you would expect if outright assertion required knowledge. Now take (2): the speaker fails at distancing themselves from what they are saying because they also claim that they know. Again, this failure is precisely what you would expect if outright assertion required knowledge. A speaker claiming knowledge thereby signals that they are well-positioned to assert outright, so hedging becomes somewhat pointless. This is a very promising argument in favour of the knowledge norm of assertion.
In a recent paper, I have argued that this argument may prove too much. At closer inspection, we find similar patterns with both stronger and weaker norms. First, certainty:
(3) The Sagrada Familia is up that street. I think, but I’m not certain that it is.
(4) #The Sagrada Familia is up that street, I think. Indeed, I’m certain that it is.
Next, belief:
(5) The new bakery in the neighbourhood sells apple cake, they say, but I don’t believe it.
(6) #The new bakery in the neighbourhood sells apple cake, they say. Indeed, I believe it.
These new data support two competing hypotheses: that outright assertion is governed by a stronger certainty norm and that outright assertion is governed by a weaker belief norm. We started with a smaller set of data comprising only examples with knowledge. Now we have expanded the set and added more examples with certainty and belief. So, where does all of this leave us? Plausibly, since we observe similar patterns with both stronger and weaker norms, this richer set of data is no longer best explained by the knowledge norm. It may well be that outright assertion obeys to a knowledge norm, but looking at how we hedge outright assertions in ordinary language does not seem to secure this conclusion.
We have reached a familiar point: linguistic practices are suggestive but ill-suited to settle a philosophical issue. In a passage that I particularly like, John Langshaw Austin warns us of this limitation. He writes:
Certainly, then, ordinary language is not the last word: in principle it can everywhere be supplemented and improved upon and superseded. Only remember, it is the first word.
The same may apply to the connection between hedging, knowing and asserting. Linguistic data are only the beginning of a story that ties the three together. In the long run, different considerations may supplement, improve, and supersede these data.
To read more, check out my paper ‘Evidentially Hedged Assertions and the Knowledge Norm of Assertion’, Erkenntnis (2025). https://doi.org/10.1007/s10670-025-01015-6
Dario Mortini is a María de Maeztu postdoctoral fellow at the Barcelona Institute of Analytic Philosophy, where he works on a range of topics at the intersection of epistemology, philosophy of language, and philosophical methodology.
Open Philosophy 