Kant and Herder on Reason’s Relation to Language

Abstract

This paper offers a reading of the infamous mutual critique between Kant and Herder by criticising the standard account of their tense relation, which attributes a priority of reason to the former and a priority of language to the latter. As Kant thinks that judging can only be realised through its expression in language, and Herder conceives of a linguistic act as the self-conscious positing of meaning, they equally reject any sharp separation of thought from its expression in language. The central difference lies in their opposing accounts of the relation between reason’s striving for metaphysical knowledge and the latter’s linguistic guise.

1. Introduction

It is well known that Kant and Herder were not only colleagues and intellectual companions, but also rivals and sometimes harsh critics of each other’s work. While it is obvious that there are fundamental differences between them, the way in which the literature has usually approached their parting of ways can often seem unsatisfactory, especially when it comes to the question of how language and reason relate to one another. Taking for granted on the one hand that Kant was not interested in natural language and its relation to thought (Markis 1982, Villers 1997, Mosser 2001), and on the other that Herder misunderstood the critical project of assessing reason’s claims to knowledge independently of any historic or linguistic contingencies (Berlin 2000, Simon 2010), many scholars end up in a rather unproductive taking of sides between either Kant or Herder, and, by the same token, between the primacy of either reason over language or language over reason.

In this paper, I propose that Kant and Herder in fact hold complementary views concerning the interdependence of thought and language. If one looks carefully enough, it becomes quite clear that Kant was not ignorant of the fact that judging always involves expressing one’s thought in language. As we will see, he was actually very explicit in his critique of the rationalistic tradition that tried to separate inner judgement from its outer expression in a sentence. In this, I follow the growing consensus in recent scholarship that Kant recognised the linguistic character of thinking (see Ehrsam 2016, Capozzi 2021, Buroker 2021, Saarni 2023, Ehrsam 2023), though I will try to propose stronger arguments as to why this is not only not contrary to the critical project but also an integral part of it. At the same time, any reading of Herder’s influential Treatise on the Origin of Language that lives up to the demands of the novel account of linguistic meaning proposed there must take into account the self-conscious nature of the linguistic acts of the human being (see Tritschler 2024: 7–11). It implies that even perception is connected to the use of sensual signs in order to express meaning and hence cannot be understood separately from our rational-linguistic capacities (see Trabant 2009). Hence, both authors alike should be seen as opposing a dualistic account of thought’s relation to language.

The paper has four parts. First, in Section 2, I will show how the alleged opposition between reason and language is an outcome of the very polemics Hamann and Herder levelled against Kant’s critical philosophy. In Section 3, I will argue that contrary to this misguided assessment, the critical Kant was already opposed to a dualist account of how judgements relate to their linguistic expression in sentences. Section 4 will then centre on Herder’s idea that any linguistic act must be a self-conscious transformation of something sensible into a meaningful sign, which I take to be in agreement with Kant’s critique of the dualist account of reason’s relation to language. In Section 5, I will end by suggesting that the more fundamental disagreement between the two authors is to be found in their diametrically opposed positions concerning the possibility of metaphysical knowledge.

2. Kant and Herder – reason vs language?

The relation between Kant and Herder is a complicated one that lived through many phases of recognition, critique, and mutual misunderstandings (see Borsche 2006). Since it is a well-known story that has been historically thoroughly investigated (see, among others, Gaier 2006: 118–123, Shell 2010: 108–112), I will just hint at how and on what grounds they criticised each other.

In his Metacritique, Herder criticizes Kant’s critical project in the Critique of Pure Reason for searching in the wrong place when he tries to analyse where reason’s contradictions, in which all metaphysical attempts inevitably end up, stem from, and how they could be prevented. He specifies that a ‘large part of the misunderstandings, contradictions and inconsistencies that are attributed to reason are probably not due to it, but to the inadequate or poorly used tools of language’ (FHA 8: 320; my translation). ¹ Instead of trying to artificially create a variety of sets of conditions of the possibility of a priori knowledge, as the different parts of the Critique allegedly all do, Herder suggests that reason should return to the ‘origin of its possession’ (FHA 8: 342), which can only be found in the expression of our concepts through language. What Kant appears to do instead is to make it seem as if reason, in its search for a priori knowledge, should separate itself from everyday language use, which leads Herder to ask the rhetorical question about the terminology used in transcendental philosophy ‘whether there can be any misuse of language worse than this one’ (FHA 8: 343; my translation). It goes without saying that everyone familiar with Kant’s work can easily see that these polemics are not based on an adequate interpretation of the critical project.

On a slightly different topic, but on a similar level of polemical contempt, Kant writes in his review of Herder’s Ideas for the Philosophy of the History of Mankind, on the latter’s speculations about an original unifying force of nature, that it amounts to ‘the attempt to explain what one does not understand from what one understands even less’ (RezHerder, 8: 53; my translation). ² In order not to further mix up metaphysical speculation with the natural sciences, Kant then expresses his hopes that his former pupil, in finishing his ambitious book project, would start to impose some form of discipline on ‘his vivid genius’ and that philosophy would guide him ‘not by mere hints, but precise concepts’ to carry out the work of ‘cautiously exercised reason’ (RezHerder, 8: 55).

These polemics obviously do contain a fundamental disagreement as to how the two authors think metaphysical questions should be approached, and the standard account has located the central root of this dissent in their respective view of the relation between reason and language. As seems to be clear from the two passages quoted, Herder suggests focusing on our experience with the concepts we already have in our use of language, instead of searching for a priori knowledge that precedes said experience. Kant on the other hand warns that only critically informed philosophy is able to prevent us from ending up in unfruitful speculation fuelled by an excessively creative use of language, if only we separate justified from unjustified knowledge claims through an investigation of their origin in our rational as well as sensible capacities. In short, in addressing their dispute, we seem to have to choose between prioritising language over reason or reason over language. The standard account of their disagreement is happy to accept this framework and only differs when it comes to choose between these two options (e.g. Berlin 2000: 239–40, Gaier 2006: 122).

Now, the central idea of such an account, that Kant and Herder are opposed in their positions mainly due to their differing priorities concerning the relation between reason and language, is of course not new. It was, in fact, the polemical side of Hamann’s and Herder’s own critique of the transcendental project which drew a sharp line between the alleged ignoring of language in the Critique and the historically informed accounts of linguistic expression in their own writings. In his draft Metacritique of the Purism of Reason, Hamann made clear very early on what he thought went missing in Kant’s project: history, experience, and, most crucially, language (Hamann 2001: 57–64). While the largely forgotten systematic core of Hamann’s arguments is a very sharp and effective assessment of some difficult problems the critical project has to face with its focus on the a priori character of properly justified knowledge, ³ it is their polemical side that has survived and consequently shaped our idea of what Herder was after when he distanced himself from Kant. ⁴

Until recently, it was especially the clear-cut separation of reason from language, and the idea that this has to be a fundamental trait of Kant’s critical philosophy, that enjoyed long-lasting success in Kant scholarship. Of course, there are differences concerning the explanation as to why Kant supposedly ignored the so-called problem of language. While some think that Kant was somehow just blinded to the importance of language by his focus on the a priori (Gipper 1987: 76, 117–8, 121), the most ardent proponents of this line of interpretation claim that Kant was not even in the position to understand the so-called problem of language, since the critical project essentially consists in throwing ‘impure’ linguistic considerations out of the paradise of non-linguistic pure reason (Markis 1982: 110–54, Villers 1997: 3). As we will now see, these interpretations are not only historically dubious – as the above-mentioned recent scholarly work has already successfully shown – but, even more important, must subscribe to a misguided reading of Kant’s own non-dualistic position in the Critique.

3. Kant’s critique of linguistic dualism

In this section, I will show that Kant argues that all thought must be intersubjectively communicable, and thus objectively judging how things are can never be carried out without the use of some linguistic expression. The literature on Kant’s approach to the faculty of language has probably long been misled by the fact that it is not an explicit theme in the Critique as the foundational work for transcendental philosophy and seems to be marginalised by its placement as a sub-capacity of the imagination in the Anthropology from a Pragmatic Point of View, itself a work of contested status among scholars. These observations are of course not wrong, but they cannot be used to accuse Kant of ignoring language’s significance for reason’s activity altogether. In the Anthropology, Kant writes:

All language is designation of thought and, on the other hand, the most exquisite way of designating thought is through language, this greatest instrument for understanding ourselves and others. Thinking is a discoursing with oneself …; consequently, it is also a listening to oneself … inwardly. (Anth, 7: 192)

The passage exhibits several tensions in Kant’s view of language in relation to thought. If language’s function is to designate our thoughts, does this imply that the latter are themselves independent of such a designation? This seems to be further complicated by Kant’s claim that not only do we use language to communicate our own thoughts to others and to be able to receive theirs but also in order to understand ourselves. Finally, is the unusual expression that language is just the ‘most exquisite way’ of achieving such a designation of our thoughts supposed to mean that there are other means available for reaching the same goal? In short, focusing on this passage alone, we cannot be sure whether Kant thinks that language is an intricate part of thinking or rather a mere means to make thoughts explicit and clear to ourselves as well as others. ⁵

It is only once we broaden the textual basis that we can see what Kant thinks of the relation of a thought to its expression in a sentence (see Capozzi 2021: 98–9). In his reply to Eberhard, Kant states in a footnote:

The logicians are by no means correct in defining an assertion [Satz] as a judgement that is expressed in words; for, we must also make use of words in thought for judgements which we do not regard as assertions. (OAD, 8: 193f. fn.)

The context makes clear that Kant wants to guarantee that so-called problematic judgements, which are not themselves asserted, for example, the antecedent in a hypothetical judgement, can still count as judgements. To achieve this, he criticises conceiving of the distinction between merely considering a thought in a problematic judgement and also claiming its truth in an assertion or sentence (Satz), by assigning linguistic expression only to the latter and not to the former. An important consequence of this critique is that no judgement, not even problematic ones, can be conceived without being expressed in words. In the same vein, he stresses this point several times in his lectures on logic. Here is just one of these passages:

Now when the logicians say: a judgement is an assertion clothed in words; that means nothing, and this definition is worth nothing at all. For how will they be able to think judgements without words? (Log-Wiener, 24: 934)

Note that even though all of these passages are concerned with the specific problems Kant finds in the traditional understanding of the modality of judgements, he uses an argument that implies that all judgements must be expressed in words. Since all judgements need to be so expressed, the use of words cannot be what distinguishes assertoric from problematic judgements. ⁶

In fact, I will argue that these passages all speak for interpreting Kant as criticising the following position:

Linguistic dualism: Thoughts exist and can be understood independently of any expression through words in sentences. ⁷

Kant explicitly says that the rationalistic picture behind such a position, that a thought is only contingently put into words, but is in itself independent of any such expression in a sentence, is misleading. The reason he gives seems straightforward: as no one has yet achieved thinking a thought without the use of words, our concept of a thought should not exclude this fundamental trait. Linguistic dualism is wrong, according to Kant, since it ignores the linguistic nature of the act of thinking a thought.

Now, it seems that such a position can be easily criticised as mere psychologism. Even though we cannot but think in words, this could just show that there is one more layer of contingent facts about our psychology that has to be separated from the actual meaning of what a thought with objective purport is. The latter’s objective validity, as Frege’s critique of psychologism has successfully shown (see Frege 1984: 368–9), should be conceived of as being independent of which individual person thinks it at which time, and, by the same token, through which means, linguistic or other, such a grasping of its truth takes place.

That Kant was a linguistic dualist also seems to follow from the already mentioned fact that he mainly considered the issue in the Anthropology, which only treats of the human being with regard to certain pragmatic goals, which are, strictly speaking, not themselves transcendental (Anth, 7: 119). Along these lines, one could also interpret the following striking statement from the Opus Postumum:

Thinking is a speaking, and this a hearing. (OP, 21: 103)

Such passages have usually been read as Kant’s recognition of the anthropological fact that, in our infancy, we learn to think by acquiring a natural language and its expressive possibilities. As he notes in one of his reflections:

We need words, not only to become understandable to others, but also to ourselves. This capacity to use words is language, and children learn to speak. (Refl, 16: 839)

A linguistic dualist could take all these passages to support the claim that, for Kant, there is a fundamental difference between the anthropological recognition of a contingent fact about our psychology and the nature of the thoughts we grasp in thinking them. Either Kant did himself think this and expressed such a difference in the quoted passages, or he should at least in principle have recognised it, due to his own hostility towards psychologism in logic (see A54/B78, B167-8, and JL, 9: 14).

Such a reading is not to be disregarded from the outset, and it presents far stronger reasons for the separation of thought from language than the already mentioned interpreters who take Kant simply to ignore the issue. But even then, I hold that there are good reasons to think that this line of interpretation is mistaken and that Kant has in mind a much stronger relation of our rational faculties to their linguistic expression. ⁸ First, let me propose that we reconstruct Kant’s main idea here in the following way. If a sentence were only the ‘outer’ expression of the ‘inner’ judgement, how could it even be understood? It is not enough to say, ‘I can only think in words, and I need them to cause you to think the same by uttering them’. For then, it would not be clear how such a causation of your understanding by my words could be successful, as the latter would be just the expression of my private thoughts without any relation whatsoever to anybody else thinking them. Only if the act of thinking is already conceptually connected to some potential understanding by someone else can a sentence be the expression of my shareable judgement, which can only be achieved if thought is already formed through the use of language. This would imply that without the potential for being understood as a speech act, there would not even be any thought to begin with. If that turned out to be true, then the allegedly mere contingent fact that we need language to express our thoughts is an always necessary part of the act of thinking itself. Since an act of thought is for Kant nothing psychological, but rather a necessary and universal characterisation of the thought’s object, through self-consciously taking it to be so, language would in this specific sense have to be an intricate part of all acts of thought.

At this point, it might be important to stress that when Kant talks about the role of language for judging in general, he is not concerned with particular languages, such as German, Latin, or English – although he has also made some controversial claims about them, as has recently been shown (Lu-Adler 2023: 113–5). Rather, he has in mind the fact that we need to use some form of expression if we want to carry out thinking. So while he would certainly reject the claim that contingent facts about some natural language have anything to do with the universal forms of thought that we have through the categories, he also sometimes compares transcendental logic with the grammar of natural languages (e.g. P, 4: 322). Although I will not be able to focus on the role of the analogy between logic and grammar here, there has been a continuous interest in exploring this relation recently (see Mosser 2008, Buroker 2021, Capozzi 2021, D’Agostino 2023).

The above-sketched systematic argument can be textually justified by considering the transcendental deduction. One central part of the complex argument which Kant gives there is that judging is necessarily connected to the self-consciousness of this act and thus implies the possibility to reflect on the fact that the judging subject holds the judgement to be objectively true, independently of her grasping it (B141). The self-consciousness Kant has in mind by calling it the ‘original-synthetic unity of apperception’ (B132) is subjective, but only by being universal:

I also call its [the apperception’s] unity the transcendental unity of self-consciousness in order to designate the possibility of a priori cognition from it. For the manifold representations that are given in a certain intuition would not all together be my representations if they did not all together belong to a self-consciousness, i.e., as my representations … they must yet necessarily be in accord with the condition under which alone they can stand together in a universal self-consciousness, because otherwise they would not throughout belong to me. (B132-3)

An act of judgement must be able to be carried out by different thinkers, since thinking it means claiming it to be objectively true (B167-8). Of course, the universality Kant has in mind here is first and foremost the universal nature of the act of judgement. But it is an essential part of this universal nature of objective judgements that they can be carried out by different thinking subjects. If this is correct, then the judgement’s objective universality implies its self-consciousness, and the latter implies its intersubjective universality.

Understood in such a way, objective and intersubjective universality are two sides of one and the same idea: that a judgement can be thought by different thinkers on the grounds that its truth is independent of who thinks it. Judgement must be, as a self-conscious act of thought, always already shareable by different thinkers. That a thought belongs to everyone who thinks it, is not a feature added to it in a second step, but rather a condition of the possibility of understanding the identity of its intersubjective form with its objective content. This implies that the kind of intersubjective agreement we are talking about here is not, as it were, located on the level of the content of judgements, but rather consists in the ability to judge in the first place. Since the latter implies a capacity to make oneself understood in a shared language, it amounts to the fundamental agreement of self-consciously using language at all.

In Kant, we find further evidence for this position in two important passages. In the section On having an opinion, knowing, and believing, he stresses that ‘objectively’ taking something to be true must mean that ‘it is valid for everyone’ (A820/B848). He then goes on to write:

Truth, however, rests upon agreement with the object, with regard to which, consequently, the judgments of every understanding must agree … The touchstone of whether taking something to be true is conviction or mere persuasion is therefore, externally, the possibility of communicating it and finding it to be valid for the reason of every human being to take it to be true; for in that case there is at least a presumption that the ground of the agreement of all judgments, regardless of the difference among the subjects, rests on the common ground, namely the object with which they therefore all agree and through which the truth of the judgment is proved. (A820-1/B848-9)

While Kant interestingly stresses the ability to communicate what we think in order to find out whether our thoughts are objectively true by observing the agreement with other thinkers, this can only be properly achieved by measuring both acts of taking something to be true by directly relating them to the object the judgement is about. In the Prolegomena, he concludes from these considerations:

Objective validity and necessary universal validity (for everyone) are therefore interchangeable concepts, and although we do not know the object in itself, nonetheless, if we regard a judgement as universally valid and hence necessary, objective validity is understood to be included. Through this judgement we cognize the object … by means of the universally valid and necessary connection of the given perceptions. (P, 4: 298)

It is in this sense that all actual thinking is not only connected to our ability to say what we mean, but to say it in a way that makes it possible to be understood by others, as well as ourselves. The linguistic dualist holds that in order for us to achieve universality in judging, we need to abstract from all linguistic particularity. But in doing so, we can only reach the appearance of universality, since it would be the result of an abstraction from the self-conscious act of the particular thinker that thinks the judgement in language. ⁹

One might worry here that this type of argument is not in line with Kant’s focus on the objective validity of a true judgement, as opposed to its intersubjective communicability, which is only an outer sign of the former. ¹⁰ But this worry is unwarranted, since there is no difference between knowing a judgement to be objectively true and knowing it to be shareable with others. Of course, not all sentences we utter are true. But all judgements we know to be true objectively are shareable through some form of language. It goes without saying that here, I use ‘language’ non-restrictively, that is, in the most general sense possible, which includes all kinds of expressions of a thought – from rudimentary linguistic forms such as gestures to sentences in some highly abstract formal language. As already mentioned, the expression of a judgement in a string of words is only one such form, albeit for Kant a central and very important one.

If the given argument is sound, then it represents a better interpretation of the passages in which Kant criticises the rationalistic two-step picture of first grasping a non-linguistic thought and then dressing it up in some linguistic guise. Even more important, we have seen that it is in line with central tenets expounded in the transcendental deduction in the Critique, the most central text in Kant for understanding what a self-conscious act of thought, qua being contained in a judgement, is. Kant cannot be a linguistic dualist, since this would contradict his position that the objective universality of a thought is tied to its self-conscious expression through some linguistic particularity for a shared community of thinkers.

4. Herder on linguistic self-consciousness

The preceding section has shown that Kant does not belong to the camp of linguistic dualists who try to separate the universal objective meaning of a thought from the particular sentences of a natural language in which it is expressed but rather takes the position that one of the two can only be had in connection with the other. In this section I will argue that, on this question, Herder is in agreement with Kant, as the former’s account of how language is infused with and depends on self-conscious thought excludes linguistic dualism as well. As I will now show, through his transformative account of the linguistic being, Herder ultimately reaches the same conclusion by way of a different trajectory.

The problem of the origin of language Herder tackles in his pioneering Treatise on the Origin of Language was a contested topic of debate in eighteenth century philosophy (see Lifschitz 2012). To put it simply, the question is how such an extraordinarily complex system such as a natural language could have been invented by human beings without the capacity to think, if at the same time the latter also seems to presuppose at the very least the mastery of language in order to be carried out. Herder summarises the difficulty as follows:

Without language the human being has no reason, and without reason no language. (Herder 2002: 91; FHA 1: 727)

There were of course several rival approaches to an answer to this puzzle. Two of the most important strands which Herder directly engages with were those of the French sensualists and the German rationalists. While Condillac argued that the origin of language must be a merely sensible unfolding of what is already present in the communication achieved by other animals (Herder 2002: 75–6; FHA 1: 708–10), Süßmilch countered this by stressing the immense complexity of human language which could only be explained by reference to a divine teaching (Herder 2002: 69–72; FHA 1: 702–5). Herder in turn tries to find a way beyond both super-sensible (cf. Sikka 2004: 183–200) and empiricist (cf. Bertram 2006: 227–46) explanations for the origin of language.

According to Herder’s position, the human being must carry out thinking ‘in the first condition in which he is a human being’ and human self-awareness ‘must reveal itself in the first thought of the child’ (Herder 2002: 85; FHA 1: 719). To the subsequent question whether one could say that the child only displays a mere ‘ability’ to think, without exhibiting any actual ‘force’ to do so, Herder replies as follows:

Mere, bare ability which even without a present obstacle is no force, nothing but ability, is as empty a sound as plastic forms which form but are themselves no forms. If not the slightest positive contribution to a tendency is present with the ability, then nothing is present … It is sophistry that the use can transform an ability into force, something merely possible into something actual; if force is not already present, then of course it cannot be used and applied. (Herder 2002: 86; FHA 1: 720–1)

To be sure, this just seems to show that human children must first have the capacity to think, which then is being trained and actualised by language acquisition. Thus, it appears, there must be a temporal as well as a conceptual primacy of reason over language. Note, however, that Herder’s position implies that reason is not a capacity which can be acquired at a later stage in the life of a human being, but must always, in some form or another, already be in act. Now, if language is a capacity that is equally required for being human, its invention cannot be a temporal event either, but must rather be the natural unfolding of the rational being’s way of navigating the world (Herder 2002: 89; FHA 1: 724).

How can we then, against this background, account for the origin of language as an actual achievement of the human being? We are well advised to look for Herder’s answer in the ingenious first sentence of the Treatise:

Already as an animal, the human being has language. (Herder 2002: 65; FHA 1: 697)

Taken at face value, this statement seems to involve a contraction, since Herder also holds that language is a trait that is specific to human beings (see Herder 2002: 77; FHA 1: 711). A human being could then still be an animal under the very same description that is supposed to characterise it as being distinct from all animals. On my reading, Herder tries to show over the course of the whole Treatise why this very first sentence is not a contradiction, but a mere paradox whose function is to get us on the right track in order to understand the linguistic nature of human reason. ¹¹

To make a long story short, if we want to avoid ascribing a plain contradiction to Herder in his very first sentence, we should take him to hold the following position:

Transformative Account of the Linguistic Being: Linguistic signs constitute meaning insofar as they are used by self-conscious rational beings to express their sensations and thoughts. By consequence, a being that has language is a self-conscious and sign-using being through and through, i.e. in all its acts, rational as well as sensible. ¹²

According to this theory, a linguistic sign is not just a contingently chosen given part of our sensibility which is then connected to a separate meaning, but rather depends on the use that beings like us make of it. If that were not the case, then we could not even explain how a perceivable sign can ‘carry’ any meaning. But as we have seen in Kant, this is a necessary presupposition for making intelligible how we can understand language at all. Herder connects this insight to the idea that the capacities that we share with animals, especially our ability to perceive objects that are different from us, must be infused with meaning. ¹³

This is exactly what Herder hints at with his infamous tale of the baptism of a bleating sheep (Herder 2002: 88–9; FHA 1: 723–4). What happens through the act of attaching a word to an object is the self-reflective universalisation of perception: I do not only name the sheep, but I also name, or rather express, my experience of it. Both happen in one and the same act. By naming it, I underline that I already perceive it in a distinct, repeatable way that can be expressed in a universal concept:

The human being, put in the condition of awareness [Besonnenheit] which is his very own, with this awareness (reflection) operating freely for the first time, invented language. … The human being demonstrates reflection when the force of his soul operates so freely that in the whole ocean of sensations which floods the soul through all the senses it can, so to speak, separate off, stop, and pay attention to a single wave, and be conscious of its own attentiveness. (Herder 2002: 87; FHA 1: 722)

It has been duly noted that Herder’s unusual tale results at this crucial juncture in the position that, in order to posit a meaningful word that stands for an object, we need self-awareness which enables us to isolate parts of our experience for further reference.

Herder continues to elaborate on this in his extensive consideration of the role of the human being’s hearing of sounds in the creation of meanings expressible through language (Herder 2002: 96–110; FHA 1: 732–49). A central point is that, in hearing, we already reenact the articulation of the meaningful structure of what we hear (Trabant 2009: 129). If the sheep bleats, I hear that it does so by noticing the ‘expressive’ aspect of its articulation. Transforming such a noise into the sign for its source is made possible by our own capacity for speech as a carrier of meaning. Herder’s point is of course not about some onomatopoetic imitation of natural sounds, but rather that, for a linguistic being hearing a particular sound already means taking the sensible to be able to carry a universal meaning. Speaking implies naming the universal in the individual self-conscious linguistic act. Finally, thinking, as it is considered here, means first and foremost relating such a linguistic universality to other such universalities, that is, forming concepts through the connection of words in sentences. As Herder will later write in the Metacritique:

What is it that we call thought? Inner speech, i.e. to articulate the internalized mark to oneself; to speak is to think aloud. (FHA 7: 389)

What we can see at work here is some kind of inversion of the already quoted dictum from Kant’s Opus Postumum, since Herder tries to show us that, for human beings, hearing is a speaking and this speaking is a thinking. It becomes clear that, in direct comparison with Kant, Herder criticises linguistic dualism exactly from the opposite endpoint. Following Herder, we can now see that one cannot understand linguistic meaning by separating its universality from the particular, as linguistic dualism does, but only through holding both together in the sensible, and for this very reason meaningful signs that human beings use to express themselves (see FHA 1: 733).

In the course of the last two sections we have seen that, starting from different ends of the conceptual relation of language and reason, Kant and Herder reach the same inflection point: the concrete unity of the word with its meaning. To appropriate Heraclitus’ beautiful metaphor: while Kant approaches this unity on the way down through the linguistic character of thought, Herder goes the same way up through our use of language in order to express thinking. This may also explain why the literature has taken them to be on different pages altogether. If one does not take into account how the two ways reach the same idea from opposite sides, it is of course impossible to see that, after the inflection, each would be in a position to traverse the other’s route from a different direction. But as we will see in the final section, as soon as we consider reason’s linguistic activity left to its own devices, there is still plenty to disagree on, or, to stick with the picture, other pathways open for the two authors to traverse.

5. The unconditioned and reason’s linguistic nature

Until now, I have stressed a fundamental agreement between Kant and Herder. If we have thus reached a better understanding of their independent, albeit complementary critiques of linguistic dualism, what then are we to make of the obvious differences expressed in their sometimes hostile assessment of each other’s philosophical projects? I will end by suggesting that the fundamental disagreement is to be found in the way in which, in connection with the nature of its linguistic activity, they each characterise reason’s striving for knowledge of the absolute.

As we have seen, for both authors the problem of how thinking is connected to linguistic expression comes down to the question of how universality and particularity are related to each other. According to Kant, a universal judgement must be thought by a particular thinker, which implies that it is necessarily expressed in language. In this, reason plays an ambivalent role. On the one hand, it is the faculty through which we know the particular in the universal, and thus recognition of this unity is central for any critique of linguistic dualism. On the other hand, through the distortion of language, Kant holds that reason necessarily starts to misunderstand its own capacities as soon as it tries to gain knowledge from recognition of the unity in question.

As this very fact has often been overlooked, it can seem puzzling how abruptly reason’s illusion emerges in the transcendental dialectic. There, Kant distinguishes between three kinds of illusions. An illusion of the senses stems from the way in which we experience phenomena and can result in an error of judgement if one does not take into account that things can often merely seem to be a certain way (A295/B351-2). As the senses themselves do not err, a correction in the corresponding judgement does not make the sensory illusion disappear (A297/B354). A logical illusion in turn does cease to exist as soon as one understands the logical error in play (A296-7/B353). Now, Kant famously holds that there is also a third kind of illusion which he calls ‘transcendental’ (A295-6/B352) and which is supposed to combine features of the other two. It is, like logical illusion, an illusion that stems from reason and not the senses, but it is also, like sensory illusion, a natural and necessarily persisting illusion (cf. Ganzinger 2023: 641).

While the exact nature and status of transcendental illusion is contested in the literature (see Grier 2001: 102–30, Willaschek 2018: 123–6, 135–48), for the sake of my argument I only make use of the core assumption that it always consists in unavoidably taking a condition that is objectively valid for our reasoning capacity to be a condition that is objectively valid for an object of knowledge. In particular, reason necessarily attempts to take the concept of the unconditioned, which is arrived at through asking for the condition of something conditioned, to be a characterisation of the object which allegedly corresponds to it (B364-6/A307-9). Such an illusion cannot disappear like some logical error, but it can also not just stem from how things falsely appear to be, as happens in sensory illusions. Rather, the necessity of transcendental illusion is due to a confusion about the status of our own finite reason (cf. Ganzinger 2023: 647), combined with a sensual fixture of reason’s concept of the unconditioned in a linguistic expression.

It is at this juncture that language becomes critical for Kant’s rejection of uncritical metaphysics. Since it is exactly because reason enables us to recognise a particular expression to be in unity with its universal meaning that we cannot help but try to take the latter as a quasi-object in itself (A307/B364). Thus in all three types of dialectical fallacies, it is the linguistic guise of metaphysical claims that enables the confusion to take hold on our reasoning capacity. ¹⁴ According to Kant, there is no danger here as long as we use reason’s power to originate ideas of the unconditioned in order to relate judgements in inferences that in turn enable us to organise our scientific endeavours in the pursuit of knowledge (A680/B708). But if, quite naturally, we suppose that using a specific expression for the result of such an operation, by which reason goes beyond what it conditionally interrelates, entitles us to posit the unconditioned as an actual object, we mistake a universal operation of reason for an existing totality. Or, as I formulated it above, as linguistic beings, we then naturally mistake a necessary but subjective principle of our cognition for a universal and objective principle to which we attach some linguistic expression. For Kant, it is crucial to always remind ourselves that any such concept of a totality of conditions is brought about by our finite rational faculties. Since the concept of the unconditioned must be created through the act of thinking it (A481-2/B510-11), its positing as something real independent of any such act must be seen as a necessary illusion generated by the use of reason, but whose linguistic guise makes it possible for it to hide in plain sight.

Such a position is fundamentally opposed to Herder’s approach to these issues. Since he starts with the sensible unity of universality and particularity in linguistic beings that perceive the world as carrying meaning, it seems natural for him to think that such a unity must actually exist, and that we can know of its existence independently of how our reason actively arrives at its concept. Herder claims that it is in and through language that reason reaches its fullest potential, since only here does it become possible to partition the unconditioned into its actual parts (Heinz 2013: 168–9, Seebohm 1972: 67–8). According to this uncritical approach to metaphysical questions, the word itself would amount to a universal concept which comprises individual instances, but is at the same time also something sensible, and thus can seem the ideal candidate for a totality that is in fact given to us, and not only thought (see FHA 1: 552). ¹⁵

This speculative conception of reason shows up in the Metacritique, when Herder tries to argue that our senses of hearing and seeing, as well as the corresponding organs, are fundamentally connected to the historical development of language as a kind of preservation of the universal in its particular linguistic form (FHA 1: 553, 8: 419–20; Zammito 2009: 66). Already in the Ideen, the above-mentioned concept of an existing original force of nature supposed to make us understand the unity of individual organisms (DeSouza 2018: 122–5) only seems to make sense if we presuppose that nature itself is something like an encompassing super-organism (cf. Zammito 1992: 202–6 and, critically, Martin 2021: 10–11). In his most speculative work, God: Some conversations, Herder goes even further and suggests that this alleged original force in nature has to be God (FHA 4: 710), and, by consequence, that God must be identical with the totality of all that exists in nature.

From a critical point of view, all of these arguments brought forward by Herder not only display transcendental illusion but are prime examples of the linguistic nature of the mistakes involved. Be it the hypostatising of humanity as a substantial subject expressing itself through historical languages in the Metacritique, the supposition of an existing natural whole from which we can derive individual organisms by parcelling them out of the totality in the Ideen, or the idea of God as an individual, but all-encompassing thing – all of these speculations clearly show the mark of what Kant would have, and in some cases indeed has chastised as a non-critical use of language that ends in the subreption of an idea as an independently existing unconditioned being (cf. Sandkaulen 2016: 638).

Due to his critical stance towards reason itself, Kant is in no danger of falling prey to these kinds of linguistic traps. Paradoxically, at least here he seems to be in a better position to preserve Herder’s insight into the interdependence of language and reason. As I hope to have shown in this paper, Kant has the conceptual resources to pose the question of the dependence of all reason’s activity on linguistic expression. In this, his position potentially even points beyond Herder. But it was Herder that made us properly see in the first place how even the comprehensiveness of such a question already takes us beyond a dualistic conception of reason’s linguistic nature.