* Assistant Professor, Département d'économie politique, Université de Genève.

** Professor, Départment de sciences économiques, and Fellow, Centre de recherche et développement en économique (C.R.D.E.), Université de Montréal.

The purpose of this paper is to introduce noneconomists to the economics of multilingualism. This area of research is relatively little known among economists, and until recent years, few of its methods and results had been published in the journals usually read by language specialists, although sociolinguists and applied linguists had for many years been pointing out the importance of economic dimensions in their field of research. It is therefore useful to begin with a brief review of the literature, which owing to limitations of space does not report specific results, but lists the major issues that have been examined to date. More extensive surveys are available in Vaillancourt (1985a) and Grin (1994c; 1996e).

In the second part of this paper, we present an analytical framework in which the market and non-market, private and social benefits and costs associated with multilingualism can be integrated. This paves the way for a discussion of the economic rationale of language policy measures required by and affecting multilingualism. The main points of the paper are summarized in a closing section, where we discuss some of the strengths and limitations of the economic perspective on language issues.

While the title of this paper refers to the economics of multilingualism, it is appropriate to note that economists describe this area of the economics of human resources as the "economics of language" and that both theoretical and empirical work primarily focus on bilingualism, rather than multilingualism in general. The economics of language can be defined as the study of the relationships between linguistic and economic variables; in addition, it also includes the study of language-related issues where economic variables have little or no part, but which can nevertheless be examined with the concepts and methods of neo-classical economics. A more formal elaboration of this definition, along with its main theoretical underpinnings, can be found in Grin (1996b). In this paper, "economic" refers to mainstream economic theory; to our knowledge, the literature in alternative (that is, non neo-classical) economic theories has never addressed language issues, whether theoretically or empirically.

Contributions to the economics of language have mostly come from three groups of scholars: Canadian (mainly from Quebec), American and European economists. The first were particularly active in the 1970-1985 period, examining both the nature of language as an economic good and the socio-economic status of speakers of English and French. American economists were particularly active in the 1965-1985 period but have published fewer papers than their Canadian counterparts; most of their work focuses on the estimation of the returns to English for Hispanics. Finally, European economists became interested in this issue in the late 1980s, contributing an increasing number of papers on various theoretical issues as well as, more recently, some empirical work.

We have regrouped the literature under five headings. These are: defining language; language and earnings; language of work, consumption and business; language dynamics, addressing language acquisition, maintenance and shift; and language policies.

The need to define language from an economic perspective arose when multivariate techniques were first used in 1965-1970 to link earnings and individual characteristics, which include the language skills of survey or Census respondents in Canada and the United States. Additional work resulted in new approaches, so that in total four definitions were used.

The first approach treats language as an ethnic attribute or marker similar to race. This approach was inspired by Becker's (1957) theoretical model of racial discrimination, and by the empirical work undertaken in the U.S. in the 1965-1970 period to explain earnings differences between black and white Americans. Raynauld, Marion and Béland (1969) used this approach in their work for Quebec, while Fogel (1966) applied it in the United States. Such a definition, however, overlooks the role of language as a communication tool. The following two approaches address this issue, but neglect the ethnic marker dimension of language.

Carr (1985)¹, Breton and Mieszkowski (1975, 1977) and Boucher (1985) all use the analogy of international trade between countries to stress the role of language as a means of exchange between groups or individuals. Carr points out that a few currencies dominate international trade, and that the same could be expected in the case of languages, Breton and Mieszkowski compare language-related costs to transportation costs, and Boucher points out that language policies can be seen as a protectionist tariff. These analogies, while interesting, have not been used in empirical work.

Language was implicitly treated as human capital by Hocevar (1975) and explicitly described as such by Carliner (1976) and Breton (1978). This was an important development since it allowed the use of the Mincerian analytical framework (named after the economist Jacob Mincer) to calculate the returns to language in the same way as the returns to education or experience. One must note, however, that language is a supercollective good in that its communication value increases with the number of speakers (Sabourin 1985)². This raises the problem of "externalities", which will be taken up again later in this paper.

Defining language solely as a form of human capital overlooks the ethnic dimension of the first language learned, usually defined as the individual's mother tongue. Vaillancourt (1980) put forwards a model that distinguishes between the mother tongue, seen as an ethnic attribute and a form of human capital, and the other languages known by an individual, solely defined as elements of human capital. This approach is now commonly used in empirical work.

The empirical relationship between the earnings of individuals and their language attributes, and in particular the net impact of such attributes (that is, once other factors such as education or experience have been taken into account), has been examined extensively for Quebec, Canada and the United States and has been the object of some work in Australia, Germany and Switzerland. Empirical work on the issues is carried out using:

- the ordinary least squares multivariate analysis technique;

- public use samples from Censuses (Canada, United States, Australia) or survey data (Quebec, United States, Germany, Switzerland);

- more or less narrow samples with sex (men) and age (25-64) often used as constraints;

- the logarithm of earnings, or only wages and salaries as the dependent variable;

- education, experience and its square and weeks worked (if the sample is not restricted to full year workers) as control factors (independent variables), in accordance with the usual specification of Mincerian earnings equations. Other variables such as marital status, region, type of employment, etc., are also commonly used, depending on the availability of data.

In the case of Quebec, Vaillancourt has carried out work using the 1971 (Vaillancourt 1980; Lacroix and Vaillancourt 1981), 1981 (Vaillancourt 1988) 1986 (Vaillancourt 1991) and 1991 (Vaillancourt 1996) public use sample of the Canadian censuses. The variables used are as identical as possible, which facilitates intertemporal comparisons for the whole population and various subgroups. In addition, issues such as the returns to language for highly qualified workers (Lacroix and Vaillancourt 1980) or to different levels of language skills (Vaillancourt and Pes 1980) have also been examined.

In the case of Canada, there exists evidence scattered here and there on the rates of return to language attributes, but until recently no systematic intertemporal comparison had been made. Bloom and Grenier (1992a, 1994b) do so for the 1971-1986 period, and review some of the existing studies; Beach and Vaillancourt (1996) carry out similar calculations for Canada and its linguistic regions.

In the case of the United States, authors have used either Census or Survey of Income and Education data to examine first the impact of Hispanic ethnic origin (Long 1977; Carliner 1980; Reimers 1983; Chiswick 1991), and then the impact of the level of competence in English on earnings (Grenier 1984a; McManus 1985, 1990; Chiswick 1991; Rivera-Batiz 1991; Bloom and Grenier 1993).

Research in Europe on the relationship between language and earnings has begun more recently, and includes work by Dustmann (1994) for guest workers in Germany and Grin (1995a, 1996d) for Switzerland. In the latter, the data set used contains more precise linguistic information than almost all North American studies, so that both the level (nil to excellent) and the nature (oral/written, active/passive) of language skills in five different national and non-national languages can be taken into account.

(a) Language of work

Breton and Mieszkowski (1975), Hocevar (1975) and Vaillancourt (1980), have each identified various factors that influence the language used in the workplace. These factors are the language contents of goods and services produced, the language of the various markets, the language of production factors, particularly technology and the workforce, and the language of the owners or senior management. Vaillancourt, Champagne and Lefebvre (1994) and Leblanc (1994) have used administrative or survey data on the language of work and linked it to data on the ownership of firms (Vaillancourt and Leblanc 1993) and on exports to show that bilingual francophones will use more French in the workplace when owners are francophones and when the Quebec markets are important.

(b) Language of consumption and business

Hocevar (1975) argues that individuals will prefer consuming in their mother tongue out of ethnic pride. Vaillancourt (1980) agrees but adds that an individual, while bilingual, may be more efficient timewise when consuming in his/her first language. Vaillancourt (1985b) shows that bilingual francophones consumers in Quebec have a preference for French but in addition, the less bilingual they are, the more they prefer French. Surveys carried out by language boards in the case of Western European minority languages (e.g. Catalunya 1988; 1993) report similar findings.

Businesses and firms will offer goods and services in a given language when it is profitable to do so, which depends on factors like economies of scale, market size and translation costs. Pellenbarg (1993) and van Langevelde (1994) show that in the case of Friesland, the linguistic environment has no influence (whether positive or negative) on the locational choice of firms. Finally, Holden (1987) shows that language is almost systematically overlooked in the marketing literature; recent work by Grin (1994b), however, models the firm's decision to switch from unilingual to bilingual advertising.

Vaillancourt and Lefebvre (1981) and Vaillancourt and Payette (1984) for Quebec francophones and Grenier (1984b) for Hispanic Americans have examined the determinants of learning English. They show that children with better educated parents and those more exposed to English (for example, when English is the first or second language of one parent, or when respondents live a predominantly English speaking region) are more likely to learn English than others. Similar results for Canada are reported by Chiswick and Miller (1994). Recent models (Selten and Pool 1991; Church and King 1993) use a game-theoretic approach to model the link between second language learning and expected benefits .

Once a language has been learned, it can be forgotten or not passed on. Grin (1990; 1992; 1993c) examines this issue with a formal model of language use by bilinguals, showing that survival thresholds for minority languages can exist, but that they must be defined as a function of not just one, but several demolinguistic variables interacting with one another. Grin (1993a) concludes that the process of economic integration in the European Union will have different effects for different minority languages, and prove positive for some, such as Catalan, and negative for others, such as Irish.

The first two economic contributions to the study of language policies³ are due to Thorburn (1971) and Jernudd (1971), who argued that cost-benefit analysis could be useful in assessing these policies, but did not study a particular policy in detail. A set of studies carried out for a Quebec governmental commission examined issues such as the language of work (Carlos 1973; Dagenais and Van Peeterssen 1973) the availability of qualified French-speaking workers (Girard et al. 1973) and the link between francization and francophonization (Inagaki et al. 1973)⁴. These results, while useful as inputs in cost-benefit analysis, were not used to carry out a full-scale analysis of this kind. The first attempt at evaluating the costs and benefits of Quebec's language policies was made by Breton and Mieskowski (1975). One of their main conclusions is that government-mandated change in language use may be optimal to move from one situation (English dominant) to another (French dominant); in other papers, however (e.g. 1977) the same authors dismiss language planning as unjustified meddling with market forces. Important work by Pool (particularly 1991) develops a rational choice perspective on the selection of official languages in a multilingual polity; Pool (1996) uses a similar approach to outline policy alternatives in the case of the European Union. Solutions derived from Pool's analysis can serve as inputs into other models of the allocation of language rights in multilingual polities (e.g. Grin 1994a, 1996a).

The debate surrounding the adoption in 1977 of the Charter of the French language in Quebec led to analyses of its costs. A synthesis of these and other studies concludes that the transitional costs of francisation are less than 1 % of sales (Vaillancourt 1987) and that long-run costs are even lower. These studies cannot assess, however, if the benefits exceed costs or not. One notes, however, that in general in the labor market costs are borne by non-French speakers and benefits by French speakers (Vaillancourt 1992a, b) European work in this area has remained more theoretical, examining the choice between translation and language teaching in a multi-language environment (Colomer 1996) and the possibility of using contingent valuation analysis to assess the non-market value of minority languages and to establish relevant public policies (Grin 1993b, 1995c). Some work has also been carried out on the economic effects of spending incurred as part of the policy efforts aiming at the preservation or promotion of minority languages, including multiplier effect (e.g. Ó Cinnéide and Keane 1988 on Irish, or Sproull and Ashcroft 1993 on Scottish Gaelic).

Nobody much disputes the fact that multilingualism is a widespread phenomenon; in addition, several social trends concur to an increasing occurrence of these two types of multilingualism. The most significant trends are: the reassertion of local, often language-based identities, which have recently been boosted by major political upheavals like the break-up of the Soviet Union; the magnitude of contemporary migration flows; the development of supra-national entities such as the European Union, whose activities extend to an increasing number of domains; and the deepening integration of local economies into a world-wide economy.

Multilingualism and multiculturalism are often touted as forms of wealth or as "resources", but such claims usually fail to move beyond metaphor, which is of limited help in dealing with the policy issues raised by multilingualism. Multilingualism almost systematically calls for some kind of policy, because of conflicting claims to shares of power (or even to hegemony) from communities with distinct language characteristics, or because maintaining a full-fledged multilingual environment (with its attendant consequences for the provision of education, social services, etc. in many languages) places demands on society's resources, financial and otherwise that conflict with other claims. From an economic standpoint, the first question to address is how far multilingualism is, indeed, a resource (because it may generate both costs and benefits), and then how multilingualism affects society's "welfare"—or well-being.

The economic approach to the value of multilingualism and the determination of some optimal level of multilingualism does not claim to provide ready-made answers to policy problems. Nevertheless, economics addresses issues often ignored in the more standard perspectives on language problems, and can serve as an ingredient, among many others, in the selection and design of language policies. In what follows, we deliberately limit ourselves to a highly summarized presentation of the most general concepts, it being clear that its function is to provide the groundwork on which more elaborate or context-specific analyses can build.

Depending on the manifestations of multilingualism considered, benefits and costs can be of an extremely diverse nature. The number of languages in question, the field (individual, organizational, official, societal) in which multilingualism is examined, the type of skills (oral / written, active / passive), etc. add almost countless dimensions of variability. Instead of embarking on a Sisyphean typological enterprise, we shall use two simplifiying assumptions in order for some fundamental economic distinctions to emerge.

First, let us define multilingualism as the coexistence of n languages (as opposed to m) in a particular polity, where n > m. Let us assume here, for the sake of the example, that n=3 and m=1, implying that we are comparing trilingualism with unilingualism; of course, any pair of values for n and m could be compared in the same way. Second, we shall assume that each situation defines a linguistic environment. The concept of linguistic environment, described in more detail in Grin (1995b), is a theoretical construct used for analytical purposes. It subsumes in an extensive (but obviously not exhaustive) fashion all the relevant information about the status, in the broadest sense of the word, of the various languages present in a given polity at a certain time. This includes the number of speakers, individual proficiency levels in the various languages, the domains of use of each language by different types of actors (individuals, corporations, state, civil society organizations), their attitudes towards the languages considered, etc. Corpus data, insofar as they impact on language status, are also relevant to the characterization of linguistic environments. Each linguistic environment entails costs and benefits, and the latter can be broken down in four groups according to fig.1, which describes the benefits and costs of a trilingual, as opposed to a unilingual environment. The terms "cost" and "benefit" are used here in a broad sense: as shown below, these costs and benefits may be of a non-material and non-financial nature; their use need not imply any parallel with a managerial approach.

fig. 1: four types of benefits and costs of multilingualism

Cell 11 contains private market benefits (b11) and costs (c11). Most of the empirical research reviewed in the first part of the paper refers to private market benefits, which consist mainly of language-based earnings differentials accruing to individuals who know (in this case) three languages. It could also include access to lower-priced goods. No systematic research has been carried out so far on the associated costs, namely, the expenditure incurred by individuals to master languages in addition to their mother tongue.

Cell 21 refers to private non-market benefits (b21) and costs (c21). The first include the satisfaction directly derived from engaging in activities in three languages, as opposed to just one, or even the mere possibility of doing so. Such benefits are analyzed theoretically in Grin (e.g., 1990) but just like all types of non-market benefits, their estimation raises considerable difficulties; as a result, no attempt at a systematic evaluation has been made so far. Contingent valuation methods, which involve the use of surveys where respondents are asked how much they would be willing to pay in order to ensure the continued presence of some particular language in their linguistic environment, or to safeguard (in our example) a trilingual environment, could be used to yield estimates of such benefits.⁵

The private non-market costs of multilingualism may include the "disutility" (or negative satisfaction) associated by some with life in a multilingual environment, particularly if some of the languages involved are not legitimized historically or socially. Such costs can, in principle, be assessed through surveys asking respondents how much they would be willing to pay to avoid a trilingual environment.⁶

Cell 12 comprises social market benefits (b12) and costs (c12). Traditionally, social benefits are computed as the sum of private benefits. This is standard procedure in the economics of education for the estimation of the social rates of return to schooling, and it is appropriate to the extent that benefits observed at the private level indeed capture the total value created. However, the presence of externalities (as mentioned earlier) may drive a wedge between the sum of private benefits and actual social benefits. If externalities are positive, social market benefits will exceed the sum of private market benefits.

Positive externalities are benefits that result from the behavior of actors, but without the latter being able to fully appropriate them; in other words, positive externalities can be seen as unintentional gifts from the individual to the community. Suppose for example that some unilingual individual j, who speaks language A as a mother tongue, becomes trilingual by virtue of learning languages B and C. In so doing, she will earn private benefits in the form of higher labor income. However, her knowledge of languages B and C will increase the usefulness of the language skills of some trilingual individual k, because the latter's language skills will now be usable with an additional person; this may have a positive (if marginal) effect on k's wage rate.

Social market costs are the sum of private market costs (mostly acquisition costs), but exceed this sum in the presence of negative externalities, which are unintentional costs imposed of some actors as a result of the behavior of others, without compensation. For example, individual j's learning of languages B and C increases the supply of speakers of those languages; all other things being equal, this may drive down the wage premium accruing to people who already speak those languages, and hence reduce the net social value of trilingualism over unilingualism.⁷

Finally, cell 22 includes social non-market benefits (b22) and costs (c22). Benefits can be computed as the sum of private non-market benefits, but there again, externalities may drive a wedge between the simple sum and the more complex aggregate. The social non-market benefits of trilingualism arise from more harmonious inter-community relations, a stronger sense of social cohesion, or the value of diversity in its own right (which an isolated individual cannot enjoy by himself). In the same way, social non-market costs can be estimated as the sum of private non-market costs (reflecting some individuals' dislike of trilingualism). Total cost may exceed the sum of private costs in case of mutual reinforcement of some people's distaste for multilingualism, generating harmful conflicts.

In addition, social costs include the expenditure covered by the state (presumably out of tax revenue) on goods and services whose aim is to maintain multilingualism, but which individuals do not take into account in their private weighing of benefits and costs, because state expenditure (and the coresponding share of the tax burden borne by individuals) is independent of the latter's consumption of these goods and services. Examples are state provision of multilingual education, social services and the like. Depending on the goods and services considered, it is not always clear if these costs ought to be assigned to the "market" or the "non-market" row of fig. 1, hence our mentioning them in a separate paragraph. Estimates of the public expenditure linked to Quebec's language legislation have been made (see the "language policy" section earlier in this paper), but they do not state the problem in terms of increasing degrees of multilingualism. A key component of the costs discussed here is made up of expenditure on the school system; very little work has been carried out to date on the level of such expenditure, because the accounts of education systems never provide a breakdown of expenses by subject, a necessary basis for evaluating the amount of resources spent on teaching languages. Recent estimates by Grin and Sfreddo (1996), however, indicate that public spending on second, third and fourth language instruction in Switzerland accounts for some 10 % of total public spending on education, amounting to approximately 0.5 % of GDP (gross domestic product).

Are some degrees of multilingualism preferable to others? What can the economic framework presented here contribute to answer this question? Let us first recall that to a large extent, multilingualism is a "given": for example, the number of language communities in a polity at a certain point in time cannot be reduced, short of ethnic cleansing through massive deportation or genocide⁸. Hence, the unqualified quest for some "optimum multilingualism" is not a legitimate policy question. However, many other determinants of linguistic environments can be objects of language policy; for example, given a certain demolinguistic make-up, the quest for a socially optimal degree of official multilingualism is a valid policy objective. Let us therefore focus on those determinants of linguistic environments that are legitimate objects for policy. More formally, this implies that estimates of the costs and benefits defined with the help of the break-down in fig. 1 are made for a given level of those determinants which are not policy objects; the reasoning, however, remains unchanged.

As shown earlier, trilingualism, as opposed to unilngualism, entails costs and benefits. In the highly simplified framework used here, the net value of trilingualism can then be approached by summing b12 and b22, and then subtracting c12, c22, and public expenditure. Similar estimations can be made for any degree of multilingualism, that is, for any number of languages. We would then end up with a range of figures reporting benefits, costs and net value for n=1, 2, 3, 4, 5, etc.⁹ The numbers can be represented as curves in a diagram (fig. 2); for the sake of simplicity, curves are smoothed, which is an appropriate representation if the indicator of the degree multilingualism is based not only on the number n of languages present (for example, the number of languages taught in schools or receiving official status), but also on the progressive introduction of each language in an increasing number of domains; if we were to define, say, k different domains, increasing multilingualism could be arranged along a scale comprising kxn degrees (for a discussion of such scales and the corresponding institutional arrangements, see Grin, 1994a or 1996a).

Obviously, estimated benefits and costs reflect the preferences of individuals, as such and as members of social groups. These preferences are socially, culturally and historically determined; in addition, the level of benefits and costs depend on political and social arrangements that can be more or less favorable to one or the other extent of linguistic diversity, depending on the languages considered. Economists have little to say on these matters, and we shall therefore confine ourselves to a few general remarks on the shape of the curves represented in fig. 2.

fig. 2: multilingualism and optimality

We are assuming that benefits and costs both rise along with the extent of multilingualism. This assumption is borne out by common sense and by the existing results on private benefits and expenditure, although little is known as to the relative rate at which costs and benefits rise. Available evidence, however, suggests that benefits exceed costs up to a certain point, but that costs catch up and finally outweigh benefits. Moving from unilingualism to bilingualism, for example, is likely to generate considerable benefits for society as a whole, without causing prohibitive expense. However, adding one official language in a country that already recognizes ten may yield benefits smaller than the associated costs of the move. For example, if official status means that the minutes of parliamentary debate must be translated into each of n official languages, the number of translation directions to be provided is n(n-1), that is, 90 if n=10. Adding one language to the set of official languages entails an exponential increase in translation directions, which then jump to 110.

In other words, there is good reason to suppose that the net value associated with different degrees of multilingualism first increases, reaches a maximum and then decreases. This means that, in most cases, optimality takes the form of an interior solution, where optimal multilingualism is neither zero nor infinite. Only a full specification of the parameters of the problem, followed by painstaking (and necessarily incomplete) estimations can help to determine if actual optimality is in the region of two, five, ten, or any other number of languages, depending on context.

However, heated public debate on the vices and virtues of multilingualism reveals that hasty pronoucnements on the subject are not uncommon. On the one hand, some claim that the costs of multilingualism always exceed its benefits. In other words, the optimal degree of multilingualism is zero, a case better known as unilingualism. The "English Only" movement in the United States, as well as some nationalist regulations in other parts of the world, such as the new "Law on the State Language of the Slovak Republic", passed on 15 November 1995, which severely restricts the use of Hungarian by native speakers that make up over 10 % of Slovakia's population (1991 Census figures), are telling examples. On the other hand, blanket endorsements of multilingualism commonly found in the linguistics or multicultural education literature (often along with the metaphor of multiligualism as a resource) appear to assume that language rights should always be granted, irrespective of the number of claimants and the costs implied. This amounts to an assumption that benefits always exceed costs, and that the optimal degree of multilingualism tends to infinity—or at least towards some undefined maximum, presumably constrained by the number of languages currently in existence.

Simplified as it is, the analytical framework developed here can help to question such hasty pronoucements and can provide some (though, of course, not all) of the elements necessary for the selection and design of language policies that seek solutions to the many social, political and economic problems raised by multilingualism.

Summary and discussion

The economics of language span a wide range of issues, many of which can be approached using the concepts and methods of the economics of language. The overwhelming majority of questions studied stem from the presence of more than one language. Economists have at various times characterized language as an ethnic identity marker, an element of human capital, or both, in trying to assess the effect of individuals' linguistic attributes on their earnings. Literature on the link between language and economic activity examines language use at work, in consumption and advertising, or the effect of language variables on local economic development or the locational choice of business firms. Much of the existing work therefore studies the impact of language on economics; another strand of research, however, focuses on the reverse causation, and examines how linguistic variables are affected by economic ones. Finally, a growing body of literature addresses language planning problems; the emphasis then is often put on identifying and evaluating the costs and benefits associated with policy options.

Our overview of the economics of language shows that economic perspectives can supply useful elements in the study of language-related phenomena, and help to select and design language policies. However, the application of these instruments cannot dispense with the conceptual background or analytical perspective of other social sciences. For example, economists take language as a given, and drastically simplify notions such as "competence" or "attitudes" in order to make them amenable to deductive modelling and empirical testing; the contribution from applied linguists and sociolinguists helps to ensure an appropriate use of such variables. In the same way, the basic assumptions that run through neo-classical economics unavoidably steers both analysis and conclusions in certain directions. For example, the focus on costs and benefits may lead to a narrowly managerial outlook on language problems; collaboration with sociologists and political scientists helps to remember that the choice of a policy is a political and social issue, which technical approaches as such (whether rooted in economics or any other discipline) cannot solve.

The issue of multilingualism, whether explicitly or not, is therefore present in most of the work published to date. The theoretical concepts and tools developed in the economics of language, along with its empirical results, are all germane to the question of how to seek solutions to the problems frequently raised by multilingualism. The general analytical framework developed here shows how the various strands of research relate to the policy issues; it rests on the assumption that the identification and measurement of the benefits and costs associated with various degrees of multilingualism are useful elements in the analysis of language policies, if the latter are meant to increase social welfare.¹⁰ One key result is that, assuming some degree of multilingualism can be shown to be preferable to others, it is unlikely to be zero or infinite, which implies that a positive but finite degree of multilingualism is likely to yield the highest social welfare. It is typical in economics to find non-extreme solutions, for example in environmental economics where "optimal pollution" is neither zero nor infinite. Obviously, this type result has major implications for language policy.

While all this clearly poses limits the economists' contributions, the fairly demanding standards of internal consistency required in the formal theoretical or empirical models favored by economists and rational choice scientists can also guard against seductive, but misleading analogies, such as the well-known "language as currency" hypothesis that regularly re-emerges in academic discourse on languages. The economics of language also yield results found nowhere else, and thus they can help to cast new light on issues about which people can hold very stubborn beliefs.

Notes:

Annotated bibliography

References

Beach, C. and F. Vaillancourt. 1996. Income and languages skills of Canadians. 1970-1990. Ottawa: Statistics Canada. Mimeo.

Becker, G.S. 1957. The economics of discrimination. Chicago: Chicago University Press.

Bloom, D. E. and G. Grenier. 1992a. Earnings of the French minority in Canada and the Spanish minority in the United States. In B. R. Chiswick (ed.) Immigration, language and ethnicity Canada and the United States. Washington: American Enterprise Institute. 373-409.

Bloom, D. E. and G. Grenier. 1992b. Economic perspectives on language: The relative value of Bilingualism in Canada and the United States. In J. Crawford (ed.) A source book on the official language controversy. Chicago: Chicago University Press. 445-451.

Bloom, D. E. and G. Grenier. 1993. Language, employment and earnings in the United States : Spanish-English differentials from 1970 to. 1990. NBER working paper. 4584.

Boucher, M. 1985. La Loi 101. Une approche économique. [Bill 101: An economic approach]. In. F. Vaillancourt (ed.) Économie et langue. [Economics and language]. Québec: Conseil de la langue française. 101-116.

Breton, A. 1978. Le bilinguisme. Une approche économique. [Bilingualism. An economic approach]. Montréal: C.D. Howe Institute. VII.

Breton, A. and P. Mieskowski. 1975. The returns to investment in language. The economics of bilingualism. University of Toronto, Institute for Policy Analysis, Working Paper No. 7512.

Breton, A. and P. Mieskowski. 1977. The economics of bilingualism. In W. Oates (ed.) The Political Economy of Fiscal Federalism. Toronto: Lexington Books. 261-273.

Carliner, G. 1976. Returns to education for Blacks, Anglos and five Spanish groups. The Journal of Human Resources. 11.2.172-184.

Carliner, G. 1980. Wages, earnings and fours of first, second and third generation American males. Economic Inquiry. 18.87-112.

Carlos, S. 1973. L'utilisation du français dans le monde du travail au Québec, Commission Gendron, Étude E-3. Québec: Éditeur officiel du Québec

Carr, J. 1985. Bilingualism in Canada. Is the use of the English language a natural monopoly? In F. Vaillancourt (ed.) Economie et langue. Québec: Conseil de la langue française. 27-37.

Catalunya 1987. La llengua en el món del comerç [Language in the world of commerce]. Barcelona: Direcció General de Política Lingüística. [General Directorate of Language Planning].

Catalunya 1993. Resum estudi sobre la incidència de la llengua catalana en el comportament dels consumidors [Summary study on the incidence of Catalan on consumer behavior]. Barcelona: Direcció General de Política Lingüística. [General Directorate of Language Planning].

Chicwick, B. 1991. Speaking, reading and earnings among low-skilled immigrants. Journal of Labor Economics. 9.2.149-170.

Chiswick, B. and P. Miller. 1994. Language choice among immigrants in a multi-lingual destination. Journal of Population Economics. 7.2.119-131.

Chiswick, B and P. Miller 1995. The endogeneity between language and earnings: International analyses. Journal of Labor Economics. 13.2. 246-288.

Church, J. and I. King. 1993. Bilingualism and network externalities. Canadian Journal of Economics. 26.2.337-345.

Colomer, J. 1996. To translate or to learn languages? An evaluation of social efficiency, International Journal of the Sociology Language 121 (in press).

Dagenais, M. and A. van Peeterssen. 1973. L'utilisation du français comme langue de travail au Québec. Possibilités et contraintes économiques. [The use of French as a working language in Quebec. Options and economic constraints]. Commission Gendron, Étude E-22. Québec: Éditeur officiel du Québec. VIII.

Dustmann C. 1994. Speaking fluency, writing fluency and earnings of migrants. Journal of Population Economics. 7.2.133-156.

Fogel, W. 1966. The effects of low educational attainment on incomes : A comparative study of selected ethnic groups. The Journal of Human Resources. 1.2.22-40.

Girard, G., L. Brault, R. Champagne and W. Empey. 1973. La production des universités québécoises et la population de formation universitaire au Québec. [The production of Quebec universities and the university-educated population in Quebec]. Commission Gendron, Étude E-5. Québec: Éditeur officiel du Québec.

Grenier, G. 1984a. The effect of language characteristics on the wages of Hispanic American males. Journal of Human Resources. 19.1.35-52.

Grenier, G. 1984b. Shifts to English as usual language by Americans of Spanish mother tongue. Social Science Quarterly. 65.537-550.

Grin, F. 1990. The economic approach to minority languages. Journal of Multilingual and Multicultural Development 11.1&2.153-174.

Grin, F. 1992. Towards a threshold theory of minority language survival. Kyklos. 45.1.69-97.

Grin, F. 1993a. European economic integration and lesser used languages. Language Problems and Language Planning 17.2.101-116.

Grin, F. 1993b. Minority language promotion: On the practical usefulness of economic theory. In L. Dafis (ed.) Economic development and lesser usedlLanguages: Partnerships for action. Llanbedr Pont Steffan (Cymru/Wales): Iaith Cyfyngedig. 24-49.

Grin, F. 1993c. The relevance of thresholds in language shift and reverse language shift : A theoretical examination. Journal of Multilingual and Multicultural Development. 14.5.375-392.

Grin, F. 1994a. Immigrant and autochthonous language rights: A territorial approach to multilingualism. In T. Skutnabb-Kangas, R. Phillipson and M. Rannut (eds.) Linguistic human rights. Berlin: Mouton de Gruyter. 31-48.

Grin, F. 1994b. The bilingual advertising decision. Journal of Multilingual and Multicultural Development. 15.2&3.269-292.

Grin, F. 1994c. The economics of language: Match or mismatch? International Political Science Review. 15.1.27-44.

Grin, F. 1995a. La valeur des compétences linguistiques: vers une perspective économique [The value of language competence: towards an economic perspective]. Babylonia. 2/95.59-65.

Grin, F. 1995b. La valeur privée de la pluralité linguistique [The private value of linguistic plurality]. Paper presented at the Mercator Seminar "Approches économiques des langues minoritaires". [Economic approaches to minority languages]. Maison des Sciences de l'Homme. Paris, June 1995.

Grin, F. 1995c. L'identification des bénéfices de l'aménagement linguistique : la langue comme actif naturel. [The identification of the benefits of language planning: Language as an environmental asset]. In C. Phlipponneau and A. Boudreau (eds.) Sociolinguistique et aménagement des langues [Sociolinguistics and language planning]. Moncton (N.B.): Centre de recherche en linguistique appliquée. 67-101.

Grin, F. 1996a. Conflit ethnique et politique linguistique [Ethnic conflict and language policy]. Relations internationales [International Relations]. 88. (forthcoming).

Grin, F. 1996b. Economic approaches to language and language planning: An introduction. International Journal of the Sociology of Language. 121 (in press).

Grin, F. 1996c (ed.) Economics of language. (Thematic issue of the International Journal of the Sociology of Language, 121).

Grin, F. 1996d. Minority language and socio-economic status: The case of Italian in Switzerland. Cahiers du Département d'économie politique, University of Geneva. 96.10.

Grin, F. 1996e. The economics of language: Survey, assessment and prospects. International Journal of the Sociology of Language. 121 (in press).

Grin, F. and Sfreddo, C. 1996. Dépenses publiques pour l'enseignement des langues secondes en Suisse [Public spending on second-language teaching in Switzerland]. Report to the Swiss National Science Foundation. Mimeo.

Hocevar, Toussaint. 1975. Equilibria in linguistic minority markets. Kyklos. 28.2.337-357.

Holden, Nigel. 1987. Language barriers as differential constraints on the international behavior of firms. In H. Tonkin and K. Johnson-Weiner (eds.) The economics of language use. New York: Center for Research and Documentation on World Language Problems. 119-139.

Inagaki M., L. Brault, G. Girard, P. Olivier and C. Sabourin. 1973. La présence francophone et l'emploi du français au niveau des cadres du secteur privé au Québec en. 1971 [Francophone presence and use of French in private sector middle management, Quebec 1971]. Commission Gendron, Étude E-16. Québec: Éditeur officiel du Québec.

Jernudd, B. 1971. Notes on economic analysis for solving language problems. In J. rubin and B. Jernudd (eds.) Can language be planned? Sociolinguistic theory and practice for developing nations. Honolulu: The University Press of Hawaii. 263-276.

Lacroix, R. and F. Vaillancourt. 1980. Attributs linguistiques et disparités de revenus au sein de la main d'oeuvre hautement qualifiée du Québec [Language attributes and income differentials in the highly skilled workforce in Quebec]. Québec: Conseil de la langue française.

Lacroix, R. and F. Vaillancourt. 1981. Les revenus et la langue au Québec. 1970-1978 [Language and incomes in Quebec, 1970-1978]. Québec: Conseil de la langue française.

Lang, K. 1986. A language theory of discrimination. Quarterly Journal of Economics. 101. 363-382.

Langevelde, A. van. 1994. Language and economy in Friesland : A first step towards the development of a theory. Tijdchrift voor economiesche en sociale geografie. 85.1.167-77.

Leblanc, M. 1994. Une analyse économique des déterminants de la langue de travail des travailleurs francophones bilingues du Québec [An economic analysis of the determinants of the language of work of bilingual francophones in Quebec]. In Actes du colloque sur la problématique de l'aménagement linguistique [Proceedings of the language planning symposium]. Montréal: Office de la langue française. 569-588.

Long, J. 1977. Productivity, employment discrimination and the relative economic status of Spanish origin males. Social Science Quarterly. 58.3.357-73.

McManus, W. 1985. Labor market costs of language disparities: An interpretation of Hispanic earnings differences American Economic Review. 75.818-27.

McManus, W. 1990. Labor market effects of language enclaves : Hispanic men in the United States. The Journal of Human Resources. 25.228-52.

Ó Cinnéide, M. and M. Keane. 1988. Local socio-economic impacts associated with the Galway Gaeltacht. Galway: Coláiste na hOllscoile Gaillimhe.

Pellenbarg, P. 1993. Language, economy and the regional image: A case study of Friesland. In L. Dafis (ed.) Economic development and lesser used languages: Partnerships for action. Llanbedr Pont Steffan (Cymru/Wales): Iaith Cyfyngedig. 1-23.

Phillipson, Robert. 1992. Linguistic Imperialism. Oxford: Oxford University Press.

Pool, J. 1991. The official language problem. American Political Science Review. 85.495-514.

Pool, J. 1996. Optimal language regimes for the European Union. International Journal of the Sociology of Language. 121 (in press).

Raynauld, A. and G. Marion. 1972. Une analyse économique de la disparité interethnique des revenus [An economic analysis of inter-ethnic income idfferentials]. Revue économique [The Economic Review]. 23.1.1-19.

Raynauld, A., G. Marion and R. Béland. 1969. La répartition des revenus selon les groupes ethniques au Canada. [Income distribution by ethnic group in Canada]. Commission Laurendeau-Dunton, rapport de recherches, 4 vol.

Reimers, C. 1983. Labor market discrimination against Hispanic and Black men. Review of Economics and Statistics. 65.570-579.

Rivera-Batiz, F. 1990. English-language proficiency and the economic progress of immigrants. Economics Letters. 34, 295-300.

Rivera-Batiz, F. 1991. The effects of literacy on the earnings of Hispanics in the United States. In E. Menendez, C. Rodriguez and J. Barry-Figueroa (eds.) Hispanics in the labor force. New York: Plenum Publishers.

Robinson, C. 1988. Language choice: The distribution of language skills and earnings in a dual language economy. Research in Labor Economics. 9.53-90.

Sabourin, C. 1985. La théorie des environnements linguistiques [The theory of linguistic environments]. In F. Vaillancourt (ed.) Economie et langue [Economics and language]. Québec: Conseil de la langue française.

Selten, R. and J. Pool. 1991. The distribution of foreign language skills as a game equilibrium. In R. Selten (ed.) Game equilibrium models. Berlin: Springer Verlag. 64-87.

Sproule, A. and B. Ashcroft. 1993. The economics of Gaelic language development. Glasgow: Glasgow Caledonian University.

Thorburn, T. 1971. Cost-benefit analysis in language planning. J. Rubin and B. Jernudd (eds.) Can language be planned? Sociolinguistic theory and practice for developing nations. Honolulu: The University Press of Hawaii. 223-305.

Vaillancourt, F. 1980. Difference in earnings by language groups in Quebec. 1970. An economic analysis (Ph.D. Thesis, 1978). Québec: Centre international de recherche sur le bilinguisme, publication B-90.

Vaillancourt, F. (ed.). 1985a. Economie et langue. [Economics and language]. Québec: Conseil de la langue française.

Vaillancourt, F. 1985b. Le choix de la langue de consommation. [Choosing the language of consumption]. In F. Vaillancourt (ed.) Économie et Langue. [Economics and language]. Québec: Conseil de la langue française. 209-220.

Vaillancourt, F. 1987. The benefits and bosts of language policies in Quebec, 1974-198. In H. Tonkin and K. Johnson-Weiner (eds.) The economics of language use. New York: Center for Research and Documentation on World Language Problems. 71-91.

Vaillancourt, F. 1988. Langue et disparités de statut économique au Québec: 1970 et. 1980 [Language and socio-economic status differentials: 1970 and 1980]. Québec: Conseil de la langue française.

Vaillancourt, F. 1991. Langue et statut économique au Québec, 1980-1985. [Language and economic status in Quebec, 1980-1985]. Québec: Conseil de la langue française.

Vaillancourt, F. 1992a. English and Anglophones in Quebec: An economic perspective. In Survival: official language rights in Canada. Toronto: C.D. Howe Institute. 63-94.

Vaillancourt, F. 1992b. Language and public policy in Canada and the United States: An economic perspective. In B. Chiswick (ed.) Immigration, language and ethnicity - Canada and the United States. Washington : American Enterprise Institute. 179-228.

Vaillancourt, F. 1996. Le français dans un contexte économique. [French in an economic context]. In J. Erfurt (ed.) De la polyphonie à la symphonie [From polyphony to symphony]. Leipzig: Presses de l'Université de Leipzig (in press).

Vaillancourt, F., R. Champagne and L. Lefebvre. 1994. L'usage du français au travail par les francophones du Québec : une analyse économique. [The use of French at work by Quebec Francophones: an economic analysis]. In Martel, P. and J. Maurais (eds.) Langues et sociétés en contac. [Languages and societies in contact]. Tübingen: Niemeyer (Canadiana Romanica 8). 483-493.

Vaillancourt, F. and M. Leblanc. 1993. La propriété de l'économie du Québec en. 1991 selon le groupe d'appartenance linguistique. [The ownership of Quebec's economy in 1991 by language group]. Montréal: Office de la langue française.

Vaillancourt, F. and L. Lefebvre. 1981. Antécédents familiaux et connaissance de l'anglais chez les francophones du Québec [Family background and knowledge of English among Quebec Francophones]. L'actualité économique. 57.343-358.

Vaillancourt, F. and M. Payette. 1984. Antécédents familiaux et revenu de travail des hommes québécois [Family background and labor income of Quebec males]. L'actualité économique. 59.24-38.