Category Archives: Uncategorized

Le Livre d’Artiste – a Digital Scholarship project

Introduction to Le Livre d’Artiste 

This blog post features Mandy and Tiffany’s placements [for the MSc in Digital Scholarship] at the Taylor Editions, where they turn two remarkable examples of le livre d’artiste (the artist book) into digital editions. The making of the artist book is a uniquely French tradition where artists create original illustrations that accompany the texts. However, the illustrations are not merely reflective or decorative – they represent collaborative dialogues between literature and visual art as the artists interpret and visually respond to the literary content. 

Our exhibition features Poèmes de Charles d’Orléans, transcribed and illustrated by Matisse and Le Satyricon by Petronius, with original engravings and ornamental designs by Derain. These works represent two approaches in le livre d’artiste tradition to reimagining book production: Matisse’s lithography creates a manuscript-like appearance that evokes handwritten texts, while Derain’s copper and wood engravings maintain a more conventional book format with printed typography. Both works depart from traditional bookmaking and painting through their reproducibility – le livre d’artiste exists as multiple identical copies, raising complex questions about authenticity and aura of art in the age of mechanical reproduction, like Walter Benjamin once explored in The Work of Art in the Age of Mechanical Reproduction. 

Henri Matisse (1869–1954) 

Edition at: https://editions.mml.ox.ac.uk/editions/matisse

Matisse’s illustrated and transcribed work of Poèmes de Charles d’Orléans, published in 1950. Working from his studio in occupied Nice (France) during World War II, Matisse found creative renewal and resistance in the poetry of Charles d’Orléans, a medieval French poet who was captured at the Battle of Agincourt and spent 25 years as a prisoner of war in England. Thus, the warrior poet was a symbol of French cultural (and political) resistance—a powerful reminder that was much needed. Matisse’ illustrated books also transformed the French Le Livre d’Artiste tradition. Rather than simply providing illustrations to printed poetry, Matisse integrated calligraphy, ornamental framings, symbolic fleur-de-lys motifs, and portraits into unified, organic compositions. Using lithography—a technique that transfers drawings from stone plates—the Mourlot Frères workshops, who completed the printing on 25 February 1950 under the direction of Matisse, were able to preserve the delicate quality of his crayon work. The exhibition features all of the lithographs that the University of Oxford currently holds. 

André Derain (1880–1954) 

Edition at: https://editions.mml.ox.ac.uk/editions/satyricon/

Derain’s illustrated and engraved work on Le Satyricon by Petronius, published in 1951 in Paris under the imprint Aux dépens d’un amateur, stands as a bold reinterpretation of the classical Roman text. Originally commissioned in 1934 by the renowned art dealer Ambroise Vollard, the project was delayed by Vollard’s untimely death and the disruptions of World War II. It was finally completed and published under Derain’s direction in the post-war period, at a time when artists and intellectuals were actively re-engaging with antiquity as a means of cultural renewal. Le Satyricon, with its satirical and unrestrained depiction of Roman society, provided Derain with a rich, transgressive subject. His approach was not to passively illustrate but to embody the chaotic, earthy spirit of the text through raw, expressive engravings. Derain created thirty-three burin engravings and forty-three woodcut ornamental designs, which were engraved by the artisan Paul Baudier. These elements were not mere decorations but formed an integral visual rhythm that complemented the unruly energy of Petronius’ narrative. Rather than adhering to the refined, polished aesthetic of traditional illustrated books, Derain’s work is strikingly physical. His jagged lines, primitive forms, and unfiltered emotional immediacy push against the conventions of the Livre d’Artiste, in much the same spirit that Matisse revolutionised the genre. The etchings were printed by Georges Leblanc, while Baudier executed the woodcut designs with remarkable fidelity to Derain’s forceful graphic language. Derain’s Le Satyricon is not just an artistic achievement but also a significant cultural artifact: an example of post-war French artists reclaiming the classical past on their own terms, embracing its vitality and irreverence as a pathway to creative freedom. 

Encoding Matisse – Mandy 

The research challenges I have encountered grew out of the intricate nature of the poems themselves. Since the warrior-poet became a symbol of French cultural and political resistance (Lalaurie 147-148), Matisse, working during WW2, kept the Middle French spellings in his transcriptions out of homage. Thus, my research question centred on balancing diplomatic transcription principles, which would allow for the preservation of these archaic spellings that symbolise the timeless spirit of resistance, while ensuring accessibility for the readers. 

My solution was to embed nearly 100 explanatory footnotes using the <note> element that appear as pop-up annotations. Originally, I was planning on using the elements <reg> and <orig> (regular and original spellings) to represent the Middle French and the more readable modern French. However, I noticed that Taylor Editions’ stylesheet automatically shows the regularised spelling in the main text with the archaic spelling in the pop-up window (Figure 1), which would lead to a less authentic reading experience. Thus, I pivoted to using <note> element, which gives the readers access to regularised spellings upon clicking. For example, when a user clicks on the asterisk next to the word “cueur,” they immediately see “cœur” (modern spelling), creating a layered, multifunctional reading experience. 

screenshot showing the original spelling in a pop-up window

Fig.1. Taylor Editions’ pop-up window for original spelling 

Since I cannot find any critically acclaimed, modernised version of Charles d’Orléans’ poetry, I employed AI in the process of trying to capture the full extent of linguistic variations. First, I used my knowledge of French to detect irregular spellings. Then, I used Claude Sonnet 4.0 to conduct a comprehensive scan of the poems and found several more irregular spellings. These results were then double-checked by ChatGPT. I also added a note in my editorial declaration stating that if more spelling variations are found, the TEI file will be updated. 

I also added a translation page for all of the poems. For well-known pieces, I incorporated A.S. Kline’s freely available translations. For those without existing translations, I employed Claude Sonnet 4.0 to conduct the translations, which were then doubled-checked by myself and ChatGPT to ensure that there were no mistakes. The purpose of providing translations is to serve as a point of departure and reference for readers who might otherwise be excluded from engaging with medieval French text, rather than producing definitive translations. To conclude, by preserving the archaic spellings via the <note> element while providing layered interpretive support through annotations and translations, I managed to maintain cultural and political resonance while reaching contemporary audiences. 

Apart from ensuring content accessibility, I also attempted to make accessible the delicate illustrations Matisse created (high-resolution images are present on the edition’s website). I employed Claude Sonnet 4.0 to create detailed descriptions of Matisse’s illustrations to serve users with visual disabilities. I generated initial descriptions by asking Claude to only describe concrete, essential visual elements such as colours, textures, and spatial compositions to maximise objectivity. The generated descriptions were then refined by me and put under the <figDesc> element. 

To further enhance accessibility for the general public, I implemented “Iconclass”, a standard classification system that assigns numerical codes to visual subjects and motifs, to provide standardised, concise classifications for these illustrations, as I believe that consistent, clear terminology will be more accessible than art history jargon like “cartouche”. For example, the ornamental borders (similar to “cartouche”) present in various of Matisse’s illustrations are now all classified as “ornament ~ round and curved forms (+ framing decorations)” and encoded with their Iconclass numbers [e.g., 48A9813(+2)] in the TEI. Despite Iconclass’ convenience, I retained the AI-generated descriptions as they provide both accessible classification and more detailed visual descriptions (Figure 2). 

screenshot showing links to iconclass

Fig. 2 Iconclass demonstration 

I placed the classifications at the end of the visual descriptions with corresponding links to the Iconclass website (coded with a <ref> attribute). For example, when a user clicks on the link embedded in the word “Fleur-de-lis”, they will be directed to the Iconclass website (Figure 3), where users can find many images containing fleur-de-lis. 

screenshot of iconclass website showing examples of fleur-de-lis

Fig. 3 Iconclass website for Fleur-de-lis 

Thus, the power of Iconclass extends beyond simple accessibility enhancement as it would enable users to explore how Matisse’s fleur-de-lis features as part of the French visual culture, connecting Matisse’s modernist interpretations to medieval manuscripts and inadvertently echoing the sense of the enduring French cultural identity that Matisse was already expressing through his preservation of archaic spellings. This digital edition thus becomes not merely a reproduction of Matisse’s work, but an accessible portal into the history of symbolic motifs in cultural expressions. What began as an effort to make the content more accessible to the public has created a powerful, convenient research tool. 

These accessibility principles extended beyond the digital edition itself into the accompanying parts of the placement. For example, my colleague and I provided comprehensive introductions, in our exhibition in the Taylorian that features the manuscripts, that contextualise both the Strachan Collection and Le Livre d’Artiste, and English translations, attempting to minimise scholarly barriers that might exclude general audiences. 

In conclusion, I gained insight into how accessibility can be ensured while preserving the authenticity of the materials. At physical, linguistic, and scholarly levels, AI-generated (human-refined) descriptions and translations can serve users with visual impairments, reduce language barriers, and replace exclusionary jargon with standardised terminology. Thus, Matisse’s wartime vision of cultural resistance finds a new digital expression through inclusive and authentic design. 

Encoding Derain – Tiffany 

The process of creating the digital edition of Derain’s Le Satyricon posed a set of editorial challenges distinct from those of Matisse, yet similarly anchored in questions of authenticity, materiality, and cultural continuity. Where Matisse’s project foregrounded the preservation of Middle French orthography as a quiet act of cultural resistance, Derain’s Le Satyricon demanded careful attention to the visual architecture of the livre d’artiste itself—a genre where the text, typography, and illustrations form an inseparable whole. 

I faced the challenge of preserving the physicality of a livre d’artiste, a book where the text, typography, and illustrations are inseparably bound. Derain’s Satyricon is not a conventional printed text; it is a visual object, where the layout, ornamental flourishes, and page structure are integral to its meaning. My editorial decisions therefore centred on diplomatic transcription, prioritising the faithful encoding of the book’s material features—line breaks, page breaks, italics, indentations, and even selected typographical ornaments. Initially, I considered simplifying the layout to improve readability, as many digital editions tend to privilege the fluidity of text over its original structure. However, I realised that such simplification would diminish the visual and rhythmic qualities that are fundamental to the livre d’artiste tradition. Derain’s modernist vision, much like Matisse’s, resists separation between form and content. To strip away the visual scaffolding would be to misrepresent the artwork. In this sense, the encoding process became a deliberate act of preserving Derain’s artistic choices, ensuring that the visual language of the book survives in the digital space. 

While Mandy’s project involved extensive annotation of Middle French spelling variations to support linguistic accessibility, the accessibility challenges I encountered were rooted in the Satyricon’s fragmented, episodic narrative. To assist readers in navigating this disjointed structure, I embedded interpretative scene summaries using the <note> element. These summaries act as navigational aids, helping users unfamiliar with the Satyricon’s abrupt shifts to locate and follow individual episodes without disrupting the original text’s material integrity. In addition, I encoded personal and place names using controlled vocabularies and referenced established authority files to enhance discoverability and interoperability with other digital resources, such as those within Taylor Editions. This parallels Mandy’s application of Iconclass classifications in the Matisse project, where controlled vocabularies support wider connections across collections and visual cultures. In my case, this approach anchors Derain’s Satyricon within broader networks of classical and modernist references. 

Throughout the edition, I chose to preserve original spelling, punctuation, and grammatical features without modernisation, in line with documentary editing principles. While this choice may pose challenges for casual readers, it safeguards the historical and philological value of the source text. Much like the preservation of Middle French orthography in the Matisse project, the decision to retain textual irregularities in Le Satyricon recognises that these details are not errors to be corrected but artefacts to be transmitted. 

Reflecting on the placement, I came to see encoding as a form of cultural stewardship. In fact, the act of encoding transforms into an act of respect: a commitment to carrying forward the artist’s gestures, their creative resistances, into new digital contexts without erasure or flattening. Ultimately, these decisions demonstrate that digital editions are not simply vessels for content, but portals that can preserve the material, visual, and cultural weight of the original artefacts, allowing them to speak to both scholarly and general audiences in meaningful ways. 

Sustainability: SDS at Oxford  

The long-term preservation and accessibility of the livre d’artiste collection data will be supported and sustained through the Sustainable Digital Scholarship (SDS) service at the University of Oxford. SDS will host the digitised images, descriptive metadata, and research data relating to the entire livre d’artiste collection, rather than the editions themselves, which will be hosted on the Taylor Editions. The SDS platform will provide stable, permanent access to the images and metadata, and will include direct links to both the Matisse and Derain digital editions hosted on Taylor Editions. By consolidating the collection’s metadata within SDS, the project not only ensures secure preservation but also creates an expandable framework for future scholarship. As additional works in the collection gradually enter the public domain—70 years after the death of each artist—this infrastructure will allow for the seamless development of new digital editions over time. 

Sustainability is critical for any digital project, as the value of digital editions depends on their long-term stability, accessibility, and integration within wider research networks. The collaboration with SDS directly addresses this need by providing robust infrastructure that supports continued access, scholarly reuse, and future growth. This forward-looking approach ensures that the livre d’artiste collection can evolve as a living digital resource for future researchers and readers.  

screenshot of digital edition

Digitising the European Cult of Saint Margarete of Antioch

Reposted from the History of the Book blog.

“Launch” of the completed edition at the Weston Library Coffee Morning on 11 November 2022 (Lucian Shepherd presenting from the Bibelsaal of the Herzog August Bibliothek Wolfenbüttel)

One Wednesday afternoon, Week 1 of Trinity term 2022, I was perusing through the Digital Bodleian database. Having finally set aside the time for Emma Huber’s Taylor Edition course, I was in search of a text to edit. When I first came across Douce MM 493, my interest was instantly piqued with its printed illustrations and hagiographical content, namely a Middle French ‘Life of Saint Margarete of Antioch’. However, as soon as I started scanning through the other folios, I very quickly found much more than Middle French. It contained folios in Middle English, Dutch, Latin and even some woodcuts cut out of an Early Modern Spanish text, and what ties them all together is their joint focus on St Margarete of Antioch. It was at that moment that I knew — this was the text for me.

Figure 1: MS Douce MM 493, Upper flyleaf verso and fol. A2r

Digital Humanities is an area of research which I have only recently discovered. In Oxford, we are quite fortunate to be in the company of many researchers who do use and lecture on digital techniques such as textual encoding, quantitative analysis, and stemmatics. Last year, I attended a ‘History of the Book’ digital launch which showcased the work of graduate students who had digitised their own selected texts, such as Philippe de Thaon’s Bestiary encoded in TEI P5 XML by Sebastian Dows-Miller (Oxford, Merton College Library, MS 249). (For more on this text, see Sebastian Dow-Miller’s post.) This was a truly eye-opening experience for me as Digital Humanities represented a way to merge my passion for language and literature with the mathematical side of me which I had left behind at A-level. After a discussion with Emma Huber, I knew this was an area I wanted to explore.

As two Oxford terms flew by, I felt the growing sense of urgency to start the course this Trinity term before the stress of fourth year, and now that I had finally found a suitable text, I felt more enthused than ever. To summarise briefly the main contents of the miscellany, I have laid them out as bullet points below:

  • Front and rear endleaves: Two woodcuts of the life of St Margaret taken from Pedro de la Vega’s ‘Flos sanctorum’, Medina del Campo, 1578.
  • ff. A2r – B8v: A printed Middle French ‘La Vie de sainte Marguerite’ from c.1495? with woodcuts throughout.  It, however, is missing folio A8.
  • ff. 2r – 5v: Folios 2 and 5 of a fragmentary printed Middle English ‘Life of Saint Margarete’, suggested to be from 1493.
  • pp. 369 – 372: Two folios from a Dutch book ‘Den Roomschen Uylen-Spiegel’, ed. J. Lydius (Dordrecht, 1671), containing a hymn for St Margaret in Latin and in Dutch.
Figure 2: MS Douce MM 493, Lower flyleaf recto

The main reason for the eclectic nature of this miscellany is its collector, Francis Douce (1757–1834). Well known by archivists for perfecting the “cut-and-paste” technique, he often disassembled volumes and cut out woodcuts to be glued down in other copies or held in guard books for fragments. This can best be seen here with the front and rear woodcuts bound into the volume and on the lower flyleaf recto where you can see an image, the size of a stamp, glued on to the page. For more on Douce, see his entry in the Oxford Dictionary of National Biography.

When looking for a suitable text, I learned that there are a number of points which you should consider: Firstly, it should be of a suitable length to transcribe and encode over the time period of the project, in this case the eight weeks of an Oxford term. Secondly, it should be a manageable project with a clear plan from the start. You might think about whether you are going to transcribe the whole text or only an extract from it and which codicological features you are going to include, such as marginalia, shifting hands (i.e. scribes), and ownership inscriptions. Thirdly, you of course must consider any ethical and legal issues which may arise. For example, you should always doublecheck the copyright license of any text or photo you plan on using, as any derivative work, such as a digital critical edition, would still fall under the license’s jurisdiction. For this reason, I chose to work with a text held by the Bodleian libraries, which was out of copyright and already digitised so I did not need to worry about taking images of the volume myself either.

Figure 3: MS Douce MM 493, fol. A6r in Digital Bodleian

The second hurdle to overcome was how to transcribe the text. Transcription is when you record whatever is on the pages of a text, either very literally or with some leniency. I ended up choosing to make two transcriptions: one diplomatic which corresponded strongly with the text, and one standardised which I edited to make more readable and accessible. Thus, my edition could be used by an audience with a broader range of experience with Middle French. One particular advantage that this twofold approach had was that it highlighted common contemporary writing and printing practices, such as a lack of distinction between i’s and j’s, and u’s and v’s, and the frequent mixing up of u’s and n’s by the typesetter. For example, I struggled over the word ‘duue’ on fol. A6r (see the image above the title) which could be read as either ‘duve’ or ‘dune’, before realising that it was meant to be ‘d’une’. Whilst I did start transcribing entirely by “eye”, I discovered mid-way through Week 2 a platform called ‘Transkribus’ which uses Artificial Intelligence to recognise text and transcribe it. While a free version is accessible to anyone, you can also use pre-paid credits to train the software yourself to recognise a particular typeface or hand for a large corpus of texts, thus making it an effective time-saving tool for quantitative research. For MS Douce MM 493, its free AI models were very effective for the Middle English and Dutch parts, producing very few errors. Hence, I found the best approach to be a combination of the latest technology and a trained eye glancing from folio back to Word document.

Figure 4: MS Douce MM 493, p.372.

Having completed both versions of my transcription and painstakingly perfected them under the guidance of Professor Daron Burrows and my dear friend, Michael Angerer, it was time to encode. Whilst I was certainly hesitant at first, I would say to any prospective course participants, “Don’t be so scared!” Following the TEI (Text Encoding Initiative) guidelines is not as complicated as you might think. TEI is a type of XML (Extensible Markup Language) used to describe our data (here: an early printed text transcription) which can then be read and interpreted by a computer. Through a set of guidelines agreed upon by the TEI community, a consistent standard can be adhered to with fixed meanings, as defined in the online guidelines. For example, the following line describes the catchword ‘Dat’ on p. 372 of MS Douce MM 493:

<fw type=”catch” place=”bottom-right”>Dat</fw>

<fw> is an element which stands for ‘forme work’ and describes headers, footers, catchwords, etc. The type attribute specifies the type of ‘forme work’ (here: a catchword) and the place attribute describes the location of the ‘forme work’ on the page. The marked-up XML document can then be transformed via XSLT (Extensible Stylesheet Language Transformations) into various other formats such as HTML for a webpage, PDF, or EPUB for a publication. For my project, TEI enabled the text to become a searchable critical edition as opposed to mere online facsimiles. Looking to the future, both the Middle French and Middle English verse narratives can now be used for linguistic analysis to compare with other versions of the ‘Life of Saint Margarete of Antioch’. In addition, this could further lead to stemmatological research where you examine the codicological relationships between the various versions in terms of time, language, and degree of influence, etc. Hence, textual encoding is just as important as transcription as it opens up new methodologies for research in the humanities.

Figure 5: MS Douce C subt. 249, fol. clxl

Once the encoding was complete and I had checked through everything again, my digital edition was ready for publication. What at first seemed to be a challenging project was now a complete critical edition, free for anyone to access and use. Over the course of these eight weeks, I have even been fortunate enough to be able to examine my volume up close, see its beauty in real life and then try to highlight its most interesting features through digitisation. At times, this project has almost become like a detective game, questioning how accurate the catalogue entry is and hunting down other copies in Oxford for comparison (metadata can be found both in the entry for La vie de sainte Marguerite [French] and for The Life of St. Margaret [English verse]). For example, whilst the catalogue entry claims that the Spanish woodcuts are connected to MS Douce adds. 125, I can say with near absolute certainty that this is not true. Out of the three editions of Pedro de la Vega’s ‘Flos sanctorum’ held in the Bodleian Libraries (MS Douce C subt. 249, MS Douce adds. 125, and MS Balliol College Library, St Cross, 0550 e 05), only MS Douce C subt. 249 contains a printed illustration of St Margarete of Antioch which corresponds exactly with the rear endleaf woodcut (see above and the rear woodcut), thus proving that the woodcut was cut out from this version of ‘Flos sanctorum’. Through the Taylor Edition course, I have learnt a lot about the history of this miscellany: compiled in the 19th century, yet stretching from the late 15thcentury Middle French verses to the 19th century note on the upper flyleaf, and now available for all future students to read and admire for its eclectic nature.

To see my digital edition of MS Douce MM 493, click this link.

Figure 6: MS Douce MM 493, p. 372 on the Taylor Editions website

For more on the numerous ‘Vies de sainte Marguerite’, see their entry on ARLIMA (Archives de littérature du Moyen Âge)

For more on the Middle English ‘Life of Saint Margaret’, see the catalogue entry and this entry in the Digital Index of Middle English Verse. I also later found a transcription from the University of Otago with some interesting metadata.

Lucian Shepherd is a 2nd year undergraduate student at Oriel College, University of Oxford . He studies Modern Languages (French and German) and is hoping to pursue a career in academia in the future.