Set for failure: plot twist

In the discipline of Hispanic studies, and most specifically, the field of Spanish literature, success is traditionally measured by achieving three key milestones: publishing an academic book, securing tenure and contributing to the field through a series of single-author scholarly articles. More often than not, said milestones are achieved by close reading and becoming an expert in a few authors or works. However, lacking a rigid plan – as a first-generation international student who didn’t really know what she was getting into – I opted for an unconventional path: I would mix literature written in Spanish, in general, with computers to engage in more experimental readings. This divergence, which hasn’t led me to a tenure-track position nor a book, is often perceived as a failure. Yet to me, this rather different path has been a place for personal and professional growth – with its privileges and risks – offering space to experiment in my scholarship and to support inquiry-based learning and teaching.

This was not solely my decision. I was explicitly given permission to venture into more experimental academic territories in my scholarship and ‘fail’ if it came to it. This key learning moment that I’ll never forget took place when I defended my dissertation proposal in early 2016. While articulating my idea to employ social network analysis methods to analyse nineteenth-century Spanish historical novels, Matthew Jockers, the digital humanist on my committee who was helping me with the computational analysis, posed a crucial question. More or less, he asked, ‘Jenn, you keep saying “if I am able to …” or “if I get to …” So, what if you fail to extract character mentions in a manner conducive to a meaningful application of SNA and, thus, are left without results?’

Caught off guard, I found myself for a few seconds at a loss for words. Luckily, and to my complete surprise, my advisor, Óscar Pereira Zazo, quickly interjected with a reassuring response: ‘She can write about the research process and the reasons for failing at the experiment and she can still successfully graduate’. The learning process was more important than the results.

Automatic Natural Language Processing models did fail me: they were not accurate enough for my purposes back then. However, through persistent trial and error (and running against the student funding clock), I was able to find a workaround to the challenge of employing existing NLP models for character name identification, normalisation and counting. My solution involved turning a printed index of characters from a 1940s edition into a comprehensive data dictionary of characters enriched with metadata for many variables (Isasi 2017). This approach allowed an algorithm to locate the names in the novels, count their (co)occurrence, and later run social network analysis and visualisations on the data. Contrary to initial expectations, the experiment proved fruitful, as I not only avoided failing in my idea but also managed to establish meaningful connections to canonical readings of the works by established galdosistas. I think it was a pleasant surprise for everyone.

Comfortable with experimenting

Since then, I’ve been actively engaged in or advised on various data curation and research experiments that tried to replicate my dissertation model or explore new methods to analyse text-as-data. In most instances, I have had to create new datasets either independently or through collaboration due to factors such as: no previous investment in creating data about the question that crossed our mind; the absence of digitised records in an organised or clean data-collection format; the information being housed in different spaces and formats; data restrictions within academic systems for purposes of review and promotion; and more. And so, when it is time to apply a method to probe a (research or for fun) question, the only thing I have time and energy to do is something that is smaller than I envisioned. I mostly create visualisations or write some provocation, if I produce anything at all. Although I often don’t produce a traditional paper, I persist in running experiments, gaining valuable insight into both the subject matter and, most importantly, the methodology employed.

This persistence is inspired by the words of my advisor, who challenged the notion of failure in a department of Modern Languages. These words have remained central to my approach to research, mentoring, workshop instruction and service. It is the perspective through which I guide students (and faculty when they are adventurous enough), urging them to enter the world of digital scholarship in any shape or form. The phrase ‘she can write about the research process and the reasons for failing’ has evolved over the years in my various roles into a commitment to engage, document the process, acknowledge and celebrate successes, and to take note to learn from things that didn’t unfold as anticipated. As Graham puts it, this process is not just ‘messing around the Internet’ or with digital methods, because ‘there’s a cycle of exploration, documentation, experimentation, documentation, experimentation … [we] write it down. Blog it. Turn it into an article. Share’ (2019, 85). This process is a form of scholarship as well.

Teaching from lessons learned

I see the opportunity of being able to share this knowledge with others as a professional accomplishment. Whether in my official position first as a CLIR postdoc at University of Texas-Austin (USA) and now as a director of the Digital Liberal Arts Research Initiatives at Penn State University (USA), my role as editor of Programming Historian en español, a journal that shares digital methods with a global audience, or as a consultant for collaborators and colleagues – primarily friends of the global majority – I strive to guide the selection of the most appropriate and ethical project workflows and tools, taking into account context and resources for courses or projects.¹ And the way said plans are developed usually has a learning goal, rather than being a simple prescription. Often, although at the beginning most students and faculty seem reluctant to ‘play’ in class or project meetings, we together discover that embedding digital tools to explore the cultural record invariably leads us to pose thought-provoking questions, beginning with the datasets (How was this collected?) and ending, usually, with visualisations (Are they meaningful? Are they respectful?), running through questions about the subject matter.

I must admit that, sometimes, experimenting or suggesting to experiment is also accompanied by moments of frustration. How can it be otherwise? Sometimes, the dataset is not available or is dirty; or the tool you thought could be useful cannot process text-as-data in different languages; or it is no longer updated and supported because it runs out of funding, and it doesn’t work anymore (see Dombrowski’s reflection in this volume). Or you have to let the idea go, because it won’t count towards academic promotion. But this experimenting is critical in the context of a digital humanities pedagogy ‘that centers concepts and inquiry-based learning’ where mostly everyone comes into the field with different skills and knowledges (Walsh 2023, 198). And, I would add, everyone leaves the learning environments also with different skills and knowledges, forcing us to collaborate with each other and succeed or fail and learn together.

In closing: embrace the possibility of failure

Since it doesn’t always lead to a traditionally readable work, the idea that experimenting with digital scholarship methods within the discipline of literature is often deemed a failure persists among many. For example, a colleague that I had not seen in a decade asked me recently: ‘I don’t know what you do but a director [position] sounds good. Are you happy?’ The concern about my happiness was prompted, I know, because in their minds, I’ve failed. I am not a tenure-track assistant professor of Spanish; rather, I am an assistant research professor of digital scholarship.²

I am happy, indeed. As I fight for changes where we find there should be changes, personally, I am grateful every day for the opportunities I have to participate in them. In my teaching and sharing knowledge, I am happy. I learn something new every day, and in collaborating with other scholars in many fields I also learn from their wins and failures – because we now often share both cases (Dombrowski 2019, with special notice to its Addendum). The flexibility to pursue scholarly questions within my work responsibilities is also a privilege I cherish. Moreover, the landscape has changed and if one chooses or needs to publish, there are now numerous avenues to present projects at various stages – be it in the inception phase in journals like Startwords, method-focused contributions on Programming Historian, or reviewing projects on Reviews in DH. At the same time, in teaching, and as shown in What We Teach When We Teach DH, we are making strides towards fostering a more inclusive environment where success is not, and cannot be, uniformly measured. We are seeing a collective effort that is underway – such as the one seen in this volume – to challenge conventional norms of success in (our parts of) academia.

And, as much as I can, I am dedicated to advocating for the inclusion of this type of scholarship in our academic and disciplines’ system and guiding future generations. Inspired by my mentor’s advice, I embrace the idea that permission to fail is, in itself, a pathway to success. ¡Gracias, Óscar!

Notes

1. 1 ‘The CLIR Postdoctoral Fellowship Program offers recent Ph.D. graduates the chance to develop research tools, resources, and services while exploring new career opportunities’, usually in libraries (Council on Library and Information Resources).

2. 2 Needless to say, this perception is based on a widespread elitism about academic ranks that is reflected both at the individual level and in the system of benefits and opportunities.

References

• Dombrowski, Quinn. ‘Towards a Taxonomy of Failure’ (blog). 2019. Accessed 24 November 2024. https://quinndombrowski.com/blog/2019/01/30/towards-taxonomy-failure/.
• Graham, Shawn. Failing Gloriously and Other Essays. Digital Press at the University of North Dakota, 2019. Accessed 24 November 2024. https://thedigitalpress.org/failing-gloriously/.
• Isasi, Jennifer. ‘Acercamiento al Análisis del Sistema de los Personajes en la Narrativa Escrita en Español: El Caso de Zumalacárregui y Mendizábal de Pérez Galdós’, Caracteres: Estudios Culturales y Críticos de La Esfera Digital 6, no. 2 (2017): 107–37.
• Walsh, Brandon. ‘The Three-Speed Problem in Digital Humanities Pedagogy’. In What We Teach When We Teach DH: Digital Humanities in the Classroom, edited by Brian Croxall and Diane K. Jakacki, 197–206. Debated in Digital Humanities. University of Minnesota Press, 2023. https://doi.org/10.5749/9781452969558.