The Liberal Linguistic Turn: Kymlicka’s Freedom Account Revisited
This article revisits the principal argument Will Kymlicka has developed for a marriage between liberalism and multiculturalism: that the liberal value of freedom requires a cultural context of choice. I show that this freedom argument rests on a romantic philosophy of language. Critics of this freedom argument have pointed out that it is not necessarily an individual’s own culture that provides freedom: any culture could do so. I articulate a romantic-Kymlickean response to this critique by showing how individuals’ life choices come to be entwined with the particular culture that provides their context of choice. But while that safeguards existing individuals from assimilation, it does not block future generations from being introduced into the life-world of an additional cultural context. Such slow intergenerational assimilation projects are not necessarily worrisome, however. They can sometimes have the virtue of realizing non-identity values in addition to freedom.
KEY WORDS: Kymlicka, romanticism, Herder, assimilation, group-differentiated rights, multiculturalism
Like many scholars working on questions of ethnocultural justice, I have been, and remain, fascinated by the work of Will Kymlicka. His books Liberalism, Community and Culture and Multicultural Citizenship are to the field of ethnocultural justice what Rawls’s Theory of Justice is to justice in general: they constitute a paradigm that maps the normative issues and provides a comprehensive moral outlook. Scholars interested in these questions cannot ignore that paradigm.
In this paper, for a commemorative volume celebrating the 25th anniversary of Liberalism, Community and Culture, I critically examine the central argument Kymlicka gives for group-differentiated rights: the argument that individual freedom requires a cultural context of choice.
In section I, argue that the freedom argument is a liberal articulation of a romantic understanding of the value of language. In section II, I address an often-phrased objection to the freedom account: that it cannot ground a right to a particular culture. I argue that this objection can be overcome because it is in reference to a culturally specific choice-set that the life choices of already-existing people have been made and continue to be upheld. In section III, I argue that the freedom argument does however allow for legitimate nation-building efforts that seek to add contexts of choice to existing contexts.
THE FREEDOM ARGUMENT
Many liberals, especially after World War II, have denied outright the legitimacy of minority rights or other forms of group-based recognition. But Will Kymlicka has developed a distinctively liberal defence of minority rights. He argues that freedom, of key importance to liberalism, is always embedded in a cultural context of choice. In order to freely decide what is valuable in life, people need options to choose among. Yet “[i]n deciding how to lead our lives, we do not start de novo” (Kymlicka 1989: 164). We reflect upon the values, beliefs and models offered to us within our culture and language, and then choose to defend or reject particular values from that given choice-set. Culture and language thus function as a context of choice (1989: 164). Our cultural structure makes us aware of the options available to us, so that we can examine them and select the ones we find valuable. Language and culture are the “media through which we come to an awareness of the options available to us, and their significance; and this is a precondition of making intelligent judgments about how to lead our lives” (1989: 165). Without our cultural context of choice, we could not meaningfully exercise our capacity for making autonomous life choices. “Put simply, freedom involves making choices amongst various options, and our societal culture not only provides these options, but also makes them meaningful to us” (1995: 83).
Therefore, if liberals want to realize individual freedom and autonomy, they need to endorse rather than ignore the value of languages and cultures. They ought to recognize that membership of a culture is a primary good in the Rawlsian sense (1989: 166). Rawls assumed that the political community is culturally unified. But in cases where this is not the case, such as in multinational states which host more than one national-cultural group, group-differentiated rights are warranted, to enable the groups to provide their members with a context of choice. In short, the freedom argument serves as the engine for Kymlicka’s defence of group-differentiated rights. Such rights enable individual freedom.
Kymlicka does not historicize this freedom argument; he works it out relying on the work of contemporary liberals like Ronald Dworkin (see e.g. 1995: 82-84). Yet, the freedom argument has a significant historical pedigree. In particular, it was at the heart of a tradition sometimes referred to as the Hamann-Herder-Humboldt view, which also pervades the work of theorists like Hans-Georg Gadamer and Charles Taylor (Lafont 1994: 13; Taylor 2016: 48). I will call this tradition in what follows the ‘Romantic Tradition’. In the romantic tradition’s philosophy of language, language has a world-disclosing function. Language opens up a world of language- and culture-specific meanings. Language is not simply a tool by which already fully formed individuals can communicate with others. Language in fact constitutes people by cognitively influencing them: it is through language that people get the information they need to develop their own beliefs. Our native language, says Herder, is the first world that we see (Herder 1987: 336). The world is not present to everyone in the same way. Instead, languages give speakers access to a situated view of the world. As Gadamer argued, to have a world we need to have a language (Gadamer 1975: 411).
In this view, language knowledge functions like a key to a room: one needs to speak the language to access what is discussed in the room. Once inside the room that the key gives access to, one is surrounded by arguments and styles of discussing that are not readily available to people who don’t speak the language unless through translators who hold the key to other rooms. Each room is called a “life-world”: a set of shared assumptions and ideas about the world. It is this life-world that is disclosed through language. For Gadamer, and for Herder and Taylor alike, our experience of the world unfolds from within our language (Gadamer 1975: 145).
It follows that if the interest of individuals in having access to their language and cultural life-world is not respected, they cannot fully realize themselves. Herder has argued that if we lose the disposition to think in the language in which we are brought up, we lose ourselves, and also the world (Herder XVIII: 336-337). From this, theorists like Herder and Taylor also drew the conclusion that the world-disclosing function of language generates a legitimate justification for policies that seek to explicitly recognize and protect languages (see e.g. Herder XVII: 59; Taylor 1993: 46-47; 53-54). People need access to their language and cultural tradition in order for them to function well, to receive a (first) position.
Kymlicka’s liberal multiculturalist account of the value of language and cultural membership rests on this romantic view of the life-world. For Herder, language and culture function as a psychological matrix in which we are raised as social beings (Herder V: 135; Barnard 1965: 57). This is similar to describing culture (as Kymlicka does relying on Dworkin) as the ‘spectacles’ through which we experience the world (Kymlicka 1995: 54). The phrase that culture ‘provides options’, and ‘makes them meaningful to us’ (Kymlicka 1995: 83) might as well have been written by Herder. Both accounts take it that individuals need their language to be able to be full human beings, because individuals do not start and reflect upon the world from a view of nowhere, but require cultural and linguistic ‘spectacles’ through which they see their world.
So Herder (and his followers) has argued that individuals need to have access to their linguistic and national-cultural life-world, which provides them with a “horizon of meaning” that they need in order to fully realize themselves (Taylor 1993: 46-47). Kymlicka (and his followers) have argued that individuals need access to their linguistic and national-cultural context of choice, which provides them with shared options from which they can make autonomous and free life choices. Both hold that language and culture help structure what we think and choose in life and that this linguistic and cultural structuring justifies policies that seek to explicitly accommodate linguistic and cultural groups.
Yet, while the freedom argument is grounded in romantic life-world premises, Kymlicka’s version of multiculturalism is not a replica of the romantic tradition. Kymlicka articulates an explicitly liberal version of the romantic ‘linguistic turn’. As a liberal, Kymlicka integrates the romantic conception of the value of language and culture into an explicitly liberal framework. This liberalization changes the argument in two ways. First, it means that the freedom argument is articulated with liberal premises: language and cultural membership are presented as necessary for the liberal value of freedom. Since language and culture structure our life options, and since liberals care about enabling individual choices, liberals have a reason to give people access to language and culture. Second, liberalizing the argument implies imposing liberal limits on what can be tolerated in the name of securing access to linguistic and cultural membership. For example, ‘internal restrictions’, by which a group attempts to silence individuals who seek to change the character of culture, are illegitimate from a liberal point of view (Kymlicka 1995: 104-105). They are illegitimate because they are freedom-destroying instead of freedom-enhancing. The resulting liberal nationalism can be understood as a nationalism that has been stripped off anything illiberal. Nationalistic politics are not aggressive against other national identities and typically expose a ‘thin’ or open conception of national identity. So Kymlicka’s version of the freedom argument can be characterized as a liberal romanticism: it gives a distinctively liberal shape to the argument, and it sanitizes romanticism from its illiberal off-shoots by formulating liberal limits to what can be demanded in the name of multiculturalism. It would therefore be wrong to see Kymlicka as a romantic theorist pur sang. But Kymlicka and Herder (and the Hamann-Herder-Humboldt tradition more generally) are united in the account given of the value of cultural membership. This account serves for both as the ground for the political concern for cultural groups. It holds that individuals are not fully formed beings prior to having language and culture: individuals need their situated horizon. Language is not simply a tool for communication; it influences our cognitive constitution and thereby structures our life options. And individuals have a legitimate interest in the protection of their linguistic and cultural life-world. Kymlicka takes this romantic account, liberalizes it, and put it at heart of a comprehensive political theory of multiculturalism.
FREEDOM IN A PARTICULAR LANGUAGE
The romantic tradition is radically different from another tradition of linguistic justice: the French revolutionary take on language policy. Understanding the difference between both traditions is helpful to understand, and rebut, an objection levelled against Kymlicka (which I deal with below) about the reason why particular languages and cultures are to be respected. Revolutionary ideologists like Barère, Grégoire, Duhamel, Talleyrand, and others were concerned with language, and with the role language plays in the thought process. They were convinced that the success of the revolution depended in part on its ability to eradicate vague language and to purify the language from confusions and sources of misunderstanding. Their solution was to purify French and to spread it all over France. This purification entailed ‘revolutionizing’ French by weeding out synonyms and confusing manners of speech, and by producing an unambiguous spelling, grammar and dictionary. And its spreading, they thought, necessitated the eradication of linguistic diversity, especially resulting from diversity of patois and of foreign languages on French territory. The title of l’Abbé Grégoire’s report presented to and accepted by the Assemblée nationale in 1794 – Rapport sur la nécessité et les moyens d’anéantir les patois’ – illustrates this commitment to remove the patois. Because language and thought are connected, linguistic diversity within a state implies to the existence of groups with distinct political and other ideas. The patois made 30 peoples out of one people, argued l’Abbé Grégoire (2002: 337 ). This has negative consequences. It leads for example to a lack of cohesion and stability. And it is incompatible with the revolutionary ideal of equality because social equality implies linguistic homogeneity: people must be confronted with the same ideas and life opportunities. Instead France ought to be linguistically unified.
So these revolutionary theorists advocated a policy of unilingualism on the basis of the idea that knowledge is influenced by language. They wanted to liberate patois speakers from the yoke of their locality by destroying the patois and by assimilating its speakers into French. They posited that the ideas of patois speakers and speakers of foreign languages were intertwined with their local linguistic background (see e.g. Grégoire 2002: 336, 348), and they assumed the choice-structuring function of language. But they did not believe that this function was a reason for protecting local languages. Instead, they saw it as a reason to engage in the very opposite of protecting local languages: to eliminate those languages and to liberate speakers from their narrow background. They assumed that language structures choice, but they did not hold that language is therefore protection-worthy. It is with regard to this second element that the romantic tradition strongly diverges from the French-revolutionary outlook. Romantic theorists like Herder, and many contemporary multiculturalists alike, premise their argument for language recognition on the argument that people’s choice sets are bound up with their native language and culture. French revolutionaries shared that premise but instead argued for language policies that disconnect people from their life-world so as to assimilate them into another. They did so for non-identity reasons such as equality, state cohesion or the political success of revolutionary ideas. In doing so, they are similar to contemporary liberals like Brian Barry (2001), Thomas Pogge (2003) or Daniel Weinstock (2003), who argue for intervening in cultural affairs in such a way that non-identity values such as socioeconomic equality (Barry and Pogge) or democracy (Barry and Weinstock) are promoted, which for them often means promoting the state’s majority language.
This contrast between the romantic and the French revolutionary traditions of linguistic justice lies at the heart of an objection addressed at Kymlicka’s multiculturalism. One of the most frequently levelled objections to the freedom argument – the argument that one needs a cultural-linguistic context to be presented with options to choose from in life – is that it cannot ground a right to a specific culture. The freedom argument may establish the need to ensure access to a cultural-linguistic context. But how does it justify protecting one’s own language and culture, rather than assimilating people into the life-world of another culture? For example, Margalit and Halbertal (1994) have argued that Kymlicka’s understanding of culture as a precondition of individual freedom can only justify a right to culture, not a right to a particular culture. They argue that if a minority culture would be destroyed by the presence of a larger majority culture and if the members of the culture have the opportunity to assimilate (against their will), Kymlicka cannot object since the new culture can equally function as a context for individual choice (for this critique, see also Forst 1997: 66; Galeotti 2002: 208; van Leeuwen 2006). If this objection is correct, then the argument for culture is consistent with two possible outcomes: the romantic/Kymlickean outcome to protect people’s own cultures and the French revolutionary outcome to pursue assimilation into a language and culture of wider opportunities and of democracy. So the argument can cash out both as multiculturalism and as non-multiculturalism. It is therefore unclear what role the freedom argument plays in grounding multiculturalism.
Can the argument for cultural recognition based on the romantic/Kymlickean freedom argument answer this objection? I see two strategies for it to overcome the assimilation critique. The first strategy is to supplement the argument with another culture-determining argument. For example, Kymlicka has argued in a footnote in Politics in the Vernacular that identity considerations provide a basis for specifying which culture will provide the relevant context for the context of choice (or autonomy) argument:
I admit that my argument here was unclear, but what I meant to argue was that considerations of identity provide a way of concretizing our autonomy-based interest in culture. In principle, either the minority's own culture or the dominant culture could satisfy people's autonomy-interest in culture, but considerations of identity provide powerful reasons for tying people's autonomy-interest to their own culture. Identity does not displace autonomy as a defence of cultural rights, but rather provides a basis for specifying which culture will provide the context for autonomy (Kymlicka 2001: 55, n. 7).
In this view, the freedom/autonomy interest can be satisfied within any culture: minority members could realize it in the minority culture and as well as in the majority culture. But the minority members bring an extra interest to the table to show why the freedom interest is to be realized in the minority language: identity interest in the minority culture.
This is also the approach taken by Chaim Gans (2003), who takes the two as independent arguments. In Gans’s approach, people have a fundamental interest in freedom, as well as in being respected for, and not humiliated by, their identity (2003: 43). By the culture of their identity, Gans and Kymlicka mean the particular culture that people want to adhere to: it is the culture of their preference (2003: 43). So both Gans and Kymlicka refer to two arguments for group-differentiated rights: freedom and identity.
This dual approach successfully avoids the Margalit and Halbertal critique. But one problem with it is the normative status of the identity argument. It is not immediately clear why the fact that an individual or group has an identity in X is a reason for X to be required by justice. This reason may possibly be provided, but the overall argument is then no longer as intimately tied to liberalism itself. The argumentative virtue of the freedom argument is its intrinsic connection to liberalism’s core value of freedom. This allowed Kymlicka to marry liberalism with multiculturalism, by connecting the liberal emphasis on choice with culture as a context of choice: if you are a liberal, you have to care about cultures. It is however not as immediately clear how satisfying identity interest is tied to the liberal project. It is compatible with liberalism to do so but does not seem required by it. We need an additional argument to say why identity is important. By appealing to a value (identity) that is not as directly connected to liberalism, it leaves the path open for a liberal to take the freedom but not the identity interest on board. So the success of the argument now hinges on the strength of that additional argument as to why identity matters. And it is not immediately evident that that argument can be a liberal one.
Moreover, the fact that the freedom argument cannot do the work in itself in this view opens up the argument to other criticisms. For there are other auxiliary arguments that are equally compatible with liberalism and also provide a basis for determining in which culture to realize the freedom interest. The French revolutionary arguments are examples. If, as Grégoire, Barry and Pogge have it, socioeconomic equality is to provide that basis, then the right language policy towards minorities may entail doing what we can to make them realize their freedom interest in the majority language, which gives speakers better socioeconomic opportunities. Another auxiliary argument is democracy: since democracy may function better in a unilingual environment, we might seek to ensure that French speakers in Canada realize their freedom interest in English so as to ensure a smooth Canada-wide English public sphere.
In short, taking the freedom interest to ground the right to ‘a’ culture and the identity interest to ground the right to a ‘specific’ culture introduces theoretical hybridity into the argument. That hybridity decreases the argument’s liberal credentials, and, especially, it makes it a straightforward move to counterbalance the identity argument with non-identity arguments such as stability, equality or democracy that may point towards the majority language and culture as the best location determiner for the satisfaction of the freedom interest.
The second strategy, which I prefer and will now lay out, resists the move to understand the freedom interest as inherently lacking a location determiner. It argues that the freedom interest cannot be satisfied within any culture but instead requires the individual’s own culture.
In the romantic tradition, language and culture open up a life-world. They do so by situating an individual and presenting her with particular choices to choose from. Inside the room that language is a key to, one is presented with particular values, beliefs and options. For example, growing up in a language and culture, one is presented with particular traditions in literature or a specific history focus in schools. One is also introduced to particular metaphors, political catch phrases, historical narratives that help structure one’s life choices and opinions.
As a result, individuals who have grown up in a specific language and culture have selected options from that particular life-world and developed life options that are meaningful within the context of that culture. Once they have begun to make their own life choices, the intelligibility and the exercise of these choices require the continued availability of the cultural-linguistic choice-set. The person they have become depends on the particular choice set offered by their own culture, and requires that culture for their choices and personality to continue to make sense. So the life choices, desires, values and beliefs of individuals are entwined with the particular language(s) one speaks.
Therefore, to impose a different language and culture on people who are already situated in a cultural context, through linguistic and cultural assimilation, would imperil their freedom. Not only would they face linguistic difficulties to understand the new values. They would also be cut off from access to options that are meaningful to them and from the life-world in which their life choices make sense. This would result in ‘loss of world’, in alienation. They might rebuild it in some way, by reintroducing parts of their values in the new cultural group, by translating metaphors and by developing a distinctive minority accent in the new language. But it would be a form of patching up, and the loss of freedom is beyond complete repair. Once we take the idea seriously that our culture structures our choices, the freedom interest not only grounds the idea that individuals need a language, but also that they need their own language, which is the language of their parents and of their fellow speakers and cultural members.
That a life-world can be destroyed through linguistic assimilation does not mean that life-worlds cannot integrate new material. Cultural mediators and translators bring along new stories, concepts and ideas. There are also new phenomena, such as technologies or natural events. They too are added to the language, and embedded in the linguistic and cultural horizon. This adding of external or new ideas to a language and culture, however, does not call into question the ‘framework idea’ of language and culture – the fact that linguistic and cultural options come in the form of a comprehensive package.
I hope to have made it credible that, in my version of it, the freedom argument does the work on its own, and does not require an auxiliary argument. It has therefore parsimonious virtue compared with the bifurcated approach. It contains both the ‘what’ and the ‘where’: it says what culture provides to the individual (access to a situated context of choice) and has an in-built location determiner (the culture that provides the content is the one in which the individual has grown up and in connection with which she has developed and connected her life choices). French-style liberal assimilationism is thereby ruled out: the ‘romantic’ nature of the freedom argument grants people a right to their own language. Moreover, the fact that the argument for the value of culture is based purely on the freedom argument retains a tighter fit within liberalism’s core principles, and allows for the conclusion that group-differentiated rights are required by liberalism rather than being merely compatible with it.
At this point we must consider an objection to this self-standing interpretation of the freedom argument. One might question if the ‘romantic’ interpretation of the freedom account fully escapes the assimilation critique. The self-standing version only works, one might object, for currently living people. It does not apply to future people who have not yet connected their life projects with a particular context of choice. This objection can come in two versions.
The first version states that the freedom interest of new-borns could be exercised in any culture. New-borns and future unborn generations don’t have a particular context of choice yet, so they might be inculcated with any culture and language. My argument above that the freedom interest can be self-standing depends on the fact that people’s past expectations and life choices were made in reference to an already existing choice-set. But that choice-set is not there yet for future people.
At the same time, however, new-borns are dependent children who need parents. Parents do have a language and life-world, and they have a responsibility to help their children navigate their way around in their particular society or culture until they are grown up and are able to autonomously steer their life course. It is hard to see how parents could do so without transferring knowledge of a particular set of expectations, stories and metaphors. This enables the children to speak a language fluently and have a ‘mother-tongue’, to grow up with the parents’ stories, with the childhood songs that the parents know, with the often implicit values and expectations that exist within a life-world, and so on. The children have both emotional and cognitive needs that the parents must provide, and, other things being equal, the best location for the children’s freedom interest is their parents’ culture. As the case of adoption shows, a child can grow up in any language; but it is important that they know the language of the people they grow up with. For new-borns the freedom interest is realizable in any culture (as opposed to that of people who have already grown up in a specific culture) but it is the parental bond that provides the reason for locating its realization within the specific culture that the parents are raised in.
Yet, apart from new-borns, contexts of choice can and do change: the contexts of choice of current generations are not identical to that of their grandparents. The second version of the objection states that the language and culture of future generations can become structurally unrecognizable to us due to ongoing intergenerational cultural change. So the assimilation objection still retains some force with regard to the future shape of a culture’s structure. Is it not legitimate for the group’s cultural structure to change, or even entirely disappear?
To answer this we must follow the logic of the freedom argument as the theoretical engine of the liberal case for group-differentiated rights. The freedom argument holds that individuals must be given a context of choice from within which they can choose their own life options. This context of choice can change. And that change can occur both character-wise, as far as its values and beliefs are concerned, and structure-wise, as far as the cultural structure or the existence itself of the culture is concerned (Kymlicka 1989: 167). Character-wise, as far as the content and the values of the cultures are concerned, a culture could evolve from Christian to secular or from agrarian to post-industrial. As Kymlicka argues, it “is right and proper that the character of a culture change as a result of the choices of its members” (1995: 104).
But it could also legitimately change its structure (a possibility not explicitly discussed by Kymlicka, see 1995: 104-105). It is possible that, in the course of a few centuries, language A changes to such an extent that it could no longer be understood by the earlier speakers of A, while the freedom interest of its speakers is never harmed. 15th century Dutch and contemporary Dutch are not mutually intelligible, due to the linguistic evolution Dutch has undergone. From the point of view of the freedom argument, nothing is problematic about such a structural change, as long as the change occurs gradually and does not leave living speakers of the previous structural stage behind. It is also possible for a part of a language or cultural group to develop a distinct language and culture of its own, thereby radically changing the boundaries of the existing cultural structure, for example by developing a dialect of language into a self-standing language.
So the second articulation of the objection is correct. The self-standing interpretation of the freedom argument only applies to currently living people: it cannot protect a culture from changing quite radically over the course of time. There is nothing wrong with changes in the cultural values or structure. The freedom argument does not rule out such changes.
I have just given an example of a case in which the cultural character or structure changes in such a way that a significant difference exists between the culture at different points in time. It is also possible for the life-world to remain similar but for the people to shift between two different life-worlds. Indeed, over a few generations, the life-world could for example move from language A to language B, while at each single instance in time fully providing a context of choice to individuals. As Alan Patten has argued:
Imagine that language L did gradually decline in use—to the extent that it fell below the threshold in which it offers a context of choice. It would be a mistake to conclude from this fact alone that L-speakers would be left without a context of choice. It would only be unilingual L-speakers who would necessarily have lost their context of choice since multilingual L-speakers may find meaningful options and opportunities in other languages. And the very same processes that generated the decline in usage of L in the first place—the massive attraction of some other language, for instance—would help to ensure that there are very few unilingual L-speakers (Patten 2001: 707-708).
Provided the change indeed occurs at a slow enough rate, freedom-enabling contexts of choice can adjust in such a way that no-one experiences significant lacks of freedom. For example, in areas where two languages exist of which one is more dominant than the other, often the children born to parents with different first languages are bilingual, but the grandchildren already lean towards the dominant language. A language shift can then occur relatively smoothly, with at no point leaving individuals stranded without a context of choice. Van Parijs has summarized this ‘Laponce law’ of unilingualization after a few generations as: “The nicer people are with one another, the nastier languages are with each other” (Van Parijs 2000: 219; Laponce 2001: 188-189). This law partly explains the radical transformation of Brussels from predominantly Dutch at the end of the 19th century to predominantly French a century later.
In short, people’s cultures and languages can change in two ways. Either the culture or language itself evolves, as in the differences between 15th century Dutch and contemporary Dutch. Or people could slowly change their structure and values from one to another culture or language, as in Brussels many ancestors of contemporary speakers of French (often with Dutch surnames) have done. And it is possible for each generation to have a fully comprehensive context of choice, either at some point within the internal evolution of a language and culture, or at some point on the continuum between unilingualism in language A and unilingualism in language B. So the freedom argument is compatible with such changes, provided they don’t occur overnight.
That the freedom argument is compatible with such changes does not necessarily mean they ought to be encouraged; we might seek to block them. Yet such changes ought not always to be seen as regrettable. As we saw before (and as Kymlicka) acknowledges, languages and cultures not only provide a context of freedom but also serve non-identity purposes such as equality and democracy. We may sometimes judge that the non-identity reasons for sharing a particular context of choice are so compelling that it is desirable to seek to adapt the structure of the culture in such a way that non-identity purposes are served in addition to freedom interests. This will often clash with ethnocultural justice, since it may lead to the ‘revolutionary’ linguistic assimilation strategy that the freedom argument is supposed to withstand. Yet in some cases I think we should allow for such a conclusion in a way that doesn’t negate the freedom argument. I will give two examples of such cases: that of intralinguistic differences within a language in the form of regional dialects, and that of building a European identity.
What does the freedom argument say in the context of dialects? In a typical linguistic situation marked by a standard version of the language (such as standard German) that is functionally superposed over different regional dialects (such as Bavarian), the linguistic context of choice will be partly provided for by the standard version. The standard version is usually the language of the public sphere, of media, of political life, of universities, and of schools. As a result, individuals partly satisfy the freedom interest through the standard language.
Of course, there are historical reasons for why the freedom interest is already partly realized through the standard language: it is the result of the success of linguistic nation-building. The modern nation-state has over the course of its history attempted to unite the citizenry and to create a common national language. The current result of this history of nation-building and standardization is, in the usual case, the existence of a ‘diglossic’ context of choice: while for certain freedom-related functions such as family relationships the dialect provides a context of choice, for others like political engagement or for written sources, the standard is the normal choice (Ferguson 1959). It is common to have two such choice contexts at once, and even for one (the standard) to be clearly dominant in terms of status, as it is the standard version that receives the lion share of state recognition (in public schools, in the laws, in the courts, and so on).
There clearly are non-identity benefits to be had from having a standard language and from spreading its knowledge among the population. These include efficiency, democracy (which is easier when there is a shared language version that is mutually understood) as well as equality (as a larger job market, and better socioeconomic mobility, are enabled this way).
Yet, in some cases dialects may be strong, big or widespread enough to be able to in principle sustain a full context of choice in the dialect alone. Bavarian, a group of German dialects, is an example, and its 14 million speakers make it a more widely spoken language than for example Danish or Greek, which each provide full choice contexts. In such cases, the freedom argument is compatible with lending full political support to the dialect. But it is also perfectly compatible with seeking to maintain the current diglossic situation with the standard version having most of the status and state recognition. So the choice for dialectal dominance or standard dominance, then, cannot be based on the freedom argument alone. That choice can also be informed by non-identity arguments. This dialect/standard discussion shows, I think, that contexts of choice can be layered in the sense that individuals have freedom interests in two language versions: both in the standard and in the dialect. And it makes sense to seek to avoid only having such interests in the local dialect, to maintain the standard version of the language, and to spread its knowledge among the speakers of the dialect.
My second example of a case where the malleability of the freedom interest should not be seen as a problem is the attempt in Europe to build a European identity in addition to existing national identities. Kymlicka’s view is that we should not make EU institutions directly accountable. Indirect accountability is more appropriate. The result is a form of representation with the national representative as a medium between the people of the member state and the European decision-making body. This is the intergovernmental view of the EU: the EU is steered by the nation-states, such as in the European Council. It stands in contrast to the supra-national view of the EU, in which a direct relationship between individuals and EU institutions and representatives is fostered, such as by granting more powers to the directly elected European Parliament rather than to intergovernmental organs like the Council. In Kymlicka’s view, citizens debate at the national level how they want their national governments and representatives to act in intergovernmental contexts (2001: 317-326). Freedom, equality and democracy are best exercised within national-cultural-linguistic units, such as within the Danish, Catalan or Flemish units, and not directly at the European level.
While I fully agree that national-cultural choice contexts are to be protected and are entitled to political autonomy, I do not share this normative disagreement with European supra-nationalism. The national-cultural groups all came into existence through active nation-building efforts. Kymlicka also defends nation-building policies for currently existing nations, provided the policies are liberal. But he is not prepared to pursue nation-building or identity-building at levels above the nation. It is not clear to me, however, why existing nations have a privileged position here. If recent processes of globalization and Europeanization are creating a new civil society in Europe (Kymlicka 2001: 326), and also to some extent new identity structures, there is no reason within the liberal nationalist preoccupation with the importance of identities to inhibit the nascent emergence of a European supranational ‘life-world’ (enabled by for example a European culture, Europe-wide cross-national mobility, or shared European symbols) fully compatible with the continued importance national life-worlds. Yet, even if this choice context is still very tentative or fragile, there may be non-identity reasons unrelated to the freedom argument – though compatible with it – to stimulate the development of such a supranational cultural layer. Within the European context one might argue, for instance, that a strong European identity might help foster the development of a generous redistribution from richer to poorer regions, or even simply to uphold existing schemes of (limited) intra-EU solidarity as exemplified in the European scheme to uphold a plan to share the burden of the influx of asylum seekers in Europe, or support for bailing out Greece, or even simply to help sustain the solidarity between net contributors to the EU like the Netherlands or Sweden and net beneficiaries like Poland or Hungary (Cipriani 2014: 14-15).
The answer to the question whether we should stimulate an EU identity layer depends on what one thinks the EU is to do, and whether EU-wide redistribution is desirable. My (unargued for) normative premise is the desirability of extending the project of distributive justice beyond the nation-state, and I support the emergence of an EU layer of distributive justice that redistributes wealth between richer and poorer nations and individuals within the EU (while EU-distribution is in my view itself only a stepping-stone towards global distributive justice). In this view, we should support a supranational view of the EU. Other views about the normative goal of the EU may indeed lead to less ambitious takes on the need for EU identity. However, a full answer to whether EU redistribution is desirable requires a more worked out normative theory of the EU, and of distributive justice, which I have not provided here. But the point is that this choice between fostering or withstanding EU identity-building cannot be made on the basis of the freedom argument alone. That argument allows for both an intergovernmental and a supranational conclusion, for keeping identity local and for fostering an EU-wide identity. It is ultimately the non-identity arguments that normatively point to the desirable future location of the context of choice.
These two examples of dialect/standard choice contexts and the desirability of an additional European identity layer show, in my view, that it is compatible with the freedom argument, and sometimes desirable, to steer future generations’ freedom contexts in a certain direction. The freedom argument does not need a supplementary argument to block assimilation into another culture since existing people’s choice contexts are entangled with the particular culture they adhere to. But this is not true for future generations’ projects, so we do need to involve additional arguments for determining the desirable future location of the choice context. Two prominent such supplementary arguments are equality and democracy. Of course, these supplementary arguments may uniquely support the current context location. But in many cases the supplementary arguments will point in a different direction. In such cases difficult trade-off assessments must be made. The dialect/standard and the EU examples are in my view two cases where new choice contexts have been (in the case of the standard language) and can be (in the EU) stimulated, without destroying or diminishing existing choice contexts. In both cases I think the attempts are desirable.
In conclusion, the freedom argument in the self-standing version I have defended can resist the linguistic assimilation conclusion for the current generation and also for new-borns. Current generations perceive their options and have devised their life plans in reference to a specific cultural context – that in which they were brought up. Assimilating them into another culture obfuscates their options, and makes their life plans less meaningful. And it is in the interest of new-borns to be raised in the language and culture of current generations. But while the freedom interest protects the language and culture of current and near-future generations, it cannot, on its own, withstand a very long-term project of adding a cultural context. If there are good non-identity reasons for doing so, and provided such long-term changes occur slowly enough and don’t imperil currently existing choice contexts, they can be legitimate.
We can see Kymlicka’s project as having married liberalism with the romantic idea that language and culture open up a distinct life-world. Individuals who grow up in it have a legitimate interest in state recognition for their language and culture, as a means of preserving their context of choice. We do not need additional arguments to safeguard the realization of this freedom interest within individual’s own cultures as opposed to any culture, since their chosen choice set is dependent on and attuned to their own culture. Yet in some cases we may seek to add new choice contexts if there are good non-identity reasons for doing so. Provided that existing choice contexts continue to be protected, and the non-identity arguments are compelling, we ought to sometimes do so, and the liberal linguistic turn accomplished by Kymlicka allows for it.
 Moreover, for liberal nationalists like Kymlicka and for Herder alike, we are not just embedded in cultures, but in national cultures. Kymlicka uses culture as synonymous with ‘nation’ or ‘people’ (1995: 18). Herder also uses ‘people’ (Volk) and ‘nation’ interchangeably. He understands the nation as a cultural entity with a shared language as its essential feature (Herder XIII: 257-258; 363-365).
 Liberal nationalisms often acknowledge their indebtedness to Herder. Yael Tamir (1993) has argued that liberal nationalism “is a direct descendant of the cultural pluralism of Herder” (1993: 79). And Taylor, who is characterized as a liberal nationalist by Kymlicka (2001: 210) mentions Herder as one of his most important intellectual fathers (1993: 135-139).
 Herder is not generally seen as a central theorist of liberalism. At the same time, Herder upholds the importance of the rights and interests of individuals. Herders’ nationalism can also be called liberal in the sense of moderate. It is not based on racial or ethnic criteria but on linguistic and cultural ones. And both Herder’s and Kymlicka’s versions of nationalism grant group-differentiated rights such as self-government rights to national minorities, instead of seeking to assimilate minorities within a statewide nationalism. It is a common criticism that romantic political theories tend to limit the moral community to the national community and thus ignore the moral duties they have to individuals of other nations (see Larmore 1996: 54-56). But this characterization does not apply to Herder, who systematically grounds his nationalism in the universal moral equality of individuals and their shared Humanität (e.g. XIII: 346). And Herder clearly spoke against state policies that seek to crush nations within it. He instead defends the idea that each people with a shared language and national character should form a distinct state (XIII: 384-385). In fact, Herder’s nationalism is essentially anti-imperialist in nature – his passionate anti-colonialism was grounded in the idea that nations are to be self-determining. Note that this is close to the justification offered by Kymlicka for a moderate nationalism. Compare for instance Herder’s idea of a dam against being engulfed by others (ein Damm gegen fremde Überschwemmungen), with Kymlicka’s statement that “[i]t is one thing to learn from the larger world; it is another thing to be swamped by it, and self-government rights may be needed for smaller nations to control the direction and rate of change” (Herder XVIII: 236; Kymlicka 1995: 104). More generally, Kymlicka legitimates his normative proposal to grant national cultures distinct political units of their own within multinational states as external protections against the impact of bigger or more powerful nations within the same state. They allow the group “to protect its distinct existence and identity by limiting the impact of the decisions of the larger society” (Kymlicka 1995: 36). But despite the fact that Herder can be called liberal in the sense of ‘moderate’, he did not work out a defence of culture from within the liberal tradition, nor did he formulate liberal limits to what is entitled in the name of culture. Instead, in Kymlicka’s work, liberalism functions both as an engine and as a filter for nationalist demands, and both dimensions are not present in Herder. For an account of Herder’s liberal credentials, see Patten 2010.
 To be sure, they are not advocating the removal of minority languages, as Grégoire and Barère did.
 Of course multilingual people can frequent other rooms, and bring back newly acquired ideas and describe them to the fellow speakers of the language.
 Against Kymlicka’s argument, Jeremy Waldron has argued that even though options are culturally mediated, “it does not follow that there must be one cultural framework in which each available option is assigned a meaning. Meaningful options may come to us as items or fragments from a variety of cultural sources” (Waldron 1992: 783). I think it is important to recognize the possibility of having more than one linguistic-cultural framework. Indeed, in multilingual areas like in Catalonia, Brussels, or Singapore, it is far from uncommon to find people who have two or more linguistic and national ‘contexts of choice’. Kymlicka has not explicitly thematised this possibility in his theory (see for this criticism De Schutter 2011). But admitting the possibility of having two linguistic-cultural life-worlds does not damage the freedom argument in itself. From the fact that individuals may have two linguistic-cultural frameworks, it does not follow that that the idea of a linguistic-cultural framework does not make sense. A bilingual Catalan-Spanish citizen will integrate foreign elements like originally Indian yoga and Korean-produced smartphones, and both the Catalan and Spanish frameworks can integrate such elements within their web of beliefs and existing options, bestowing meaning upon it. International politics will be reported in the newspaper in Spanish and in Catalan, and both languages embed the content of the news in their own framework of metaphors and linguistic particularities. It is also possible for certain elements to be integrated more by one than by the other linguistic horizon. And when this bilingual citizen reads a Scandinavian novel, she will do so in a version translated in one of the languages she masters. As Kymlicka argues (1995: 85), that an Irish-American eats Chinese foods and reads her children Grimm’s Fairy-Tales “is not moving between societal cultures. Rather it is enjoying the opportunities provided by the diverse societal culture which characterizes the Anglophone society of the United States.” The fact that it is possible to be a full member of two societal cultures does not call Kymlicka’s freedom argument into question: it just means that freedom can be realized in two cultures. So, Waldron is right to say that meaningful options come to us from a variety of cultural sources. And there is indeed no need to assume that one can only have one life-world. But these are all compatible with holding the the importance of there being culturally and linguistically mediated life-worlds that do structure options. There must not be one cultural framework, but each framework one is a member of structures options in a particular way.
 While European identity surely is significantly weaker than national identities, it does nonetheless exist. It even exists in a way that resembles national identity positions in multinational states: many citizens have dual identities, others have only national identities and still others uniquely conceive of themselves as European citizens. 8% of Europeans self-define as ‘European only’ or ‘European and national; 38% self-identifies as ‘national only’, and 52% see themselves as ‘national and European’ (Eurobarometer Spring 2015). It is therefore not the case that the EU suffers from a problem that the member states by definition don’t have. Several EU member states are multinational states, such as Belgium, Spain or the United Kingdom: they contain more than one nation and have sought to politically accommodate such national identities in envisioning their own unity. These multinational states themselves can be understood to have a state-wide demos alongside sub-state national demoi. Citizens may be members of more than one national community at once (such as of the Spanish and of the Catalan nation simultaneously), and different individuals of the same ‘nation’ may be internally divided with regard to which context is the most relevant one (38.7% of Flemish respondents in a 2014 survey answered that they self-identify as much with Flanders as with Belgium; 31.1% self-identified only as Flemish or more as Flemish than Belgian; 29.8% only as Belgian or more as Belgian than Flemish, see Swyngedouw et al. 2015). The practices of such multinational states and the nested identities they make possible offer an example of the type of practice we can and in my view ought to pursue at the level of the European Union.