Software used:
- Python 3 (I'm using 3.9.13)
- matplotlib 3.8.2
- mpltern 1.0.2
- numpy 1.26.2
Older versions of python, matplotlib, mpltern and numpy probably work as well.
The required libraries are hopefully easily just by running pip install -r requirements.txt from this folder
or maybe python3 -m pip install -r requirements.txt to ensure you are actually using python3
To get access to the data, download the CKCC data from their gitlab or the archived data at DANS and extract it into this folder. I downloaded revision 8bff13f390e5b32bb331c86854f9c85e97438482 of the Gitlab repo, as downloading from version 2 of the archive was not possible when starting this work, but it may be functional again with version 3. After extraction you should have a ckccRestored/1677705613221-Project_Circulation_of_Kn/original/data/corpus/data folder with subfolders for the corpuses, wherein there are txt files and xml files containing the letter transcriptions and metadata.
To run the program, just run python3 main.py from this folder, most of the output will be to stdout
and the figures will be in popup windows. You need to close the window to have the program move forward. The figures are also automatically saved to separate .png files.
This project tries to answer two questions:
- How does the popularity of different languages vary for the correspondence of Dutch golden age scientists, and why?
- Which languages do specific prolific letter writers, namely Hugo de Groot and Christiaan Huyghens, use during their life and how can they be related to the writers' biographies?
While Latin was the predominant language for learned communication in Europe during the Middle Ages, French would start to replace it especially for international and diplomatic use during the Early Modern period. Similarly, European court culture was predominantly in Latin until French gradually replaced it in the 18th century (De Bom). This common culture was an outgrowth of the international Latin-literate learned communities, the 'respublica literatia' (De Bom). Scientific writing would continue to be done mainly in Latin for this whole period, but vernacular languages also gained more and more use for domestic communication. By looking at the senders of letters and the language they write in, we can gain a glimpse at what languages they expected the receiver to understand and which language they considered the correct one for their particular letter, be it due to subject matter and established terminology or their own level of language proficiency. The shift in language use was not a sudden, but rather gradual change. In the Low Countries, Latin literature started to lose relevance during the early 17th century in the south, but in the north it remained relevant during the 18th century (Deneire). During this time, Latin was used by academics, clergy, diplomats, and used for poetry, drama and scholarly and scientific work (Deneire). The same individual could use different languages for different purposes or discourses, for instance Latin for learned discourse and the local vulgar language for other purposes in the Early Modern period, as an example of code-switching (Frijhoff et al 2017, p. 12).
The Dutch golden age (ca 1580-1750) was also a high point for the Dutch language, both before and after being disregarded in favour of French among the middle and upper classes (Frijhoff 2017, 95, 98). In the 1580s, the States General of northern Netherlands switched from French to Dutch for communication, after the separation from the partly French-speaking southern Netherlands. During the 17th century there was an increased interest for Dutch as the Bible was translated to it in 1637 and standardization and use as national language was increased (Frijhoff 2017, 97-98).
This project focuses on Hugo de Groot and Christiaan Huyghens, two prolific multilingual authors and scientists active in the Low Countries during the Dutch golden age, though active at separate times. By comparing them against other roughly contemporary Dutch scientists, we can see if their language use follows that of their peers and the time period.
The data used is officially to be cited as HuygensING-CKCC, 2011, "Project Circulation of Knowledge and learned practices in the 17th-century Dutch Republic. A web-based Humanities Collaboratory on Correspondences (Geleerdenbrieven) - archived version d.d. 2013-07-23", https://doi.org/10.17026/dans-xfd-n8y5, DANS Data Station Social Sciences and Humanities, V3, UNF:6:M3CZThdwpPeElXlaFE/RLQ== [fileUNF]
The dataset comes from the CKCC project and consists of 20 020 letters and some other texts from the corpuses of nine scientists
active in the Dutch republic in the 17th century. Some letters are available as transcriptions, others have been summarized.
They all have metadata about the sender and recipient as well as date, place and the language used. This data is available in a TEI-conformant XML format.
Here is an example of a letter from Hugo de Groot's corpus (ckccRestored/1677705613221-Project_Circulation_of_Kn/original/data/corpus/data/groo001/groo001brie01_01.xml#L527-L545):
<!-- BOI -->
<div type="chapter" subtype="artifact">
<interpGrp>
<interp type="id" value="0007"/>
<interp type="date" value="1599"/>
<interp type="sender" value="meursius.johannes.1579-1639"/>
<interp type="senderloc" value="?"/>
<interp type="recipient" value="groot.hugo.1583-1645"/>
<interp type="recipientloc" value="?"/>
<interp type="language" value="la"/>
</interpGrp>
<head type="ed" id="a0007">7. [<date>1599</date>]. Van Joh. Meursius<note n="3"><seg type="source">Gedr. Io. Meursi Exercitatt. critic. I p. 244 (1599)</seg>, vóór de Notae in Plauti Truculentum.</note>.</head>
<div type="opener">
<p><persName key="meursius.johannes.1579-1639">Ioannes Meursius</persName> <persName key="groot.hugo.1583-1645">Hugeiano Grotio</persName> S.D.</p>
</div>
<p><hi rend="i">Odi puerum praecoci sapientia</hi>, aiebat ille<note n="4">Apuleius, Apol. 85.</note>; sed eo dicto, mi <persName key="groot.hugo.1583-1645">Groti</persName>, nihil censeo ineptius. Cur enim? Quippe <hi rend="i">generosioris arboris etiam planta cum fructu est</hi><note n="5">Quintilianus, Inst. or. VIII, 3, 76.</note>, quod Quintiliano iuvene dictitabatur.
Sed aurem vellit <persName>Cato</persName>, qui ‘senilem iuventam signum esse praematurae mortis’ pronuntiat<note n="6">Plinius, Nat. Hist. VII, 52.</note>. Quod utinam falsum? exemplis enim et experientia comprobatum, non diurnare hoc hominum genus. Dolendum certe: et quod de Caesare<note n="7">S. Aurelius Victor, Epitome 28.</note> olim dictum, ‘non nasci eum debere, aut Reip. non tam cito eripi’, idem de praecocibus et praestantissimis istis ingeniis videtur; quibus si ad frugem pervenire daretur, quanto usui Reip. erant futura? Tua caussa hoc dico, cui in ipsa fere pueritia tam praeclare sapere contigit, ut <hi rend="i">μελλέϕηβος</hi> in ingenii et eruditionis admirationem cunctos erexeris. Cave autem te efferas, aut ames nimis propter haec: quin Dei potius in te beneficentiam agnosce, illique ut gratus sis sedule labora. Gratus autem eris - vide quam sollicitus sit hic affectus, quem non contemnes - si dona, quae tibi largiter suppeditavit, nominis ipsius sanctissimi solius gloriae, Reique publicae utilitati impendere studebis. Quod nisi facis, infelix es, immo nihili. Hoc cogita, et me ama: atque haec ad Plauti Truculentum examina.</p>
</div>
<!-- EOI -->
and one from Caspar Barlaeus (ckccRestored/1677705613221-Project_Circulation_of_Kn/original/data/corpus/data/barl001/barl001-corpus.txt#L1-L26):
### 0001.xml
<TEI>
<teiHeader>
<meta type="id" value="0001"/>
<meta type="date" value="1615-08-31"/>
<meta type="sender" value="barlaeus.caspar.1584-1648"/>
<meta type="senderloc" value="?"/>
<meta type="recipient" value="vorstius.conradus.1569-1622"/>
<meta type="recipientloc" value="?"/>
<meta type="language" value="la"/>
</teiHeader>
<text lang="la">
<body>
<div type="letter">
<div type="opener">
<p lang="la">Clarissime D. Doctor,</p>
</div>
<p>Memor officii mei & promissi exemplar hoc <persName key="bogerman.johannes.1576-1637">Bogermanni</persName> ἐλεγχομένου [<hi rend="i">confutati</hi>] ad te transmitto, ut ejus lectione aliquam doloris istius, quem adversariorum tibi infligit pertinax improbitas, partem soleris. Est enim argumento quoque tuus, quia de te, aut potius adversus antagonistarum tuorum rabiem pro te. Quanquam in ea nos incidisse tempora doleam, in quibus etiam, quod verum est, dicere suspectum est & noxium. Quod an hujus nostri seculi proprium sit, an vero veritatis & innocentiae individuum concomitans, nondum decernere potui. Certe quicquid sit, non est despondendus animus, praesertim iis, quos recti conscientia erigit, quique ad hoc vocatos se norunt, ut imagini Filii Dei conformes reddantur.Sed quid ego haec apud te Vir Clariss. qui nec me monitore opus habes, & illum hactenus animum gessisti, quem nec heroum offensa, gravis alioqui, nec plebeiorum insultus, nec fratrum odia a recto dimoverunt, quem ut tibi porro infractum praestet Deus consolationum, etiam atque etiam rogo. Hoc tamen scito, minime mihi satisfactum fore, si pro hoc nostro munusculo gratiarum actionem transmiseris. Duo siquidem mihi cum <persName key="@Simonides (556-468 v.Chr.)">Simonide Poeta</persName> sunt scrinia, unum Gratiarum actionum, alterum largitionum & donorum; quorum illud vacuum nobis semper & inane est. Quare verba mihi dari nolo, rem postulo, hoc est donum pro dono, & quidem, ut quam libere apud te agam videas, Anti-Sladum tuum. Tunc enim lucrum me fecisse rebor, & χρύσεα χαλκείων [<hi rend="i">aureum aere</hi>] commutasse, si hoc potitus fuero.</p>
<div type="closer">
<p>Vale vir clariss. cum uxore & liberis, & meliora a Domino spera.
<date>Prid. Cal. Sept. 1615</date>. Raptim. Festinante en jamjam abitum ad te parante <persName key="molino.domenico.1573-1635">Molano</persName>.</p>
</div>
</div>
</body>
</text>
</TEI>
For this project, I am only looking at the metadata, that is the <interp> elements inside <interpGrp> elements in the XML files and <meta> elements inside <teiHeader> elements for the TXT files. I am also limiting the data only to those letters which have a known sender, approximate date, and listed language for the whole letter. Since this corpus is from several different institutions and presented in slightly different XML markup, I am limiting the data to those that have the language listed in the metadata section and not as an attribute for the transcription itself. I identify letters in the XML by searching for <div> elements of the subtype "artifact", for the TXT files I select <teiHeader> elements having <meta> children. Another subtype "replaced_artifact" may also refer to letters, possibly lost or not available or not transcribed, but those elements are excluded for now to make the parsing easier.
Anthoni van Leeuwenhoek's corpus (leeu027) is discarded since it seems to be from a bilingual Dutch-English collection of his letters with no indication as to what the original language of each letter was. While Constantijn Huyghens' (huyg001) letters do follow the expected XML format, most of them are discarded due to not having language info on a letter level in the metadata. Descartes' corpus (desc004) is also discarded due to the structured XML data not being present in the archive.
This leaves us with the corpuses of Hugo de Groot, Christiaan Huyghens, Constantijn Huyghens, Caspar van Baarle, Isaac Beeckman, Dirck Rembrantsz van Nierop and Jan Swammerdam, which should be 19 011 letters in total according to the CKCC website. de Groot and Christiaan Huyghens get analyzed separately, the rest are grouped together into what I call "The Others".
I parse all the XML files and parse and transform the TXT files to harmonize the XML data in them. I then combine them into one under a common root, after which I go through the children of all <interpGrp> tags to extract senders, dates and languages. I discard all letters where any of these values are marked as missing. The letters are dated with a varied precision, I only extract the year and in case of a multi-year period (presumably mostly two), the first year.
After discarding letters which lack sender, language info or date, we are left with only 11 776 letters since the vast majority of Constantijn Huyghens' corpus is rejected. For de Groot, 7845 letters from 1594–1645 from 312 senders in 9 languages, for Christiaan Huyghens, 3046 letters from 1636–1695 from 230 senders in 6 languages, and for the rest 885 letters from 1612–1685 from 114 senders in 7 languages. The main languages are Dutch, French, and Latin. The minor languages are German, English, Italian, Spanish, Portuguese, and Ancient Greek. Mostly there are only a handful of letters in each minor language, but for de Groot there is a sizeable German proportion and for the other corpuses there is a fair amount of English and Italian.
For the timelines of language variance, I simply plot the amount of letters in Latin, French, Dutch, German, English and Italian per year in my extracted list of years and for the related stack plot I calculate their relative proportions.
I then go through all smaller languages (not Latin, French, Dutch) and print all senders with letters in that language together with a count of their letters in that language and the total amount of letters they have sent.
After that I go through the amount of letters they have sent in Latin, French and Dutch and calculate the relative proportions to create a ternary scatterplot and a backup 3d scatterplot that is easier to rotate and zoom in on. I only looked at letters in these three languages, since those are the three languages that are most present in the CKCC corpus and get their own columns in the project's table on language variation. I also print out the top 100 most prolific authors in these languages in text format before I limit the data to only contain those authors who write in more than one of these three languages. This limitation is mostly to make the scatterplot readable, since a mass of points at the three extremes would require a different representation in form of a heatmap. I also print out that list of "multilingual" authors and the language distribution of their letters.
Throughout de Groot's life, Dutch and especially Latin remain important languages for letter writing, both by him and his correspondents. During the latter half of his life, he started to receive more letters in French and German, though he rarely wrote in French before the 1630s and not very much after that either. This may just be due to the majority of the correspondence in the corpus being from the 1630s onwards, though (see the graphs for sent and received letters). de Groot lived in exile in France 1621-1632, which may explain why he is starting to both receive and send more letters in French. In 1632 he moved to Hamburg and then back to Paris 1634-1644 as a Swedish ambassador. The timelines also show that letters in German are almost exclusively in the same period of ca 1630-1645 from where a huge portion of de Groot's corpus comes. The German material is predominantly from Georg Keller (1589-1652) who was the secretary to the Swedish envoy to Hamburg, Johan Adler Salvius (according to this note), so it seems possible. Salvius is the second-most prolific German writer in this subset. Using German as a (presumably) diplomatic language seems surprising, but it may say something about the language profiles of the Swedish diplomatic corps. The ongoing Thirty Years' War (1618-1648) presumably continued to make the Swedish elite more focused on the German-speaking areas. de Groot himself continued to write predominantly in Latin throughout his life, though some 30 % of letters penned by him are in Dutch and from around 1630 onwards, French. His use of French is presumably now related to his diplomatic work within France. Some interesting outliers in the data are 1611 and 1619 where there is a spike of letters in Dutch, but these are not analyzed in detail here.
Christiaan Huyghens wrote predominantly in French, especially from around 1660 onward which also was his most active period as a letter writer (timelines for sent and received letters). Before that, his yearly output switched often between French and Latin, and he occasionally wrote in Dutch and Latin throughout his life. 1645-1650 seems an interesting French-dominated period both for his received and sent letters. This corresponds with his studies at Leiden University (1645-1647) and Orange College (1647-1649). Huyghens had many correspondents in France and lived in Paris 1666-1681 as a member of the Académie des sciences. After this he returned to The Hague for the remainder of his life. These moves are also reflected in the data, where his Paris period means he almost exclusively wrote in French and after his return to the Netherlands, he started writing in Dutch again. Many of his scientific discoveries were published in the 1650s and 1680s, though he may well have discussed these and more in correspondence throughout his life. This first active period is reflected in the majority of his correspondence being in Latin in the 1650s, but this is not the case for the later period, since his scientific works during this period was written in Dutch, French and Latin instead of exclusively Latin, and it would make sense that correspondence about his discoveries would be in the same language as his publication to be able to use the same terminology. His dataset also contains some interesting outliers in the 1670s: 1676, where he sends many letters in both Dutch and English and 1677, where Latin spikes for both sent and received letters, but a close reading of these years are left outside of this analysis. Most of the Italian letters in the dataset are from Leopoldo Medici to Christiaan Huyghens. Leopoldo was a patron of the arts and interested in science. The remaining Italian letters are from Italian optician-astronomers like Galileo Galilei, Carlo Roberto Dati and Giuseppe Campani.
The remaining dataset (stack plot and timeline) is very much smaller and scattered, with some years lacking any letters, but it does show an intriguing pattern of Latin being predominant from the 1620s to the 1640s, before English becomes predominant in the 1650s and in turn is replaced by Dutch in the 1660s as well as French to some extent, especially in the 1670s. The English portion is perhaps attributable to close connections to members of the Royal Society (to be founded in 1660), with Constantijn Huyghens (1596-1687) writing 31 letters in English, as well as Royal Society members like Robert Hooke and William Brouncker corresponding with Christiaan Huyghens almost exclusively in English. Comparing Christiaan Huyghens' corpus against the Others' corpus shows that for Christian and his correspondents, English is used most actively in the 1660s and 1670s, so anything related to the Others' corpus for for c. 1640-1670 must be treated with extreme caution as it may well just be due to the low amount of extant letters from that period in the corpus. The predominance of French is perhaps due to France's rise as a superpower, being at its height during the reign of Louis XIV (r. 1643–1715, personal rule 1661-1715). The persecution of the Protestant Huguenots during his reign probably also caused some of them to emigrate to the Dutch republic or petition their brothers in faith to help them, in French.
The ternary plots (for de Groot, Huyghens and Others) show many multilingual authors writing in varying amounts of French and Latin exclusively, and some using Latin and Dutch exclusively, not very many using French and Dutch exclusively. Very few people are using all three languages, among them Hugo de Groot (1583-1645), Constantijn Huyghens (1596-1687), Christiaan Huyghens (1629-1695) and Jan Swammerdam (swam001, 1637-1680) who are people whose corpuses are used, so presumably they have the best coverage of letters written in different languages for different receivers and uses. Other interesting trilingual letter-writers are Constantijn Huyghens (1628-1697), Queen Kristina of Sweden, Johannes Uytenbogaert (1557-1644) and Johan Brosterhuyzen (1596-1650). Further research could go into this correspondence for a close reading to find out whether all these letters are written by the senders or by a secretary, and what their contents are per language.
de Groot, being of a previous generation to Christiaan Huyghens, represents a mostly Latin-and-Dutch letter writer, whereas Huyghens has a more polyglot approach also including writing in English, but with decidedly less Latin in the 1660s onwards, which seems in a rough agreement with the other scientists in the dataset. The Latin respublica literatia is replaced with overlapping French/Dutch/English/Italian networks of science communication. Latin's lesser relevance during the latter half of the 17th century seems in accordance with Deneire and Frijhoff 2017, though a close reading of the letters would be needed to categorize what kind of subjects were discussed in which language.
The dataset is dominated by Hugo de Groot's correspondence. This problem also existed in the complete CKCC corpus, but there Constantijn Huyghens also contributed with almost as many letters as de Groot, and my subset of data also exclude some other corpuses. Additionally, de Groot is a bit of an odd duck in the CKCC corpus, being more of a humanist, jurist, and diplomat than a natural scientist. Christiaan Huyghens is the other dominant figure in my dataset, and all the other remaining corpuses combined have less than a third of his correspondence. Some authors and letters may be duplicated between the corpuses, for instance Constantijn Huyghens (1596-1687) appears both in his own corpus as well as his son Christian's, though I have not checked whether the letters are duplicates. Trying to analyse 17th century use of language by counting only one language per letter is not very representative of the actual practice at the time, where one letter could contain paragraphs or sentences or even sentence fragments in many different languages. There is a case to be made for analyzing the use of German, English and Italian in more detail, since roughly 200 letters in my subset are in German, roughly 80 are in English and roughly 50 in Italian, which are maybe surprising based on the existing research. Alternatively, they could have been completely ignored as the focus should be on the major languages French, Dutch and Latin and they are mostly the output of a handful of correspondents other than the corpus owners. I keep these minor languages for the timelines and top author lists per language, but I exclude them from the scatterplot since trying to do a 6d scatterplot was beyond my ability.
It should be pointed out that the CKCC corpus consists of scientists well-known enough to have archived their correspondence and receive later publications of it. It cannot be said to be representative of all Dutch letter-writers of the period, not to mention the people they were in contact with.
It is entirely possible that I am misunderstanding how some of the XML files are structured and that my filtering of letters to analyze is discarding interesting categories of letters. Since I developed the program starting with de Groot and looking mostly into the structure of his corpus, and I got to a point where I could analyse almost all of his letters, I have just assumed that the remaining corpuses behave mostly the same. Since most of the non-epistolary material is concentrated to Constantijn Huyghens' corpus and I exclude most of it, I assume I am not including a lot of e.g. minutes of a meeting that the CKCC project mentions.
I am blindly trusting the CKCC corpus data to be correct, though the project website does mention many problems it encountered with the dataset. Presumably many of them were fixed during the course of the project, such as the different spelling variations of names.
No proper statistical analysis of the timeline is done, no correlation analysis or interpolation or proper figuring out of how the very varying amount of letters per year affects the analysis. This is mostly due to lack of time. I am also not doing any proper cluster analysis or heatmaps or the like for the scatterplots, just judging by eye, so any results should be taken as extremely preliminary.
The scatterplots are littered with the labels for each writer. This is a bad compromise where the main point is trying to make the writers identifiable. I could have made them anonymous points and judge them as a nameless mass, or picked a subset of them, maybe the top 20 writers, but I do not know if that would be more representative or more misleading. Another alternative would have been to use some other plotting library and make the labels appear only when the mouse cursor hovers over a point, but this was not done due to time constraints.
The analysis is pretty surface level and does not go into detail on how to classify "monolingual", "bilingual" or "trilingual" authors into interesting national/societal groups.
The output could be saved in eg. a csv file instead of directly in stdout for re-use and more data analysis.
De Bom, Erik. "Diplomacy and Court Culture". In Brill’s Encyclopaedia of the Neo-Latin World edited by Craig Kallendorf et al. Brill Reference Online. Web. 2 Feb. 2024.
Deneire, Tom. "Neo-Latin Literature—The Low Countries". In Brill’s Encyclopaedia of the Neo-Latin World edited by Craig Kallendorf et al. Brill Reference Online. Web. 2 Feb. 2024.
Frijhoff, Willem. "4. Multilingualism in the Dutch Golden Age. An Exploration" In Multilingualism, Nationhood, and Cultural Identity: Northern Europe, 16th-19th Centuries edited by Willem Frijhoff, Marie-Christine Kok Escalle and Karène Sanchez-Summerer, 95-168. Amsterdam: Amsterdam University Press, 2016. https://doi-org.libproxy.helsinki.fi/10.1515/9789048530007-005
Frijhoff, Willem, Escalle, Marie-Christine Kok and Sanchez-Summerer, Karène. "Languages and Culture in History. A New Series" In Multilingualism, Nationhood, and Cultural Identity: Northern Europe, 16th-19th Centuries edited by Willem Frijhoff, Marie-Christine Kok Escalle and Karène Sanchez-Summerer, 7-14. Amsterdam: Amsterdam University Press, 2016. https://doi-org.libproxy.helsinki.fi/10.1515/9789048530007-001