1. Part I: Herder as a constitutivist

In Herder’s theory of language, language has constitutive impact on the content and manner of how we think. This is a running thread throughout his works, such as in his early Fragments on Recent German Literature and in his Ideas on the Philosophy of History of Mankind. It is also at the core of his critique of Kant in his mature and untranslated Metacritique of the Critique of Pure Reason (Metakritik zur Kritik der reinen Vernuft). The contrast with Kant in the Metacritique is an excellent context to specify his constitutivism.

In this strident antikantian book, Herder criticizes Kant’s critique of pure reason for its a priori transcendental rationalism. This may be surprising given the attempt by Kant to take on board elements from both empiricism and rationalism, by combining sensory input and a priori cognitive forms. Certainly, for Kant, experience is structured a priori by the human mind and ‘we can cognize of things a priori only what we ourselves have put into them’ (CPR: Bxviii). Yet knowledge of objects of experience requires those sensory inputs, which our cognitive system then processes: ‘we can have cognition of no object as a thing in itself, but only insofar as an object of sensible intuition, i.e. as an appearance’ (CPR: Bxxvi): there is no knowledge of objects beyond experience. Reason is limited by experience: ‘the voyage of our reason may proceed only as far as the continuous coastline of experience reaches’ (CPR: A396).

Despite this emphasis by Kant on the role of experience, for Herder, Kant still errs on the side of rationalism. For Herder, Kant’s theory is an unfortunate return to a ‘Cartesian’ way of thinking (Bojda and Gutschmidt Reference Bojda, Gutschmidt, Bojda and Gutschmidt2022: XXXV) about the emphasis on reason, albeit now as part of an idealism. Herder’s main critique is focused on the existence of transcendental a priori, which Herder denies.

Herder’s alternative is a position that combines (i) realism (against Kant’s idealism), (ii) a constitutive view of the role of language for cognition, and, ensuing from this, (iii) the view that reason is historically situated (against Kant’s a priori transcendentalism), and (iv) the idea that reason can be used to trim language and prevent its misuse (against Kant’s idea that reason generates antinomies). I now discuss these in turn.

(i) On the first issue, Herders’ realism, we can be brief. Herder does not claim originality here. He holds that there are truths out there, inserted into the world by God, and it is our task to seek to unravel them. He rejects the idea that what we know is just a reflection of ourselves, of laws of nature that we put into things (Herder Reference Herder, Bojda and Gutschmidt2022: 143). Instead, we must learn how the world itself is constituted. ‘It is a world, my son, which you have not, cannot and should not create; get to know her, become useful to her’ (Herder Reference Herder, Bojda and Gutschmidt2022: 6).Footnote ¹ So, rather than imposing our own laws, we strive to know the world as it is. ‘I want to recognize the universe as it is given to me and as I am given to it’ (300).

Herder thereby rejects Kant’s view about the limits of our knowledge, and specifically about knowledge of things in themselves. For Herder

So for Herder, there is a truth out there about the world as it is, and we can access that truth.

(ii) However, and this is the second point, to get to know this world that we endeavour to know as it is, we need to pass through language. This is because thought and language are connected. It is on this issue – the constitutive role of language for thought – that Herder puts most of the emphasis in his critique of Kant, who remains silent about language in the Critique of Pure Reason. As Herder writes in the first pages of the Metacritique:

For Herder, cognition requires language. Language is a collection of words that we use to organize ideas in our mind. Language expresses what is in the mind, but, at the same time, what is in the mind is already linguistically mediated. That is why Herder endorses what has come to be called constitutivism: language plays a constitutive role in the thought process. Words are not external representations of non-linguistic thoughts – thoughts formed without language. That way of thinking would be fundamentally non-constitutive, or instrumental. On the instrumental view, words are instruments with which we express thoughts, as if we already had perfectly formed thoughts and were just looking for external means to express them, or, in the helpful formulation by Isaiah Berlin, ‘as one looks for a glove to fit a fully formed hand’ (Berlin Reference Berlin2013: 234). By contrast, for Herder, words do express thoughts, but the thought process itself is already constituted by words. This constitutive view of the role of words, according to which cognition requires language, implies that thinking, because it involves words, can be understood as inner speaking:

Thinking is thus inward speaking, and speaking is outward thinking. When we think about an object, for example, we focus, within the ocean of possible things that we could focus on when an object presents before our senses, on some focal points. We do that with the help of Merkmale, characteristics of the object that we choose to focus on.

In doing, so, we have an inclination to both generalize and particularize the object before us. Generalization occurs by categorizing a given object under a class that we already know by using the Merkmal for that class. For example, when we encounter a specific tree, we may use the Merkmal and its corresponding word ‘tree’ to represent it (248). That means that, mentally, we classify this object as a ‘tree’, even though it is a specific tree.

But while we generalize, by finding similarities, we also have the urge to particularize, by making distinctions. In fact, we generalize with the intention of particularizing in the end (250). Human beings like to recognize an object as a specific member of the class. That is the act of recognizing a thing precisely. We like to understand a thing by finding the specific Merkmal for it that distinguishes it from other things we know. When we encounter a new thing, then, we select a specific Merkmal for it. We can think of a particular tree as a brown tree, for example, or a tree that gives apples.

Both when we generalize and when we particularize a thing, we use a Merkmal to refer to that thing. And since a Merkmal is the basis for a word, with the written or spoken word being the external representation of the Merkmal,Footnote ² using Merkmale for thinking implies that thinking occurs linguistically. (Of course, the Merkmale themselves have been passed on to us by culture and tradition, as I will now seek to show.)

(iii) This brings us to the third point and to the heart of Herder’s critique of Kantian pure reason – the historicity that comes with language. Herder’s point is not just that we think in language but also that the way we think with language makes us dependent on how language was formed before us. Indeed, as we just saw, when I seek to subsume a given thing under a category, I will use the word for that category. But, apart from new distinctions that I make myself, the words for categories are not mine; the words of my language were given to me by my parents and teachers. And people speaking different languages emphasize different things, and thus categorize differently.

This implies that, for Herder, we cannot uncover a priori cognitive universals: since language is necessary for thought, and the words in language have come to us historically, we think in culturally and historically specific terms. To arrive at a priori universals, we would have to bypass language, but since thought is mediated by language, doing so is impossible. As Schalow and Velkley put it, for Hamann and Herder, the Kantian ‘transcendental exposition of the principles of knowledge (…) hinges on another level of discourse’ rather than being ‘self-grounding’ (Schalow and Velkley Reference Schalow and Velkley2014: 5). That level is the level of language, which, in the way Herder understands it, involves also culture and history.

This point about the historicity of language is conceptually distinct – and therefore does not follow from – the point about the constitutive linguisticality of cognition (the second point above): it is distinct because, purely on the basis of that constitutive linguisticality point, one might think of the a priori concepts of reason as linguistic concepts. If one pursued that line of argument, one might arrive at a Chomskian idea of a universal grammar of language, the idea that transcendentally, we are linguistically endowed in a specific way. We could then think of the ‘table of categories’ as the table of the fundamental words or grammatical rules with which we think, for instance.

Of this idea, it is possible to find traces in Kant. While Kant does not theorize language in the Critique of Pure Reason (Kant Reference Kant, Guyer and Wood1998), several scholars have begun to explore the role of language in Kant’s writings (see Schalow and Velkley Reference Schalow and Velkley2014). Some of these scholars have even argued that Kant endorsed parts of the constitutive view of language prior to and after the Critique of Pure Reason (Forster Reference Forster, Schalow and Velkley2014; Lu-Adler Reference Lu-Adler2023). Of particular importance here is Kant’s idea of a ‘transcendental grammar’, which indeed suggests exactly this idea:

But while endorsing the constitutive role of language is in principle compatible with holding on to an idea of universal a priori principles of linguistic cognition, this is certainly not Herder’s point. In the Metacritique, Herder says that the way we understand things is not something that nature sent us into the world with but that it is an acquired way of thinking, that we have an educated reason (Herder Reference Herder and Forster2002: 300). Indeed, in the book, he repeatedly writes that ‘no Prius is without a Posterius’ (90). By this adage, Herder means that our principles of understanding are not universal a priori but are derived from our interpretation of reality. Since we select Merkmale, which we then speak out as words with which we interpret and categorize objects, those Merkmale are not a priori features. They are themselves derived from our interpretation of the world. And since the words (which capture Merkmale) in the language that we think with come from our ancestors – and not from nature; we are not born with them – cognition does not occur through transcendental a priori principles. Indeed, for Herder, to understand how a mind works, we need to understand the linguistic repository that the mind employs to think. Agreeing with Leibniz, Herder states:

Language is a historically formed storehouse of our thinking, which we use to think. Objects thus get classified by us using the words of our language.

The corollary is also that our thoughts are shaped by the language of our nation. The linguistic forms are not transcendental a priori concepts of thinking but come to us a posteriori, based on the words in the language of our nation. Every people has its own language-specific form of reason: ‘Every people has its own so-called reason, which expresses itself in language and constitution, in customs and institutions’ (Herder Reference Herder, Bojda and Gutschmidt2022: 293). So the mind is surely shaped by forms of thought, but rather than transcendental and a priori, these thought forms are national and a posteriori. They are historically, nationally, and linguistically inculcated ways of interpreting the world.

(iv) The above does not imply that we are imprisoned by the language with which we reflect. That is the fourth idea that Herder stresses: reason can tell us when language goes astray and when words and meanings must be corrected. In the Metacritique, Herder uses this idea to reject Kant’s idea that reason gives rise to antinomies. In Herder’s view, Kant’s antinomies are the result of bad usage of language, which leads to word quarrels and misunderstandings. Such seeming antinomies are not the result of reason and understanding itself, but from the ‘abuse of language’ [Mißbrauch der Sprache], from the ‘indeterminate grasp of concepts and their designation by deceptively obscure schemata’ (121). In order to preserve true reason from its problematic transcendental appearance, we must correct that kind of language, by clearly defining the terms that we use. Reason itself is true and pure; what Kant thinks of as antinomies are only the result of abuse of language. In short, the culprit is not reason itself but badly used reason resulting from badly used language.

This confirms that Herder holds that progress is possible. We can be led astray by our faulty use of language, but we can and should also correct language and reach progress (see also Cloeren: Reference Cloeren1988: 46-52). In his earlier work, such as in the Ideas as well as in the Treatise on the Origin of Language, Herder develops a theory of the universal progress of reason: humanity has one shared reason. For Herder, human reason expresses itself in a myriad of ways in different cultures, languages, and periods. But that is for him compatible with the idea of a universal progress of reason. And this progress also expresses itself in and through language. In fact, as is clear from above, the two – reason and language – develop in sync: as he writes in the Treatise, there is ‘progress of language through reason and of reason through language’ (Herder Reference Herder and Forster2002: 125). Herder argues throughout the Metacritique that Kant’s metaphysical verbiage sets language, and progress, on a wrong footing. But we can correct language, define words clearly, and thus straighten the path of progress of reason and language.

2. Part II: Condillac as a constitutivist

Above, I have outlined Herder’s linguistic constitutivism: thinking relies on language. Moreover, our concepts are not universal but specific to the language we speak and the historical period we speak it in. In what follows, I will argue that Herder’s constitutivism is less innovative than often thought. In particular, the idea that thought was dependent on language, and on the specific language we speak, was far from exceptional among late-18th Century philosophers in France. This idea was for example articulated by Étienne Bonnot de Condillac, the French Enlightenment scholar who engaged, just like Herder did in the Treatise, in the question of the origin of language. His constitutive views, and those of several others, had significant impact on several idéologues of the French Revolution who referred to Condillacian ideas on language. One important example of the latter group is l’Abbé Grégoire, the drafter of a famous language policy plan that was adopted by the Convention nationale in 1794, whose reliance on Condillacean constitutive views I will indicate in part III of this paper.

The claim that a theorist like Condillac already endorsed, 25 years before Herder did, a theory stressing the constitutive role of language for cognition (for this claim also see Aarsleff Reference Aarsleff1982: 146–209), is remarkable in the light of the fact that, frequently, the theories of Hamann and Herder are contrasted with a more instrumentalist view of thinking that came before them.

Charles Taylor is a case in point. As I discussed above, Taylor portrays the HLC school as endorsing an instrumentalist view of language as consisting of words as tools to name objects or thoughts. By contrast, the HHH tradition replaces linguistic instrumentalism with ‘constitutive-expressivism’, the idea that words are not just tools that make visible thoughts and feelings that we already had, but instead that language brings a wide range of human feelings and ideas into existence.

More specifically, Taylor argues that the contrast between those two traditions finds its origin in the critique by Herder of the position developed by Condillac in the debate about the origin of language. In the 18th century, philosophers like Condillac, Rousseau, Maupertuis, Süßmilch, and Herder contributed to the question how language has or could have come into existence. Condillac’s 1746 book Essay on the Origin of Human Knowledge (Essai sur l’origine des connaissances humaines) was one intervention in this debate, and so was the book that made Herder famous, the Treatise on the Origin of Language, published in 1772.

Taylor conceives of Herder in this origin debate as a ‘hinge figure’, who ‘originates a fundamentally different way of thinking about language and meaning’ (Reference Taylor1995: 79–80) in the critique by Herder of Condillac’s account of the origin of language. Taylor argues that Herder ‘brings about a rotation of our thought about language’, and that ‘many of our contemporary modes of thought are implicit in the steps he took’ (92). He identifies two main steps that Herder takes in contrast to Condillac: (i) constitutive expressivism and (ii) holism. Both ideas enabled other transformations emphasized by Taylor, such as the idea that we are both prisoners and creators of language, the idea that language brings transformations in our thinking, the non-priority of prose, and the dialogical nature of language. In what follows, I argue that, pace Herder and Taylor, those two main ideas are also articulated and endorsed by Condillac. If I am right, then it is misguided to classify Condillac as an instrumentalist or an atomist about ideas; he is squarely in the constitutive camp, as Hans Aarsleff has also argued in his analysis of Condillac (Aarsleff Reference Aarsleff1982: 146–209). After that I will also show that Condillac’s revolutionary ‘followers’ were constitutivists, which has implications for the contrast between the two schools.

(i) Take the first and main issue: constitutive expressivism. This is the idea that the expression plays a constitutive role in the thought process. For Herder, Taylor claims, ‘Speech is the expression of thought. But it isn’t simply outer clothing for what could consist independently’ (Taylor Reference Taylor1995: 92). The latter – speech as outer clothing – is what, according to Taylor, the older Condillacian outlook assumes: we already have our thoughts, sensations and feelings, and then we seek ways to utter them. Rather, speech for Herder is ‘constitutive of reflective (linguistic) thought, of thought which deals with its object in the linguistic dimension’ (Taylor Reference Taylor1995: 92).

I think Taylor is right in imputing this idea to Herder. My sketch of Herder’s constitutivism above corroborates this claim. But this claim does not constitute a fundamental break with the tradition of Condillac. On the contrary, the same vision applies to Condillac.

While at the start of his Essay on the Origin of Human Knowledge, Condillac applauds John Locke for dedicating the third book of Locke’s Essay Concerning Human Understanding to language, he still finds Locke lacking for not have paid sufficient attention to language (Condillac Reference Condillac and Aarsleff2001: 7–8). Whereas Locke assumes that the mind gets ideas through sensation which it can then endlessly mix and re-arrange into more complex ideas, Condillac argues that ‘it is well established that in infancy we had sensations long before knowing how to turn them into ideas’ (7). Indeed, he says, the mind does not immediately possess all the operations of understanding (such as memory and imagination). We need to explain, better than Locke did, how our mental operations develop and get exercised (7). The main explanatory factor here is signs, a claim for which he argues throughout the book, and also in its opening pages, by asserting:

Indeed, ‘Our mind fixes its ideas only by signs’ (Reference Condillac and Aarsleff2001: 90). So the development of the human faculties of memory and imagination ‘depends entirely on the connection of ideas’ (187), and this connection requires signs. This has grave consequences: we need signs – and thus language – to develop our mental faculties. We also need them to organize the ideas in our minds, and a differential organization of our signs would make us think differently.

This implies for Condillac that a literary hero like Corneille was dependent on the progress of his language for the expression of his genius: therefore, if he had lived earlier, he would not have been able to show us his talent because the French language had not yet reached the stage that it had in Corneille’s age (188). Another example: the scientific success of Isaac Newton was dependent on the progress of the English language:

Had they lived earlier than they did, Corneille and Newton could not have come up with the same results because of the state of French and English. It is only because of the progress of their languages that these geniuses could fully their talents in the manner that they have done now.

Could it be replied, Condillac subsequently asks, that great poets such as Corneille might have bypassed the limitations of French and English, by directly writing and thinking in Latin, surely a very advanced language? No, answers Condillac, and his answer is revealing: ‘having been accustomed to conceive of things in the same way as they were expressed in the language they had grown up with, their minds were naturally constrained’ (188). In other words, Corneille needed to have lived in the right time, within the particular developmental stage of 17th century French, for him to have been able to showcase his particular genius.

All this is remarkably close to what Herder argued vis-à-vis Kant. Based on what Condillac is saying, there could not be transcendental a priori structures of the mind: minds must be developed and that development requires language. Indeed, Condillac’s take on the impossibility of the development of the same genius in Corneille and Newton at a time when their language was not sufficiently developed, is an illustration of Herder’s point that reason progresses through language (Herder Reference Herder and Forster2002: 125). Indeed, for Condillac, we need signs to think, to arrange the ideas in our mind, and not only to express them. Rather than linguistic instrumentalism, Condillac endorses constitutivism.Footnote ³

(ii) The second fundamental change that Herder initiates, according to Taylor, is holism. ‘One of the most important and universally recognized consequences of Herder’s discovery was a certain holism of meaning. A word has meaning only within a lexicon and a context of language practices, which are ultimately embedded in a form of life’ (Taylor Reference Taylor1995: 93).

Does Condillac endorse this view? He does not explicitly mention the term ‘holism’ (and neither does Herder). Perhaps using the term holism in this context has become more common after the philosophies of language of Wittgenstein, to whose form of life view Taylor refers here, and Derrida, who emphasized the fact that words and meanings contain traces of other words and ideas, to which they are indebted for their meaning, rather than being stand-alone essences.

That said, Condillac does stress that the words in a language are influenced by the character of the nation. For him, language is not just important for individual cognitive processes but also for the thought process of collectivities. Languages are ‘like a painting of the character of each nation’s genius’ (Reference Condillac and Aarsleff2001: 194). He dedicates an entire chapter to this notion of the ‘Genius of Languages’ and to the claim that a people expresses its ideas in language, and, in turn, gets shaped by that language. By using signs, nations combine different ideas according to their particular genius (193–194). This implies that one needs to understand a word in the context of the character of the background nation that uses it: words do not have standalone meanings but require the background of the national language within which they are situated.

As an example of how the ‘character of nations influences that of languages’, Condillac speaks of the fact that a people will add ‘accessory ideas’ to its words (185). In Latin, he says, agricultural terms ‘imply ideas of nobility’ (185), which they do not in French. This has to do with the fact that the Romans found agriculture of high value as an important art, whereas, Condillac argues, the

This passage shows not only that words emanate from a national and historical speech community but also that the meaning, intent, and use of individual words co-depend on the ‘accessory ideas’ from that background community, which situate those words. In other words, words do not stand alone. They are embedded in the way of thinking of a nation and imbued with the accessory ideas that that nation attaches to it. So Condillac is a holist, in the way Taylor understands the word, namely: ‘a word has meaning only within (…) a context of language practices, which are ultimately embedded in a form of life’ (Taylor Reference Taylor1995: 93). Indeed, if a language structures the mind of a person in accordance with the national character of the people, for Condillac one-word lexica cannot exist; we would need the entire background understanding to properly comprehend and use each word.

In summary, Condillac is not the instrumental atomist of meaning that he is made out to be. He starts from the position that the mind needs words to think. We need words to exercise our mental operations such as imagination and reason and to combine the ideas in our mind. These words must be understood in the context of the national speech community: the background assumptions and associated ideas are part of what it means for a word to function. Rather than instrumentalism and atomism, therefore, Condillac endorses a constitutive view of language, and a form of holism – understood as the idea that words make sense as part of a national manner of thinking and speaking. For Condillac, just like for Herder, transcendental a priori could not exist since cognitive processes must be developed through language, and once developed, each nation cognizes in its own particular way. Both Condillac and Herder argue that cognition requires language, and that cognition and language develop simultaneously. Both also appeal to the notion of a national people with which the language is intricately connected.

3. Part III: Herder as a recognitionalist; Condillac as a non-recognitionalist

I have argued that the originality of Herder does not lie in his constitutivism about language and cognition, since the earlier French theorist Condillac, who is taken by Taylor and others as one of the main protagonists of the instrumentalist tradition, also held constitutivist and holistic claims about language.

Having said that, I do think that there is a relevant difference between a Condillac-inspired and a Herder-inspired take on language with direct relevance to contemporary language rights debates. Herder took an additional step that Condillac did not take. In what follows, I suggest that step is to be called recognitionalism. By recognitionalism, I mean the normative view that nations or languages are entitled to cultural and language recognition, for example in the form of language rights or by granting national cultures political autonomy. A particular recognitionalist normative theory will need to advance an account of what recognition entails. It will also need to add further conditions and principles of justice for calls for cultural and language recognition to be justified, and much of the political philosophy domain of multiculturalism, linguistic justice and nationalism is concerned with the nature of those conditions and principles. But to qualify as a recognitionalist theory, that theory must normatively hold that cultural and language groups are in principle – and subject to further conditions – entitled to accommodation in the form of national, cultural, or language protection or in the form of self-government. Recognitionalism holds that, in addition to civil and human rights, there are also cultural and language rights.

What is the relationship between constitutivism and recognitionalism? A constitutive view can cash out as a recognitionalist position, but it does not have to. I will now argue for two claims: (i) Condillac endorses constitutivism but does not take the recognitionalist step and (ii) Herder combines constitutivism and recognitionalism, by grafting the latter onto the former.

(i) That these two ideas – constitutivism and recognitionalism – are distinct and that the former can be held without the latter can become clear from zooming in on French revolutionary language politics. In 1794, almost half a century after the publication of Condillac’s Essay of 1746, the French Convention nationale, the revolutionary assembly governing France, approved a plan by the Abbé Grégoire, to ‘annihilate the patois and to universalize the use of the French Language’ (Grégoire Reference Grégoire, de Certeau, Julia and Revel2002). The goal of the plan was to destroy the patois and the regional languages spoken in France and to universalize the use of French in France. So Grégoire was certainly not a recognitionalist; instead, he defended the very opposite of recognition for patois and local vernaculars: the plan called for widespread linguistic assimilation of all patois and minority language speakers into French.

What makes his plea interesting for our present discussion is the way he justified it. Assimilation into French was not just seen as a matter of efficiency or pragmatism. That assimilation is the normative goal is in fact evident, Grégoire argued, once we realize the very important role of language for cognition. While referring to Condillac, he argues that words are the ‘ties of society’ and the ‘depositaries of all our knowledge’ (Grégoire Reference Grégoire, de Certeau, Julia and Revel2002: 314). When a people gains new understandings, new words will follow: ‘words are the exchange bills [lettres de changements] of understanding’ (304).

But the opposite is also true: understanding is also dependent on words, just like Condillac and Herder also argued. This dependence can have very negative consequences, according to Grégoire. If the language one speaks is imperfect, a great number of errors in reasoning ensue (314). This is exactly what happens to the speakers of the patois in France. The ‘poverty of language’ of a patois ‘constricts the mind’ (304) of its speakers. It is therefore not desirable to render the revolutionary message into the patois. Addressing his fellow revolutionaries, he argues that the poverty of these languages would ‘mutilate your addresses and decrees, if it does not make them intranslatable’ (304). These local languages are full of prejudices and ignorance, which can only breed counter-revolutionary machinations (304). Grégoire’s radical conclusion was to eradicate the patois and to assimilate all patois speakers into French.

This theory of assimilation is built on constitutivist foundations: it is because language structures the mind, that we can, and should, liberate the mind of countrymen by destroying their language and by assimilating them into French. This is remarkable. It would equally have been possible to ground anti-recognitionalism in the belief that languages are cognitively neutral, that Frenchmen and patois speakers just think alike, and that therefore it makes a lot more sense, from an efficiency or deliberative democracy point of view, to converge on one shared language. This is for example the view of Brian Barry, who intervened in the multiculturalism debate with his 2001 book Culture and Equality: An Egalitarian Critique of Multiculturalism. Barry rejects recognitionalist theories, such as those proposed by Charles Taylor, Will Kymlicka, as well as Herder. He grounds his anti-recognitionalism in a version of linguistic instrumentalism: ‘it can be said of language as of no other cultural trait that it is a matter of convention’ (Barry Reference Barry2001: 107). Barry does not fully work out the underlying philosophy of language on which this statement depends. He just holds the view that, since the language we speak is conventional, we just need to agree on which convention we endorse. Since we need to settle societal matters through collective public conversations, we just need to pick one and the same language to communicate our thoughts in, and ‘any language will do as the medium of communication in a society as long as everybody speaks it’ (107).

While Barry vigorously rejects any multicultural endorsement of minority rights and language rights – although he does not go as far as Grégoire in arguing for the annihilation of minority languages – he does not ground his anti-recognitionalism in constitutivism but in fact in its opposite: instrumentalism. Grégoire, by contrast, was not an instrumentalist: he does not think any language will do as a medium of communication or that the specific language we use to communicate and think in is ultimately irrelevant. Cognition depends on language, and languages shape the minds of their speakers. French is a far better medium (though it is not perfect, and we need to reform and improve it), and if we decide, which he thinks we should, to assimilate all patois speakers into French, we will do patois speakers a service: assimilation into French will extend their minds. Grégoire, then, is a constitutive anti-recognitionalist. He argues for non-recognition of the patois and for assimilation into French on the basis of a deeply non-instrumental understanding of the relationship between language and thought.

Condillac himself did not go that far. He never pleaded for assimilation of minorities into French. But he did go in a direction of which Grégoire defends an extreme version. As we saw, Condillac also argued that a language can be in a stage where it does not allow for the full emergence of the genius of certain individuals such as Corneille or Newton.

This idea that a language can constrict the mind relies on a view of language similar to the one outlined by Grégoire: Grégoire argued that the patois constricted the minds of Frenchmen, and that French was better suited for reasoning clearly. Similarly, Condillac argued, half a century earlier, that French used to be not suitable to the development of the genius of Corneille: if Corneille had lived in a different, earlier, age, his mind would have been less progressed.

On this basis, Grégoire draws the extreme normative conclusion that we should eradicate those impoverished languages and, in the process, liberate their speakers. What was Condillac’s normative solution? Condillac did not add a political philosophy, including a language policy plan or a theory of linguistic justice, to his language theory. So we do not know what his position would be on minority language protection. That said, Condillac does have some normative suggestions when it comes to using language. Building on his theory of the dependency of the progress of reason on language, he dedicates the last section of his essay to ‘the means by which error can be avoided’ (Condillac Reference Condillac and Aarsleff2001: 6), and to ‘how to conduct ourselves in the search for truth’ (196). He there argues that the origin of our errors in reasoning lies in the defective reasoning in language. The solution is to, in a Lockean manner, pay careful attention to the way we connect signs to ideas. In common language usage, contingent circumstances have determined which ideas have been categorized under the same sign (200). Therefore, we should sometimes radically go against common usage and reorder the connection between signs and ideas on more transparent grounds. Doing so is not needed for simple ideas, whose signs are often unequivocal. But the rejection of common usage is often necessary for complex ideas (200). The goal should be to have a ‘determinate signification’ for each sign (204) and to make language as precise as it can be (207). And the education of children should, ideally, occur by using precisely defined signs. Never should we confuse children’s minds by ‘entangling children in paralogisms, sophisms, or other forms of bad reasoning’ (83).

(ii) What, then, is the difference with Herder? It is not that Herder never wants to go against usage. As we saw above, Herder thinks there is misuse of language, just like Condillac does. Language frequently has a negative influence on our understanding. We should correct the ‘abuse of language’, which obscures concepts (Herder Reference Herder, Bojda and Gutschmidt2022: 121). When we do so, we will for example see that the antinomies that Kant talks of can be reasoned away, that these result from the bad use of language. If we use language properly, they vanish. So, also on this issue, Herder and Condillac are connected: both hold that the imperfect use of words leads to imperfect thinking, and that we can and should make clear which ideas are implicated by which words.

But Herder does make a move that Condillac never did. He stakes out a recognitionalist position, arguing for the idea that nations and languages should not be run over, assimilated, or colonized by other nations. Ideally, each national language group is to have a separate state:

It is the same thought that makes Herder come out very strongly against colonialism. Each nation is to develop its progress of reason in its own manner, undisturbed by others.

Nations should not impose their culture on other nations but should allow those other nations to find progress by themselves. After all the world is big enough: ‘Does the earth not have space for us all? Does not one land lie peacefully beside the other?’ (Reference Herder and Forster2002: 379). This is how God has wanted it, by installing the diversity of languages as a barrier against simply assembling people, as a ‘dam against foreign inundations’ (ein Damm gegen fremde Überschwemmungen), so as to enable each people to maintain its own national character (2002: 385; XVIII: 236Footnote ⁴). Each culture and language has its own way of being. We should not assimilate other languages and groups; we should instead let them develop human reason in their own way. For Herder, anti-colonialism is the upshot of his nationalism.

In short, we should not impose our values on other cultures because those other cultures should steer their own course. But this argument is only part of Herder’s normative case for recognitionalism. He has a related but still distinct argument for not imposing our language and culture on others: doing so would cause cognitive disturbances. As we saw above, Herder holds that we think in a culturally and historically situated language. Our mental world is dependent on the national language that we speak. This view allows Herder to reach the conclusion that assimilation by another language is problematic because it would disturb our thought pattern. He addresses this issue of assimilation head-on in his imaginary dialogue on the language policy of Joseph II, who made German compulsory in his multilingual empire. Herder rejects this language policy because it does not respect the languages of non-German peoples. In the language of a people ‘lives its entire wealth of thought [Gedankenreichtum] in tradition, history, religion, and basic principles of life, all its heart and soul. To deprive such a people of its language or to degrade the language is to deprive it of its only immortal possession, which passes from parents to children’ (Herder XVII: 58). To repress someone’s language is to rob them of their reason (Herder XVII: 58).

Losing one’s language is therefore a form of cognitive dissonance: one is not in touch any longer with the cognitive framework from within which one thinks and speaks. Indeed, a man who loses his national language (Vaterlandsspräche) loses his national spirit and thereby ‘himself and the world around him’ (Herder XIII: 337). Herder holds that people can and should learn many other languages, but one must always have and keep one’s own national language (336–337). So linguistic assimilation (without maintenance of one’s own language) is cognitive disturbance. To force one nation to assimilate to another is a form of injustice, which we can understand, in part, as a cognitive injustice.

In short, Herder argues not only that a language is constitutive of the cognitive process (which Condillac also held); he adds the normative position that assimilation is a cognitive wrong. And he argues for the latter on the basis of the former: because we think in language-specific ways, forcing another language onto us leads to cognitive disturbance. In this way, then, Herder grafts his pro-recognitionalist position onto his constitutivist theory.

Before concluding, one final issue must be thematized: if Condillac and Herder are both constitutivists but only Herder turned constitutivism into a recognitionalist stance, where does this leave Kant, Herder’s intellectual adversary in the Metacritique? As discussed above, despite Kant’s silence on language in the Critique of Pure Reason, some scholars have pointed out that there are traces of constitutivism in his work. But is Kant also a recognitionalist? Or should he be counted as a Condillacian on this: someone whose views could be, but have not been, combined with either pro- or anti-recognitionalism? One clue to answer this question can be found in the last piece of writing that Kant published himself: his postscript to a German-Lithuanian dictionary (see Shell Reference Schalow and Velkley2014; Mininger Reference Mininger2005). In that very brief – but dense – essay, Kant supports the dictionary initiative. He does so by arguing that the Lithuanian language deserves to be preserved and fostered. This position clearly puts him on the road to recognitionalism. But I do not think he really arrives at a full recognitionalist stance grounded in a Herder-style intrinsic respect for people’s constitutive linguistic embeddedment, for two reasons. First, his argument does not apply to all languages; it is restricted to Lithuanian. And second, while his recognitionalist view applies to Lithuanian, Kant’s motivation is not ultimately based on the importance of allowing the Lithuanian people access to their culturally and linguistically situated cognition. Kant instead grounds his pro-Lithuanian stance in ‘the utility that the state can draw from the assistance of a people of such character’ (translated in Shell Reference Schalow and Velkley2014: 187). The Lithuanian character is described as consisting of courage, pride without arrogance, absence (or lesser presence than in neighboring peoples) of ‘slavish obsequiousness’, a disposition to have equal relationships with others, and ‘open-heartedness’ (Shell Reference Schalow and Velkley2014: 182). In other words, it is first and foremost because the state can derive justified benefits from the civic qualities of such a people – it is good for the state to have such qualities in its inhabitants – that we ought to preserve its language. At the end of his essay, Kant adds a second argument: ‘the advantage that the sciences – especially the ancient history of the migration of peoples—might draw from the still unmixed language of a folk lineage that is ancient and confined to a narrow region and as it were isolated’ (Shell Reference Schalow and Velkley2014: 187). This alone provides a reason for ‘preserving its particularity’ (187). Kant’s argument, then, is ultimately not grounded in the need of Lithuanian speakers for cognitive access or access to their own lifeworld, but in the instrumental value they bring to the state (given the Lithuanians’ civic qualities) and to science. While this is a recognitionalist stance, in contrast to Herder’s recognitionalism, it is not universalizable – since it only holds for a specific people and language – and it is not grounded in constitutivism.