The Original Position

Rawls's “original position” attempts to neutralize the effect of particularistic biases in the process of determining which principles of justice ought to guide society, by stripping individuals of particularistic traits, such as gender, age, race, or talent. To that end, the individual parties to that hypothetical agreement behind the veil of ignorance possess only the knowledge of “the general facts about human society. . .[and]. . .are presumed to know whatever general facts affect the choice of the principles of justice” (Rawls Reference Rawls1999, 119). This carefully crafted thought experiment raises immediate difficulties when we consider the question of language. Specifically, if the original position is a decision-making (or agreement-reaching) process, which linguistic mechanisms facilitate it? Language may potentially be neutralized, along with race or ethnicity, to the extent that it is perceived as a divisive identitarian feature. But, without any language at their disposal, how are the parties to reach any kind of meaningful engagement at all? The conceptual infrastructure of human cooperation, particularly the type upon which the Rawlsian notion of cooperative justice is premised, requires a fairly complex and nuanced level of semantic interaction, one that could accommodate a meaningful discussion of concepts such as “freedom,” “democracy,” and “fairness.” Such concepts, however, are not a stand-alone ethical vocabulary. Rather, they are rooted in the particularistic tradition of Anglo-American political philosophy. (Mis)perceiving Anglo-American political semantics as “the human norm” (Wierzbicka Reference Wierzbicka1997, 32), by perceiving English to represent a “universal logic, identical in all times and all places” (Cassin Reference Cassin and Cassin2014b, xviii), ironically nurtures a particularistic bias rather than successfully avoiding it.

Anna Wierzbicka (Wierzcibka Reference Wierzbicka2006, chaps. 4–5) for example, shows how terms such as “fair” and “reasonable,” which are central to Rawls's political philosophy, present an undeniably “Anglo” cultural bias. The meaning of the English word “reasonable,” for example, involves “limiting one's claims on others and at the same time appealing to reasons” (Wierzbicka Reference Wierzbicka2006, 135), attitudes which, Wierzbicka points out, emerged from specific aspects of British history such as the British Enlightenment. Despite its supposed equivalence, the French “raisonnable” does not fully correspond to the “family of meanings” (Wierzbicka Reference Wierzbicka2006, 140) of “reasonable,” further encompassing, for example, aspects of the notion of “sensible” (Wierzbicka Reference Wierzbicka2006, 139). Similarly, “fairness,” with its ideas of “social cooperation” and of “a public and agreed-upon standard” (Rawls Reference Rawls2001, 6), can be traced back to John Locke's political philosophy and, more generally, “to the post-Enlightenment move away from metaphysics and to the shift from an ethics based on religion to a ‘procedural morality’ and to an ethics based on ‘reason’, ‘social cooperation’, and ‘each participant's rational advantage’” (Wierzbicka Reference Wierzbicka2006, 154).Footnote ³ For these reasons, Wierzbicka claims, there exist no exact translations of “fairness” in French and German (and, arguably, in many others of the world's roughly 6,000 languages). Even when approximate translations are available, speakers of different languages are nevertheless “nudged” to focus on different connotations owing to different semantic overtones.

Wierzbicka's critique, which also includes an analysis of Rawls's contextualised political semantics (Wierzbicka Reference Wierzbicka2006, 143–4, 152–5), introduces an epistemic dimension into the original position by identifying linguistic relativity effects among the deliberating parties behind the veil of ignorance. English, after all, is not the only language in which moral and political debates take place. Concepts such as Japanese wa (和, “harmony,” “peaceful group conformity”) (Wierzbicka Reference Wierzbicka1997, 253, 279) or Bantu ubuntu (“humane-ness,” “sharing a universal bond of humanity”) (e.g., Louw Reference Louw, Sullivan and Tifft2006, 161–73) seem just as pertinent for discussing the just state of society.

This does not imply, however, that the Rawlsian original position is irremediably biased due to Rawls's use of Anglo-American political semantics. After all, the parties in the original position are essentially motivated by prudential reasons. Their goal, that is, is to maximize the good of the citizens they represent, who have “higher-order interests” (Rawls Reference Rawls1999, xiii) in cultivating and employing their moral powers, and who require certain primary goods. Since the notion of the “good” is not specific to the moral vocabulary of English (Wierzbicka Reference Wierzbicka2014, 34–5) there exists a scope, within the original position, for neutralizing the potential negative effects of linguistic relativity. Crucially, however, in their deliberations the parties should seek to refrain from using terms that belong exclusively to Anglo-American (or to any other parochial) political semantics as part of distancing the original position from specific human particularities. These terms include, for example, “freedom,” “democratic equality,” and “fairness.”

Take freedom. As Wierzbicka (Reference Wierzbicka1997, 125–55) has cogently shown, and as we illustrate more extensively in a subsequent section, the English concept of “freedom” cannot be easily translated across linguistic boundaries, since it is rooted not simply in the English language but also in the particular political culture in which it has emerged, which can hardly be described as universal in any empirical sense. Yet this concept is central to the deliberation among the parties in the original position, especially with regard to the formulation of the first principle of justice, which is concerned with the protection of citizens’ basic liberties. How can the parties in the original position reach any just decision with regard to what basic liberties (if any) ought to be protected by the state, if they use the English linguistic conception of “freedom,” which is grounded in a specific and partial political epistemology?

Similarly, the concept of “democratic equality” is central to the deliberation among the parties in the original position in connection with the second principle of justice, which is concerned with the arrangement of social and economic inequalities, and especially with the formulation of the “difference principle” (Rawls Reference Rawls1999, 65–73). Yet the concept of “democracy” cannot be easily translated across linguistic boundaries (e.g., Dupuis-Déri Reference Dupuis-Déri2004, 118–34), at least without being sensitive to the risk of conflating its empirical realities with its universal normative presuppositions. As Frederic C. Schaffer argues, “Xhosa speakers today talk of idemokrasi, Chinese students demonstrated for minzhu, and Václav Havel attempted to institute demokracie. These examples are hardly trivial. Translating minzhu, demokracie, or idemokrasi, by ‘democracy’, as journalists and scholars regularly do, is potentially problematic because the cultural premises that infuse American practices and institutions may not be universal” (Schaffer Reference Schaffer1998, 14; see also Freeden and Vincent Reference Freeden, Vincent, Freeden and Vincent2013, 8). But then how can the parties in the original position deliberate about the best way of arranging social and economic inequalities on the basis of the idea of “democratic equality,” if the English concept of “democracy” contains connotations, or semantic “overtones,” that are not shared by its literal translations in other languages?

With regard to the concept of “fairness,” we have already highlighted its Lockean post-Enlightenment roots. One might then observe that the prudential and self-interested reasons that motivate the parties in the original position make it unnecessary for them to have a concept of fairness in order to be able to deliberate about the principles of justice. However, this objection overlooks the fact that for Rawls, while the parties in the original position are indeed self-interested, they are also “presumed to be capable of a sense of justice and this fact is public knowledge among them” (Rawls Reference Rawls1999, 125). A sense of justice involves the willingness to comply with fair terms of cooperation and “it means that the parties can rely on each other to understand and to act in accordance with whatever principles are finally agreed to” (Rawls Reference Rawls1999, 125). It is “a tendency to answer in kind” (Rawls Reference Rawls1999, 433) which “would appear to be a condition of human sociability” (Rawls Reference Rawls1999, 433). According to Rawls, a purely self-interested kind of social cooperation would not result in stable institutions and society. It is therefore necessary not only that “the principles of justice are agreed to in an initial situation that is fair” (Rawls Reference Rawls1999, 11) but also that the parties in the original position share the same (linguistically codified) concept of fairness.

Another objection might be that the idea of fairness is in fact not as parochial as we claim. For example, one might argue that the widely shared golden rule “do unto others as you would have them do unto you” is, in fact, a concept of fairness. However, consider that “the golden rule shows two major sides: one promoting fairness and individual entitlement, conceived as reciprocity; the other promoting helpfulness and generosity to the end of social welfare” (Puka Reference Puka2010). Rawls's work, we have seen, grants significant importance to the idea of reciprocity as a key aspect of fairness. Yet it does not assign any significant importance to attitudes such as helpfulness and generosity, which are central to the golden rule. The latter, therefore, presents different connotations from the Rawlsian concept of fairness. It can also be argued that, although the golden rule is assumed to be an ethical universal, such universality nevertheless requires empirical grounding. The fact that the golden rule exists in both Judeo-Christian and Confucian traditions does indeed mean that it is not exclusive to either of them, but it does not conclusively follow from this that it is a universal property of human morality.

A final objection that we would like to consider is the claim that fairness, for Rawls, is a highly complex and technical philosophical concept, i.e., one that is distant from the everyday speech of ordinary citizens, regardless of their native language. To this, we would like to respond, first, that Rawls's theory is not as distant from everyday political and moral judgments as this objection suggests. After all, Rawls claims that the rationale for the original position thought-experiment also involves the idea that we should “see if the principles which would be chosen match our considered convictions of justice or extend them in an acceptable way” (Rawls Reference Rawls1999, 17). Our everyday political and moral concepts therefore play an important role, according to Rawls, in the formulation of a theory of justice. Second, we would like to stress that the analyses of complex philosophical terms and of everyday political speech are not mutually exclusive but complementary. Examining the implications of linguistic relativity for such a sophisticated philosophical framework as the original position, that is, does not prevent us from also considering its effects on real-world deliberation. Indeed, the latter task is the one we aim to carry out in the next section.

For now, we would like to conclude the present analysis by stressing that the main challenge raised by the effects of linguistic relativity on the original position is the establishment of a semantically-grounded epistemology that is sufficiently “calibrated,” from which a shared framework of cooperative justice could eventually be generated. A better epistemically informed version of the original position should seek greater epistemic inclusiveness in discussing political ethics, a discussion that may therefore only take place with the incorporation of language into the process.

What would such an alternative version of the original position look like? Wierzbicka's critique of the tendency to mistake the English political vocabulary for the human norm is supported by a developing project on what is called the “Natural Semantic Metalanguage” (NSM) (Wierzbicka Reference Wierzbicka2014, 33), a “decompositional system of meaning representation based on empirically established universal semantic primes” (Goddard Reference Goddard, Henie and Narrog2010, 450), which aims to identify through cross-linguistic research the core set of semantic universals. This developing list of primes comprises, in effect, the “semantic. . .Lego” (Wierzbicka Reference Wierzbicka1995, 149)Footnote ⁴ of human language, with local exponents identified and tested across a broad range of languages, including English, Russian, Polish, French, Swedish, Malay, Japanese, Korean, Ewe, and East Cree. The English exponents of the NSM include substantives such as “I,” “you,” and “someone,” descriptors such as “big” and “small,” evaluators such as “good” and “bad,” actions such as “do,” happen,” or “touch,” and several others (Wierzbicka Reference Wierzbicka2014, 34–5). This “semantic. . .Lego” (Wierzbicka Reference Wierzbicka1995, 149), in turn, enables the construction of more complex social and political notions, such as “cooperation,” “altruism” (Wierzbicka Reference Wierzbicka2014, 104–7) and “tolerance” (Gladkova Reference Gladkova and Goddard2008). Such complex and localized concepts cannot be the starting point of the discussion over shared values, because they fail to draw on an empirically common ethical and political language (Wierzbicka Reference Wierzbicka2014, 66–7).

This does not mean, of course, that the English exponents of the NSM should be preferred over their counterparts from other languages. The whole point of the NSM is that terms such as “I,” “you,” “someone,” “good,” and “bad,” unlike terms such as “fairness,” “freedom,” and “democracy,” do have substantive cross-linguistic literal translations, as empirical research shows.Footnote ⁵ Therefore choosing the English version of the NSM instead of any of the other languages for which this framework has been tested and verified involves a negligible amount, if any, of epistemic or cultural bias. The choice between the different versions of the NSM is, in other words, as epistemically neutral as present research allows, and may be dictated by pragmatic reasons without unintentionally prioritizing one's particular epistemic standpoint. The use of the NSM would allow individuals who do not necessarily share a native language to nevertheless form and develop an epistemically informed version of the Rawlsian original position, by using their language-specific NSM exponents as a linguistic “common ground.”

An epistemically informed original position therefore begins not with particularistic ethical vocabularies (e.g., “fairness”), but rather with these semantic “molecules” that are more easily shared, recognized and deliberated across linguistic and epistemic boundaries. Importantly, this proposed revision of the Rawlsian thought experiment intends to negotiate a middle ground between the seemingly effabilityFootnote ⁶ principle of the Rawlsian experiment (and, more broadly, of Anglo-American political philosophy) on the one hand, and the strong version of the linguistic relativity principle (i.e., linguistic determinism) on the other. The revised thought experiment is thus premised on the assumption that epistemic differences can and indeed ought to be bridged in the process of a collective reflection on the just state of society. Such bridging, however, requires closer and direct attention to linguistic epistemology, rather than the attempt to eliminate it from the picture altogether. For the original position to deliver on its promise to select principles of justice in a way that is genuinely unbiased, the NSM offers a theoretically and methodologically capable tool for taking the linguistic pluralism of contemporary societies—and its epistemic effects on moral and political semantics—seriously.

“Re-coding” the original position, whether in English or in any other NSM-tested language, would be a very complex and time-consuming endeavor, and as such a task that is understandably well beyond the scope of this article. It seems plausible to think that it might result in the same consensus illustrated by Rawls, or in a consensus on different principles of justice. The key point is that this would be a nonbiased (or only minimally biased) kind of consensus, because the thought experiment would be grounded in an epistemically informed linguistic framework. There is sufficient flexibility in the way the original position is designed, we believe, to accommodate the revisions we advocate.

Public Reason

The idea of “public reason” is central to Rawls's later work (Rawls Reference Rawls2005a; Rawls Reference Rawls2005b). In political liberalism, public reason is “the reason of equal citizens who, as a collective body, exercise final and coercive power over one another in enacting laws and in amending their constitution” (Rawls Reference Rawls2005a, 214). The goal of public reason is to guarantee the public justifiability and legitimacy of coercive legislation concerning fundamental matters in societies characterized by a pluralism of reasonable comprehensive doctrines. Public reason places citizens under a moral “duty of civility” (Rawls Reference Rawls2005a, 217) to only appeal to political values, rather than to their comprehensive doctrines and conceptions of the good, when making decisions about “‘constitutional essentials’ and questions of basic justice” (Rawls Reference Rawls2005a, 214), e.g., issues concerning “who has the right to vote, or what religions are to be tolerated, or who is to be assured fair equality of opportunity, or to hold property” (Rawls Reference Rawls2005a, 214).

Given that public reason applies to the real deliberation of real people in the real world, the kind of NSM analysis that, we argued, can contribute to a better epistemically informed version of the original position, cannot realistically be applied to it, at least in most cases. For example, “no translator/interpreter will ever be allowed a 21-page parenthesis in the middle of a European Union speech by the German Chancellor” (Collin Reference Collin2013, 17) in order to explain the difference between two German terms for “homeland,” i.e., Heimat and Vaterland, or to consult other useful materials, such as the Dictionary of Untranslatables (Cassin Reference Cassin2014a). Admittedly, not all multilingual deliberative forums are truly deliberative. For example, in the United Nations (UN) General Assembly, government officials from different states normally act as delegates and present positions that have already been worked out at the national level. In this sense, they are not involved in a true process of deliberation. Nevertheless, in many other cases actual multilingual deliberation does take place, as for instance in meetings of state leaders in the European Council, of ministers in the Council of the European Union, and of MEPs in the European Parliament.

Some scholars, especially political philosophers, might be tempted to argue that misunderstandings due to linguistic barriers can normally be overcome through deliberation, especially since the latter allows moving beyond word-by-word translations in order to explain the meaning of single words through more complex and developed utterances. However, while this might be possible, for example, in the context of academic seminars, where participants possess, alongside a comprehensive textual common ground, sufficient time, expertise and (one would hope) open-mindedness, the same is unlikely to apply to deliberations among politicians and, even more so, to everyday deliberations among ordinary citizens.

What implications, then, does linguistic relativity have for public reasoning? If linguistic diversity entails epistemic diversity, and if the use of the NSM in public deliberation is in most cases not a practical option, then public reasoning may only be possible if members of a political community share one language, and share it at a native or near-native level of competency. This is because “each native language has trained its speakers to pay different kinds of attention to events and experiences when talking about them. This training is carried out in childhood and is exceptionally resistant to restructuring in adult second-language acquisition” (Slobin Reference Slobin, Gumperz and Levinson1996, 89). For example, English native speakers are “nudged” by the English grammar to focus on the progressive aspect of an action, whereas Spanish native speakers tend to focus on the distinction between perfective and imperfective aspects of an action, and Turkish native speakers make a distinction between witnessed and non-witnessed actions (Slobin Reference Slobin, Gumperz and Levinson1996, 73–4). What is the significance of these differences for normative political thinking?

Take the latter example offered by Slobin, concerning the distinction between witnessed and nonwitnessed actions in the Turkish language. This property, which linguists call evidentiality, namely “the grammatical means of expressing information sources” (Aikhenvald Reference Aikhenvald2004, xi), enables (and sometimes obliges) speakers to make refined distinctions between information obtained, for example, firsthand or nonfirsthand, information that is received directly, inferred or assumed, and information obtained visually or nonvisually. As well as marking speakers as credible and trustworthy based on their competency in the use of evidentials (Aikhenvald Reference Aikhenvald2004, 9–10, 357–8), evidentiality systems have also been reported to develop a higher degree of metalinguistic awareness. The importance of evidentials for metalinguistic awareness is “reflected in the ways people can discuss evidentials and explain why one evidential and not another was used in a particular circumstance” (Aikhenvald Reference Aikhenvald2004, 360). Such metalinguistic awareness has a clear political dimension since it enables speakers to construct—and contest—social and political articulations. For example, many Macedonian speakers have criticized a Macedonian politician's choice to opt for a frequent firsthand evidential in a book on Alexander the Great, which seemingly suggested him as the direct heir to Alexander (Aikhenvald Reference Aikhenvald2004, 317). This metalinguistic awareness is also enhanced by being aware of the lack of corresponding systems in other languages through language contact (e.g., speakers of Turkic and Balkan languages with proficiency in English). Switching between languages with and without evidential systems is a demanding cognitive task, as “those who speak a language with evidentiality find it hard to adjust to the vagueness of information sources in many familiar European languages such as English, Portugese, and varieties of Spanish other than those spoken in the Andes” (Aikhenvald Reference Aikhenvald2004, 360).

Similarly, speakers of different languages are often “nudged” by their language to focus on different aspects of a certain ethical or political concept. For example, the terms “citizenship” and “nationality” are often translated into other languages as if they were interchangeable (Collin Reference Collin2013, 291–2), despite the legal connotations of the former and the ethnic connotations of the latter. Likewise, as Audard and Raynoud note, “liberal designates a progressive or social-democratic attitude in the United States, but in France the word signals an opposition to the welfare state” (Audard and Raynoud Reference Audard, Raynaud and Cassin2014, 570). Other key concepts of political deliberation that do not necessarily “travel well” cross-linguistically include “democracy” (Dupuis-Déri Reference Dupuis-Déri2004, 118–34), “peace” (Ishida Reference Ishida1969, 133–45), and “freedom” (Wierzbicka Reference Wierzbicka1997, 125–55).

Let us consider, for example, the term “freedom,” and imagine that English and Russian native speakers are engaged in public reasoning in English about legislation concerning free speech and privacy. The English word “freedom” differs quite significantly in its meaning from the Russian word svoboda, even though it is normally used as the latter's standard translation in English. More specifically, svoboda conveys the idea of “a ‘loosening’ of some sort of material or psychological straitjacket” (Wierzbicka Reference Wierzbicka1997, 140), linked to “the need to ‘spread out’, to ‘overflow’ any bounds like a flooding river” (Wierzbicka Reference Wierzbicka1997, 142) in order to achieve “an exhilarating sense of well-being” (Wierzbicka Reference Wierzbicka1997, 141; see also Vasylchenko Reference Vasylchenko and Cassin2014). The strongly anarchic connotations of the term svoboda differ significantly from those of the English term freedom, which “is not incompatible with restrictions and constraints; on the contrary, it suggests a perspective from which constrains imposed by the law can be seen as necessary to guarantee the inviolability of everyone's personal space” (Wierzbicka Reference Wierzbicka1997, 144).

When applied to public reasoning about free speech and privacy, this linguistic epistemic gap can have significant implications. More specifically, while English native speakers using the word “freedom” will be “nudged” to think that guaranteeing freedom of speech is in principle compatible with imposing legal constraints on it (e.g., in order to guarantee people's privacy, the infringement of which can legitimately be considered a harm, and an infringement on other people's freedom), Russian speakers (who are non-native English speakers) will assume that the word “freedom” is a more or less problem-free translation of svoboda. They will therefore be “nudged” to think that in English “freedom of speech” implies a total lack of constraints on the speaker, e.g., even if their speech potentially infringes upon the privacy of other people.

Or consider, to use a different example, the English expression “family values,” which among native English speakers (especially in the U.S.) is normally and almost unconsciously associated with “the ‘traditional’ nuclear family and. . .[with disapproval]. . .of homosexual marriage or gay adoption or sexuality outside marriage” (Collin Reference Collin2013, 295). This expression is normally used as a literal translation of the Danish term familieværdier which, however, presents different connotations and emphasizes “society's responsibility to provide daycare centres, parental leave after childbirth, and medical care for children, a far cry from opposition to gay marriage” (Collin Reference Collin2013, 295). Thus, when engaged in public reasoning in English with regard to the state provision of free childcare, many English and Danish speakers will be using at some point the English term “family values” in order to support or reject that measure or to specify how it should be implemented in more detail.Footnote ⁷ Regardless of their specific position on the issue (e.g., some of them may be socially conservative, others more liberal), native English speakers will be “nudged” to associate “family values” with what is normally considered as the traditional nuclear family. Similarly, many Danish speakers will also be using the English term “family values” during that deliberation. However, many of them are likely to use that expression as the literal translation of the Danish word familieværdier as non-native English speakers who quite possibly, like their English counterparts, have no particular knowledge of the specific semantic baggage of that word in English (i.e., they have not been taught English in an epistemically informed way). Therefore, regardless of their specific position on the issue (e.g., some of them may be egalitarian, others libertarian), those Danish speakers will be “nudged” to think that “family values” almost implicitly involve the state provision of free childcare and they will not draw any distinction between married and unmarried or gay couples when using that term, because that distinction is not central to the semantic baggage of the term familieværdier.

While these examples are hypothetical, they are nevertheless grounded in an empirical analysis of cross-linguistic moral and political keywords. However, many non-hypothetical examples can also be offered. For instance, Szalay (Reference Szalay1981) shows how the English term “corruption” presents different connotations from the Korean term pup'ae (부패). While both terms present negative connotations, Szalay explains, the English term emphasizes the immoral and criminal nature of the act, whereas the Korean term only stresses the negative effects of the action on the smooth function of political and social institutions (Szalay Reference Szalay1981, 141). This semantic discrepancy can have significant effects on negotiations between English and Korean native speakers (Cohen Reference Cohen2004, 28).

Likewise, Raymond Cohen (Reference Cohen2004) shows that when political negotiations involve “compromise” and “concessions,” participants should be aware that the literal translation of these English terms in other languages do not always have the same (mainly positive) connotations that they have in English. In Arabic, for example, terms such as Khaleena nitjaham or Musawama refer to mutual concessions which involve reciprocal compromise. The term Tanazol, instead, involves one-sided concessions and presents a “negative, shameful connotation” (Cohen Reference Cohen2004, 108). Cohen therefore points out that it is important for Anglophone negotiators to specify what kind of concessions they would like to discuss with their Arabic-speaking counterparts, and translate key terms carefully and accurately, as this may affect the very structure of negotiations (Cohen Reference Cohen2004, 109).

Similarly, as Brigid Maher (Reference Maher2002) points out, the Italian term racommandare is inadequately captured by its usual glossing as “recommend” or “entrust,” as neither of these lexical elements conveys its meaning as “saying to someone that you would like them to help a certain other person in need of some kind of assistance or protection. Often this assistance may be in the form of lenience regarding a punishment, or favour in a search for work, with exams or with some complicated bureaucratic process” (Maher Reference Maher2002, 42).

One final example is that of “martyr,” particularly in the context of the Israeli-Palestinian conflict, where it is a longstanding topic of political and ideological tension. The Arabic notion of shaheed ( ) “is broadly used in everyday discourse to refer to anyone on the receiving end of violence in a situation of conflict, whether or not they choose to be involved in that situation” (Baker Reference Baker2007, 66). The Hebrew qadosh (שודק), by contrast, “[has] become standard usage for Jews being killed by non-Jews in a variety of contexts” (Hasan-Rokem Reference Hasan-Rokem and Lambek2008, 593), one important of which is the Holocaust, whose (Jewish) victims are often referred to as qedoshim (pl. of qadosh) (Hasan-Rokem Reference Hasan-Rokem and Lambek2008, 594). Israeli Hebrew speakers are often unaware that shaheed “is not necessarily. . . part of a narrative of violent resistance” (Baker Reference Baker2007, 66), despite the fact that the same term is actually used by the Druze community to refer to Druze soldiers in the Israeli army who were killed in action. However, the presumed semantic (and moral) equivalence between shaheed and qadosh, with the former referring, among other things, to Palestinians voluntarily involved in acts of terror, and the latter referring to their victims, is the source of deep tensions in Israeli political discourse.

Given the problems raised by linguistic relativity for real-time public reasoning and deliberation, we therefore argue that political liberalism paradoxically demands the teaching of a common civic language from an early age to all its citizens and residents (e.g., at nursery or primary school level at the latest), if the latter are expected to be able to participate in public reasoning on equal epistemic terms. This is because, as we have seen, deconstructing linguistic utterances through the NSM is simply not an option in real deliberation (e.g., in town hall meetings, during electoral speeches, in parliamentary debates, etc.). Speaking a common civic language at a native or near-native level means that, at least, citizens and residents share the same unreflective understanding of ethical and political concepts such as fairness, wa, or ubuntu in their respective linguistic civic spheres.

Four objections should be considered at this point. First, one might observe that most political concepts are “essentially contested” (Gallie Reference Gallie1956, 167–98; see also Chilton Reference Chilton and Wodak2008, 228) intralinguistically as well as interlinguistically. Yet linguistic relativity renders contestation over the meaning of political (and other) concepts much more complex due to its unconscious epistemic effects, and it is especially this that risks undermining public reasoning. Moreover, we would like to stress again that even when speakers become aware of their linguistically nudged epistemic bias, it is quite unrealistic for them to resort to the NSM lexicon in order to accurately translate concepts in different languages in real-world and real-time public deliberation.

Second, and relatedly, one might also observe that there are often significant regional and class differences in the way different people learn and speak the same language, and that this may lead them to apprehend the same political concept (i.e., its core meaning) in different (linguistically codified) ways, which nudge them towards different non-core meanings. It is not implausible, therefore, to argue that this will result in intralinguistic linguistic relativity effects on political languageFootnote ⁸ and, consequently, that this may have negative effects on public reasoning. We acknowledge this problem and recognize that it might be difficult to fully avoid it. However, as we will show in the next section, it should be noted that the teaching of a common civic language, on its own, could never eliminate the effects of linguistic relativity, due to the changing linguistic landscape of contemporary diverse societies in an age of increased human mobility. The presence of intralinguistic linguistic relativity effects, therefore, should not be seen as a threat to an otherwise epistemically coordinated demos.

Third, it might be observed that our proposal for a shared civic language amounts to a form of linguistic homogenization, which would be especially problematic in officially bilingual or multilingual states (e.g., Canada, Belgium, etc.). In response to this point, it should be noted that such states already often display a high level of linguistic homogenization within each of their constituent linguistic communities (e.g., English-speaking and French-speaking communities in Canada, Flemish-speaking and French-speaking communities in Belgium, etc.). This kind of homogenization can potentially be oppressive and unjust for speakers of the nonofficial language(s), especially when grounded in the adoption of territorial (rather than personal) language rights that often characterizes less accommodating language regimes (e.g., Belgium and Switzerland) (e.g. see De Schutter Reference De Schutter2008). A regime of territorial language rights implies that “languages should be territorially accommodated, such that on each particular territorial unit only one language group is present or officially recognized” (De Schutter Reference De Schutter2008, 105). Our solution, therefore, does not raise any more problems than such accounts of linguistic justice as those offered, for example, by Philippe Van Parijs (Van Parijs Reference Van Parijs2011) and Will Kymlicka (Kymlicka Reference Kymlicka2001), who endorse the principle of linguistic territoriality. Furthermore, even within states that embrace official bilingualism or multilingualism (i.e., the official state recognition of more than one language), it may be necessary for members of different linguistic communities to share at least one common civic language for deliberation and public reasoning beyond (native) linguistic boundaries in a process of “linguistic retooling” (Safran Reference Safran2004, 12–3). For example, in India over 20 official languages are officially recognized across the various states, but Hindi and English are also widely spoken at the federal level. Last and relatedly, as we argue in the next section, the teaching and learning of a shared civic language does not prevent (and, in fact, should be accompanied by) the teaching and learning of other languages in order for all citizens to acquire linguistic repertoires. It is therefore important to stress that while we are indifferent, for the purpose of the present analysis, to the question whether states should adopt official monolingualism or multilingualism (as long as citizens also share at least one common civic language), we do endorse personal multilingualism, i.e., the ability of individual citizens (and noncitizens) to develop a linguistic repertoire that comprises more than one language.

Fourth, one might point out that by advocating the need for a shared civic language we are simply reintroducing a bias similar to the one associated with the Anglophone political vocabulary that we have previously criticized. After all, if English political language has been deeply shaped by liberalism, is it not likely that any language chosen as the shared civic language of a political community will be equally shaped by some (liberal or nonliberal) moral vision? This objection, we believe, slightly misses the point of our analysis. The problem raised by linguistic relativity effects on political language is an epistemic rather than a moral one. That is, it aims to highlight the epistemic moral “frictions” between members of a political community who do not necessarily share a single language as native speakers, rather than to arbitrate which view is morally preferable to the others. Sharing a common civic language would nudge members of a political community to think about political and moral concepts in the same way but it would not necessarily make them agree on the value of those political and moral concepts.

As we mentioned in one of our earlier examples, for instance, Danish native speakers are nudged to think about familieværdier in the same way, i.e., as a term that connects family values with the state provision of social welfare, but do not necessarily agree that this welfare-based approach is valuable or just. Neoliberal, free-market-oriented Danish native speakers, that is, are likely to criticize familieværdier. Or take, to use another example, gender equality. It is well-known that the English language is less gendered than other languages, as shown for example by terms such as “friend” or “politician,” which are translated, respectively, as “amico” (male) or “amica” (female) in Italian, and “politicien” (male) or “politicienne” (female) in French (Collin Reference Collin2013, 293). There is also an ongoing endeavor to further reduce the gender bias of the English vocabulary, as testified for example by the gradual replacement of terms such as “actress,” “stewardess,” and “policeman” with the gender-neutral versions “actor,” “flight attendant,” and “police officer,” or by the 150-year (and going) search for English gender-neutral pronouns (Baron Reference Baron2010). Yet it would be wrong to claim that being a native English speaker makes a person automatically a supporter of gender equality. We believe that while cultural changes can affect changes in the lexicon and grammar of a language, and the latter can nudge us to understand certain concepts in certain (morally and culturally influenced) ways, this does not imply that speaking a certain language will cognitively “program” us endorse the moral and cultural views that have produced those changes.