Language Projections for Canada, 2011 to 2036
Chapter 1. Context and methods
Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.
1.1 What projections can and cannot show us
Can an analysis of the evolution of the language characteristics and behaviours in Canada over the past 25 years provide an overview of their possible evolution between now and 2036? If we were to answer yes, this would suggest, on the one hand, that we have a full and detailed understanding of the factors that influence or will influence this evolution and, on the other, that past demolinguistic trends and the mechanisms that produced them are indicative of the population’s future language portrait. However, this is not exactly the case. Of course, in the field of demography—and demolinguistics in particular—the interrelations between the underlying phenomena of demographic change within a society can usually be understood. It is also possible to explain the large processes that affect the structure of populations. Natural increase (births minus deaths); international, interprovincial and intraprovincial migration; and intergenerational and intragenerational linguistic mobility are all factors that allow us to explain the evolution of language communities or groups.
Lachapelle and Henripin referred to these scientifically-oriented projections that deal with a population’s demographic past likely to bind and limit the future as the “predetermined future.” So, while it is possible to apply a certain number of “mobility” assumptions or “transitions,” in a broader sense, to a known initial population structure, nothing guarantees that these “transitions” will repeat themselves in the future. Indeed, many demographic events or transitions are relatively stable or evolve slowly (e.g., a language’s rate of transmission or mortality). Others, however, are more volatile and marked by uncertainty, making it particularly difficult to forecast a population’s evolution in the short and medium terms. These changing phenomena require the development of multiple assumptions for their projection, to the extent that they are likely to affect the demographic change in the projected populations. The (international or intranational) geographic mobility of populations is an example of this. It is sensitive to various economic condition effects and can therefore influence such demolinguistic phenomena as net intergenerational and intragenerational linguistic mobility. Political decisions and action can also influence the evolution of certain demographic factors. The identification of national immigration thresholds, receiving refugees, changes in immigration policies or family-related policies are good examples of this.Note 1
That being said, and as emphasized by Termote (2011, p. 62), despite these cyclical variations, in many cases demographic trends are particularly heavy and difficult to reverse due to the weight of the past and the fact that a population’s initial structure often exerts great inertia. This is especially true when we examine how the age structure of a given language group might influence this group’s future migration models, as older populations, for example, are less likely to migrate.
This study presents the results of demographic projections, not predictions (see the study “Immigration and Diversity: Population Projections for Canada and its Regions, 2011 to 2036”, Statistics Canada 2017a), as it is not within our power to foresee the demolinguistic future of the Canadian population, or of the population of a particular province. As we will see later on, instead, these projections make it possible to examine and shed light on what the demolinguistic future might look like should the developed assumptions and scenarios be true. For example, there is no way of knowing if fertility will increase in the future or what this increase would be. Nor do we know how many new immigrants Canada will accept in 20 years or how these immigrants will be distributed throughout the country. That said, based on the chosen assumptions (low or high fertility, low or high migration, same or different immigration composition, etc.) and the examined scenarios (combinations of assumptions), it is possible to present a plausible range for the situation’s evolution and for the factors likely to influence it, and to quantify the sensitivity of this evolution to the different scenarios.
1.2 Previous language projection exercises
Contrary to the work carried out in the area of demographic projections, little work has been done in the past on demolinguistic projections. Termote is an exception, having led no fewer than five forecasting exercises in almost 20 years (1994, 1996, 1999, 2008, 2011) based on the 1986, 1991, 1996, 2001 and 2006 censuses.
While Maheux (1968) and Charbonneau, Henripin and Légaré (1973) conducted a few forecasting studies on the francophone population of Canada and Quebec in particular, Lachapelle and Henripin (1980) were the first to produce demolinguistic projection results for Canada. Because of the very limited computerized methods and available language data at the time, other than mother-tongue census data, Lachapelle and Henripin were only able to use the information on home language collected during the 1971 Census. As a result, the projected initial demographic structure was limited to four variables: age, region, mother tongue and home language. Furthermore, three regions were chosen: Montréal, Quebec excluding Montréal and Canada outside Quebec.
For their part, Termote and Gauvreau (1988) developed demolinguistic projections for Quebec and its regions using data on home language from the 1971 and 1981 censuses, as the 1976 Census did not include a question on the language spoken at home. Moreover, these authors started from an initial structure characterized by a period of exceptional interprovincial migration, as we will see later in our study.
Prior to the release of Termote’s first forecasting exercise in 1994, three other forecasting exercises were conducted on the Montréal region (Benjamin and Baillargeon 1977; Veltman 1989; Paillé, 1990). It should be noted that Veltman examined a high number of linguistic mobility assumptions that demonstrated the phenomenon’s limited relative influence. Paillé’s work, however, essentially focused on the evolution of the French-mother-tongue population on the Island of Montréal.
While all these exercises were carried out using a traditional projection model based on aggregate data (cohort-component model), Sabourin and Bélanger (2014, 2015) opted for a microsimulation approach. Basing themselves on the work of Bélanger and Caron-Malenfant (2005) as well as Caron-Malenfant, Lebel and Martel (2010), and using the 2006 Census population as an initial structure, they projected mother tongue and language spoken most often at home with a certain number of fertility, mortality, internal- and international-migration, and intragenerational- and intergenerational-language-transfer assumptions and scenarios.Note 2
1.3 Project what, project whom? Which projected language characteristics?
All projection exercises related to language in Canada use a certain number of assumptions and axioms. Projecting the evolution of language characteristics and behaviours, even the language situation, based on census data or, in the case of our study, 2011 National Household Survey data, forces us to acknowledge, from the outset, that these projections are inevitably limited by the available information on the topic, even though it is already very detailed in Canada compared with international practices.
Not only does the Canadian census (or the 2011 NHS) include seven questions or sub-questions on language, but with regard to language statistics, Canada is one of the few countries in the world to collect information on each of the “fields” recommended by the United Nations: a) mother tongue, b) usual language (defined as the language currently spoken, or most often spoken, by the individual in his or her present home), and c) ability to speak one or more designated languages (United Nations, 2009).Note 3
The question on mother tongue, which has remained pretty much the same since the 1941 Census, was worded as follows in the Census and the 2011 NHS:
“What is the language that this person first learned at home in childhood and still understands?
If this person no longer understands the first language learned, indicate the second language learned.”
In 2011, the question on the language spoken most often at home was worded as follows:
“What language does this person speak most often at home?”
Finally, the question on knowledge of English and French has changed little since the 1971 Census. In 2011, it was worded as follows:
“Can this person speak English or French well enough to conduct a conversation?”
Four possible answers are given: English only, French only, English and French, Neither English nor French. This is a subjective question, the answer to which depends on the assessment of the respondent—and undoubtedly in many cases one person in the household fills out the questionnaire for everyone.
Despite this abundance of information on this topic in Canada, having data on the languages used at work (main or secondary language) or the secondary use of one or more languages at home, for example, does not necessarily allow their use for projection purposes. This would also be the case if we had detailed data on language use in the public sphere (in stores, services to the population, etc.). This is largely because of the complexity of integrating them with a projection model based on a multitude of factors and characteristics that interact with one another. In other words, the number of interdependent transitions is generally the main factor that causes a problem, as well as the availability of data on each of these transitions.Note 4 For example, to project the secondary use of a language at home, we need to better understand all the factors and conditions that cause an individual to start (or stop) speaking this language as a secondary language and be able to model and predict each of these factors that may interact with one another. These factors may include such elements as language use outside the home (at work, at school, with friends, etc.), migration from one province to another, change of employment, meeting a spouse, language spoken as a secondary language prior to emigrating to Canada, and so on. Furthermore, the changes to the multiple responses on these practices over time fluctuate and are difficult to foresee.Note 5 Lastly, the language(s) used at work in areas where there is contact between different languages is (are) generally associated with specific industry sector and profession characteristics; such information is very difficult to project and not available in the Demosim model.Note 6
Regarding our initial question, i.e., what we are projecting, it would be arrogant to believe we can project the overall “language situation” from now until 2036 based on the considered transitions and initial structure used. Indeed, “language situation” implies a set of complex interlinguistic relationships and dynamics that cannot fully be taken into account by census data or large-scale surveys. The same holds true for the idea of knowing whom we are projecting. In this regard, the choice of terms and criteria used to define the language communities or groups is always partially arbitrary. Their use must therefore take into account the fact that their generally accepted meaning is not completely neutral, and that they do not make it possible to fully grasp the complexity of the reality they serve to define.
In our study, we repeatedly use the notion of language group. This many-faceted notion is necessarily limited, even though it is heuristic in nature. For example, we refer to the people reporting the same mother tongue and living within a given territory as a specific language group because of this common trait. The same is true if we use the criterion of language spoken most often at home (main home language) instead, or that of the first official language spoken (FOLS), which we will discuss below. Some might prefer to define a group based on the predominant use of one language in the public sphere, or even based on the ability to conduct a conversation in a given language. This is the criterion used by the International Organisation of La Francophonie to define the world’s francophone communities.Note 7
Lachapelle (1991) defines a language group as a grouping of human beings who share certain linguistic affinities. We could also talk about a group of people who share certain specific characteristics associated with language or with a particular language. We know these “linguistic affinities”Note 8 may be based on a multitude of criteria—demographic, sociological, political, cultural, etc.
In certain cases, and for simplification purposes, groups are even defined based on criteria they do not share with other communities. For example, a growing number of people’s mother tongue or main home language is neither English nor French, the country’s official languages. In this sense, the only linguistic affinity, so to speak, shared among them is the fact that neither English nor French is their mother tongue or home language. When the focus is on English or French, and because almost 200 languages are reported in the census, common practice has generally been to refer to those other languages as “non-official languages.” In French, the term langue tierce is increasingly used to refer to languages other than English or French. This group of languages comprises both Aboriginal and immigrant languages.Note 9
In addition to the terms “francophone” and “anglophone”, the term “allophone” is also widely used, particularly in Quebec, to refer to this “other” language group. This term was created during the Commission of Inquiry on the Position of the French Language and on Language Rights in Québec (1968–1972). This neologism, from the Greek allos (“other”) and phônê (“voice,” “sound”), usually refers to any person whose mother tongue or main home language is a language other than English or French.Note 10 The term has gradually made its way into political, scientific and even vernacular discourse, especially in Quebec, but it is also perceived by some as being tinged with essentialism (“being” allophone as opposed to “being” anglophone or francophone).
Our study will therefore intentionally avoid talking about anglophone, francophone or allophone communities or groups. Instead, we will talk about French-, English- or other-language groups, or official-language or non-official-language groups, and we will always specify the criterion to which these terms refer. Specifically, we are projecting, from now until 2036, the numbers and the relative shares for the groups whose mother tongue or language spoken most often at home is French, English or another language (also referred to as a third language).Note 11 We will also project the number and percentage for populations whose FOLS is English or French, as well as populations able to conduct a conversation in English or French.
1.3.1 Language questions used for projection purposes
Why project populations by mother tongue, main home language and FOLS?
First, because of their availability since the beginning of the 20th century and their more widespread use since the 1960s,Note 12 data on mother tongue have made it possible to study the evolution and the situation of different language groups throughout the country. In fact, for many years, the notion of mother tongue has been the main criterion used to designate French-, English- or other-language groups in Canada and in Quebec in particular. This choice stems from historical considerations. Not only have mother-tongue-based statistics long had the advantage of being approximately comparable for more than half a century, but section 23 of the 1982 Charter of Rights and Freedoms also uses the criterion of mother tongue as one of the conditions allowing parents to send their children to primary or secondary school in the minority official language.
However, the mother tongue of individuals gives no information on their use of the languages. In its 1967 report, the Royal Commission on Bilingualism and Biculturalism (Laurendeau-Dunton) stated that the census data on mother tongue were a generation behind the facts and suggested that information on the usual language of Canadians be gathered. The Commission called for the addition of a question to the census, which “would deal precisely with the main language of each Canadian, and would enable us to tell which language he speaks most often at home and at work.”Note 13 In the 1971 Census, Statistics Canada acted on this suggestion by adding a question on the language spoken most often at home. Essentially, this choice resulted from the fact that the Commission believed it would be useful to obtain information on the current use of the languages, which would complement the information obtained through the question on the first language learned in childhood (Lachapelle 1991).
The question on the language spoken most often at home, added to the census questionnaire in 1971, therefore gathered information on the main language used in the family or private sphere. Also, combining it with the information on mother tongue made it possible to study the transfer phenomenon (or language shifts), that is, the adoption of a language other than the mother tongue as the main home language. Furthermore, because the language spoken most often at home is generally the language transmitted to children (Lachapelle 1991; Marmen and Corbeil 2004; Lachapelle and Lepage 2010), adding it to the census would make it possible to study and measure the phenomena of intergenerational and intragenerational linguistic mobility.
For its part, the question on knowledge of either English or French—also a census question since 1901—made it possible to measure English–French bilingualism and, during the first decades of the 20th century in particular, knowledge of English, the language of trade and business, among French Canadians and new immigrants with a mother tongue other than English or French (Houle and Cambron-Prémont 2015).
Because of the surge and diversification in international immigration since the mid-1980s, with more and more immigrants coming from non-European countries and a very high majority having a mother tongue other than English or French (other mother tongue),Note 14 the question arose as to what the approach should be with regard to these new citizens to determine the official language in which they are likely to request services from and communicate with the federal government. Several government actors and stakeholders began to wonder about these individuals’ “first official language.” The longer these immigrants stay in Canada, the more they tend to use English or French at home or at work, and less than 2% of the Canadian population reports no knowledge of English or French from one census to the next.Note 15
At the request of the federal government, the Treasury Board in particular, in 1989 Statistics Canada defined different variants of the notion of “first official language spoken (FOLS)” to estimate the francophone (Canada and each province outside Quebec) and anglophone (Quebec) minority populations. Statistics Canada (1989) stated that “depending on the assumptions used and the order in which the three linguistic variables [knowledge of the official languages, mother tongue and language spoken most often at home] of the  census are taken into account, a number of values may be obtained.”
In December 1991, the method, called method I, adopted by the federal government was included in the Official Languages (Communications with and Services to the Public) Regulations. Section 2 of the Regulations describes the method used to determine the FOLS, namely the first of the two variants presented in Population Estimates by First Official Language Spoken (Statistics Canada 1989), “which method gives consideration, firstly [and successively], to knowledge of the official languages, secondly, to mother tongue, and thirdly, to language spoken in the home.”Note 16
As mentioned in this publication, the notion of FOLS comprises two specific dimensions. On the one hand, the epithet “spoken” refers to the ability to conduct a conversation in the first language in question. On the other hand, the adjective “first” has two distinct meanings. Among the population whose mother tongue is English or French, it designates the first language learned in childhood (mother tongue). However, among people whose mother tongue is neither English nor French, the term first official language instead designates their primary official language, the one that they know best at the time of the census or the one spoken most often at home.
Application of the FOLS to the Quebec situation certainly has heuristic value when measuring the language that the other-mother-tongue population is most likely to use in the public sphere, or is more comfortable using to communicate. So, during the Survey on the Vitality of Official-Language Minorities (SVOLM) conducted by Statistics Canada in 2006 in collaboration with 10 or so federal government departments and agencies, it was discovered that of the adults in the census metropolitan area of Montréal, where an overwhelming majority of Quebecers with English as the FOLS is concentrated, 94% of the English FOLS population reported being more comfortable communicating in English than in French, whereas 95% of the French FOLS population reported being more comfortable speaking in this language. Projecting the population up to 2036 based on the criterion of FOLS can therefore provide useful and relevant information, provided that, for a significant portion of the other-mother-tongue population, this criterion does not correspond to main home language but rather to the official language with which it is more comfortable.
1.4 What the Demosim microsimulation model can do
Demosim is a microsimulation model created and maintained by Statistics Canada and used to make demographic projections of many characteristics of Canada’s population, for example, generation status, visible minority group, Aboriginal identity and mother tongue.Note 17
Like traditional projection models (i.e., cohort-component models), this model uses the components method; however, instead of projecting aggregates of individuals or cohorts, it projects one individual at a time (microsimulation). The base population of the current Demosim version was derived from 2011 National Household Survey (NHS) microdata files, adjusted to take into account net census undercoverage and incompletely enumerated reserves in the NHS.Note 18
Start of text box
The data for the base population are taken from the 2011 National Household Survey (NHS), adjusted for net undercoverage. The three language variables projected using Demosim (as well as the variable derived from FOLS) were also asked in the 2011 Census. However, other than the three language variables, the 2011 Census, although sent to every home in Canada, included only a few demographic variables such as sex, date of birth (age) and marital status. Unlike the 2011 NHS, the census that same year did not include any questions on immigrant status or year of immigration, internal migration, highest level of education, etc. The 2011 NHS was thus used as the base population for the details it provides on the composition of the Canadian population at the beginning of the projection period.
End of text box
Cohort models have generally been used to project only one language variable at a time, either mother tongue or language spoken most often at home. However, with the current version of Demosim, we can project three language variables simultaneously and generate a fourth language variable during the projection. Mother tongue, language spoken most often at home, knowledge of official languages and FOLS are the four variables projected by Demosim.
Microsimulation makes it possible to project these language variables, which, thanks to the available data sources and the estimation methods used, are supported by numerous other sociodemographic variables. The projection parameters for changes in each of the three language variables are calculated, for the most part, using logistic regressions (binomial and multinomial), which take into account the effect of many sociodemographic (schooling, age, generation status, etc.) variables that are projected at the same time.
In our study, the approach to project the language variables comprises the following elements:
- Two sets of models enable the initial assignment (or transmission) of language characteristics to newborns based on the mother’sNote 19 characteristics (intergenerational linguistic mobility), namely mother tongue, language spoken most often at home and knowledge of official languages. Mother tongue and language spoken most often at home are simultaneously transmitted to the child through a first set of models, whereas knowledge of official languages is assigned through the second set of models.Note 20
- Two other sets of models measure individual language changes from birth to age 50 (intragenerational linguistic mobility), i.e., changes in the language spoken most often at home and changes in the knowledge of official languages.
- These four series of models are stratified by region of residence, immigrant status, language spoken most often at home and, if applicable (for transitions), mother tongue and knowledge of official languages. Stratification generates more homogeneous subpopulations and enables a greater number of interactions between these subpopulations and the models’ independent variables.
- Once assigned to children at birth, mother tongue does not change during an individual’s life.
- Language spoken most often at home and knowledge of official languages can change over a lifetime, producing language transfers (or shifts), proficiency in the official languages, English–French bilingualism or unilingualism.
- The independent variables taken into account by the models include age, sex, education level, region of residence, English–French bilingualism, mother tongue, generation and, specifically for immigrants, duration of residence in Canada, age at landing and region of birth. These variables determine the parameters for the linguistic mobility of each subpopulation defined at the stratification stage.
- The intrinsic correlation between the language variables is taken into account by the stratification by language characteristics and the inclusion of mother tongue and/or English–French bilingualism in the list of independent variables of each initial assignment model (transmission) and transition model.
- Demosim calculates and updates the FOLS during the projection based on knowledge of official languages, mother tongue and language spoken most often at home.
The microsimulation approach therefore simultaneously and consistently projects a very large number of characteristics, including four language characteristics, for each individual. This approach captures changes in the population’s demographic composition that occur during the projection, and takes into account the differential behaviours of individuals based on their characteristics.
The entire estimation mechanism for linguistic mobility is made possible by access to 2011 NHS microdata on the one hand (see Box 2) and, on the other hand, by the availability to Statistics Canada of record linkages for the 2001 and 2006 censuses as well as record linkages for the 2006 and 2011 censuses. The NHS data are used to assign language characteristics to newborns, and the linkages make it possible to identify the main individual language change flows (e.g., for a given individual, learning a language, transitioning toward a new main home language, etc.). The 2001–2006 linkage made it possible to model the changes in the language spoken most often at home, whereas the 2006–2011 linkage was used to model changes in the knowledge of official languages (see Box 3).
Start of text box
The Methodological Document on the 2011 Census Language Data (Houle and Corbeil 2013) brought to light the fact that compared with the 2001 and 2006 censuses (long-form questionnaires), the 2011 Census underestimates the relative share of mother tongues and languages other than English or French spoken most often at home. In 2011, basic language information was collected with a short questionnaire that had only 10 questions, instead of the 53 questions asked in the long-form questionnaire in 2006, which had this unexpected effect on the answers to the questions on mother tongue and language spoken most often at home. However, knowledge of official languages does not seem to have been affected by the questionnaire change, nor was the derived variable of FOLS.
As the NHS language counts were weighted to reflect the 2011 Census counts, the NHS data on mother tongue and language spoken most often at home also produce numbers that cannot be compared with previous censuses. Though the NHS, which is used as the base population for Demosim, was corrected for net undercoverage, it was not corrected to make the language variables comparable with previous censuses. However, we made sure that the projection of language variables was consistent with past trends, regardless of the bias caused by the questionnaire change between 2006 and 2011 (see Box 3).
To compensate for the comparability problems, we examined the changes observed between 2001 and 2006. This approach stemmed from the question we asked ourselves of whether the 2011 NHS could be used without risk to estimate the parameters of the transmission of language characteristics from mothers to newborns. These parameters were calculated with the 2006 Census and the 2011 NHS, respectively, and the results showed there was no significant difference between the two: the tests conducted using either parameter set showed no major difference in the projected numbers for the different language groups.
End of text box
Start of text box
The transition rates for the language spoken most often at home and knowledge of official languages were estimated using regression models. Two separate census linkages were used for this purpose: a linkage between the 2001 and 2006 censuses, and one between the 2006 and 2011 censuses. For the transitions in the language spoken most often at home, we used the linkage between the 2001 and 2006 censuses and not the one between 2006 and 2011 because of the gaps in the comparability of the results for this variable between 2006 and 2011.
However, the linkage between the 2006 and 2011 censuses was used to estimate the transition parameters for knowledge of official languages, despite the questionnaire difference. This linkage has two major advantages compared with the linkage between the 2001 and 2006 censuses. First, the linkage method and quality are superior: the resulting linkage rate was 70%. Second, because the 2011 Census (short-form questionnaire) had the three language variables of interest to us, we were able to link the 2006 long-form questionnaire and the 2011 short questionnaire, which made it possible to considerably increase the linked population size to 4.5 million people. A test with the linkage between 2001 and 2006 confirmed the soundness of this choice.
End of text box
The linkages make it possible to distinguish many single flows of linguistic mobility (intra- and intergenerational, acquisition or loss of knowledge of an official language) made up of a combination of two origin language characteristics observed during a given census and a destination language characteristic observed five years later at the next census.Note 21 Linkages make it possible to isolate all the gross mobility flows and disaggregate the population into subgroups that may display their own language practice behaviours. As Termote said so well, “the more the population is disaggregated into homogeneous categories, the more reliable the forecast will be,” [translation] (Termote 2011, p. 71), which is perfectly relevant for subpopulations defined by their language characteristics.
Intragenerational mobility parameters were estimated using 20 regression models for the mobility of language spoken most often at home representing 90% of all transitions, and 28 models for the mobility (acquisition or loss) of knowledge of official languages making up 97% of all individual transitions.
What holds true for language changes is also true for the assignment of language characteristics to children at birth (transmission). Mother-child intergenerational flows were disaggregated much like the language change flows, and for the same reasons. Modelling for these flows more or less involves the same independent variables as the change models, which ensures consistency between the two forms of mobility. Overall, the parameters for the transmission of home language and mother tongue were estimated using 10 regression models, and those for initial knowledge of official languages, with 12 regression models.
Compared with the cohort method, microsimulation considerably increases the analytical potential of projections, especially the ability to target the evolution of language characteristics in certain subpopulations (e.g., French-speaking immigrants living outside Quebec) or certain phenomena (e.g., language transfers). Microsimulation also offers more possibilities and flexibility for the development of assumptions and scenarios on future population trends.
1.5 Scenarios analyzed in our study
For our language projections, 16 scenarios were analyzed, including nine base scenarios largely modelled on those presented in the report Immigration and Diversity: Population Projections for Canada and its Regions, 2011 to 2036 and six scenarios prepared specifically for this language projection exercise which will support certain specific aspects of the analysis.
The assumptions of the nine base scenarios analyzed in our study are presented in Table 1.1.Note 22 These nine scenarios have a different reference period for the measurement of internal migration than the scenarios used for the diversity and immigration projection. In the diversity and immigration projection, internal migration was derived from 2001 and 2006 census data and the 2011 National Household Survey (NHS). In the language projections, internal migration was derived from the 2006 Census and the NHS only. The reason for this is, from 1996 to 2001, internal mobility was exceptional for Quebec’s French-speaking populations. Indeed, during this period, the economic context prompted a large number of French-mother-tongue Quebecers to leave the province and settle mainly in Ontario, Alberta and British Columbia.Note 23 This exceptional internal migration pattern from 1996 to 2001 led us to develop a specific scenario for it (see below).
The reference scenario (Scenario 1.1) and the low and high immigration scenarios (1.3 and 1.4) are used extensively in this report. The base scenarios also include four other scenarios that differ from the reference scenario by only one component: immigration level (Scenario 1.2, zero immigration after 2016), geographic distribution of immigrants (Scenarios 1.5 and 1.6) and country of origin of immigrants (Scenario 1.7). These last four scenarios are analyzed in Chapter 3. Immigration is one of the demographic factors that has a major influence on the demographic evolution of language groups in Canada (see Chapter 2). The zero immigration after 2016 scenario is only theoretical and implausible, but it makes it possible to illustrate the “absolute” effect of the international immigration phenomenon on the evolution of Canada’s language portrait.
Two additional scenarios have an effect on not only the international immigration component, but also the fertility level and life expectancy, resulting in (Scenarios 1.8 and 1.9) a weak growth scenario and a strong growth scenario. These two scenarios are also analyzed in Chapter 3.
Scenarios 1.10 and 1.11 illustrate the effect of changes in the internal migration pattern. Scenario 1.10 illustrates that, relative to the surrounding periods, internal migration from 1996 to 2001 was characterized by an increase in interprovincial migration of French-speaking people from Quebec to the rest of Canada. This scenario, which is favourable to French-speaking populations outside Quebec even though it was exceptional in the 25 years before 2011, has a certain heuristic value given that it occurred just 15 or so years prior; it therefore does not seem impossible that it would repeat itself under certain conditions. Scenario 1.11 incorporates internal migration for the periods 1996 to 2001, 2001 to 2006 and 2006 to 2011.
Note that “the reference scenario is designated as such not because of its better predictive capacity (…), but because it is a central scenario on which the other scenarios were constructed.” (Statistics Canada 2017a, p. 19) The reference scenario combines, among other things, an average immigration level of 8.3 immigrants per 1,000 population,Note 24 a provincial or territorial distribution of new immigrants upon arrival (representative of the distribution observed between July 2010 and June 2015), medium emigration, a medium fertility rate of 1.67 children per woman, medium growth in life expectancy, internal migration models representative of the average observed during the 2001-to-2006 and 2006-to-2011 periods, changes in language spoken most often at home observed during the 2001-to-2006 period and, lastly, changes in the knowledge of official languages observed during the 2006-to-2011 period. Furthermore, note that the transmission or assignment of language characteristics to newborns, i.e., mother tongue, language spoken most often at home and knowledge of official languages, is based on the 2011 NHS.
Five other scenarios, based on the assumptions of the reference scenario, were developed specifically for the language projections. These scenarios are analyzed in Chapters 3 and 4.
These scenarios differ from the reference scenario only by one component: the transition rate for the knowledge of official languages (Scenarios 2.1, 2.2 and 2.3), the language composition of immigration (Scenarios 2.4) and the intergenerational transmission rate of French outside Quebec (Scenario 2.5). Table 1.2 presents a summary description of each of the five scenarios.
The purpose of the English–French bilingualism scenarios (Scenarios 2.1 to 2.3) is to test the effect that certain changes in acquisition and retention rates (which could result, for example, from an increase in the number of students registered in a French-as-second-language program and from measures designed to promote a higher maintenance of second language skills) could have on the evolution of English–French bilingualism in Canada between now and 2036. With scenario 2.1, we first examined the assumption under which learning rate of the other official language could double among children aged 5 to 14 in Canada outside Quebec. We know that bilingualism is growing during this period of life, and doubling the odds of becoming bilingual, or at least increasing them, is not in itself totally unrealistic.
We then asked ourselves, with Scenario 2.2, to what extent maintaining learned bilingualism among English-speakers after age 18 could contribute to raising the entire Canadian population’s level of bilingualism. Lepage and Corbeil (2013) showed that bilingualism peaks when young English-speakers outside Quebec attend school, but this skill then erodes as they age. Certain measures or mechanisms could undoubtedly encourage more young English-speaking adults in Canada outside Quebec to maintain their English–French bilingualism. Chapter 5 discusses the significant increase in French-immersion program attendance in Canada outside Quebec, which tends to lead to higher retention of French skills than attending regular second-language programs.
Scenario 2.3 combines hypotheses developed specifically for scenarios 2.1 and 2.2 presented above. It is therefore a scenario that is generally favorable to the growth of bilingualism in the country.
Scenario 2.4 is a simulation of the number of French-speaking immigrants (by first official language spoken) required each year from 2017 to 2036 to hold the demographic weight of minority French-speaking populations in each province constant at the 2016 level. We calculated the number of immigrants needed to stop the weight of the French-speaking population, non-immigrants and immigrants combined, from declining starting from a reference point in time. As immigration from 2011 to 2016 is already complete and has been integrated into Demosim, we began our simulation as of 2017.
Scenario 2.5 refers to changes in behaviour in the French-speaking populations in Canada outside Quebec. We know that language transfers to English are significant in these populations and that transmission of French to children is incomplete (see Chapter 2). If the French-speaking population in Canada outside Quebec had almost complete transmission rate to French, what effect would this have on the demographic dynamics of their community? An almost complete transmission level equals the transmission rate observed among endogamous couples of over 90%.Note 25 This scenario is purely theoretical, as such behaviours change very slowly. Nonetheless, it makes it possible to estimate the importance of the phenomenon of intergenerational linguistic mobility on the evolution of the French-language population in Canada outside Quebec by 2036.