The bonds of institutional language: a discursive institutionalist approach to the Clerk of the Privy Council's annual report.

Dutil, Patrice A. ; Ryan, Peter Malachy

In the policy-making and administrative hierarchy of the Canadian government's bureaucracy, no position outranks that of the Clerk of the Privy Council. The "Clerk" occupies three critical functions that put the occupant at the nerve centre of government: as deputy minister to the prime minister, the clerk is the first advisor to the government on both policy substance and implementation. As clerk, the person is the chief executive of the Privy Council Office, the most powerful central agency of government. As Secretary to Cabinet, the clerk is charged with ensuring that the consensus of Cabinet is captured and conveyed to the administration that reports to her or him. For indeed, as "Head of the Public Service," the clerk also commands the public service.

When these three functions were formally recognized by Parliament with the passage of the 1990 Public Service Employment Act, an added responsibility was placed on the shoulders of the clerk: he or she must "submit a report on the state of the public service in each fiscal year to the Prime Minister, and the Prime Minister shall cause a copy of the report to be laid before each House of Parliament on any of the first fifteen days on which that House is sitting after the Prime Minister receives it" (Public Service Employment Act, S.C. 2003, c. 22, ss. 12, 13: part 8, s. 127). One of the best observers of this role has noted that the annual report is critically important not because it reports on "the state of the public service" but because it "contains a number of messages to the bureaucracy: it commits him [the clerk] to actions before the public service and before the wider public" (Bourgault 2008: 62).

Blending standard discourse analysis and the word frequency results of digital humanities software, this study comes to a different conclusion. The influence of the Clerk of the Privy Council's annual report (CPCAR) is debatable because it fails in its communicative mission. First, the priorities of each annual report are not the same as their emphases. Second, the reports are highly repetitive from year to year. Finally, they are decontextualized. The CPCARs reveal that clerks and PCO perceive themselves as distant from the world around them, particularly at a time of massive change, such as the rise of "New Public Management," dramatic cutbacks in funding followed by budget surpluses, the growth of the Internet, and issues of globalization such as trade, war and environmental protection.

The clerk's report is unique in commonwealth practice. The United Kingdom's Cabinet Office has produced annual reports since 1998, which focus strictly on its operations. The Secretary of the Department of the Prime Minister and Cabinet in Australia has also issued an annual report since 1998 to describe its performance. Neither of these is profiled as personalized reports from the Cabinet Secretary, and yet both are far more comprehensive and more detailed in their reporting of public sector activities than the Canadian model.

This study also sheds a new methodological light on how the "discursive institutionalism" approach to understanding public policy and government can capture the evolution of an organization's view of itself and of the world in which it operates. Frustrated by the limits of various "new" schools of institutionalism, which are much more adept at describing existing structures than in explaining how or why they have changed over time, scholars have postulated that transformation (or lack thereof) could be tracked differently. Exponents of discursive institutionalism posit that the language used by organizations should be more closely examined because it can reveal beliefs, ambitions, assumptions and policy priorities (Schmidt 2008, 2010: 3), thus connecting the work of institutionalists to the longstanding field of discourse analysis (Michel Foucault 1969, 1991; Searle 1995; Wood and Kroger 2000; Wetherell, Taylor and Yates 2001; Fairclough 2003; Bhatia 2004; Blommaert 2005; Gee 2005; Grant et al. 2005; Lange 2006). Their objective has been to show how organizations can change their minds, or refuse to do so, even while acknowledging that their environment is being transformed. They look to language as clues to how institutions are "are constituted, framed, and transformed through the confrontation of new and old discursive structures--that is, systems of symbolic meaning codified in language that influence how actors observe, interpret, and reason in particular social settings" (Campbell and Pederson 2001: 9).

As such, they borrowed the theoretical hunch pursued for decades by discourse analysis scholars, which emphasized the importance of ideas (as opposed to interests) and blended traditional methods of content analysis to illustrate the "path dependency" notions elaborated by historical institutionalists. Vivien Schmidt most recently identified discursive institutionalism as an approach "that takes account of the substantive content of ideas and the interactive processes of discourse that serve to generate those ideas and communicate them to the public" (Schmidt 2011). While illuminating in terms of the theoretical underpinnings of discursive institutionalism, this approach has not defined what methodology could best demonstrate it. This study argues that a mixed method of discourse analysis can track trends in how institutions see themselves and the world around them.

Methods

This article examines the discourse in the Clerk of the Privy Council's annual reports in two ways. First, it provides a brief description of the key priorities of each report, based on a subjective interpretation of what was presented as most urgent in the introductions and conclusions. The second method is far more objective, using software analysis techniques, collectively known as the digital humanities, both to discern the emphases of the reports and to demonstrate how the Clerk of the Privy Council has used its power of narrative. These computer-assisted textual analysis methods have been employed in the past to uncover the top issues in the Ontario budget speeches (Dutil, Ryan and Gossignac 2010); to frame strategies in policy networks (Devereaux et al. 2009); to identify authorship (Craig 2004; Schreibman, Siemens and Unsworth 2004); and to track thematic insights in digital political texts (Elmer et al. 2007). Digital humanities analysis techniques, in particular, have the potential to allow researchers to go beyond the theoretical constructs that have practically stalled innovative thinking in institutionalism and examine with more precision how ideas are "framed"--or unexamined--over time. By "mining" the discourse, different forms of analytic software can demonstrate a consistency in assumptions of what matters and, just as well, what is ignored. It reveals key anchor concepts as well as a certain migration of ideas despite the often repetitious nature of annual reports.

A key epistemological question must be answered: what discourse could be considered to be representative of an institution's thinking? In the logic of discursive institutionalism, official communications must be considered as they articulate the ideas held by the leaders of an organization. As Jurgen Habermas put it, there are four tests of validity in an "ideal speech act":

--The content of the communication is considered to be factual or true.

--The speaker is considered to be honest and sincere in what is conveyed.

--What is communicated is linguistically intelligible and comprehensible.

--What is communicated is right or appropriate in light of existing norms or values (adapted from Cukier, Bauer and Middleton 2004).

The CPCARs, as the only legally mandated, annual personal report commissioned by the government, meet this test. They are valid candidates for an assessment of discourse because they are prepared for the prime minister and Parliament. They are unique in the Canadian system in that they are perceived as the work of one accounting individual and are easily accessible to the average reader (i.e., the documents do not simply present data).

Evolution of priority themes

Paul Tellier submitted the first Annual Report of the Clerk of the Privy Council and Secretary to the Cabinet on the Public Service of Canada on 30 June 1992. (1) He argued that "much has been done by way of reform and renewal" and was eager to show "renewal" as the vision for the years to come (Canada, Privy Council Office 1992). He also emphasized that deputies had "to attach special importance to communicating with their employees" and "ensure they understood where the process of reform was headed" (1992). He noted that various efforts were being made to improve communication within departments and especially between headquarters and the regions to do this effectively.

No report was issued in the spring of 1993, and no excuse was officially made as to why the legislative requirement was not respected. The second report was delivered to a new prime minister by Glen Shortliffe almost two years later, in March 1994. The message explicitly trumpeted a change in language: where "management" was once a concept reputedly scorned inside the public service, it was now esteemed; where there had been mention of "administration," the idea of "leadership" was advanced as were notions of service and innovation (Canada, Privy Council Office 1994).

The third report was delivered by Jocelyne Bourgon, the first woman to occupy the position of clerk, in March 1995. The 28-page report was twice as long as the first report delivered by Paul Tellier and emphasized the global pressures for change on the public service of Canada. It also argued that the "strategic Policy capacity of the federal public service must be strengthened" and that "client service is what counts" (Canada, Privy Council Office 1995). Buried midway in the text was the line "the 1995 budget and Program Review will mean 45,000 fewer public service jobs over the next three years." This same report accentuated "Restoring Pride and Respect."

The fourth Clerk's report, Bourgon's second, has often been noted for its "crisis" description:

There is a "quiet crisis" under way in the Public Service today. It is quiet because few people are aware of the crisis and even fewer people have started to do something about it. The responsibility to act rests first with the Public Service. Public servants must take charge and do all in their power to remedy the situation. Their actions will provide the necessary credibility for them to ask elected officials and Canadians to join the effort (Canada, Privy Council Office 1997).

The report also noted that "globalization, new information technologies, fiscal pressures and the changing fabric of society remain the primary drivers of reform" (1997). The report made much of the fact that reform was taking place; it was "already clear that an exceptional story about reinventing the role of government is being written in Canada today" but without detail or elaboration of how this was being accomplished.

The sixth report, for the year ending 31 March 1998, was Bourgon's third and last, and was delivered on 7 December 1998, nine months late. If the public service had been in crisis in 1997, things had apparently stabilized one year later. "A year later, while the symptoms remain, the crisis is quiet no more," noted the clerk, while observing that "Public sector reform was carried out in a typically Canadian way - calmly, competently, without much fanfare" (Canada, Privy Council Office 1998). This report's priority was to promote the idea of a Canadian model of reform based on the belief that government is essential, and rejected the notion that "one size fits all." It encouraged experimentation and proposals for diverse institutional models.

Mel Cappe's first report (the seventh) was delivered in March 2000. The opening letter noted that the "the challenges facing Canada and the Public Service are significant. But we are doing everything we can to meet them" (Canada, Privy Council Office 2000). He continued Bourgon's themes of policy building, service delivery and "people." The document also shared a similar rhetoric about leadership: "We are developing leaders at all levels" (2000). Modernizing leadership in management and accountability were a particular focus in this iteration.

Cappe submitted his second report in March 2001, highlighting a new approach to human resource management in the public service. His third report (March 2002) was cast in the shadow of the terrorist attacks of September 2001, and emphasized that public servants responded to duty even when on strike. "After the tragic events of September 11," the report stated, "many questioned whether the Government would, or even should, continue the effort to modernize our human resources management regime" (Canada, Privy Council Office 2002). The report again focused on progress and modernization. The report also emphasized "a strong foundation" in public sector values, echoing the 1996 Tait report's perspective. Cappe's report noted that progress was slow on improving human resource management, which had been a goal for several years. It was candid in its concern for morale, stating that "there is a great deal of skepticism and cynicism across the public service, a feeling that those tools will never be picked up and used. I understand the disillusionment that comes from hearing a lot of talk without seeing very much action" (2002). The tools to which Cappe alluded (with an unmistakable air of pessimism) were responsibility, authority and accountability.

The tenth report was submitted by Alex Himelfarb on 31 March 2003. He sought to address the features of the Canadian model that Jocelyne Bourgon had posited five years earlier. His first report noted: "The 1990s were turbulent times for the Public Service of Canada. Our resilience and agility were tested. Despite the occasional strain, we enter the 21st century as a healthy, vibrant, secure national institution" (Canada, Privy Council Office 2003). He noted that "the reputation of our institution has been damaged in recent years. Core competencies have been criticized and questioned [...]. This scarring may not be permanent, but it shakes the pride we have in our institution. It undermines the contract of trust between citizens, parliamentarians and us. We must be vigilant at all times, being as careful as required, but operating in a manner that does not stifle innovation" (2003).

The eleventh report, Himelfarb's second, was presented in nine generously spaced pages. It was presented in the light of revelations about the sponsorship scandal: "We were all dismayed by what we have heard about incidents of serious mismanagement and, most disturbingly, breaches of the public trust" (Canada, Privy Council Office 2004). The following year, Himelfarb submitted his third report in seven pages on 31 March 2005. The theme was service transformation, especially around Service Canada. The CPCAR also placed renewed emphasis on the impact of the new Canada School of Public Service and the demands of transparency and risk management (Canada, Privy Council Office 2005).

Kevin Lynch submitted his first report, a six-page document, on 31 March 2006. He called it "very much a personal statement of how I see my duties as Clerk in relation to the vital national institution that is the Public Service of Canada" (Canada, Privy Council Office 2006). Lynch emphasized the importance of leadership. "There are outstanding leaders across the Public Service," the report noted, "and at all levels of the institution. We have a good base upon which to build, but much more can be done to support leadership" (2006). The report also took stock of improvements in internal management.

Lynch's second report, dated 31 March 2007, harkened back in style and length (eighteen pages; thirty-two pages with appendices) to Bourgon's longer CPCARs. In this report, Lynch emphasized many of the themes raised by his predecessors around policy building and maintaining reliable services, and insisted that "renewal" was something the public service had to commit to indefinitely. The clerk felt a need to identify with colleagues, adding a somewhat personal note:

Most of them have lived through a somewhat bumpy process of change over the past 15 years or more. They have sustained programs and generated new policies for new governments, while perhaps wondering how external forces will affect them and their careers, and even asking what the very concept of public service means today. I want every one of them to know that we are in this process of renewal together, and for the long term (Canada, Privy Council Office 2007).

A fourth section identified short- and medium-term goals as priorities: planning, recruitment, employee development, and providing an "enabling infrastructure." The fifth section identified issues "that will preoccupy the Public Service in the coming years": simplification of process, innovation, risk management and leadership. The purpose here was not to cast a work plan in stone, but rather to give public servants and those who care about the public service a sense of the longer term agenda, to recognize that circumstances can change, and that, "as always, they must be prepared to adapt their efforts to meet new challenges" (2007).

The fifteenth CPCAR was submitted by Lynch on 31 March 2008. Its dominant theme was renewal. The report was comparatively technical, reminiscent of documents heretofore published by the Public Service Commission. It was heavy with statistics about employees and employee demographic trends, and again underlined the incoming wave of retirements. The report highlighted a few examples of innovations designed to improve recruitment and instances where "enabling infrastructures" made improvements, such as the completion of generic job descriptions for the Computer Science group, executive development at Citizenship and Immigration Canada, and the establishment of the Service Canada College. The last eight pages of the twenty-three page document reported not on past accomplishments but on plans for future years (Canada, Privy Council Office 2008).

In March 2009, three months before his departure, Kevin Lynch delivered his last report, again centred on the theme of public sector renewal. The twenty-page report was augmented by six appendices covering topics such as deputy minister performance evaluation models and a study on the changing demographics of the federal public service. Lynch pointed to the radically changed economic context in 2009, stating he was submitting his last report in an economic environment that had been upended through the last half of 2008. "The world is experiencing the first synchronized global recession in more than 60 years," this CPCAR noted (Canada, Privy Council Office 2009). The report described the budgetary response and stated that "at no time has the Government needed a professional, non-partisan public service more than today, as we face the most difficult international economic circumstances in recent history" (2009). Like the fifteenth CPCAR, the 2009 report recognized public service renewal as a means of reinforcing the responsibility of deputy ministers for managing the public service.

The motive behind creating CPCARs may have been to provide a mechanism to monitor the progress of the Public Service 2000 initiatives, but until Lynch's recent reports, they rarely moved beyond providing summaries of selected internal changes to the public service. Prior to the increase in accountability initiatives, the clerks offered generic visions in each report, and a standard content review demonstrates that little direct research or quantitative measures were used to document any of the changes or to situate the public service in its socio-economic context from 1991 until 2007.

The selective highlighting of themes plays a useful role in the rhetorical analysis of texts such as these, but it is necessarily incomplete. It can also be misleading, for some documents may give particular importance, or "priority," to certain aspects of the bureaucratic experience without emphasizing them. The thrust and style of the CPCARs can be captured by multiple methods of textual analysis, examining both the structure and generic elements of each text and identifying where the true emphasis of words really lies.

Ideas as issue units: evaluating emphases through digital humanities

Software developed to distinguish the archeology and genre of texts was applied to the sixteen CPCARs by mathematically identifying words that represented key issues. Using this technology (which eliminates approximately 300 standardized words in English that offer little meaning, such as articles, prepositions and connectives like "at," "to," "the," "a," "of,"), the top twenty words used in the CPCARs were analyzed by decade. The words identified as "issue units" give indisputable evidence of the issues that actually dominated the reports, notwithstanding what their authors chose to accent in terms of representative institutional "issue networks" (Marres 2006). Table I presents an aggregate of the raw frequencies for the top twenty words (or possible issue units) in the sample without any synonyms aggregated into the totals, and no attention to word/sense disambiguation.

The words used most frequently over the years seem fairly typical of a bureaucratic organization, demonstrating overwhelming concern for the organization itself. All the same, some differences between the two decades can be discerned. For instance, the CPCARs produced in the 1990s showed a greater concern for innovation; the words "new" and "change" were heavily used. In contrast, the word "management" appears far more frequently in the 2000s, as does the issue of "renewal" as compared to "policy." The concept of "recruitment" is also frequently used in the 2000s, clearly reflecting a growing concern over the impact of the retirement of baby boomers from the public service.

The context in which these words appear can also be tested. Disaggregating the reports, Tables $1 and $2 present the new and unique issues delimited by an asterisk ("*") for each period in each report. The table that captures the 1992-1998 CPCARs shows how Tellier's report introduced words like "change," "progress," "renewal" and "reform." Bourgon raised the "future" in 1995 as well as "role" and "policy." Her last two reports raised issues around "research," "model" and "organization" as well as "think" and "better."

New concerns emerged in the 2000s. Himelfarb's emphasis on "accountability" in the 2005 report was a departure. The fifteenth report by Lynch (Canada, Privy Council Office 2008) introduced an unprecedented emphasis on "committees," "executives" and "performance," even though the last could not be viewed as a dominant theme by simply reading the report's table of contents or major headings. Lynch's focus on "committees" was primarily reflected in the priority he gave to the Prime Minister's Advisory Committee on the Public Service.

These few examples demonstrate the complementary aspects of using mixed methods and how "issue units" represented as "key words" can be identified and tracked to discover semantic trends in leadership communication strategies and styles. The CPCARs were tested further on the inclusion of particular issue units. Drawing on a subjective list of issues, Tables S3 and S4 demonstrate the frequency of invocations of words that reflected social and economic considerations of concern to the public service.

The charts tell many stories, and each data point invites its own line of speculation. It is interesting to note, for example, that Lynch's 2008 report mentioned "women" no less than twelve times, when they had scarcely been invoked before. It is remarkable that for both decades, Aboriginal peoples (the subject of a major Royal Commission in the 1990s), people with disabilities, and women were rarely discussed. Issues like child care and health care were not invoked, and none of the current major international players or comparative commonwealth countries was discussed (see Africa, Asia, China, and India in Table $3); even the United States and the United Kingdom were only mentioned twice each. Canada's military efforts in Iraq, Bosnia, Serbia and Kosovo were not mentioned; Afghanistan received a lone mention in Lynch's 2008 report, but only in passing (see Table $4).

If comparisons of actual texts can be made, it is also possible to measure the evolution of discourse by other standards. Donald Savoie, for example, identified paradigm words for PS2000 initiatives in Governing From the Centre (Savoie 1999: 210). The PS2000 words were mined out of the complete WordList program frequencies (see Table $4).

Table 2 was produced from the pure raw frequency scores using truncated variations of the words listed therein. The truncated word and their variations were checked in context using the HyperPo open source software. (2) For example, "based," "communication," "creative," "oriented," "risk" and "action" were checked to ensure that their use was consistent with the semantic intentions identified by Savoie (1999). Although some words proved to be red herrings, there was a clear increase in the use of PS2000 words after the millennium, even though the post-millennial sample has fewer words than the 1990s reports. Table 2 reveals that, in the era defined by the rhetoric of the "New Public Management" (NPM), the CPCARs did indeed adopt key concepts as time wore on. "Innovation" was invoked more frequently in the 2000s, as was "diverse," "task," "risk" and "action." "Old culture" concepts were seldom used in the 1990s and essentially dropped in the 2000s. The word choices in the CPCARs testified to the changes in thinking about the public service, but only slowly. It is telling of the style of administration then prevalent in the Privy Council Office.

A final set of subjective word choices based on leadership language in the private sector was created to test the extent to which reports in the age of NPM borrowed terminology from the business sector. Tables $5 and $6 present the frequencies of words of interest commonly used in modern public sector management (again broken down into two decades).

These tables reveal that business terms like "imagination," "vision" and "inefficiency" were avoided, and idioms such as "customer" and "client" and "change" were gradually phased out while words like "Canadians" and "citizens" have been emphasized. Clearly, NPM--at least in terms of its rhetoric - was not trumpeted in these reports. The word "change" dominated the 1990s and then fell out of use, only to rise again in Lynch's 2007 report; the new millennium is also notably the age of "excellence," where "imagination" is no longer invoked.

Factor and cluster analysis: the test of originality

Digital humanities research has demonstrated the power of measuring patterns in language through the recurring use of words and the variety of words by reducing their quantitative counts to standardized coefficients (Biber 1988; Craig 2004; Schreibman, Siemens and Unsworth 2004). For instance, factor and cluster analyses have revealed the level of originality among texts, differences in genre, and author attribution (Craig 2004). Using factor analysis, Figure 1 illustrates the level of originality in each CPCAR by simultaneously comparing the words used and how they are repeated. Documents that, in previous studies, have appeared in the upper left quadrant were highly original (i.e., using a rich vocabulary and a creative--less consistent, more surprising, more exact--use of language). Hugh Craig, for instance, plotted William Shakespeare's key protagonists Hamlet and Macbeth in this top left quadrant based on only the top 50 most common words (Craig 2008). Less involved characters in Shakespeare's historical dramas appeared in the lower left quadrant because they used a rich vocabulary but delivered fewer lines with less variety than the main characters. Documents plotted in the upper right quadrant use a less rich vocabulary and show an even greater degree of consistency. In other words, they are less creative. The CPCARs plot in this quadrant, signaling a consistent, institutional style over sixteen reports and six different clerks.

[FIGURE 1 OMITTED]

The visualization of Figure 1, using 250 variables (3), demonstrates the style. This number of variables creates a consistent pattern, showing clear visual clusters from the factor analysis.

The third panel in Figure 1 highlights the factor analysis plot of the CPCARs. Since there are sixteen reports spread out over seventeen years, a consistent pattern is clearly revealed, all the more remarkable given the number of different authors and teams drafting the reports. This panel also demonstrates that the CPCARs clump together using the top 500 principal components of the reports. Beyond Bourgon and two of Cappe's reports (in the circle), the CPCARs are densely packed together, demonstrating the absence of variation in the style of authorship teams over the years. This suggests that the ideas articulated in all of these reports were fairly consistent, emphasizing particular concepts and consistently ignoring others. There was very little differentiation in terms of the authors and periodization of styles in the CPCARs by decade. For example, the Lynch 2008 (the longest one) and 2009 reports marked a departure from his first two reports and are positioned, in terms of a distinct style, nearer to Tellier's 1992 report. Lynch's newer reports offered a mixture of charts, statistics and, like Tellier's, focused on renewal. Bourgon's CPCARs ranged a little more widely, showing slight changes of style and emphases from year to year. The reports produced under her aegis distinguished themselves from the pack and overlapped only with some of her successor's (Mel Cappe) reports. The rest of the CPCARs cluster tightly, regardless of the year written.

ReseauLu relational mapping visualizations

Relational network maps have also proved useful in revealing patterns of institutional issue networks and their representative discourse. The ReseauLu software plots the top twenty words in each report drawn from the same tabular data as the factor analysis, but instead produces two- dimensional relational maps. Large nodes indicate where there are more links among the CPCARs. The lengths of the lines do not have any significance, and they are adjusted manually by researchers to improve the visualization.

ReseauLu reveals the top framing issues in each report and illustrates the relations (or families) between CPCARs. For example, the main keywords common to all of the clerks are centered in the visualization, and those words are not surprisingly "Canada," "Canadians," "government," "public" and "service." Figure 2 highlights Bourgon's focus on "policy," while Tellier and Shortliffe focus on "renewal." This software demonstrates that Bourgon's first three CPCARs cluster together with the reports produced under Cappe and Himelfarb. It again demonstrates that Lynch's final two reports emphasized concepts similar to the Tellier and Shortliffe CPCARs.

[FIGURE 2 OMITTED]

Moving clockwise around the secondary clusters with the reports situated by author, the nodes increase in size according to the higher frequency of words associated with each report. Looking at the single items at the periphery, some minor differentiation in themes among these seven reports can be discerned: Lynch focused on business and culture, Himelfarb focused on accountability, ethics and innovation, while Cappe focused on the economy and change.

Discussion

This study of the annual reports of the Clerk of the Privy Council opens a window into the phenomenon of institutional discourse and behaviour, showing how students of public policy and administration can make use of digital humanities techniques to chart the evolution of thinking of political, bureaucratic and stakeholder groups on particular issues. This study of the first sixteen CPCARs, written over a period of two decades of change, reveals several phenomena. First, many reports raised distinct themes, as shown in the brief content review of each report's self-identified priorities. But it would be a mistake to assume that the reports were substantively different. The CPCARs were remarkably similar in their choice of frames and style, even though written under six different clerks. The frames analysis demonstrated that a narrow range of themes was pursued in combination with a high degree of consistency in word usage. In other words, while the reports may have highlighted different concepts from year to year, they showed little innovation and a strong institutional hand. Indeed, this analysis reveals the CPCARs' distinctive style.

By giving words equal weight, digital humanities software shows that the reports demonstrated very little originality in either the choice or the use of words. Though the melodies were changed a little, the lyrics and rhythms remained the same. It is striking how words associated with critical national and global trends were hardly mentioned, as if the CPCARs were written in an institutional environment far from the realities faced by the public service and the population it serves. In the third Himelfarb report, this reality was inadvertently acknowledged: "Each year I set out my corporate priorities for the public service. These are matters that, in my view, require special attention from deputy ministers. For the coming year, I have decided to maintain last year's priorities: official languages, diversity, learning, and modern comptrollership" (Canada, Privy Council Office 2005: 5, emphasis added).

The striking consistency and decontextualization of the CPCARs in the face of change undoubtedly reflect an institutional reflex and an unspoken pact between the executive of the public service and successive governments. If Paul Tellier had hoped that the report would speak "to problems and challenges as I see them before us" (Canada, Privy Council Office 1992: 1), the CPCARs have clearly fallen short of their objective. They have not described the complex environment in which the public service of Canada works nor have they shown how it has responded to the realities and priorities of Canadians. Judging by the CPCARs, the clerks and their Privy Council Office executive--as an institution--seem to have determined that there was no need to explore the challenges of the public service in terms of a changing context, instead leaving that territory to the political authorities.

Moreover, a significant opportunity has been lost in communicating the realities of the public service both in terms of its accomplishments and its weaknesses. A repetitive report that does not accord itself with the real environment of the public service cannot be effective. As a result, the CPCARs can hardly be perceived as significant policy instruments or as effective accountability instruments. The 2008 Government of Canada employee survey revealed that only fifty per cent of respondents agreed that "essential information flows effectively from senior management to staff;" that only forty-two per cent agreed that "senior management in my organization makes effective and timely decisions;" and that a bare majority (fifty-three per cent) has confidence in the senior management of their organization (Canada, Treasury Board Secretariat 2008). It certainly leaves open the door for the CPCAR to contribute significantly in making the senior executives of the Government of Canada more effective communicators and relevant leaders.

In dissecting the discourse of an institution and revealing how institutions communicate assumptions, strategies and objectives on critically important policy issues, the digital humanities approach to the CPCARs reveals that the language of the Clerk of the Privy Council did evolve within the narrow band of subjects it emphasized but that the CPCARs have created a style of their own. The emphases, to borrow the language of historical institutionalists, demonstrated a strong "path dependency" in terms of word use and the ideas expressed. This approach to discursive institutionalism offers students of public policy and administration a new, objective perspective on how and when institutions retain the bonds of past concepts and when they adopt new ideas. At the same time, it sheds a light on what were, and what were not, the articulated priorities of the leaders of the public service of the Government of Canada.

References

Bhatia, Vijay J. 2004. Worlds of Written Discourse. London: Continuum.

Biber, Douglas. 1988. Variations Across Speech and Writing. Cambridge: Cambridge University Press.

Blommaert, Jan. 2005. Discourse. Cambridge: Cambridge University Press.

Bourgault, Jacques. 2008. "Clerks and secretaries to Cabinet: Anatomy of leadership." In Searching For Leadership: Secretaries to Cabinet in Canada, edited by Patrice Dutil. Toronto: University of Toronto Press, pp. 41-85.

Campbell, John L., and Ove K. Pederson. 2001. The Rise of Neoliberalism and Institutional Analysis. Princeton: Princeton University Press.

Canada. Privy Council Office. 1992. First Annual Report to the Prime Minister on the Public Service of Canada (Paul M. Tellier: Clerk of the Privy Council and Secretary to the Cabinet). Ottawa: Minister of Supply and Services Canada* Available from http://www.pcobcp.gc.ca (accessed on 21 October 2011).

--.--. 1994. Second Annual Report to the Prime Minister on the Public Service of Canada (Glen Shortliffe: Clerk of the Privy Council and Secretary to the Cabinet). Ottawa: Minister of Supply and Services Canada. Available from http://www.pco-bcp.gc.ca (accessed on 21 October 2011).

--.--. 1995. Third Annual Report to the Prime Minister on the Public Service of Canada (Jocelyne Bourgon: Clerk of the Privy Council and Secretary to the Cabinet). Ottawa: Her Majesty the Queen in Right of Canada. Available from http://www.pco-bcp.gc.ca (accessed on 21 October 2011).

--.--. 1997. Fourth Annual Report to the Prinw Minister on the Public Service of Canada (Jocelyne Bourgon: Clerk of the Privy Council and Secretary to the Cabinet). Ottawa: Her Majesty the Queen in Right of Canada* Available from http://www.pco-bcp.gc.ca (accessed on 21 October 2011).

--.--. 1998. Fifth Annual Report to the Prime Minister on the Public Service of Canada (Jocelyne Bourgon: Clerk of the Privy Council and Secretary to the Cabinet). Ottawa: Her Majesty the Queen in Right of Canada. Available from http://www.pco-bcp.gc.ca (accessed on 21 October 2011).

--.--. 1998. Sixth Annual Report to the Prime Minister on the Public Service of Canada (Jocelyne Bourgon: Clerk of the Privy Council and Secretary to the Cabinet). Ottawa: Her Majesty the Queen in Right of Canada. Available from http://www.pco-bcp.gc.ca (accessed on 21 October 2011).

--.--. 2000. Seventh Annual Report to the Prime Minister on the Public Service of Canada (Mel Cappe: Clerk of the Privy Council and Secretary to the Cabinet). Ottawa: Her Majesty the Queen in Right of Canada. Available from http://www.pco-bcp.gc.ca (accessed on 21 October 2011).

--.--. 2001. Eighth Annual Report to the Prime Minister on the Public Service of Canada (Mel Cappe: Clerk of the Privy Council and Secretary to the Cabinet)* Ottawa: Her Majesty the Queen in Right of Canada. Available from http://www.pco-bcp.gc.ca (accessed on 21 October 2011).

--.--. 2002. Ninth Annual Report to the Prime Minister on the Public Service of Canada (Mel Cappe: Clerk of the Privy Council and Secretary to the Cabinet). Ottawa: Her Majesty the Queen in Right of Canada. Available from http://www.pco-bcp.gc.ca (accessed on 21 October 2011).

--2003. Tenth Annual Report to the Prime Minister on the Public Service of Canada (Alex Himelfarb: Clerk of the Privy Council and Secretary to the Cabinet). Ottawa: Her Majesty the Queen in Right of Canada* Available from http://www.pco-bcp.gc.ca (accessed on 21 October 2011).

--2004. Eleventh Annual Report to the Prime Minister on the Public Service of Canada (Alex Himelfarb: Clerk of the Privy Council and Secretary to the Cabinet). Ottawa: Her Majesty the Queen in Right of Canada. Available from http://www.pco-bcp.gc.ca (accessed on 21 October 2011).

--2005. Twelfth Annual Report to the Prime Minister on the Public Service of Canada (Alex Himelfarb: Clerk of the Privy Council and Secretary to the Cabinet)* Ottawa: Her Majesty the Queen in Right of Canada* Available from http://www.pco-bcp.gc.ca (accessed on 21 October 2011).

--2006. Thirteenth Annual Report to the Prime Minister on the Public Service of Canada (Kevin G. Lynch: Clerk of the Privy Council and Secretary to the Cabinet). Ottawa: Her Majesty the Queen in Right of Canada. Available from http://www.pco-bcp.gc.ca (accessed on 21 October 2011).

--2007. Fourteenth Annual Report to the Prime Minister on the Public Service of Canada (Kevin G. Lynch: Clerk of the Privy Council and Secretary to the Cabinet). Ottawa: Her Majesty the Queen in Right of Canada* Available from http://www.pco-bcp.gc.ca (accessed on 21 October 2011).

--2008. Fifteenth Annual Report to the Prime Minister on the Public Service of Canada (Kevin G. Lynch: Clerk of the Privy Council and Secretary to the Cabinet). Ottawa: Her Majesty the Queen in Right of Canada* Available from http://www.pco-bcp.gc.ca (accessed on 21 October 2011).

--2009. Sixteenth Annual Report to the Prime Minister on the Public Service of Canada (Kevin G. Lynch: Clerk of the Privy Council and Secretary to the Cabinet). Ottawa: Her Majesty the Queen in Right of Canada* Available from http://www.pco-bcp.gc.ca (accessed on 21 October 2011).

--Treasury Board Secretariat. 2008. "Results on leadership" [website]. 2008 Public Service Employee Survey. Available from http://www.tbs-sct.gc.ca/pses-saff/2008/highl- faits-eng. asp#sec01 (accessed 21 October 2011).

Craig, Hugh. 2004. "Stylistic analysis and authorship studies." In A Companion to Digital Humanities, edited by Susan Schreibman, Ray Siemens and John Unsworth. Oxford, U.K.: Blackwell Publishing.

--2008. " 'Speak, that I may see thee': Shakespeare characters and common words." Shakespeare Survey 61: 281-88.

Cukier, Wendy, Rob Bauer, and Catherine Middleton. 2004. "Applying Habermas' validity claims as a standard for critical discourse analysis." International Federation for Information Processing 143: 233-58*

Devereaux, Zachary, Wendy Cukier, Peter Ryan, and Neil Thomlinson. 2009. "Using the issue crawler to map gun control issue-networks." APSA. Available from http:// papers.ssrn.com/so13/papers.cfm?abstract_id=1449612 (accessed on 21 October 2011)*

Dutil, Patrice, Peter Ryan, and Andr6 Gossignac. 2010. "Thoughts into words: The Budget Speech, 1968-2003. In Searching For Leadership: Secretaries to Cabinet in Canada, edited by Patrice Dutil. Toronto: University of Toronto Press.

Elmer, Greg, Peter Ryan, Zachary Devereaux, Ganaele Langlois, Fenwick McKelvey, and Joanna Redden. 2007. "Election bloggers: Methods for determining political influence." First Monday. Available from http://www.firstmonday.org/issues/issue12_4/elmer/ index.html (accessed on 21 October 2011).

Fairclough, Norman. 2003. Analysing Discourse: Textual Analysis for Social Research* London: Routledge.

Foucault, Michel* 1969. The Archaeology of Knowledge and the Discourse on Language. New York: Harper Colophon.

--1991. "Politics and the study of discourse*" In The Foucault Effect: Studies in Governmentality, edited by Graham Burchell, Colin Gordon, and Peter Miller. London: Harvester Wheatsheaf, pp. 7-26.

Gee, James P. 2005. An Introduction to Discourse Analysis: Theory and Method. London: Routledge.

Grant, David, Cynthia Hardy, Cliff Oswick, and Linda Putnam. 2005. Sage Handbook of Organizational Discourse. London: Sage.

Lange, Bettina. 2006. "Researching discourse and behaviour as elements of regulatory law in action." In Dreaming of the Regulatory Village; Speaking of the Regulatory State, edited by Michael McConkey and Patrice Dutil. Toronto: Institute of Public Administration of Canada, pp. 108-29. Available from http://www.ipac.ca/documents/RegulatoryState.pdf (accessed on 21 October 2011).

Marres, Noortje. 2006. "Net-work is format work: Issue networks and the sites of civil society politics." In Reformatting Politics: Information Technology and Global Civil Society, edited by Jodi Dean, Jon W. Anderson, and Geert Lovink. New York: Routledge, pp. 3-17.

Savoie, Donald. 1999. Governing Front the Centre. Toronto: University of Toronto Press.

Schmidt, Vivien. 2008. "Discursive institutionalism: The explanatory power of ideas and discourse*" Annual Review of Political Science 11: 303-26.

--2010. "Taking ideas and discourse seriously: Explaining change through discursive institutionalism as the fourth 'new institutionalism.' " European Political Science Review 2 (1): 3-40.

--2011. "Reconciling ideas and institutions through discursive institutionalism." In Ideas and Politics in Social Science Research, edited by Daniel B61and and Robert Henry Cox.

Toronto: Oxford University Press. Schreibman, Susan, Ray Siemens, and John Unsworth. 2004. A Companion to Digital Humanities.

Oxford, U.K.: Blackwell Publishing. Searle, John. 1995. The Social Construction of Reality. New York: Free Press.

Wetherell, Margaret, Stephanie Taylor, and Simeon Yates. 2001. Discourse as Data: A Guide for Analysis. London: Sage.

Wood, Linda A., and Rolf Kroger. 2000. Doing Discourse Analysis. London: Sage.

Supporting information

Additional supporting information may be found in the online version of this article at the publisher's website (http://onlinelibrary.wiley.com/ journal/10.1111/(ISSN)1754-7121)

Table $1. Disaggregated Reports: Top Word Frequencies, 1992-1998

Table S2. Disaggregated Reports: Top Word Frequencies, 2000-2008

Table S3. Words of Interest, 1990s

Table S4. Words of Interest, 2000s

Table $5. Key Words, 1990s

Table $6. Key Words, 2000s

Notes

(1) All annual reports of the Clerk of the Privy Council are now available online at http://www.clerk.gc.ca. The authors of this article used older print versions of the report for their analysis.

(2) The version of HyperPo software used for this study is available at http://tapor.mcmaster.ca/~hyperpo/Versions/6.0/index.cgi?&delta_iLang=en.

(3) This is the most that MS Excel will allow to be calculated when transposing the data from Intelligent Archive into a suitable format for the SPSS suite for this number of texts*

Patrice Dutil is professor, Department of Politics and Public Administration, Ryerson University, Toronto, Ontario. Peter Ryan is instructor and instructional designer, MacEwan University, Edmonton, Alberta; he recently completed his doctorate in the joint York/Ryerson Communications and Culture program, Toronto, Ontario. They would like to thank the anonymous reviewers and the editor of the journal for their helpful suggestions. Table 1. Top Twenty Unique Words by Decade 1990s (6 reports) 2000s (10 reports) Words Frequency Words Frequency public 815 public 1006 service 664 service 823 government 356 Canada 276 policy * 243 government 228 Canada 203 management * 191 new 167 Canadians 177 Canadians 151 renewal * 161 departments 144 servants 152 services 143 work * 145 sector 138 new 138 servants 128 report * 126 governments 120 recruitment * 123 federal 110 development 118 delivery 109 people 112 need 104 employees * 109 people 101 need 107 citizens 101 human * 106 change * 99 federal 103 development 97 resources * 102 programs 96 programs 94 MS WORD: 44799 MS WORD: 45062 Total Words Total Words Aggregate Words Frequency public 1821 service 1487 government 584 Canada 479 policy 334 Canadians 328 new 305 servants 280 management 277 work 232 renewal 222 services 221 report 215 development 215 federal 213 people 213 departments 212 need 211 programs 190 sector 189 MS WORD: 79637 Total Words * indicates unique term for the period. If needed, coefficients can be calculated by dividing the word frequency by the total number of words for each sample. Table 2. PS2000 Key Words (Identified by Savoie) by Decade 1990s: Tellier, Shortliffe and Bourgon 1990s words (old culture) Frequency Controlling (control * = 23) 0 Rigid 0 Suspicious 0 Administrative 20 Secret 0 Power based (power * = 12) 0 Input/process oriented 0 (process * = 89) Preprogrammed and repetitive 0 Risk averse (risk = 4) 3 Mandatory 0 Communicating poorly (poor = 1) 0 Centralized (central* = 26) 1 Uniform (static = 0) 0 Stifling creativity (creative * = 8) 2 Reactive 2 TOTAL (variations = 150) 28 2000s: Mel Cappe, Himelfarb and Lynch 1990s words (old culture) Frequency Controlling (control * = 11) 0 Rigid 2 Suspicious 0 Administrative 5 Secret 0 Power based (power * = 6) 0 Input/process oriented 0 (process * = 56) Preprogrammed and repetitive 1 Risk averse (risk = 40) 0 Mandatory 0 Communicating poorly (poor =1) 0 Centralized (central * = 27) 0 Uniform (static = 3) 0 Stifling creativity (creative * = 21) 3 Reactive 1 TOTAL (variations = 118) 11 1990s: Tellier, Shortliffe and Bourgon PS2000 words (new culture) Frequency Empowering (power * = 12) 0 Flexible 3 Trusting (trust * = 11) 0 Managerial (manager * = 26) 0 Open 23 Task based (task * = 33) 15 Results oriented (results * = 27) 7 Capable of purposeful "action" 30 "Willing" to take risks 3 Optional (option * = 15) 1 Communicating well 9 Decentralize 1 Diverse 28 Encouraging innovation 37 Proactive 0 TOTAL (variations = 124) 157 2000s: Mel Cappe, Himelfarb and Lynch PS2000 words (new culture) Frequency Empowering (power * = 6) 4 Flexible (inflexible = 2) 5 Trusting (trust * = 20 / distrust =1) 0 Managerial (manager * = 43) 1 Open 14 Task based (task * = 13) 35 Results oriented (results * = 44) 0 Capable of purposeful "action" 48 "Willing" to take risks 28 Optional (option * = 7) 0 Communicating well (well * =37) 2 Decentralize 1 Diverse 52 Encouraging innovation 62 Proactive 1 TOTAL (variations = 156) 241