This paper is based on the study of text reuse in the Finnish press from 1771-1920. In the Computational History and the Transformation of Public Discourse in Finland (COMHIS) project, we found 61 million occurrences of similarity, which formed 13.8 million clusters of reuse. This material also included strikingly slow processes of repetition, and the longest reuse cases were almost as long as the time span of the project. In sum, 2.03 million clusters, 15 per cent of the total amount, were longer than 12 months. As well, 76,259 clusters spanned over 20 years or more. The longest span was 146 years. The paper explores the volume and nature of this long-term text reuse in the Finnish press and analyses three distinctive features of slow repetition: newspapers as a site of memory, newspapers as an archive and the political ramifications of reuse. The paper argues that the habit of reprinting old texts aimed to bridge the gap between past and present, emphasising the continuity between old and new. On the other hand, there were cases where past texts were activated precisely for the opposite purposes, to obscure the past and to show how different the bygone world was.
In the project Computational History and the Transformation of Public Discourse in
Finland (COMHIS) [1], we studied text reuse in the Finnish press. Our work is based
on the digitised corpus of newspapers and magazines from the collection of the
National Library of Finland. Our time span covers 149 years, starting from the
publication of the first issue of the Tidningar Utgifne af et Sällskap i Åbo in 1771 and end
There has been growing interest in the study of text reuse or, more generally, the overlaps between texts and text corpora. Lincoln Mullen analysed biblical quotations in American newspapers [3]; Marco Bühler et al. studied, for example, Homeric quotations in ancient texts [4]; the project Knowledge, Information Technology and the Arabic Book (KITAB) explored text reuse in the corpus of Arabic texts from the period 700–1500 [5]; and, in their Viral Texts project, Ryan Cordell and David A. Smith analysed text reuse in the American press in the nineteenth century [6]. The present Trans-Atlantic Platform project Oceanic Exchanges, led by Cordell, aims to study textual migration on a global scale and bring together researchers from six countries: Finland, Germany, Mexico, the Netherlands, the United Kingdom and the United States [7].
The culture of copying has many historical ramifications. In the nineteenth century, newspapers could circulate content that was, if we might say so, ‘born journalistic’, that is, written by the editors and authors of newspapers themselves. Papers could of course also reuse other kinds of textualities, from biblical quotations to telegrams, from paid advertisements to public announcements and from circulars to readers’ letters. Our COMHIS project has concentrated on those computational methods that can identify similar passages within the corpus of newspapers and magazines to allow for an analysis of different forms of text reuse within the Finnish press. In this article, we concentrate on one particular feature that has become evident in the project: the long-term reuse of texts. In nineteenth- and early twentieth-century Finland, viral news spread quite rapidly across and throughout the newspaper network, but there were also strikingly slow processes of repetition. The longest text reuse cases in the material are almost as long as the time span of the project. Our questions are therefore: What has been the significance of the long-term reuse of content, what were its main characteristics and why were particular texts republished over a long period? We start by discussing the overall character of this form of reuse; thereafter, we present three historically important features of long-term reuse. 2
In the COMHIS project, we developed a special solution for detecting text reuse, which is based on NCBI BLAST (National Center for Biotechnology Information, Basic Local Alignment Search Tool), a software originally created for comparing and aligning biomedical sequences. In our method, we encoded the data into protein sequences, which then could be read by BLAST to identify regions in the sequences that overlap. These pairs were clustered based on their offset values so that overlapping passages were interpreted to be part of the same cluster. A cluster then contains all found occurrences of a particular reuse. These may be fragmented, i.e. one article is separated into different clusters, due to such issues as the original pagination of texts, a lack of accurate article segmentation or too-substantial OCR (optical character recognition) noise, causing BLAST to detect only a shorter part of the reuse.
The method has been described in an article in 2017 [8], and the software code, as
well as the necessary guidelines, is openly available online [9]. This method proved
highly effective, even in cases where OCR noise makes the text almost unreadable:
according to our evaluation of the recall, 60 per cent of the characters could be wrong
and the similarity could still be recognised [10]. In total, we found 61 million
occurrences longer than 300 characters from
From the perspective of long-term text reuse, the span of the cluster is especially important. In our database, span simply means the temporal distance between the first and last hit of a cluster. Gap, in turn, refers to a break in the repetition chain. In a case with only two hits, span and gap are equivalent. There can however be clusters with several gaps. To give an example, our database includes a cluster with 48 hits (ID 14099508). The span of this cluster is 100 years, from 1820 to 1920, 36,546 days to be exact. The original Finnish-language text was published in the Turun WiikkoSanomat on 8 January 1820, and it was written to promote reading among Finns. In our corpus, the last hit of the text is from the newspaper Huittinen from 30 January 1920. Over time, the text was printed in 40 different newspapers in 22 unique locations. The cluster included long gaps in its repetition chain. After its first publication, the text was published in 1851, after a gap of 31 years. It was again published 10 times in 1895 after a silence of 44 years, which is the maximum gap of the cluster. Thereafter, the text was republished several times in 1904, 1909, 1912, 1915 and 1920 (Fig. 1).
In this example, text reuse unfolded over a period of 100 years, which can
obviously be thought of as ‘long-term’ repetition [12]. However, what does ‘long term’ mean
in our corpus? If we count the span and the gap of all clusters, we notice that 85 per
cent of all clusters are shorter than one year. The major part of reuse clusters can thus
be seen as fast repetition. In previous research,
In total, 2.03 million clusters lasted longer than 12 months. While this on the other hand supports the idea that text reuse predominantly took place rather quickly, sometimes in a viral manner, it also shows that a substantial amount of clusters travelled longer over time. If we extract clusters whose span or gap varies between one and nine years, the amount is still counted in millions. However, after a gap or span of 10 years, the amount starts to diminish. In 259,757 clusters, the span is 10 years or more. In turn, 76,259 clusters spanned over 20 years or more, and 5,888 clusters spanned over 50 years. Furthermore, 289 texts travelled over a century. The longest span in the database is 146 years, and the longest gap is 145 years.
We will continue by discussing three historically distinctive features of long-term reuse in the Finnish press. These are not categories of the most common cases but, in our view, historically relevant aspects of long-term reuse that need to be considered when analysing the cultural ramifications of the press in the nineteenth century. Furthermore, they exemplify the potential of computational methods in assessing the role of the press over a long period. 3
The volume and capacity of the Finnish press expanded towards the end of the nineteenth century. Information circulated with an ever-accelerating rhythm, which becomes obvious in our database of reuse. The majority of reuse clusters started at the end of the period. There were, as already pointed out, also texts that seem to have floated in time, and, evidently, the newspapers not only republished texts that had been recently printed, but they also drew on the volumes of previous years and decades. This is reminiscent of the notion of lieux de mémoire, proposed by the French historian Pierre Nora: there are sites of memory that participate in constructing a sense of community. These sites are not necessarily physical places but also symbolic sites that bridge the gap between past and present [14]. Although newspapers, or ‘dailies’, can be interpreted as epitomes of the present tense, the now, they have also become sites of memory. Newspapers filtered themes from the past to the present and participated in constructing cultural memory. At the same time, papers were not only a repository for remembering issues outside the press; the history of newspapers themselves also became an object of memory.
Already, the first papers of the late eighteenth century offered texts for later reuse.
A poem by two professors of the Academy of Turku, Pehr Adrian Gadd (1727–1797)
and Henrik Gabriel Porthan (1739–1804), was published in the first issue of the
Tidningar Utgifne af et Sällskap i Åbo on 15 January 1771. This poem was republished in
six papers between 1853 and 1871. When the Litteraturblad för allmän medborgerlig
bildning printed
The editors of the newspapers were conscious of the history of the medium. Another example is offered by the first Finnish-language paper Suomenkieliset TietoSanomat. Its sample issue was published on 1 September 1775, including an article on the nature of the Finnish language and the importance of using it for publishing activities. This text was the first hit of cluster 11250221 that included 21 occurrences (Fig. 3). The span of the cluster was 145 years.
As the previous examples show, newspapers participated in commemoration practices by revitalising issues from the past, for example, after 50 or 100 years. The habit of remembering the past in cycles is not a self-evident cultural practice. It seems that in newspapers, these cycles became increasingly important during the nineteenth century.
Gaps within clusters can be seen as indicative of these commemoration practices,
although it must be noted that there can be several gaps in one cluster. Still, the cycles
stand out clearly in the material when the amount of clusters is counted for each gap
in years. For example, there were 4,216 clusters with a gap of 25 years, but only 2,214
clusters with a gap of 26 years. Figure 4 shows the amount of clusters according to the
length of the gap, from 39 years to 105 years. Gaps of 40 and 50 years are the highest
peaks, and there is a shorter peak indicating a gap of 100 years. These cases of
commemoration included various kinds of materials, such as articles with topics ranging
from agriculture to rare natural phenomena. Some news items were also later marked
as historical events. In the Turun Wiikko-Sanomat, published on 15 January 1820, a
short notice reported that Napoleon Bonaparte had been imprisoned on the island of
St. Helena. The same story was republished on the centenary of this historic moment
Various kinds of stories and anecdotes travelled through time within the Finnish press. It seems the editors of newspapers around 1900 used earlier papers as an archive of possible material to be employed and reprinted. One motivation for this kind of selection of old stories to be circulated was likely the need to fill the columns of the paper, as the capacity of the press increased rapidly. On the other hand, the editors had no doubt been looking for stories that would have entertainment value among readers.
Among the long-term reuse clusters, several cases of both Swedish- and
Finnishlanguage anonymous stories have been reprinted various times in different temporal
rhythms. For example, late nineteenth-century papers showed interest in a prayer that
was published in the Åbo Tidningar in February 1791. The speciality of this prayer
was the fact that it was reported to be a prayer for married women. Its content
included passages where a wife was affirming her obedience to her husband. This prayer
was republished by seven Swedish-language newspapers, first in 1862 and 1877 and
then three times during the 1880s and finally in 1890. The original context of the
prayer, according to the Åbo Tidningar, was a divorce case in Germany during the
1780s, in which the wife had complained that she was forced to pray each night in the
presence of her husband. The prayer was apparently copied and translated from a
German magazine, as the same prayer was published in the
In addition, Finnish language anecdotes and stories were shared over a long period. On 2 January 1830, the Oulun Wiikko-Sanomia published a story about the spontaneous meeting between a former prisoner and his ex-guard. In this story, the guard is afraid of the possible revenge of the prisoner but finds that the ex-convict only wanted to buy him a beer. Seventy-four years later, this entertaining story was rediscovered and republished by several newspapers. The first reprint was in the Oulu-based Kaiku on 10 February 1904. After the first reprint, six other papers circulated the same story during February and March.
In some cases, the publishing of old stories from past papers created distance
between the past and present. For instance, on 12 February 1820, the Finnish-language
Turun Wiikko-Sanomat published a news clip reportedly taken from a ‘paper in St.
Petersburg’, which in turn had used a German newspaper as a source. This newsprint
was about large sea serpents that had surrounded a ship somewhere in the open sea.
According to the story, the mariners had shot one of the creatures. The editor of the
Turun Wiikko-Sanomat did not want to take any stance concerning the truth-value of
this report. He only stated that this is not the only newspaper report on sea serpents.
Seventy-eight years later, on 5 November 1898, the Keski-Suomi reprinted this old
text and introduced it as an example of a canard from the past. The next time the same
piece of news was found
Old newspaper material was republished for political reasons as well. We have found examples in which newspapers and periodicals around the turn of the nineteenth century benefited from the material of older papers when commenting on the state of affairs in their contemporary world.
In 1776, the Suomenkieliset Tieto-Sanomat published an announcement by the Royal Swedish Academy of Sciences on a prize awarded to a peasant who seeded the largest amount of root crops in one field. The motivation behind the prize was to promote innovations in agriculture, including crop versatility (instead of only grain crop). This announcement was found 122 years after the first publication, and the span of this cluster of reprints is quite high at 142 years (cluster ID 6317820, Fig. 5). In most reprints (total count value is 19), the old text was simply copied without any further comments. In 1908, however, the working-class paper Sosialisti added its own interpretation when publishing the old newspaper clip. According to the Sosialisti, this case—the awarding of a prize to a hard-working farmer—only showed how ‘Counts’ and ‘Barons’ took advantage of the work of the peasantry and how the Crown wanted to expand its lands by making the peasants clear (and burn over) new areas of woodland for cultivation. This way, an eighteenth-century prize call, motivated by the Enlightenment ideals of the advancement of knowledge, became a part of socialist interpretation of the historical class struggle.
Another example shows how the ‘words’ of Emperor Alexander I were being reprinted and interpreted by late nineteenth- and early twentieth-century newspapers. In this case, the source was the Åbo Allmänna Tidning from 21 April 1810, which published an official notice from the Emperor directed to the people of Finland. In this notice, Alexander assured that he would rule the land by honouring its freedom, constitution and rights. This formulation was quite near to the words Alexander directed to the participants of the Diet of Porvoo in 1809, gathered after Finland’s annexation to Russia [19]. During the second half of the nineteenth century, the Helsingfors Tidningar reprinted passages of this notice in 1861 (20 August) as a part of a polemic the paper was having on the interpretations of the Diet of Porvoo with another newspaper. At this point, the Finnish Diet had not been summoned for over 50 years, and the discussion of Alexander’s words was a way to comment on this situation. Later on, pieces of the same text were reprinted between 1871 and 1913 by several papers (cluster ID 5351089: total number of unique titles 14) in different connections.
In this paper, we have discussed the volume and nature of long-term text reuse in
We have discussed three features of long-term text reuse. Newspapers were a site of memory, where past phenomena were activated for present readers through the process of reprinting. The press also participated in consolidating mnemonic operations that were cyclic in nature: texts were reprinted 25, 50 or even 100 years after their original publication. The second feature is that newspapers served as an archive for the editors. They offered raw material for the expanding press that simply needed texts to be published. Old papers were a repository of stories, anecdotes and other ephemera often neglected by newspaper historians. This material lived long, as our examples indicate. The third feature deals with the political ramifications of reuse. This was particularly important in Finland, which was a part of the Russian Empire until 1917, and the editors often tried to find ways of circumventing censorship regulations. Sometimes, the original political intentions were turned upside down, and the quotations could be used to criticise past conditions.
The cases of long-term text reuse could further be explored by concentrating on the strategies of bringing past news to the present. Based on our research, it is obvious that the habit of reprinting old texts aimed to bridge the gap between past and present, emphasising the continuity between old and new. On the other hand, there were also cases where past texts were activated precisely for the opposite purpose, to obscure the past and to show how different the bygone world was. This tension reveals the ambiguous nature of the press as a vehicle of modernisation in the nineteenth century.
The work was supported by the research consortium Computational History and the Transformation of Public Discourse in Finland, 1640–1910 (293239) and the research project Viral Culture in Early Nineteenth-Century Europe (302433), funded by the Academy of Finland. Computational resources were provided by the CSC–IT Centre for Science, Espoo, Finland, and the data by the National Library of Finland.