Introduction

1In Canadian history, many sociopolitical conflicts have arisen from the coexistence of two different peoples in a single land. For instance, one can think of Canada's Conscription Crisis in 1942, its October Crisis in 1970, or its failure to conclude the Meech Lake Accord in 1990.2 Rival nationalism is often called upon to explain these conflict situations between French and English Canadians. According to sociologists Bourque and Duchastel (1996, 315), until 1960 two nationalisms clashed with one another: that of the French community, based on the French-Canadian "race" and the Roman Catholic faith, and that of the English community, based on the Anglo-Saxon "race" and the Protestant faith. With the changeover to a welfare state in the 1960s, these nationalisms were replaced by a Quebec nationalism and a Canada-wide nationalism. Despite the numerous disputes between francophones and anglophones in Canada, Canadian prime ministers have always laid emphasis on the benefits of having people from French and English backgrounds living side by side in one country. William Mackenzie King, for example, stated in the Parliament of Canada in 1942:

1 In the whole history of the world, no two peoples have ever lived so long in such close association with so little friction as those of French and British descent in Canada. In a century of political union we have built a nation which stretches across half a continent. We have conquered the wilderness. We have achieved great material progress. We all have become deeply attached to one common homeland.

Il n'est pas dans l'histoire du monde deux autres peuples qui aient vécu aussi longtemps en étroite association et avec moins de conflits que les peuples d'origine française et anglaise au Canada. Dans un siècle d'union politique nous avons édifié une nation qui couvre la moitié d'un continent. Nous avons conquis la solitude. Nous avons réalisé un grand progrès matériel. Nous nous sommes tous profondément attachés à une patrie commune.

2This excerpt (1) which celebrates the association of francophones and anglophones in Canada, was paradoxically taken from a political speech delivered in a national crisis situation in 1942, during which the unity of the country was challenged by a clash between these two speech communities. Moreover, in comparing the English quotation with its French translation, one notices a translation shift ("wilderness" translated by "solitude") which creates two different images of the same event. These contradictions merit closer examination, and we will here investigate the strategies put forward by the federal government in Canada to protect its institutional discourse.

3In this paper, the following question will be considered: in Canada, what role has the translation of political speeches been playing in the construction of a discourse on Canadian unity? The translated speeches investigated were delivered in national crisis situations in the twentieth century. With such a history-related corpus, where the translated speeches were crafted to deal with acute tensions between the francophones and the anglophones of this country, it will be possible to contrast the slightly different messages sent to these speech communities when serious and controversial issues have taken place. Moreover, it is hoped that ideological translation shifts will help to reveal the value system of the Canadian society at given points in contemporary history.

Corpus and theoretical framework

4The term "political speech" is here defined as a formal talk delivered with a political purpose to a specific audience. The umbrella term "political discourse" includes not only political speeches, but also all forms of political texts. The corpus studied here comprises seven speeches, delivered in both French and English in national crisis situations. These speeches fall under the category of "addresses to the nation," which have been used throughout Canadian history when the unity of the country was at stake, or when the prime minister wanted to speak to all his fellow Canadians at the same time. Addresses to the nation are usually pre-recorded in French and in English, and then simultaneously broadcast on radio or on television. The speeches in the corpus include:

• one by William Mackenzie King during the Conscription Crisis in 1942;
• three by Pierre Elliott Trudeau at the time of the October Crisis in Quebec in 1970, when the Parti Québécois came to power in 1976, and during the first referendum on Quebec independence in 1980;
• Brian Mulroney's speech in 1990 when the Meech Lake Accord was defeated;
• and finally, two speeches delivered by Jean Chrétien, before and after the second Quebec referendum in 1995.

5As a basis for argumentation, let us state that the translation shifts in this corpus will presumably be related to the Canadian federal government's institutional ideology. This hypothesis is based on existing studies of the translation of political discourse in Canada. For instance, both Charron's research (1997) on the translations of Lord Durham's Report and our own research on the translation of political speeches in Quebec and in Canada (Gagnon 2002) seem to indicate that translations of Canadian political discourse reflect the ideology of the translating institution. Hence, in Canada, the translation of political discourse plays a major role in the reproduction of ideologies and contributes to certain discourses as promoted by different social groups.

6As stated by Chilton (1997, 181): "What discourse analysis can make clear is that 'nations' and 'ethnic identity' are not natural or universal phenomena that have simply 'resurfaced' after being suppressed. They have to be constructed and promoted through discourse." The notion of discourse is quite important here. Indeed, a discourse is a way of speaking or writing which represents the attitude of a social group or its expression with respect to certain areas of social life (Hatim and Mason 1997). Ideology and discourse are closely related concepts, ideology meaning a worldview based on the values and interests of a person, a group of people or an institution (Hatim and Mason 1997). Van Dijk's views on ideology (1998) differ slightly: for this scholar, ideologies are acquired through a socialization process, meaning that an individual's ideology is only the expression of a group's ideology at the personal level. Ideologies are not exclusively found in circumstances of hegemony, but they certainly are easier to pinpoint in such contexts. For instance, in a crisis situation, power struggles are at peak level, and divergent ideologies are strongly reflected in the confrontation.

7The present research is based on a theoretical model of Critical Discourse Analysis (CDA), studying ideologies and institutional discourses. As remarked by Fairclough (1989, 40), power struggles play a prime role in the process of discourse reproduction or transformation. He also observes that when power struggles are "stable," the reproduction of ideologies is carried out in a conservative way. However, when the situation becomes unstable, such as the one in this corpus, the discourse must be transformed if it is to remain in a dominant position.

8Although it has not been part of the traditional toolkit in historiography, discourse analysis is nonetheless an important asset for the historian, as argued by Struever (1985, 250):

Investment in discourse analysis technique... is an economical expenditure for the historian.... It functions as a tool of inquiry in the traditional task of interpretation of source, the exploitation of the archive of pertinent discourses that the historian uses to reconstruct the past. It is a formalist project that promises direct access to significant social process; the formal description of the functioning of a discursive practice is at once the description of the structures and processes of social action.

9Critical or not, discourse analysis provides an interesting framework with which to look at historical texts, and it helps to understand the social, economic, linguistic, and political issues in such texts. Moreover, Critical Discourse Analysis offers another beneficial feature: its transdisciplinarity. Indeed, CDA attempts to bridge the gap between social sciences and textual analysis, as explained by Fairclough (2003, 2–3):

My own approach to discourse analysis has been to try to transcend the division between work inspired by social theory which tends not to analyse texts, and work which focuses upon the language of texts but tends not to engage with social theoretical issues.

10One cannot help but relate this statement to translation studies' own dichotomy between postmodern or cultural studies approaches on the one hand, and descriptive or linguistic approaches on the other (see Chesterman and Arrojo 2000). The translation scholar Calzada Pérez (2001) suggested that the solution to this "clash" could lie in translation studies research based on CDA. The present work should be considered as another contribution to such a theoretical/methodological pathway.

11This case study will be structured as follows. First, the concept of translation shifts will be discussed with reference to Chesterman's taxonomy (1997). Second, translation shifts will be analyzed using CDA as a theoretical framework. Third, and as a conclusion, this paper will briefly go over some of the causes and strategies related to the shifts studied.

Description of translation shifts

12Among the translation scholars who have looked at the translation of political discourse, (including Baumgarten 2001, Calzada Pérez 2001, Gagnon 2003, Hatim and Mason 1991, and Schäffner 2003), many have chosen a linguistic typology based on text linguistics, on critical discourse analysis, or on functional grammar. Examples of textual markers studied in translated political discourses are transitivity (Calzada Pérez 2001), cohesion (Hatim and Mason 1997, 143ff), metaphors (Al-Harrassi 2001) or lexical choices (Schäffner 2003). The results obtained using these typologies are of the utmost importance for the study of translated political discourse, since they represent the first attempts at describing translation shifts in political discourses. However, two problems arise with these classifications. First, they do not explicitly refer to certain issues specific to translated texts, such as omission/addition, explicitation/implicitation, or the translator's visibility. In fact, until now, most translation scholars working on translated political discourse have adapted linguistic typologies to their work in translation studies. These typologies were not primarily developed to account for translational phenomena. Second, only targeting linguistic markers traditionally used to find ideologies in texts might result in overinterpreting the ideological relevance of these markers.

13For these reasons, Chesterman's taxonomy (1997, 87–112) will be used here as a test case study to describe shifts in translated political speeches.3 Chesterman entitles his classification "translation strategies," but it will here be referred to as "translation shifts" in order to avoid confusion. Indeed, there is a general lack of consensus in translation studies on the concept of translation strategy (Molina and Hurtado Albir 2002). It is argued here that translation strategies go beyond the description level of analysis, since they help to explain the translator's behaviour. We do not agree with Chesterman's statement that strategies are text-based. Rather, strategies are procedures used by translators to solve problems (Molina and Hurtado Albir 2002), and they are inferred from texts. In contrast, a "shift" is text-based, and it should be taken as a difference in meaning or in structure between a translation and its original. However, Schäffner's research (1997) has shown that with texts produced in multilingual settings, it is quite difficult to differentiate source and target texts, and multilingual institutions often present their translated political discourses as "originals." The speeches in this corpus are also presented as bilingual originals by the Government of Canada, and the concept of "shift" will here refer to a difference in meaning or structure between two language versions of a text.

14Chesterman's heuristic model includes three levels of analysis: syntactic, semantic, and pragmatic. Shifts at the syntactic level modify the textual form (such as clause structure change); shifts at the semantic level modify textual meaning (as does synonymy); and those at the pragmatic level modify the textual message itself (for example, visibility change). Let us emphasize that these three groups overlap, and that shifts at the pragmatic level often involve shift(s) at the syntactic and/or semantic level. Also, different shifts can occur at the same time. As Chesterman himself states, his model is a useful conceptual tool to discuss translations and translation choices. For the study of translated political speeches in particular, Chesterman's classification is general enough to target some of the traditional linguistic markers used to unveil ideology in translated political discourse, yet it is not restricted to these features.

15Although Chesterman's model is quite comprehensive and well structured, there seems to be a lack of coherence between the first two levels of analysis and the third one. Indeed, at the syntactic and semantic levels, the shifts mainly fulfill a description function; that is, they label the changes occurring between source text and target text. At the pragmatic level, the functions of the shifts are rather mixed, some describing the linguistic features of translations, such as information change or explicitness change, while others present a deeper level of analysis, such as cultural filtering, interpersonal change or illocution change. In fact, the latter are examples of strategies as defined above. To deal with this inconsistency, we have slightly modified Chesterman's work: interpersonal change and illocution change have been removed from the classification, and the concept of cultural filtering has been replaced with the concept of adaptation,4 inspired by the book Translation Terminology (Delisle et al. 1999, 114). Hence, adaptation means a shift "where the translator replaces a sociocultural reality from the source language with a reality specific to the culture of the target language in order to accommodate the expectations of the target audience." Unlike Chesterman's concept of "cultural filtering," "adaptation" does not necessarily entail the ideological issues related to Venuti's work (1995). The reader will appreciate that the slight changes to Chesterman's work introduce a classification focusing almost exclusively on description features. The modified taxonomy is briefly detailed in Appendix 1.

16In order to describe the translated speeches in linguistic terms, the French and English versions of the speeches were first analyzed, and shifts were identified at the textual surface. A corpus-based study was carried out as well, using the monolingual concordancer WordSmith Tools 4.0 and the bilingual concordancer Multiconcord 1.53. These software systems were useful for locating patterns among the identified shifts. Finally, the shifts were labelled using Chesterman's modified taxonomy.

17For this case study we retained only shifts related to the lemma "Canada" (such as Canada, Canadian[s], Canadien[s], Canadienne[s]), and the words surrounding it. According to Duchastel and Armony (1993), the lemma "Canada" is a "macropolitical entity." The macropolitical entity designates great societal ensembles (the concepts of nation, people, government, country, and so on) and it is one of the key social components in the production of political discourse, since political discourses portray the world in a way that reflects "us vs. them" frameworks. Another argument in favour of focusing on "Canada" lies in Bourque and Duchastel's study of Canadian political discourse (1996). Indeed, these two scholars have established a clear link between the frequency of the lemma "Canada" and the production of a national discourse on identity. Other studies are related to the translation of the word "Canada" (Delisle 1993, 340–341; Mossop forthcoming), but for reasons of space they will not be dealt with in this paper.

Discussion of results

18This analysis of translation shifts aims at understanding the evolution of Canada's institutional discourse in Québécois and Canadian society. It is worth stating here that a rigorous study should include a great variety of shifts in order to account for ideological dimensions in translations. Consequently, this study looks for patterns in translation shifts rather than examining isolated cases. However, following CDA methodology, this research is more qualitative than quantitative in nature. It is of course impossible to assume a one-to-one relationship between translation shifts and ideology. That is why the relations established between shifts and ideology are always potential ones.

19The identified translation shifts have been divided into two main categories. The first category includes syntactic and pragmatic shifts, and is related to the lemma "Canada." From these shifts, it was found that the relationship between speaker and hearers, that is, the prime minister and his citizens, changes from one language version to another. In the second category, the shifts belong to the semantic and pragmatic group, and are associated with the surroundings of the lemma "Canada." Such analysis tends to prove that the ideological presuppositions related to the Canadian identity differ from one language version to another. The next two sections will deal in turn with these two categories of shifts.

The prime minister and his fellow citizens

20Looking at the different translations of "Canada," it seems that the observed shifts had an effect on the prime minister's position towards his francophone or anglophone audience. In some cases, personal pronouns or indefinite pronouns were used in French, whereas the words "Canada" or "Canadian(s)" were used in English. These syntactic shifts contributed to changes in the cohesion of the discourse. Cohesion is seen as a feature which affects, among other things, intratextual reference, substitution or pronominalization (Chesterman 1997, 98). Examples from two speeches delivered in 1970 and 1976 by Pierre Elliott Trudeau are provided here [our emphasis]:

2. Syntactic shift, cohesion change and transposition
Notre présomption était peut-être naïve, mais elle s'expliquait aisément, parce que la démocratie est solidement enracinée CHEZ NOUS, et parce que NOUS avons toujours attaché le plus grand prix à la liberté individuelle.
Our assumption may have been naive, but it was understandable; understandable because democracy flourishes IN CANADA, understandable because individual liberty is cherished IN CANADA.

3. Syntactic shift, cohesion change and transposition
Le scrutin du 15 novembre au Québec a fait naître CHEZ LES UNS beaucoup d'espoir et CHEZ LES AUTRES une grande inquiétude, mais tous se posent beaucoup de questions. . . .
To some CANADIANS last week's election in Quebec has given rise to many hopes. To many other CANADIANS it has been a cause of great concern, but to all it has posed many questions. . . .

21As stated by Duchastel and Armony (1993), in any political discourse the speaker adopts a certain position using certain linguistic operations. Such positioning defines his or her political role. For instance, in passages (2) and (3) in English, the Prime Minister placed himself in the Canadian context. In French, however, the location was implicit, even ambiguous. Indeed, the pronominal form was used to talk to the French Quebec audience, and the statement could have been received from either a Canadian or Québécois perspective. This double meaning was only possible because the Liberal Prime Minister Pierre Elliott Trudeau was a Québécois. One possible explanation for this shift is that ever since 1960, most people in French Quebec have identified themselves as "Québécois" rather than (or along with) "French Canadians." Moreover, nationalist issues were quite a sensitive point both in 1970 and 1976; in 1970 nationalist terrorism struck the province of Quebec, and 1976 marked the first election of a nationalist party in the Quebec government. Trudeau's speeches were delivered in reaction to these events. Pronouns, however, have not solely been used to create ambiguity in the different French versions. They sometimes did quite the opposite, and the prime minister then presented himself as a Canadian speaking to fellow Canadians. Interestingly, these shifts were more common in speeches from the 1990s. During this period, Brian Mulroney's Conservative government tried twice to modify Canada's constitution in order to meet Quebec's traditional demands, and these initiatives partly contributed to a slowing down of the separatist movement. However, when Mulroney's first attempt (the Meech Lake Accord) failed, polls indicated that nearly 60% of the Quebec population wanted separation from Canada. This eventually led to the second referendum in 1995. The following examples were taken from speeches delivered after the defeat of the Meech Lake Accord in 1990 and after the Quebec referendum five years later (our emphasis).

4. Syntactic shift, cohesion change and scheme change
NOUS, CANADIENS, avons toujours surmonté dans le passé les obstacles à notre unité, et nous le ferons encore.
CANADIANS have always overcome challenges to our unity and we shall do so again.

5. Pragmatic shift, information change
À tous les Canadiens, je dis qu'une majorité de Québécois ont choisi le Canada en partie grâce à l'extraordinaire témoignage d'amitié et de bonne volonté que vous avez manifesté la semaine dernière. DES CANADIENS COMME VOUS ET MOI, du Québec et de tout le Canada, se sont levés et nous ont montré ce que représente ce pays.
To all Canadians, I say that a majority of Quebecers have chosen Canada in part because of the incredible outpouring of good feelings and goodwill demonstrated over the last week. ORDINARY CANADIANS, inside Quebec and throughout Canada, stood up and proved what this country is all about.

22In French, Prime Ministers Brian Mulroney and Jean Chrétien clearly stated their Canadian identity and tried to identify with their French-Canadian audience. In the English speeches, however, the word "Canada" is more iconic than the inclusive "nous." In fact, in the entire corpus, the French "nous" is more frequent (149 times) than the sum of "we" and "us" (133 times) in English. As the psycholinguist Íñigo-Mora (2004) argued, there is a strong connection between personal identity (in terms of inclusion or exclusion from a group) and pronominal choice in political discourse. Looking at examples (2) to (5), it seems that the repositioning of sociopolitical relationships represented an important translation strategy adopted by the Canadian federal government. Through changes in pronominal features, this positioning was either used to avoid sensitive issues (examples 2 and 3) or to create solidarity (examples 4 and 5). Furthermore, results from examples (2) to (5) are useful for identifying three influential factors in the translation of political speeches: the audience, the prime minister's "nationality," and the historical period during which the speech was delivered.

23William Mackenzie King's case seems to corroborate these early findings. Shifts modifying the relation between speaker and hearers were almost non-existent in his speech from 1942. This Liberal leader did promote Canadian unity, but more from an Anglo-Saxon perspective than a Canada-wide one. Further, King was not Québécois and did not speak French at all. It would have been quite difficult for him to convince the French Canadians⁵ that he was part of their group. This does not mean, however, that under his leadership the translated speeches were free of any potential ideological shifts. Let us consider the following example [our emphasis]:

6. Pragmatic shift, adaptation
Looking across the Pacific, they ask... why should the Japanese attempt to come and install themselves... in BRITISH COLUMBIA? Et tournant les yeux du côté du Pacifique, ils se demandent pourquoi les Japonais... voudraient tenter de venir s'installer... dans la COLOMBIE CANADIENNE.

24In Canada in the 1940s, the proper noun "British Columbia" was sometimes translated in French by "Colombie canadienne" in protest against British imperialism. In fact, when King's speech was printed in full in the French-Canadian press, La Presse kept "Colombie britannique," whereas the nationalist newspaper Le Devoir, from which the above excerpt is taken, chose "Colombie canadienne." It would seem that Canadian newspapers sometimes changed information in political speeches according to their editorial line. That is to say that Canadian media also play an important role in the translation and diffusion of national political speeches. Hence, a fourth factor is identified for the translation of political speeches: the place of publication.

About Canada and Canadians

25In this section, the translation shifts in the surroundings of the lemma "Canada" are investigated. In particular, it is observed that presuppositions towards the Canadian identity differed between French and English versions of the prime ministers' speeches. It is also argued that these translation shifts contributed to a "face-work" process (Goffman 1967), where the hearer's self-image (face) was maintained in all versions of the speeches.

26Let us start with the case of Pierre Elliott Trudeau's speech after the first Quebec referendum in 1980. The French version related to an identity quest, whereas the English version showed the prime minister's pride in and fidelity to Canada.

7. Semantic shift, synonymy
Et c'est sur cette volonté de changement qu'il faut tabler pour renouveler la fédération canadienne et REDONNER à tous les Québécois comme à tous les citoyens de ce pays, LE GOÛT D'ÊTRE ET DE SE PROCLAMER CANADIENS.
It is upon this desire for change that we must build a renewed Canadian federation, which will GIVE to the people of Quebec and the whole country MORE REASONS TO PROCLAIM PROUDLY THAT WE ARE CANADIANS.

8. Semantic shift, synonymy
Et les Québécois ont choisi majoritairement de rejeter la souveraineté-association et d'OPTER POUR LA VOIE DE LA FIDÉLITÉ AU CANADA.
A majority of them have decided to reject sovereignty-association, and to EXPRESS THEIR LOYALTY TO CANADA.

27These examples seem to indicate that the francophones did not feel themselves to be Canadians as strongly as did the anglophones. In fact, a statement in French such as "donner aux Québécois une raison de plus de se proclamer fièrement Canadiens" could not have sounded right to a French Québécois ear in 1980. It even could have been considered as a face-threatening act (FTA), meaning that the statement would have challenged the hearer's self-image (Brown and Levinson 1978). In the first Quebec referendum, around 40% of the Quebec population voted for the independence option, hence proving that many of them were discontented with Canada. A strong and direct challenge to the Quebec nationalist discourse might have denied Québécois' need for distinctiveness and rejected their "basic claim to territories, personal preserves, right to non-distraction — i.e.,... freedom from imposition," or, in other words, threatened their negative face (Brown and Levinson 1978, 66). One can assume that the context around the 1980 referendum was so emotionally entangled that a single FTA could have triggered serious consequences.

28In examples (7) and (8), Pierre Elliott Trudeau implicitly or explicitly referred to a "desire for change" in his speech. The third example in this section, by Brian Mulroney, also deals with the idea of "change," in terms of promise for change.

9. Semantic shift, synonymy
Nous allons établir des programmes destinés à les rapprocher et à jeter des ponts entre les solitudes dans lesquelles tant de nos concitoyens anglophones et francophones DEMEURENT CONFINÉS.
We will initiate programs to bring Canadians together and bridge the solitudes in which so many English and French-speaking Canadians STILL LIVE.

29In example (9), it is clear that in French, the "solitude" problem was related to isolation, whereas in English, it was a problem the citizens had to live with. The connotation was more negative in French. In the corpus, there are no counter-examples where the surroundings of the lemma "Canada" are more negative in English than in French. It could be argued that a negative statement about Canada in English would have diminished the Canadian nationalist discourse, and as such, threatened the (English-) Canadian people's positive face: their "positive consistent self-image or 'personality' (crucially including the desire that this self-image be appreciated and approved of)" (Brown and Levinson 1978, 66).

30Using speech act terminology, the speeches in the first three examples could be defined as "commissive" speeches, where the prime minister had to explain what he and his government intended to do in order to resolve a national crisis. In the following and last example, the speech is "directive," in that the prime minister was trying to minimize a crisis by urging the Quebec population to vote "NO" in the 1995 referendum on independence. Although the stakes were somewhat different, the following shift nevertheless reflects differences in presuppositions.

10. Pragmatic shift, adaptation
Pourquoi SOMMES-NOUS SI BIEN au Canada?
Why DOES Canada WORK?

31The Liberal Prime Minister, Jean Chrétien, described Canada in French as a homely and pleasant country, whereas in English, he described it as a coherent ensemble. Knowing that the Québécois did not feel themselves to be Canadians as much as English-Canadians did, it makes sense that Jean Chrétien chose to focus, in his French version of the speech, on the "comfort" features of the country. Furthermore, the fact that a second referendum was taking place in Quebec fifteen years after the first one implied that for the Québécois, Canada still did not work. To ask "Pourquoi le Canada fonctionne-t-il si bien?" would have imposed a discourse on the Québécois that did not meet their belief. And the impact could very well have been reflected in the outcome of the referendum. After the first referendum, it was said in the French press that the controversy surrounding a speech by the Quebec minister Lise Payette had a strong impact on the referendum vote results (Godin 2001). In 1995, a "slip of the tongue" in Jean Chrétien's case could have been very costly, and he was fully aware of it.

32All in all, face-work can be seen as another translation strategy put forward by the Canadian federal government. As argued by Paul Chilton and Christina Schäffner (2002, 14), politicians' potential FTAs are "strategically formulated to lessen the affront," and "in political situations, [an] FTA is likely to have variable value for different groups of hearers; so the linguistic formulations are chosen carefully." Chilton and Schäffner's statement referred to the speaker's choice, but it certainly can be applied to the translator's choice, especially in a case where source and target texts cannot be differentiated. Examples (7) to (10) confirm this assumption.

33Interestingly, the translation strategies identified in this paper revolved around identity representation. In addition, it appears that some cases of sociopolitical repositioning (such as in examples 2 and 3) involved face-work, suggesting that there are intricate relationships between the translation strategies used to build institutional discourses.

Conclusion

34To sum up, potential ideological shifts have here been identified from a corpus of Canadian translated political speeches in crisis situations. It would appear, then, that these shifts have been caused by a power struggle between the French-Canadian or Québécois nationalist discourse on the one hand and the Anglo-Saxon or Canada-wide nationalist discourse on the other. A careful analysis of these ideological shifts has singled out two translation strategies adopted by the Canadian government in its effort to disseminate a discourse on Canada's unity. These strategies are face-work process and sociopolitical repositioning. Four factors have also been identified as playing a key role in the translation of political speeches in Canada: the historical context of the speech, the place of publication, the targeted audience, and the "nationality" of the prime minister. Further research is needed to confirm and elaborate on these results, using a larger corpus and different kinds of political speeches. Nevertheless, this study seems to prove that there are (potential) ideological shifts in Canadian translated political speeches and that these shifts are closely related to different identity redefinitions as they have occurred in the course of Canadian history.

35In fact, the question of identity redefinition can also be extended to other areas in translation history. Still in the Quebec/Canada context, one can draw an interesting parallel between the translation of political speeches in Canada and drama translation in Quebec. For instance, Annie Brisset's corpus (1990) on drama translation shows that in Quebec between 1968 and 1988, the social discourse focused on Quebec political and national identity. Around the same time, Pierre Elliott Trudeau delivered speeches in French where he introduced a somewhat confusing setting, and his position as a member of the Quebec community potentially preceded his role as Canadian prime minister. Since the 1990s, English-language plays translated in Quebec have not been as ethnocentric as the ones in Brisset's corpus (Gagnon forthcoming; Ladouceur 2000). Again, around the same period, Prime Ministers Brian Mulroney and Jean Chrétien claimed with much conviction their pride in their Canadian identity, and particularly in the French versions of their speeches. These similarities further suggest that the Quebec nationalist discourse and the Canadian nationalist discourse are evolving at the same pace, and greatly influence each other.

36That translation has been used throughout history as a vehicle for ideologies is beyond question. What this paper has shown is that it is important to understand how this has been done and what the repercussions of such ideological dissemination are. This in turn will eventually lead to a better understanding of both translation history and national histories.