In early January 2012 Jan Baas died. Baas was a member of the Dutch Senate (Eerste Kamer) for the conservative-liberal VVD (People’s Party for Freedom and Democracy) from 1960 to 1981, and he gained nationwide repute following a conflict with Hendrik Adams of the Boerenpartij. The argument became so heated that blows were exchanged. In the Senate Baas had requested the floor for “personal business” (een persoonlijk feit), a special regulation in the Rules of Procedure of the States General. Baas recounted his personal experiences with Adams during the Second World War (WWII). The newly sworn-in senator Adams, said Baas, had been his colleague at a state agricultural college during the war, where he had been known as a supporter of the German occupier. During an argument about Adams’ pro-Nazi conduct, Baas recounted, Adams had even threatened him with deportation. Baas had contemplated at length whether he should stir up these issues, which had occurred more than twenty years earlier:

Can it be justified to lay bare in public, during their lifetime, the past of persons who were “wrong,” politically or otherwise? After much hesitation I have answered this question affirmatively, if these persons seek a position in the social order that, in terms of responsibility and standing, they are, in my opinion, no longer entitled to. [1]

This, in short, was more than personal vindictiveness. Baas felt that Adams had not rehabilitated himself sufficiently to be an equal participant in political decision making. [2]

Boerenpartij leader Hendrik Koekoek has systematically emphasized that the big parties tried to isolate him by referring to the Boerenpartij as a kind of National Socialist movement that was dominated by former Nazi-collaborators. Exposing Adams in the Senate seemed to confirm this image. But was Koekoek right? Was his party disqualified as “wrong” in Parliament and if so, how was this done? In order to answer these questions we have used a new research method that is possible because the Proceedings of the Dutch States General were digitized some years ago, and the recently concluded project War In Parliament made systematic and advanced searches possible (see paragraph: methodological innovation). [3] What is impossible by hand – finding every fragment in the parliamentary debates that links the Boerenpartij and “wrong” – does become possible with a digital approach. This article is intended to contribute to the benefits of digital methods to historians by exploring the advantages of using an advanced search engine that can search large quantities of data. However, before we go into the methodology in more detail, we have to introduce the Boerenpartij in the historical and political setting of the Netherlands in the 1950s and 1960s.

From the rise of Fascism in the 1920s in Italy to the Nazi-occupation of the Netherlands (1940-1945), and then from the liberation up to the present day, Dutch politicians have referred to the dangers of Fascism and Nazism. Referring to Fascism and Nazism is still considered a tried-and-tested method to stigmatize fellow-politicians and it could result in the effective exclusion of parties. [4] References to Fascism and Nazism were also used to call attention to social wrongs. [5]

Our focus in this article is on the Boerenpartij, founded in 1958, that entered the House of Representatives (Tweede Kamer) with three seats in 1963. For the first time since the fleeting electoral gain of the CPN (Communist Party Netherlands) in 1946 the traditional centre parties ARP (Anti-Revolutionary Party), CHU (Christian-Historical Union), KVP (Catholic People’s Party), PvdA (Labour Party) and the conservative-liberal VVD (People’s Party for Freedom and Democracy) had lost a considerable number of seats, this time to a right-wing newcomer to the political stage. Earlier (extreme) right-wing attempts had failed prematurely. Especially when parties were portrayed in the news as assembly points for former Nazi-collaborators their fate was quickly decided. [6]

Hendrik Koekoek had, before he entered the political stage as leader of the Boerenpartij, been a candidate for one of the earlier right wing political attempts (the National Opposition Union, NOU). One of the pacesetters within the NOU was engineer S.D. Rispens, a former member of the National Socialist Movement, founded in 1931 and the only legal party during the war years because of its pro-Nazi attitude. Rispens was convicted after the war and deprived of the right to vote or run for office until 1957. Another candidate was the previously mentioned Hendrik Adams, also an engineer and convicted of collaboration. After the NOU fiasco Rispens and Adams searched for opportunities for a new party, this time with men without a “tainted past” as figureheads. [7] A man like Koekoek was totally compatible with this picture. [8] Koekoek, depicted in 1958 in a secret internal memo of the Ministry of Agriculture as an “a-political character” who was “intellectually inferior” to Adams and Rispens, [9] could not be linked to a “wrong” political past. He managed to leave his mark on the Boerenpartij. This success was partly due to the unrest in the Dutch agriculture sector, which was caused by the state reforms towards economies of scale and modernization. This unrest came to a head just before the parliamentary elections of 1963.

In March 1963 an uprising broke out in a small village in the eastern province of Drenthe, called Hollandscheveld. Three farmers, who had refused to pay their contribution to the Agricultural board (het Landbouwschap), were threatened with eviction from their farms. The Agricultural board was a statutory industrial organization, established in 1954, a collaboration between the agricultural sector and the government to reform agriculture and horticulture. The resulting taxes and regulations generated resistance, especially among the small farmers. Whereas before these so-called “free farmers” had been punished by imposing fines, the Agricultural board switched to more draconian measures in 1963. The courts sanctioned a decision to remove the three farmers and their families from their farms. This resulted in fierce opposition. Led by Farmers’ Party leader Koekoek hundreds of sympathizers gathered in Hollandscheveld. They were met by police who greeted them with tear gas and truncheons. Television images reached many Dutch households. This did little for the reputation of the Agricultural board and it definitely benefited the Boerenpartij. [10]

Koekoek presented himself as the political outsider who voiced the feelings of discontent of the “average” Dutch citizen. [11] Not only the small farmer could count on his support, as became clear during the election campaign, but also others who were dissatisfied with the political establishment and had right-wing sympathies. Koekoek’s political agenda (against state bureaucracy, pro severe punishment of criminals and against foreign “fortune seekers”) proved successful not only in the countryside, but also in large cities. Predictions in the summer of 1966 said that in the case of new national elections the party could win as many as twenty seats. [12] In the sixties this would have meant a revolution, and contrary to popular opinion, not a revolution from the left.

Twenty seats would never be realized. In the States-Provincial elections of March 1966 the Boerenpartij did win 6.4% of the votes, a considerable number that stirred great upheaval. In the municipal elections two months later the success continued and the Boerenpartij joined the councils of several large cities, including Amsterdam and Rotterdam. The elections were regarded as heralds of great success in the coming parliamentary elections. In the turbulent year 1966, when the newly founded political party D’66 (Democrats ’66) also created a stir on the left wing, the traditional political system seemed to falter.

The Boerenpartij received a lot of media attention. Mostly criticism came from the left, but the right side openly expressed support. Also in academia the Boerenpartij became a hot topic. Academics concentrated on the question of whether the ideas of the Boerenpartij were fascist or not. The sociologist A.T.J. Nooij wrote a dissertation based on the “F(ascism) scale” developed by Theodor W. Adorno. Nooij used interviews with supporters of the Boerenpartij and an ingenious computational model to deliver hard scientific evidence of whether the supporters of the Boerenpartij were right-wing extremists and whether they constituted a fascist peril. In the end Nooij did not quite succeed; his problem was finding the right definition of Fascism. [13] Jaap van Donselaar, who published his book Fout na de oorlog (Wrong after the war) years later, had the same problem. According to him there was no doubt that the supporters of the Boerenpartij leaned towards fascist ideas, but whether the Boerenpartij could be defined as fascist remained difficult to determine. [14]

More recently historian Koen Vossen positioned the Boerenpartij in the changing political landscape of the 1960s. He did not zoom in on the left, but rather on the right. According to Vossen the Boerenpartij should be viewed as one of the parties “in the broader spectrum of right-wing populist parties as it existed in the 1960s and that could, due to societal changes, once again flourish.” [15] Vossen’s focus was not so much on the question of how “wrong” the Boerenpartij was; rather, he pointed out the depth of the impact of the Adams affair on the Boerenpartij. Without this affair, Vossen writes, the party might have won the predicted twenty seats in the elections. But:

In the heated climate of 1966, when the 1940-1945 traumas kicked up again, the Boerenpartij epitomized the presumed continued effect of the fascist mentality, especially for progressive young people. In public meetings of the Boerenpartij the presence of young demonstrators who wanted to nip the fascist danger in the bud was now almost standard. [Vossen 2005, 246]

So Vossen attributes the beginning of the decline of the Boerenpartij to extraparliamentary actions. In this article we take a look at the parliamentary side of the story by investigating whether and if so in what way the Boerenpartij was linked with fascism and national socialism (“wrong”) in the parliamentary debates. In this way we hope to gain more insight into the degree to which the parliamentary discourse may have contributed to the fall of the Boerenpartij.

The National Library of the Netherlands (Koninklijke Bibliotheek) in The Hague has digitized the Proceedings of the States General. However, searching this huge database (approx. 2.5 million pages) has its limitations because it does not allow certain complex queries. With the search engine that was developed in the earlier mentioned War in Parliament project this is now possible. [16] In this project the parliamentary debates from 1930 through 1995 were opened up, which means that text fragments can be “harvested” on the document level (from a specific debate) as well as on the person level (belonging to a specific speaker).

To make this possible all debate transcripts were converted into semi-structured XML documents. In these documents the debates are recorded in a hierarchy of debate subjects (topics), consisting of a series of speakers (scenes), which in turn are made up of speeches (chronological, starting with the main speaker and followed by any interruptions and rebuttals). Every speech consists of paragraphs. This structure of topics, scenes, speeches and paragraphs is introduced by a computer-directed analysis of implicit information present in the original documents, such as blank lines, standard formulations, headings and typeface. In the older documents the topics are composed directly from speeches (without the intermediate layer of scenes). Each fragment from the documents generated in this way has its own unique identifier (up to the paragraph level), which can be used to determine its position in the hierarchy as well as to retrieve the fragment in question from the database.

All speeches were subsequently annotated with information regarding the speaker, political party, and role of the speaker. With the combination of speaker’s name from the original document, date of session, and chamber (Senate - House of Representatives) we managed to find in 98.5% of all speeches a unique speaker in the politician database we constructed with data from www.parlement.com. Based on biographical information a party name could then also be added. All data was placed in an open-source database (eXist) with open-source search technology (Lucene). This enables users to carry out searches on the fully indexed text. Together with the hierarchical structure of the data we can search on the word level and filter results on politician name, name of a party, speaker’s role, date, Senate and House of Representatives, and make combinations.

Capturing the historical information we were looking for was a challenge, because references to WWII were imprecise and often hidden. To put it differently, WWII was (and is) most of the time not a “natural” topic in parliamentary debates, due to the fact that WWII is part of the collective memory of Dutch society, thus references to this period could be found in every debate. This is the reason why we experimented with two different Boolean search strategies, instead of using more complex classification methods (i.e. machine learning techniques) which proved to be unsuccessful.

The first broad search strategy consisted of formulating a number of search queries that were linked. The goal of this strategy was to combine result sets that would yield a single result set of relevant hits. We cast a wide net, so to speak: all potentially relevant words as they could occur in the spoken text when the subject was “wrong” were combined in the query:

fascis* OR NSB OR “ politiek delinquent ” OR “ politieke delinquent ” OR “ politieke delinquenten ” OR collaborat* OR “ nationaal socialisme ” OR “ nationaal socialistisch ” OR “ nationaal socialistische ” OR antisemitis* OR oorlo* OR Hitler OR Mussert OR Roskam OR Boerenleider OR Jeugdstor* OR NSK* OR Waffen-S* OR Landstand

This yielded two result sets for the period 1963-1981 (the entire parliamentary term of the Boerenpartij). The first result set contained all references to “wrong” by the speakers of the Boerenpartij and yielded 178 hits. The second result set contained all references to “wrong” of all other parties and yielded 7662 hits. It should be mentioned that there can be more than one hit in a debate (topic) or even in a speaker (scene), because each of the queries mentioned above is “harvested” at the speech level. Also important to mention is that during the writing of this article (July/August 2012) it is not yet possible to reproduce the query via the publicly accessible web interface of the search engine. This is due to a limit on the use of the “wildcard” (*) that is necessary at present to avoid server overload, a technical obstacle that will be remedied in the near future.

A second widely cast net led to a result set of all references to the Boerenpartij by others than Boerenpartij members:

Boerenpartij OR Adams OR Koekoek OR Voogd OR Brake OR Harmsen OR Harselaar OR Bossche OR Koning OR Kronenburg OR Leffertstra OR Nuijens OR Verlaan

This query resulted in 12,204 hits.

The combination of both queries “wrong” and “ Boerenpartij ” excluding the Boerenpartij (which we already had with 178 hits) yielded 179 hits.

The set of 179 hits we subsequently examined for relevance. This was necessary because search terms like “oorlo*” (war), for example, did not result in relevant fragments (“war” yes, but not the Second World War). Noise was not restricted to war-related search terms.

The search for persons (the members of the Boerenpartij) also generated noise. In a query a distinction must be made between searching for persons who have said something, and searching for persons who are mentioned. The first is relatively easy. As the structure of parliamentary reports has remained fairly consistent through the years, it was possible - despite the necessary OCR (optical character recognition) errors - to detect the announcement of speakers in nearly every case, and identify the speaker by matching speaker name and party name with a database of politicians with precise session days and party memberships. In this way a distinction can be made between the Christian-Democrat politician Jan de Koning, the liberal H.E. Koning, and J. de Koning of the Boerenpartij. In the second case, in which the search is for persons mentioned in the spoken text this is not yet possible, because such identification will depend on the context of what is said. A search query on “Koning” will yield many non-relevant results. The search engine is not able to distinguish between the three “Konings” solely on the basis of the text.

Noise is a major obstacle in this study, partly as a result of the Boolean search method. What was impossible during the term of the project due to the one-year time limit, i.e. limiting the noise, is still an ambition for the future. This could be done for example, through more advanced search techniques, such as using statistical language models on the content of the debate text in order to make a classification of speeches that do or do not address a specific topic. Another way of addressing this problem could be clustering speeches based on the relevance of words within a speech with regard to a specific topic, where different algorithms are available to calculate this “relevance.” However, this is not off-the-shelf technology and requires more time to implement.

However, even that would not solve the problems. Qualitative analysis of the results will still remain a prerequisite to finding an answer to the question if and especially how the Boerenpartij was linked with “wrong,” as we will explain below in the analysis of the results. It is the context that determines the relevance of a reference. Concretely, and this is essential, this means that every hit must be examined, making the (too) extensive result sets (thousands of hits) too labour intensive.

Knowledge of historical details on the subject also proved necessary. The search term “Holocaust,” for example, does not yield results before 1978, because this name for the murder of the Jews only became established after the American docudrama series Holocaust. This does not mean, however, that the persecution of the Jews was not discussed before 1978. A term like “political delinquent,” on the other hand, was used mainly in the period 1945-1955 and slowly vanished over the following years to be replaced with words like “ex-national-socialist” or even “war criminal.”

After our first search strategy and the assessment of the hits on the combination of the queries “wrong” and “ Boerenpartij, ” the question emerged whether we had now found all relevant hits. To check this we employed a second search strategy. It consisted of a series of specific queries that were continually adapted and compared:

This method of searching and comparing results eventually produced twenty relevant hits, with a clear peak in 1966, the year of the Adams affaire, with eight hits. As the query was small every time, it was easy to experiment, which resulted in interesting and not directly sought results. References to “Fascism” for example, came from the communist CPN side a disproportional number of times: of the 499 references to Fascism between 1958 and 1982, 177 were made by the CPN, meaning that this small party was responsible for a third of the total number. The Boerenpartij referred to Fascism 18 times. Not once did the Boerenpartij refer to “national socialism” in the relevant period and only 3 of the 208 times was “anti-Semitism” mentioned by the Boerenpartij.

Both the specific and the broad search strategy finally resulted in the same (relevant) hits and can therefore be considered (reasonably) reliable; however, it should be mentioned that 100% completeness (recall) cannot be guaranteed. The search engine indexes the source documents on the word level, and errors can occur due to the earlier mentioned OCR errors, unknown spelling variations, and ambiguity due to wrongly interpreted punctuation marks and hyphens. One example is the comparison that senator Algra from the Anti Revolutionary Party (ARP) made on 25 May 1965 between NSB and Boerenpartij. The search engine missed this hit in both strategies because in the text NSB was followed by a comma instead of a full stop (N.S.B,).

The fact that we found this debate anyway underscores that results by search engines retrieved from scans of historical documents to some degree remain more unreliable even after the most careful OCR corrections than born digital documents. Scanned documents are as unique as their originals: one small compositor error, a small layout change, is copied on the scan and is not always detectable even with the latest of techniques and best corrections. Unlike the current digital Parliamentary reports, which are much easier to read for search engines, any piece of paper that was scanned by the Koninklijke Bibliotheek has its own uniqueness, as does anything that was preserved in the pre-digital era. This is important when the possibilities and impossibilities of digital historical research are at issue: if a relatively standardized corpus like the Proceedings is accompanied by so much uncertainty, then digitized archives will present an even larger challenge.

What did the result sets yield with regard to content? The first link between “wrong” and the Boerenpartij in the Proceedings was 19 November 1963, during the approval of the budget of the Ministry of the Interior. Koekoek complained that the Dutch Secret Service (BVD) had listened in on him and that a letter had been stolen. He presented himself as a victim of the establishment that discredited him as a right-wing extremist. [18]

Koekoek did not present clear evidence, but was indirect and not very concrete. This resulted in an awkward debate between him and the conservative-liberal Minister of the Interior Edzo Toxopeus, in which Toxopeus rejected Koekoek’s allegations as vague and insinuating. For Koekoek, however, the issue was that he was made out to be “wrong” and a danger to the state. J.H. Scheps, a World War Two resistance fighter and a heavyweight within the labour party, intervened by pointing out that nobody thought Koekoek was “wrong,” but that Koekoek did have to back up his allegations with evidence. Otherwise, according to Scheps,

We will return to the atmosphere we knew in the years when the N.S.B. started and when all manner of forces weakened the essential value of parliament. All those individuals who are reputed to stand up for the people at that time proved what it means when the powers of parliament are weakened. [19]

Here the connection with “wrong” consisted of a referral to a mentality and procedure that were “wrong,” not to “wrong” political positions or ideologies. This observation clearly underlines the importance of qualitative analysis. In Scheps’ eyes the Boerenpartij acted “wrong” rather than was “wrong.”

In May 1965 anti-revolutionary representative H. Algra, also a former resistance fighter, tried a slightly different tack in the Senate. He compared the success of the Boerenpartij to the success of the NSB in the 1930s, a consequence of feelings of distrust and dissatisfaction with government policies among part of the electorate. Algra hastened to say that this obviously was not enough reason to put both parties on a par, but felt a warning was in order. [20]

Marcus Bakker, leader of the communist party (CPN), went a little further during the Budget Debate on 12 October 1965. In his opinion the government elicited fascist thinking from the Boerenpartij with its policies, thereby stimulating political instability. [21] Koekoek replied the next day that the Boerenpartij had nothing to do with Fascism:

Mr Bakker brought it up. Now let him prove – he has every opportunity to do so – that we have anything to do with the N.S.B., and that we have anything to do with fascism. I can sit back and wait confidently, because I can declare in advance, that we do not. Regardless of how well he can speak his piece, he will certainly not be able to get out of this one, even if he has an hour to do so. [22]

The remarkable dynamics of accusing and demanding evidence, that would prove characteristic of the debates regarding Koekoek and his party, started to take shape.

Up to that point attacks on the Boerenpartij in the debates had focused on its unparliamentary conduct and rabble-rousing that was reminiscent of the NSB. But the Adams affair that started after the 1966 summer recess was of a different nature: here a Boerenpartij senator’s national socialist past was exposed. This was no longer about NSB-like methods but about a former collaborator who thought he could return to the political arena without reflection on his war past.

Without an advanced search engine we could also have found the debates regarding Adams. What we would not have discovered, however, is that the Adams affair was a (once-only) break with the way in which members of parliament had treated the Boerenpartij so far. Only this affair was literally about “wrong” Boerenpartij-politicians and the key question was on what conditions a former political delinquent was allowed to participate in national politics again. All other references (before and after) were insinuations about the political conduct of the members of the Boerenpartij that leans towards or plays into the hand of Fascism.

In fact, our project is a clear example of a combination between a quantitative (give me all references to x) and qualitative (what does x mean by close reading) approach.

The State-Provincial elections on 20 September 1966 resulted in the installation and swearing in of the new members of the Senate. Immediately following this official ceremony VVD-senator Baas was given the floor on “personal business.” It stood to reason, Baas argued, “ that we would welcome the members of [the Boerenpartij] and treat them like our new colleagues. ” [23] However, the war past of one of them made this impossible. This person was Hendrik Adams, whom Baas had met in his position as teacher at a state agricultural college. In class, Baas said, Adams had “systematically propounded his ideas” on the importance of collaborating with the occupier. In 1944 the situation escalated and Adams had threatened Baas with the words: “I, Adams, will make sure you are deported as soon as possible.” These words echoed in his mind for many years, Baas said: “More than 20 years later our paths cross again. You will understand, Mr President, that I am not able to be amicable towards Mr Adams” [24]

The truth was out: Baas could not accept Adams as a colleague; the rift caused during the war was not yet mended. According to Baas, Adams had in no way expressed regret for his “wrong” attitude during the war. “If I believed,” said Baas, “that Mr Adams understands why he was convicted and lost the right to vote, and that he had come to the realization that he was wrong, then I might have been able to forgive him.” [25]

The bomb exploded after the Senate session ended. “Boiling with anger [Adams] entered the coffee room of the Senate,” Baas later explained, “and stormed towards me. ‘You miserable oaf…’ he shouted, among other things. And then I thumped him.” [26] The scuffle, which was widely reported in the newspapers, shows the emotion of the renewed confrontation between people who had been each other’s enemies during the occupation years under a very different balance of power. At moments like these the question of conditions for renewed forms of living together between former “right” and “wrong” citizens became urgent.

On 4 October Adams was given the opportunity to defend himself, and was unsuccessful. Adams evaded the accusations, believed that there was a conspiracy against the Boerenpartij, and refused to apologize for his pro-Nazi behaviour during WWII.

For Baas a demonstration of regret was a necessary precondition to be able to accept a former collaborator as a colleague in parliament. During the “personal business” speaking time Adams had been given the opportunity to do this: distance himself from his “wrong” past and show that he had reformed. As he had not, Baas felt the “former political delinquent Adams” lacked “every personal right to be a member of this Senate”: “ For other political parties too things would have been different had Adams admitted that he was ‘wrong’ during the war. ”

Instead, Adams emerged from the debate as an incorrigible national socialist collaborator and cowardly traitor, as well as an anti-Semite [during the war he wrote for an anti-Semitic magazine] of the worst kind. In fact, Adams’ political death sentence was already signed. On 25 October 1966, in the Senate, Adams’ letter of resignation was read:

The unworthy and above all unlawful treatment, which I have been forced to endure during the previous two sessions, in a hostile atmosphere from all Dutch political parties and under your & – not very authoritative & – leadership, is unacceptable to any honest Dutchman and it is only on the basis of this consideration that I hereby inform you of my withdrawal as officially elected representative of the Boerenpartij in this Senate. [27]

The Senate took note of Adams’ missive. Although the turmoil continued outside parliament for some time and schisms led the Boerenpartij into hot political waters, there were very few references to Adams in either the Senate or the House of Representatives. [28] The dividing lines were drawn clearly: former collaborators who did not publicly renounce their war past were not accepted in the highest political bodies. And, as shown by the motion of no-confidence against Koekoek, politicians who did not appreciate this could count on severe disapproval. Did Adams’ retreat also mean the end of the connection that was made between the Boerenpartij and “wrong?” It’s time to draw some conclusions.

First of all, up to the Adams affair in the autumn of 1966 the Boerenpartij was not associated in Parliament with a fascist ideology, but was “merely” accused of a “wrong” mentality; the conduct of Koekoek and associates allegedly showed the non-acceptance of parliamentary mores and rules, which was reminiscent of the undemocratic conduct of the NSB in the 1930s.

Secondly, the Adams affair meant a rift in the political discourse about the Boerenpartij. An alleged “wrong” mentality was no longer the issue. The affair demonstrated a totally different aspect of the link between “wrong” and Boerenpartij: the discussions about Adams showed the unwritten conditions that former collaborators had to meet in order to be allowed back into political decision making. Adams was unable to stand his ground because he would not renounce his past. In fact, he made it clear he did not regard his war past as “wrong.” The fact that Koekoek stood by him and attempted to expose politicians in other parties as “wrong,” almost resulted in political suicide.

The third conclusion is that after 1966 hardly any links were made in Parliament between the Boerenpartij and “wrong.” This begs the question whether the rejection of Adams also served another purpose, namely to render the Boerenpartij harmless. Disqualifying political opponents with a reference to the Second World War was (and is) an effective means to do so. Based on our findings we may state that the association with a “wrong” mentality by itself had little political effect. The Boerenpartij continued to grow in the period between 1963-1966. However, when a concrete person was convincingly exposed as “wrong” and the party did not unambiguously distance itself from this individual, electoral success evaporated.

As the first two conclusions are the result of interpretation of the gained results (close reading), the third – no links between the Boerenpartij and “wrong” – couldn’t have been drawn without a digital approach, and is promising for further historical research. More research would benefit our understanding of political transition periods and the relationship between electoral success of newcomers to the political scene and the attempts of the “old” parties to disqualify their intentions.

Unfortunately, digitized sources are generally used in much the same way as the originals. The advantage of digitization is then limited to easy access (no more trips to the university library) and to finding what you are looking for (a particular debate or speaker) more quickly. The state-of-the-art search engine that was developed in the War in Parliament project enabled the combination of several search queries in order to systematically check if, and if so, in what way the Boerenpartij was linked to “wrong” in the parliamentary debate. In other words, in this project we did more than just speed up our research by formalizing our research question. During this process the “War in Parliament” pilot project has taught us some valuable lessons.

The first lesson concerns the nature of the information that one looks for in the context of studying references to WWII in general, and in this particular case the relationship between the Boerenpartij and “wrong.” With the exception of debates that explicitly concern WWII (debates about the release of war criminals, pensions for war victims etc.). WWII tends to be a very implicit topic in the parliamentary debates. References are often made in passing and some are very hard to capture. This is one of the reasons why machine-learning techniques with which we experimented (see above) failed. The second lesson was that the relevant phrases are time-specific (Holocaust, political delinquents) and semantically ambiguous. To give just one example of semantic ambiguity: “ bezetting ” (occupation) in the sense of the German occupation of the Netherlands between 1940 and 1945 also harvested references to the occupation of departments, hospital beds etc. So, the need for close reading remains a prerequisite to gaining the relevant results. [29]

The big challenge is therefore how to improve retrieval performance. In a proposal we submitted to the Netherlands Organization for Scientific Research this year, and in which the digital approach is crucial to harvesting our results, we build explicitly on the experiences gained in War in Parliament. [30] In this proposal, and in addition to the current state-of-the-art Boolean query suggestions, we plan to add three key extensions aimed at improving the suggestions and thereby the retrieval system’s effectiveness. One way is to include synonym structure, either user-contributed, corpus-generated or extracted from a dictionary or back-of-the-book-indexes (e.g. knowledge based enrichment). The second addition is semantic in nature; from experiments involving media studies researchers, we know that semantically informed search suggestions (based e.g., on people, organizations and locations) are key to exploratory searches in digital humanities. We aim to integrate such suggestions within the proposed Boolean query suggestion environment. [31] We will visualize such semantic suggestions in term clouds to analyze them. Finally, the third addition, is to make Boolean query suggestions time-aware. [32] References to WWII were, as reported above, not only implicit in many cases, they were also subject to trends (changes in vocabulary). Terms that make good query suggestions during one period may not be very effective during another. The insights gained in time-awareness will be integrated into the query suggestion process through a combination of time series analysis (of term frequencies) and co-occurrence analysis. With these improvements, we hope we’ve found a solution for the main problems experienced in the War in Parliament project.

This is the more important, because WWII has been the most powerful benchmark of political morality in post-war Western societies, and political elites were (and still are) in permanent need to relate to the defining experiences of warfare, political oppression and genocide. An interdisciplinary approach will provide us with models for a better understanding of that past. Extensive usage and further development of existing text-mining tools will enable us to address questions about political transition periods (in this article a new right-wing party entering the political stage) and whether or not there is a relationship between the utilization of WWII as a political argument and these transitions. A digital approach, as we have seen, facilitates the analysis of large text corpora over long time periods.

We want to conclude this article with a remark on the conditions which are needed to conduct digital historical research. This is all the more important because it is a much neglected subject, not only in the scholarly literature but also by technical utopianists who push for a revolution in the humanities. The Dutch Academy for Sciences (KNAW), for instance, launched an innovative programme for the humanities that primarily focuses on tools. In this journal, the Swedish scholar Patrick Svensson, who can hardly be accused of scepticism towards digital innovations, wrote in 2012: “‘While the ideas of grand challenges and big humanities certainly have attraction and require forward thinking in order to identify complex problems and large-scale visions, we should be careful not to uncritically accept the frame of big humanities, which, for instance, has a tendency to be coupled with a positivist agenda and a homogenization of the humanities.’” [33]

Besides the danger of homogenization of the humanities, which means here the desire to emphasize on quantitative criteria in the humanities and thus to squeeze out the qualitative (and therefore less measurable) ones, we need to realize that digital source material is a prerequisite to apply text-mining techniques. This may sound like we are stating the obvious, but we are not; in practice much of the material is still in analogue form, and can therefore not be searched using text-mining techniques. At NIOD, for example, less than three percent of the total archival material has been digitized. If the large-scale version of the methodological innovation is to be successful, mass digitization of paper collections is inevitable. This step is in danger of being skipped because there is so much focus on the more “sexy” development of tools.

Furthermore, much archival material is in formats that either cannot be transcribed at all as text [e.g. graphical material] or which has only been digitized as a page image but has not been transcribed as text [e.g. page images of manuscript materials or documents that for various reasons cannot be processed with OCR]. This means that digitization of this material can help us retrieve certain information more easily, but it will remain necessary to analyse results by hand at a rather low level. But given these challenges, the success of methodological innovation first and foremost depends on digitization of paper collection for which there is often no money or no will (or both). However, if this isn’t done, only a small number of historians will get involved in the much needed discussions on the improvement of digital techniques to conduct historical research.