IDENTITY AND CONSTITUTIVE ATTACHMENTS

Will Kymlicka was arguably the first contemporary liberal theorist to defend the right to cultural–national preservation on the basis of an assumption concerning the relation between identity formation and cultural–national communities, and his work has had a significant influence on debates concerning liberal nationalism. Kymlicka's assumptions concerning identity formation are problematic for two primary reasons. First, Kymlicka's assumption concerning the nature of so-called “constitutive” attachments supports a counterintuitive and unappealing conception of the liberal autonomous self, one that plays into the hands of critics of liberalism who contend that liberalism presupposes just such a flawed conception. Second, Kymlicka's conception of the liberal autonomous self is radically at odds with his views concerning the importance of cultural–national identity and membership, which presupposes precisely the kind of constitutive attachments he denies.

A key component of Kymlicka's defense of cultural–national rights rests upon his conception of personal autonomy¹

and his conviction that autonomy is the core liberal value. The fundamental freedoms liberalism grants to each individual are important because liberal freedom “grants people a very wide freedom of choice in terms of how they lead their lives…. It allows people to choose a conception of the good life, and then allows them to reconsider that decision, and then adopt a new and hopefully better plan of life” (Kymlicka 1995, 80). The importance of culture, specifically what Kymlicka (2001, 53) calls a “societal culture”—“a set of institutions, covering both public and private life, with a common language, which has historically developed over time on a given territory”—is that it “provides people with a wide range of choices as to how to lead their lives.” Thus societal cultures are characterized as “contexts of choice” (1995, 89). The idea is that (for most persons at any rate) societal culture forms a kind of “authoritative horizon” of shared meanings such that ends can, for the most part, meaningfully be chosen intraculturally but not interculturally.

What Kymlicka (1995, 81) identifies as the fundamental importance of our being able rationally to choose and revise our conceptions of the good, to choose from an array of culturally determined ends in exercising our personal autonomy, depends on the view that

To assess Kymlicka's view of identity, it is necessary to consider briefly why Rawls held this view in relation to particular conceptions of the good and why he abandoned it.²

Rawls's assertion that persons are not inevitably tied to a particular conception of the good relates to his conception of moral personality as presented in A Theory of Justice (1971). There, moral personality is characterized by two capacities: one, the capacity rationally to choose and revise a conception of the good—a version of personal autonomy; the other, the capacity for a sense of justice—a version of Kantian autonomy (505).³

When realized, the former is expressed by a “rational plan of life”; the latter, by a “regulative desire to act upon certain principles of right” (561). Throughout A Theory of Justice, Rawls assumed that as part of a rational plan of life individuals choose their dominant ends: “[A] moral person is a subject with ends he has chosen…. [T]he self is prior to the ends that are affirmed by it; even a dominant end must be chosen from among numerous possibilities” (569). According to critics, this appeared to commit Rawls to a view of a “temporally prior” fully formed and fixed self that then chooses its dominant ends, and what this seemed to imply was that the identity of the self was in some way fixed or determined apart from all of its ends (Sandel 1982).

The first to raise this sort of critique with regard to Rawls's theory was Bernard Williams (1976), in his discussion of “ground projects”: “A man may have, for a lot of his life or even for just some part of it, a ground project or set of projects which are closely related to his existence and which to a significant degree give a meaning to his life” (208–9). Williams commented,

In “Kantian Constructivism in Moral Theory” (1980), Rawls attempted to respond explicitly to the criticism raised by Williams. In elaborating on the capacity of persons to have a conception of the good, Rawls writes, Kymlicka is critical of the notion of people's self-understandings being tied to certain ends or conceptions of the good and criticizes Rawls for having modified his views concerning the relation between the self and its ends. Kymlicka argues (1989, 58) that Rawls “gives no real examples of people whose self-understandings are bound to a particular complex of ends” and endorses instead Rawls's (1974, 61) earlier view that persons “do not think of themselves as inevitably bound to, or as identical with, the pursuit of any particular complex of fundamental interests that they may have at any given time.”

In describing what he takes to be central to the “liberal” view of the self and to one's sense of identity, as opposed to the “communitarian” view of the self, which he attributes to Michael Sandel, Kymlicka demonstrates why he rejects Rawls's revised conception of the possible relationship between persons' self-understandings and their “complex of ends,” and it is a view that he has consistently reaffirmed:

Although it may not be easy or enjoyable to revise one's deepest ends, “[n]o end is immune from such potential revision” (Kymlicka 1995, 91). What does it mean to assert that people's self understandings, their sense of identity,⁵ is not bound to a particular complex of ends, when this complex of ends includes what Rawls (1980, 544–45, 1985, 241, 1993, 31) identifies as religious, philosophical, and moral convictions, as well as enduring attachments, loves, and loyalties?

In stating what is central to the liberal view of the self, Kymlicka amalgamates two distinct claims. One is that “no end is immune from potential revision,” that one's deepest ends can and sometimes do change over the course of a lifetime. This is an empirical generalization and from the third person or “external” perspective: It is the claim that experience shows us that in the course of time, persons may come to abandon ends, even those that perhaps, at some given time in the past, they could not have conceived of themselves as being without, or adopt ends that they could not have conceived of themselves as having. It is undeniable that persons' dominant ends can and do change in this way, but in and of itself this tells us nothing about people's self-understandings or sense of identity as it relates to their dominant ends at any given point in time: for it is perfectly compatible with this that after a sufficiently profound change in ends, persons' sense of self will be radically altered; they will view themselves as, in effect, different persons; e.g., on the road to Damascus Saul of Tarsus becomes Paul the Apostle.

The second distinct claim Kymlicka advances here is a claim about person's self-understandings or sense of identity from the first-person perspective at any given point in time. It can be stated succinctly as follows: No particular ends or attachments contribute to, or are a part of, a person's sense of identity, inasmuch as we can always envision ourselves without our present ends. As Kymlicka points out, we can never conceive of ourselves as devoid of all ends or as “unencumbered” by them, for we must always choose on the basis of something (i.e., some end) and groundless choice is no choice at all. Therefore, “the process of ethical reasoning is always one of comparing one ‘encumbered’ potential self with another ‘encumbered’ potential self” (Kymlicka 1989, 53). Hence for the envisioned potential self in such reasoning to be recognizably me, it must be encumbered with some of the ends I currently have. According to Kymlicka—and this is the crucial point—this does not require that the envisioned self be encumbered with any particular ends: It is not the case that “we can't perceive our self without some specific end or motivation” (Kymlicka 1989, 53, original emphasis).

It follows that if in our ethical reasoning no specific end or attachment must be given with the self to make any imagined “potential self” recognizably me (i.e., to preserve its identity), then no specific end or attachment from amongst all the ends a person might have can be essential to one's sense of identity. To repeat: What is required to preserve one's sense of identity in ethical reasoning according to Kymlicka is that some one end or attachment or motivation be “given with the self,” but no particular end or attachment or motivation. This nonprivileging of any end or conception of the good or any attachment or loyalty or motivations is, I want to argue, at odds with our own deepest self-understandings as well as Kymlicka's own defense of rights of cultural–national membership.

All individuals have plural ends, using the term “ends” in the broadest possible sense to include projects, plans, relationships, commitments, motivations, and the like. It is also the case that for most individuals (or for all normal individuals) some ends are more important than others. Some may involve mere “preferences,” while others may rise to the level of Williams's (1976, 208–9) ground projects or be what Charles Taylor (1989, 50) calls “constitutive concerns,” concerns that enter into the “full definition of someone's identity, and include an individual's stand on moral and spiritual matters and some reference to a ‘defining community.”’ For many people, religious, philosophical, and moral convictions and commitments perform such a role, and in any moral individual, one would expect the relation between these convictions and sense of identity to be particularly strong. Such convictions may form what John Kekes (1983) has called “identity-conferring” commitments. To have such a commitment, moral or otherwise, means that from the first person perspective it is immune from potential revision precisely because the individual cannot conceive of him or herself as being the same person without it. Consider the common expression “I could never do that,” e.g., betray my friends and family or commit murder for money. By using such language one is making a statement about one's future self, claiming, in effect, that if one could do such things at some point in the future, one would no longer be oneself.⁶

Of course, perhaps the individual in question will come to do such things. Misfortune or despair can turn saints into sinners and sinners into saints, and reflection and experience can change one's beliefs as to what is a valuable form of life—perhaps for the better, but by no means necessarily so. Sometimes, failure to adhere to an ideal or principle, to be faithful to it, is seen as an instance of self-betrayal. The concept of self-betrayal makes sense only if one assumes an identity relationship between one's sense of self and certain convictions and attachments, for what one betrays in the process of self-betrayal is not an ideal or set of principles in the abstract, but oneself. Consider in this regard Harry Frankfurt's (1999, 139; cf. Royce 1969). characterization of the betrayal of an object of love:

A rupture to inner cohesion as Frankfurt characterizes it means a loss of integrity. As Christine Korsgaard (1996, 102) points out, etymologically, integrity is “oneness” and integration is what makes a person one: On a view in which there can be no identity-defining or identity-conferring relationships or commitments or attachments, there can be neither self-betrayal nor integrity.

Since Kymlicka is critical of Rawls for revising his view of the relation between the self and its ends, it is worth noting in this regard what Rawls says in A Theory of Justice concerning our moral ends and motivations. When Rawls (1974, 641) comments that “persons do not regard themselves as inevitably bound to, or identical with, the pursuit of any particular complex of fundamental interests that they may have at any given moment”—which Kymlicka (1995, 81) quotes with approval—he is referring specifically to interests associated with a person's autonomous selection of a “plan of life.” Regarding moral autonomy and persons' moral interests or moral motivations, however, things stand otherwise. According to Kymlicka (1989, 53, 58) persons' sense of self is not tied to any particular “motivation,” and he maintains that this was Rawls's view in A Theory of Justice. In his account of moral motivation in that work, Rawls explicitly links certain moral motives—our interest in acting justly—to the identity of the self:

As Rawls (1971, 443) points out, shame “implies an especially intimate connection with our person”; it “springs from a feeling of diminishment of the self.” What Rawls came to acknowledge then, in his writings after A Theory of Justice, was not the existence of identity-defining motivations—which he already assumed in relation to our motive for acting justly—but that persons may have a constitutive relationship to their conceptions of the good, or a host of other religious or moral ends or motivations.⁷ One need not accept Rawls's earlier Kantian account of moral motivation to acknowledge the intimate connection between who a person is, or takes himself/herself to be, and his/her moral motivations.⁸

Kymlicka defends his autonomy-based conception of liberalism on the basis of an untenable conception of the self. It is a conception that in the name of personal autonomy requires an impossible “distance” between one's sense of self and one's ends and motivations, and it is a view that plays into the hands of those critics of liberalism (in particular, communitarian critics) who contend that liberalism is an untenable political conception precisely because it depends upon such an implausible and unattractive conception of the self (MacIntyre 1981; Sandel 1982, 1996; Taylor 1979).

IDENTITY AND NATIONAL CULTURE

Kymlicka's insistence that ends, motivations, convictions, loves, loyalties and attachments are not constitutive of our sense of identity is particularly perplexing given that he argues vigorously for the constitutive nature of our cultural–national communities and posits a strong identity relationship between individuals and such communities. Kymlicka argues for minority group-based rights to preserve minority societal cultures within liberal societies on the basis of the constitutive dimension of cultural heritage. Some argue against any special rights to preserve minority cultures on the grounds that through assimilation, minority cultures will be able to enjoy the cultural membership of the larger liberal society into which they are assimilated. Against such views, Kymlicka (1989, 175, 1995, 105) argues that it is extremely difficult for persons to integrate into another culture because of the role of cultural membership in defining persons' identities:

In support of this view of the constitutive nature of cultural and national identity, Kymlicka (1989, 175–77) cites such factors as the reluctance of persons to give up cultural–national associations even when faced with negative costs of membership and the well-documented sense of despair, purposelessness, and loss of identity that ensues when persons are forcibly deprived of their cultural membership. Cultural–national membership, as Kymlicka characterizes it, is an attachment that shapes our very identity, that defines who the person is, and it is also a constitutive attachment that for the most part (or for most individuals) is both unchosen and cannot be changed: “[I]t is, and will remain, a constitutive part of who that person is” (Kymlicka 1989, 175).

According to Kymlicka, cultural–national membership is an attachment of the sort that, in Rawls's (1980, 544–45, 1985, 241, 1993, 31) words, persons “cannot stand apart from” and “regard it as unthinkable to view themselves without.” Yet as noted, Kymlicka criticizes Rawls for making precisely this assumption about the relationship between persons and their ends, attachments, and loves. If, as Kymlicka argues, Rawls is wrong in suggesting that certain attachments and loyalties can be an integral part of an individual's self-conception, inasmuch as all ends are open to rational revision and we can always conceive of ourselves without our present (and particular) ends, then why can we not always conceive of ourselves without the attachment of our particular cultural–national membership? Why is not cultural–national membership always potentially revisable in the same way that Kymlicka insists all of a person's other ends and attachments are?

The only way to resolve this contradiction is to assume the following: Kymlicka should not have said that Rawls was wrong to assume a constitutive or identity-defining relationship between persons and their attachments such that they cannot conceive of themselves without them but, rather, that Rawls was wrong with regard to every possible attachment except one. Cultural or national membership is the only attachment or loyalty or end that can be constitutive of, or an integral part of, a person's sense of identity. And in effect, this is what Kymlicka (1995, 89) says: Cultural-national identity is particularly suited to serving as the primary focus of identification. If this is Kymlicka's claim, however, it faces a serious difficulty.

Any argument that can be used to demonstrate the so-called constitutive dimension of culture vis-à-vis the individual can be applied equally to argue for the constitutive dimension of ends vis-à-vis the individual. If, as Kymlicka (1989, 175) expresses it, upbringing cannot be erased but remains a constitutive part of the individual, then an individual whose primary ends and ground projects are connected in some way to his or her upbringing within a particular cultural–national community—as is often the case—will equally be constituted by those ends. And if the cultural–national community is a form of traditional society in which individuals are generally assigned certain ends, primary projects, or roles as a part of their normal upbringing and socialization, then the two perspectives, community–self and ends–self, cannot always be clearly distinguished.

In certain kinds of traditional cultures, the cultural life of the community is organized around one overarching conception of the good, often a religious conception. Part of the normal process of upbringing in such cultures that, according to Kymlicka (1989, 175), “is and will remain a constitutive part of who that person is” and that “affects our very sense of personal identity” will be the holding and pursuit of certain religious ends, so that such ends, by Kymlicka's own logic, can be equally constitutive of the self. It is beside the point to object that traditional cultural communities are illiberal because they do not promote personal autonomy, for what is at issue is the question of whether or not persons can have constitutive attachments to any particular ends or complex of ends at all, regardless of whether or not the culture in question is liberal or illiberal.

Clearly not all cultures are autonomy-promoting contexts of choice, a fact that Kymlicka (e.g., 1995, 163–72) himself recognizes. So perhaps in denying that persons can ever be constituted by their ends in the same way that they are constituted by their cultural–national communities, Kymlicka is referring to liberal persons rather than persons in any absolute sense of the term, that is, persons who embrace the ideal of personal autonomy Kymlicka espouses. The assumption might be that in traditional socieities that do not enable autonomy, persons can have a constitutive bond to certain ends associated with the community, its specific religious norms or values, for example. Perhaps Kymlicka intends that in social environments such as these, what he identifies as the communitarian conception of the self as bound to or identified with a given complex of ends does apply. In support of such an interpretation, one finds assertions such as the following:

The relevant “we” in this passage would be autonomy-embracing liberals, to be contrasted with the “naïveté” of once existing—or still extant—traditional cultures. In communities that encourage personal autonomy, persons will not be constituted by the culturally determined array of ends both because they will be viewed as potential objects of choice rather than predetermined or culturally ascribed, and because persons will be encouraged to “stand back” from whatever ends they choose, to maintain a critical distance as it were, and continually assess and reassess the value of these ends before the court of individual reason.

On this reading, the distinction between liberal and communitarian conceptions of the self would be a sociological rather than theoretical one. It is worth noting that liberal advocates of a conception of personal autonomy such as Joseph Raz have emphasized the extent to which the capacity for the exercise of autonomy is a culturally dependent variable. In particular, Raz (1986, 369–70) associates autonomy with the social conditions of industrial and postindustrial society:

The assumption that persons can be constituted by their cultural–national communities but not by any of the ends available within their cultures is both supported by, and parallel to, the manner in which Kymlicka defines culture itself. For just as the (sense of) identity of the self “exists” independently of any of the ends a person might have at any given time, ends that one might normally associate with the character of an individual, Kymlicka (1989, 168; cf. 1995, 84–93; 2001, 25–26) maintains that a culture can be said to exist distinct from its character at any given time:

Kymlicka employs this distinction to argue that while the norms of a culture may be illiberal to the extent that they are not egalitarian and/or diminish the opportunities for personal autonomy, what the liberal is committed to preserving is not the culture's “character” (which includes its norms) but, rather, its “structure” or “existence.”

This is a problematic distinction. It is not clear in what sense people can be said to share a common culture if they need not share, in at least some minimal sense, a culture's norms, values, and attendant institutions, all of the things one would associate with a common cultural “way of life.”⁹

At times, Kymlicka seems to associate the structure or existence of a culture with its history and language, but to talk of the history of a cultural–national community presupposes the antecedent existence of such a community: It begs the question what it means for a culture or nation to be said to exist (and to have a history) apart from its values, norms and attendant institutions. If the cultural structure is to be associated solely with language, we are faced with the problem that groups may in fact share the same language, but because of significant cultural differences (i.e., differences in their values, norms, and attendant institutions) be held to be (or consider themselves to be) members of distinct national cultures. It is not my intent here to concentrate on the difficulties associated with Kymlicka's distinction between the existence of a culture and its character (see Tomasi 1995) but, rather, to note the additional difficulties this makes for Kymlicka's insistence that while persons' sense of identity can be constituted by their cultures, they cannot be constituted by any of the norms, values, and attendant institutions associated with their cultures. By what, then, are they constituted? What features of the culture determine persons' identities?

IDENTITY AND NATIONALISM

Kymlicka's view concerning cultural–national communities as our only constitutive attachments leads to an illiberal privileging of identities. Specifically, it prioritizes nationalist identity over and above all of the other “identities” that an individual might have and the nation over and above all other possible cites of identity formation. Such prioritizing represents the advancement of a substantive conception of the good in violation of Kymlicka's own principle of liberal neutrality concerning conceptions of the good. Kymlicka (1995, 88) associates what he identifies as the liberalization of a culture with a thinner and less distinctive cultural identity:

Since Kymlicka equates cultural identity explicitly with national identity, let us turn to the concept of national identity to discover precisely the kind of identification Kymlicka has in mind.

Recalling all that Kymlicka says about the absolutely privileged status of cultural–national identity, we are faced with the following troubling conclusion: In those cases where the cultural community is conceived as a nation (presumably by its members), nationalist identity is the only attachment or loyalty or love or end that individuals cannot conceive of themselves as ever lacking, that can be constitutive of their sense of self, and that can define who or what they enduringly and unchangeably are. Nationalistic identity means that individuals are “essentially” part of a greater whole—nationalist identity must always be communal—and if individuals' well-being is so fundamentally connected to the well-being of their nation (as the source of their sense of self), it is a short step to assume that persons find their own highest good in and through the flourishing of the nation. A view such as this is a view of the highest good or the ends of life: It is a “thick” or “substantive” or “comprehensive” conception of the good derived from a view as to what our “true nature” is and what ultimately fulfills that nature.

According to those versions of liberalism that take, or strive to take, value pluralism seriously, such a substantive conception of the good neither can be the locus of liberal politics or a liberal “way of life” nor can serve as a justification for liberal principles. What Rawls (1993, xix) calls the “fact of pluralism”—that a free society will result in a plurality of radically diverse religious and comprehensive conceptions of the good over time—precludes liberal society from being based on one comprehensive conception of the good, something that could be achieved only through the coercive and repressive use of state power. Not justifying liberal principles on one such conception is part of the liberal aspiration that all aspects of the social world—in particular, the coercive use of state power—be justified, as far as possible, to all members of society (e.g., Ackerman 1980; Larmore 1987, 1990; Rawls, 1993).

Kymlicka (2001, 331–36) is at pains to deny that the value of national identity as he presents it amounts to a particular conception of the good or that the promotion of such identity by the state would amount to a violation of the liberal principle of neutrality concerning conceptions of the good, which he endorses.¹⁰

He characterizes national identity as a “thin” good and maintains that if the state promotes such a thin good and a thin communal identity it is not violating principles of liberal neutrality (Kymlicka 2001, 334).¹¹ This is in contrast to communitarians, who seek to promote a “thicker” conception of the common good, “even if this limits the ability of individual members to revise their ends” (Kymlicka 1995, 92). The question of how the state goes about promoting the common good, however, is entirely distinct from the question of the nature of the good that is being promoted. Presumably even the “nonrestrictive” promotion of a substantive or “thick” conception of the common good would violate Kymlicka's principle of liberal neutrality.

In what sense can the common good of national identification be characterized as a “thin” good or a “thin” identity in Kymlicka's theory? The good of one's “sense of nationhood” (Kymlicka 1995, 88) is the only good for Kymlicka that is identity-defining; it is our only constitutive attachment, the only identity that cannot (for most persons) be changed, the only identity loss of which results in loss of identity and agency, the only attachment beyond rational revision, and the primary focus of identification. In support of his argument concerning both the constitutive and the “transcendent” nature of our national identifications, Kymlicka (1995, 90) refers to Tamir's (1993, 72) assertion that through national membership our individual accomplishments take on an “additional meaning” by becoming “part of a continuous creative effort;” and he refers to Benedict Anderson's (1983, 11) view that “national identity enables us to transcend our mortality, by linking us to something whose existence seems to extend back into time immemorial, and forward into the indefinite future.”

According to Anderson (1983, 11), the idea of the nation-state arose historically with the “ebbing of religious belief,” the “disintegration of paradise,” and the “absurdity of salvation,” when what was needed was “a secular transformation of fatality into continuity, contingency into meaning.” Anderson's view of the quasi-religious function of national identification is a view that finds expression in writings such as Fichte's (1922 [1806]) Address to the German Nation, where the nation is unambiguously characterized as the secular transformation of fatality into immortality: “He who regards his invisible life as eternal, but not his visible life as similarly eternal, may perhaps have a heaven and therein a fatherland, but here below he has no fatherland, for this, too, is regarded only in the image of eternity.” The nation, as a secular substitute for religion, is as “thick” a conception of the good as religion itself.

Let me forestall the following objection: I am fundamentally misunderstanding what Kymlicka means by the constitutive nature of our cultural–national attachments. These attachments are unique because it is only by having such attachments in the first place, by being members of one or another cultural or national community, that we can have a sense of self at all. To be a self, to be a person, is to engage in practices of one sort or another; but all practices can have meaning only if they are in some sense culturally defined or given meaning within a cultural “background” or “horizon”:

All of these assumptions are, I believe, relatively unproblematic as a matter of sociology or anthropology or of a “phenomenological” account of the self (e.g., Geertz 1973; Mead 1967; Taylor 1989), but if this is all that Kymlicka is in fact claiming, very little follows from it.

First, it does not follow from the fact that ends have culturally defined meanings that there must exist one single cultural framework in which each end is assigned a meaning or that ends can meaningfully be chosen only intraculturally and not interculturally. As a “cosmopolitan” liberal such as Jeremy Waldron (1992, 783) notes, “Meaningful options may come to us as fragments from a variety of cultural sources.” Second, even if we were to assume that there must exist one cultural framework for each complex of culturally defined ends, this tells us nothing as to how people understand themselves vis-à-vis their cultures. Kymlicka's point is not simply, or primarily, that practices must have culturally defined meanings—a proposition that a cosmopolitan liberal such as Jeremy Waldron readily accepts; nor is his point simply a causal one, i.e., that one's cultural surroundings, the societal culture of one's upbringing, determine to a greater or lesser extent what ends or options can be meaningful or viable for a given individual. In taking about persons' sense of identity and in making a claim that cultural–national identification is a defining component of person's self-understandings, Kymlicka is discussing a strong feeling or attachment that people have to their national heritage, of which they are aware, and of their intense desire to preserve their national identities, i.e., their nationalistic sentiments.

Such sentiments are exactly what one would expect when talking about nationalism; thus Weber's (1948, 179) referral to a nation as a “community of sentiment” or Renan's (1992 [1882], 17) definition of a nation as “a grand solidarity constituted by the sentiment of sacrifices which one has made and those that one is disposed to make again.” By talking of sentiment I am not endorsing the view that nationalism necessarily “appeals to our tribal instincts, to passion and to prejudice” (Popper 1962, 49). Rather, the point is that to conceive of one's national identity—one's “sense of nationhood” as Kymlicka (1995, 88) phrases it—as a central part—the “primary focus”—of one's identity, entails holding a subjective belief about the overriding value of one's nation and the centrality of one's national identity for his or her self-understanding.

To clarify this point, consider a distinction between identity and identification. Peoples' identities are doubtless shaped by innumerable factors, including the specific linguistic, socioeconomic, normative, and institutional features of their national communities. But this causal fact concerning formative influences does not necessarily mean that persons necessarily identify with their national communities in any significant way: They may be completely alienated from them; the survival of the community may be a matter of complete indifference or they may wish its demise or assimilation even though their identities may have been shaped in certain ways by membership in the nation in question. The point is that one cannot assume the existence of nationalistic sentiments, of strong identification with the national culture, from the bare fact that certain features of the nation—its language or social customs, for example—have exerted a formative influence upon the individual. To talk, as Kymlicka does, about the overriding value of an individual's sense of nationhood is to talk not about formative influences, but about something the individual identifies with and values, i.e., it is to talk about an individual's conception of the good. Hence in assuming the existence of, as well as advocating the promotion of, the common good of a sense of nationhood characterized as such a profound and overriding value, Kymlicka advocates a shared conception of the good in violation of his own principles of liberal neutrality.

Kymlicka (1989, 165, 175, 1995, 83, 2001, 209–10) places a good deal of emphasis upon language as constitutive of identity and links language with national culture (even though he acknowledges [1995, 218, n. 29] that language cannot be, in and of itself, that which differentiates cultures or nations). The emphasis on language illustrates well how assumptions concerning identity formation can easily slide into assumptions concerning what I have termed identification. Kymlicka (1989, 175) is surely correct to argue that persons' identities are constituted in some very important way by their native languages. Language is the very medium through which persons form a conception of themselves and represent, to themselves and others, who and what they take themselves, and the world around them, to be. Language is not a neutral medium of communication or self-representation; it is a “marker of the societal goals, the large-scale valueladen arenas of interaction that typify every speech-community” (Fishman 1972, 4; quoted in Kymlicka 1989, 175). This does not mean, however, that persons thereby consciously identify with or embrace their languages as a source of pride or consciously value them as a central part of their identity. Some do, but for others, language has no such connotations—it is simply something they know and use, largely unreflectively, like any other valuable and familiar tool. To be sure, language is, and always has been, an important part of nationalism and nationalist movements (e.g., Anderson 1983; Gellner 1983, and Hobsbawm 1990), but it is nationalists, not native speakers, simply by virtue of being native speakers, who view their language as a source of pride and distinctiveness and as what binds them, in some important way, to their cultural or national community.

IDENTITY AND LOYALTY

A nationalist prioritizing of identity may be used to support a parallel prioritizing of claims of loyalty. Such a prioritizing of loyalty claims, however, is problematic for liberalism. It also rest on an assumption that identity can give rise to loyalty claims. Neither identity nor “identification,” however, can, in and of itself, give rise to claims of loyalty or any obligations whatsoever.

If claims of loyalty are related in some way to an individual's identity (e.g., it is my duty to act thus because I am an American, a Christian, a parent, a rational being), such that disloyalty is also a form of self-betrayal, and if there can be only one, unchanging form of identity—one's nationalist identity—then it is possible to develop on that basis an absolute hierarchy of loyalty claims: Claims made on behalf of the nation must override all other possible claims made on the individual, whether of religious commitment, family, friendship, or personal conscience, because it is only in betraying the nation that one can betray oneself. Commentators have often defined nationalism precisely in terms of the priority of claims made on behalf of the nation to all other claims (e.g., Berlin 1979, 345; Connor 1993, 156; and Smith 1991, 74).¹²

According to Brian Barry (1987, 352–54), nationalism is a doctrine that “claims that all people should give their highest loyalty to their nation” and that “in politics … the pursuit of national interest should subordinate all other interests.” A theory of liberal nationalism, however, cannot maintain—and still be liberal—that people should always give their highest loyalty to their nation, that it is their supreme loyalty.

To clarify this point, consider Michael Sandel's account of the birth of what he terms the liberal (American) “procedural republic.” According to Sandel (1996, 12), underlying the procedural republic is a commitment to a form of liberal Kantianism that opposes any moral view that

Sandel (1996, 17) argues that the contractualist vision of society that flows from such a conception of the self can never be “a community with which I identify, but rather the claims of an arbitrarily defined collectivity whose aims I may or may not share”: “[T]hose whose fate I am required to share really are, morally speaking, others, rather than fellow participants in a way of life with which my identity is bound.” Sandel (1996, 53–54) discusses the case of Minersville School District v. Gobitis (1940) and quotes with approval Justice Frankfurter's opinion, which supported compulsory flag salutes in public schools as a legitimate way of cultivating the communal national identity of citizens.¹³ Gobitis was overturned three years later in West Virginia State Board of Education v. Barnette (1943), which Sandel (1996, 54) decries as the “birth” of the procedural republic. The problem with Sandel's story is that he ignores the fact that the Jehovah's Witnesses were also “encumbered selves,” bound by their identities as members of a church, and were obligated to fulfill ends that they believed were given by God.¹⁴ Is it the case that in a clash between the claims of religion (or religious communities) and the claims of national unity (or national communities), the claims of nationalism must always take precedence?¹⁵

Kymlicka would, perhaps, reject Sandel's argument concerning Gobitis and Barnette on the basis of what he terms “polyethnic” rights, rights “intended to help ethnic groups and religious minorities express their cultural particularity and pride without hampering their success in economic and political institutions of the dominant society” (Kymlicka 1995, 31). Yet he is by no means opposed to policies that promote national identity: “For liberal nationalists … it is a legitimate and essential task of government to protect the ongoing viablity of national cultures, and, more generally, to express people's national identities” (Kymlicka 2001, 42). In his defense of polyethnic rights, Kymlicka stresses the extent to which such rights foster the integration of ethnic/religious groups into the mainstream national culture: The policy of allowing Sikhs to wear turbans in the Royal Canadian Mounted Police “is intended, not to allow Sikhs to withdraw from the larger society, but precisely to modify the institutions of mainstream society so that Sikhs can integrate into them as fully as possible” (Kymlicka 1995, 177); and he stresses that religious groups that seek to withdraw from the larger society—“Christian and Jewish groups such as the Amish, Hutterites, Quakers, and Hasidim”—hence subverting the apparent culturally integrative purpose of polyethnic rights by having “been granted exemption from the usual requirements regarding integration”—are “an exeption” and “atypical” (Kymlicka 1995, 177–78). If the exemptions sought by the Jehovah's Witnesses are seen as not fostering their integration into the national culture, then perhaps Kymlicka would not support them, but however Kymlicka would decide this case, as I have argued, his arguments concerning the nature of national identity provide a good justification for overriding the claims of religious identity should the two conflict.

If we assume that persons' sense of identity can be constituted by a wide range of ends and commitments, what follows from this? Let me begin by considering what does not follow. What follows is neither a specifically communitarian conception of the self nor a communitarian politics. Kymlicka, however, believes that this does follow. In criticizing Sandel's communitarian conception of the self, he argues that if some of our ends are constitutive of our sense of identity, then “there is no reason why the state should not reinforce people's allegiances to those ends, and limit their ability to question and revise these ends” (Kymlicka 1995, 91). By the same logic that Kymlicka (91) attributes to Sandel, however, Kymlicka's view of national culture as a constitutive attachment would allow the state to reinforce that allegiance and limit persons' ability to question and revise, if not the nation's norms (i.e., its “character”), then its existence as a nation. Yet what if people question or reject their nation as a whole?

This entire logic, however, is faulty. It does not follow from the fact that people have such attachments that this confers upon the state a right of this sort. Without rehearsing a familiar story about the rise of liberalism from the advent of religious pluralism and the wars of religion, it was, arguably, a recognition that people do in fact have constitutive (religious) attachments of such a sort that they could not envision themselves without them, and would give their lives rather than betray them (as well as slaughter those of other faiths in their name), that led to the liberal view that it was precisely not the role of the state to enforce peoples' allegiances to their (religious) ends. The acknowledgment of such constitutive attachments combined with the advent of religious pluralism led to the liberal view that the state was not entitled to enforce one comprehensive (religious) conception of the good—hence Locke's (1983 [1689]) principle of the separation of Church and State.

If constitutive attachments do not give the state the right to reinforce them, can they give rise to duties on the part of individuals? For example, does the fact that an individual's nationality is an important part of one's sense of self give rise to an obligation to be loyal to the nation? That someone views something as an integral part of one's sense of self does not give rise to any claims of loyalty or obligations whatsoever. This is not to deny that there exist special obligations or “associative duties” based on ascriptive membership in families and clans, nations and states, races and religion, cultures, communities, and classes (e.g., R. Dworkin 1986, 195–216; Gewirth 1988; and Scheffler 2001)—parents have special obligations to their child, not to all children—as opposed to general obligations, grounded on impartiality, that are owed by anyone to anyone. Associative obligations are linked to identity in the sense that it is one's identity as a mother, or an American, or a Jew that gives rise to special obligations to one's children, or country, or coreligionists. As Samuel Scheffler (2001, 48–65) argues, any relationship we have reason to value can give rise to obligations whether or not it is voluntarily incurred.

The reason why identity alone cannot give rise to obligations or claims of loyalty is that persons can and do struggle against their own identities, and sometimes for good reasons: “A person can be alienated from his values, too; and he can be alienated from them even as they continue to grip him and to influence his behavior-as for instance, when someone recoils from his own materialism or his own sense of sin” (Wellman 2000, 134). Persons may reject a feature of themselves that they come to view as worthless or evil, and the need to reject it may be all the more imperative out of a recognition that what they now reject is indeed a part of who they are.¹⁶

In this regard, consider the following comment of Charles Taylor (1985, 34) concerning one's national heritage: “[M]y lineage is a part of my identity because it is so bound up with certain qualities I value, or because I believe that I must value these qualities since they are so integrally part of me that to disvalue them would be to reject myself.” The first alternative here certainly has normative force: Individuals' lineage is a part of their identity and it is linked with qualities they value (and, presumably, have reasons to value). And it is also relatively unproblematic to assume that the reason why they especially value their particular heritage—as opposed to any other heritage—is because their heritage is in some sense a part of their identity. The second alternative, however, reads more like a description of a psychological defense mechanism than anything giving rise to a normative claim. Perhaps there are cases in which morality does require that I reject myself, or a part of myself; but does fear of this act of rejecting mean that I must value those qualities (what is the normative force of “must” here?) and perhaps pretend to value what I really reject—or what I should reject—out of a desire to avoid psychic pain or “cognitive dissonance” or an “identity crisis”?¹⁷

The case of Alfred Dreyfus is instructive here. According to Bernard Yack (1996, 34), “Dreyfus was so intent on demonstrating his loyalty to the French nation because that community was an important part of his personal identity.” This may be true, but as Judith Shklar (1993, 191) has argued, such loyalty to a nation that completely rejected his Jewish identity and betrayed him in the name of an anti-Semitic French nationalism was hardly to be praised as an instance of one being “true to oneself”:

Was Dreyfus a hero or was he “crazy” for having clung to his identity as a French nationalist and remained loyal to the French nation when it became apparent that French nationalism included a heavy dose of anti-Semitism?

There are, admittedly, complex questions concerning the nature of associative obligations. According to Dworkin (1986, 202–6), for example, associative obligations are owed only to moral associations. Against this view, Tamir (1993, 101) argues that since associative obligations flow from “relatedness and identity,” “they are independent of the normative nature of the association.” She goes on to argue,

This formulation appears to make a feeling of belonging and “connectedness” the necessary and sufficient condition for the existence of associative obligations, which means that when one no longer has a feeling of belonging or no longer identifies with the community in question, then one no longer has any associative obligations. Yet if, for example, a parent no longer identifies with the role of parenthood, do the associative obligations of parenthood thereby end? Furthermore, by resting associative obligations on a feeling of belonging, then no matter how horrendously an individual is treated by a community—any community—as long as one identifies with or feels “connected” to it, one is obligated to it. In certain circumstances, however, such identification can verge on being pathological, and it is hard to see how moral obligations could be derived from identifications of such a sort.

BEYOND DICHOTOMOUS CONCEPTIONS OF THE SELF

It might appear that what has been argued thus far is a somewhat uncomfortable mix of communitarian and liberal conceptions. On the one hand, I have argued for the reality of constitutive attachments and that a large range of communities, ends, and motivations may be involved in identity formation. On the other hand, I have argued against any supposition that solely because something is a part of one's sense of self that it eo ipso gives rise to obligations either to oneself or to the nation. In fact, the stylized and stereotyped distinction between liberal and communitarian conceptions of the self with which Kymlicka works needs to be abandoned. Kymlicka is working with a false dichotomy that associates what he identifies as the liberal autonomous conception of the self with the lack of constitutive attachments (all, that is, except one) and the communitarian, “nonautonomous” conception of the self with the existence of such attachments.¹⁸

Liberal theory must move beyond a number of limiting and artificial dichotomies: that which places liberalism, autonomy, and the nonexistence of constitutive attachments on one side and communitarianism, the lack of autonomy, and the existence constitutive attachment on the other.

The assumed opposition between the liberal autonomous self and constitutive attachments is false because so-called constitutive attachments can equally be the result of autonomous choice. As Tamir (1993, 33) argues,

Tamir's point is not that we choose our cultural-national identities from an Archimedian point outside of culture, but rather, for all that our sense of self may be shaped by our upbringing in a particular national community, we can still choose to embrace this community and identify with it or reject it and adopt a new national identity (Tamir 1993, 20–34). Kymlicka appears to deny this and to assume that our national identities are largely fixed and that most people will strongly embrace and identify with their native national communities. The problem with Kymlicka's account is that it seems unable to account for the experience of immigrants, whom Kymlicka (1995, 14–15, 76–80, 2001, 275–89) expects to be able largely to integrate into the national culture to which they have emigrated, i.e., change, at least partially, their national identities. This is so despite his insistence that it is “rare” and “difficult” for persons to move between cultures (Kymlicka 1995, 85).

The claim here is not, however, that Tamir's approach is superior to Kymlicka's because she views national identity as the result of autonomous choice. Both Tamir's and Kymlicka's descriptions may accurately describe the way in which individuals identify with a given national culture. The mode of identification may itself be related to the presence or lack of the conditions of autonomy, i.e., in a more traditional culture that does not necessarily value individuality and personal autonomy, one would expect people to have a much harder time, if not the complete inability, autonomously to chose their ultimate nationality, or their religion, or a host of other things. In more cosmopolitan, pluralistic environments, such choices may be more feasible; but this is doubtless an oversimplification. People may lead “traditional” lives in the midst of cosmopolitan environments (consider the Lubavitcher Hasidic Jews in Brooklyn New York); relatively secluded Amish children can and do decide to leave their faith and their traditional ways of life; and one and the same “cosmopolitan” individual may be “autonomous” in relation to some important aspects of life, e.g., career and marriage, but not others, e.g., religion.

That people can identify with their cultural–national communities both autonomously and nonautonomously, however, misses the point, for what both Kymlicka and Tamir are advancing is a moral ideal of the person to complement their conceptions of liberal nationalism. Both are “autonomy-based” liberals.¹⁹

Kymlicka (1995, 81) defends the value of autonomy in part on the assumption that it is by leading an autonomous life that individuals can satisfy what Ronald Dworkin (1989, 486) calls the “endorsement constraint”: “No component contributes to the value of a life without endorsement…. [I]t is impossible to think that someone can lead a better life against the grain of his profound ethical convictions than at peace with them” (Dworkin 1989, 486). As Kymlicka (1995, 81, 1990, 203–4) phrases it, a necessary precondition for leading a good life is that “we lead our lives from the inside, in accordance with our beliefs about what gives value to life,” and he equates endorsement with autonomous choice—that is, one endorses what one chooses (and vice versa).

Liberal proponents of an ideal of autonomy often argue as if endorsement is equivalent to autonomous choice or can be exhibited only through such choice,²⁰

but such need not be the case: People can endorse what they do not choose. As one proponent of the ideal of autonomy acknowledges, An individual whose identity is constituted by membership in a given community—national, religious, familial—that has never been viewed as an object of choice and that he/she has never contemplated leaving—and hence never “chosen” to remain affiliated with—can still be an individual freely leading life on the basis of his/her beliefs concerning value. Kymlicka's own characterization of the value of societal cultures can be seen as a least a partial affirmation of this position, in opposition to Tamir. For while societal cultures, according to Kymlicka's characterization, are not something that persons autonomously choose from a range of options, they are something that people for the most part value and willingly embrace.

According to Tamir (1993, 21), Kymlicka confuses choice with change when he asserts that it is not the case that the more choice we exercise the better: “We do not suppose that someone who has made twenty marriage choices is in any way leading a life more valuable than someone who has no reason to question or revise their original choice” (Kymlicka 1990, 209). Tamir (1993, 21) argues that Kymlicka overlooks the fact that staying in a marriage also reflects a choice: “How do we know that we have no reason to question and revise our original choices unless we reflect on them and decide them anew?” The occasion for choosing to stay in a marriage—or any other union or association—is precisely when one is contemplating leaving it, so that a marriage one was continually choosing to remain in would be a marriage totally lacking in commitment (contrast choosing to remain married, a Christian, an American, with affirming one's love, faith, patriotism).

The advocate of autonomy-based liberalism might respond that it is obvious that persons can endorse, and be committed to, all sorts of things, but the fact of endorsement alone is insufficient to establish the value of a mode of life—consider the unsettling possibility of a “happy slave” (Herzog 1989, 215–47). It certainly is not sufficient to establish its value, but neither is the fact that a mode of life is the result of autonomous choice sufficient to establish its value, inasmuch as advocates of personal autonomy acknowledge that—in contradistinction to the exercise of Kantian moral autonomy—persons may autonomously choose all manner of immoral, or degraded, or worthless forms of life.²¹

As Thomas Spragens (1999, 125; cf., e.g., Raz 1989, 380, or Feinberg, 1986, 45) writes in a defense of personal autonomy:

That autonomy can be used in the pursuit of worthless or immoral ends, however, is not a refutation of the claim that personal autonomy still adds something of value to life. Mill (1975, 55) thought the exercise of personal autonomy necessary for the development of certain distinctive capacities: “The human faculties of perception, judgment, discriminative feeling, mental activity, and even moral preferences are exercised only in making a choice,” by which Mill meant a choice of a mode of life. Yet it seems unreasonable to assume that all such faculties cannot be employed in leading a more traditional or customary mode of life. Surely the Calvanite, whom Mill (1975, 58–59) decries as an exemplar of the evils of a nonindividualistic, customary mode of life, uses all of the distinctive capacities Mill lists in leading the life he/she does (and perhaps moral preference to a greater degree than many).

It might be argued that endorsement cannot be “authentic” if made under conditions that distort or pervert both preference formation and the individual's will: Desires can be “authentic” only if formed under conditions of freedom. Conditions of freedom, yes; but a key question for liberals is whether freedom should be equated with personal autonomy. The implication of equating freedom with personal autonomy is that a life lived in a culture that does not value individuality and personal autonomy is thereby unfree. Assuming, with Locke (1983 [1689]) that faith—or any other belief—cannot be coerced, I am also assuming that one who embraces a traditional form of life and believes in its value—even if not aware of alternatives—is not being “compelled” to do so.

Sometimes it is argued that the coercion or compulsion in customary societies is the process of socialization itself. For example, according to George Kateb (1984, 343; cf. Spragens 1999, 121–32), “One's dignity resides in being, to some important degree, a person of one's own creating, making choosing, rather than in being merely a creature or a socially manufactured, conditioned, manipulated thing: half animal and half mechanical and therefore wholly socialized.” The assumption here is that those who lead customary modes of life are not really leading their lives at all: The relevant causal agent is something external to the individual, his/her culture or “socialization,” and the individual is, as it were, the passive receptacle of what is transmitted to him/her. In contrast, the particular type of choice associated with personal autonomy appears to locate agency squarely in the individual. The choice of ends and forms of life reflect the agent's own determination about value and what are to count as valid reasons; the validity of reasons is not determined for him/her by his/her cultural environment.

This is an assumption concerning the relationship between autonomy and human agency. An assumption such as this, however, rests upon contestable assumptions as to what is necessary for a belief, or mode of life lived on the basis of a given belief or system of beliefs, to be “one's own.” For a belief to be one's own, it need not be unique; all that is required is that one be able to give a reason for it. “Traditional” reasons, such as believing in something on the basis of a belief in the authority or knowledge of one's ancestors, or elders, or priest, or community, or the authority of a divine text, or God, do not make what one believes any less one's own or a life lived on that basis any less an expression of one's beliefs concerning value, i.e., any less an expression of human agency (see, e.g., G. Dworkin 1988, 38). Beliefs held in common by members of a community are still the individual beliefs of every single member of that community.

Of course, any individual may be wrong about the value of the way of life he/she is leading. It is in acknowledging that one can be wrong about a way of life—that our beliefs about the good life are fallible—that assumptions about the importance of autonomy as “rational revisability” (Buchanan 1975) enter the debate. For Kymlicka (1989, 18), rational revisability is important “precisely because our goals can be wrong, and because we can revise and improve them.” Yet just as personal autonomy may be one path to error about the value of particular form of life (see quote from Spragens, above)—and nothing precludes that it be a road to error after error—so it may be a path to a valuable form of life. Liberals generally assume, on the basis of a belief in value pluralism, that there can be a plurality of valuable forms of life (e.g., Berlin 1998; Raz 1986, 395–97). Why assume, then, that the road to valuable forms of life must be unitary? Or, put another way, why assume that only autonomous forms of life can be truly valuable? To view value pluralism as a principle concerning the variety of ends and forms of life that may be autonomously chosen is to restrict the concept to such an extent as to call into question the use of the term “pluralism” in the phrase “value pluralism.”