Table of contents
Breaking Your Research Into Chunks: Phasing, Targeting, Funding and Working with Collaborators
Building your research strategically: Layering Outcomes
Moving Forward: How do I Apply a Phased, Layered Approach to My Own Research
Funding in the Digital Arts & Humanities: Research-as-Process
Introduction
In this workshop, we want to forward a more holistic approach to thinking about fundingin the digital humanities (DH). For scholars who are just beginning DH projects it can be difficult to wrestle with the years-long timescale common to digital research, and it can be difficult to propose complicated, labor-intensive projects without first prototyping a technology or approach. This workshop will give an outline to think about the iterative processes that link research funding and publication.
Scholars are trained to think of a research pipeline that operates as follows: 1) Acquire funding; 2) Use that funding to acquire resources both material (equipment, corpora) and immaterial (labor); and 3) Turn those materials into an argument in a publication.However, this process does not take into consideration the various funding streams and deliverables that may be necessary for a digital project.
- Many times, we need to apply for funding multiple times.
- Our projects may change based on the technologies we acquire and the challenges we find along the way.
- Publications do not always mean scholarly journal articles or books, but can manifest in many different forms for different audiences (corpora, code, databases).
- Publications can (and should) happen before the project is finished.
Instead, we should be thinking about how we can infuse and integrate each of these discrete steps into one another. To do this, we need to reconceptualize the co-constitutive relationship between our funding streams and our publications:
- Publications are part of the funding and resource acquisition process, not the end.
- We need to expand and rethink what “publication” means—including exhibits, code, corpora, articles, PhD dissertations, appendices etc.—because these materials have impact (other people can build on them for their research),and they can shape our funding strategies.
Patrícia’s team developed the following publications, classes, talks, and dissertations through this work:
Journal Articles
Patrícia Amaral, Hai Hu and Sandra Kübler. 2022. Tracing semantic change with distributional methods: The contexts of algo. Diachronica, International Journal of Historical Linguistics, vol. 39. Published online first, Available: benjamins.com/catalog/dia.21012.ama
Patrícia Amaral, Zuoyu Tian, Dylan Jarrett and Juan Escalona Torres. 2022. Tracing semantic change in Portuguese: A distributional approach to adversative connectives. Journal of Historical Linguistics. Available: doi.org/10.1075/jhl.21028.ama
Hu, Hai, Patrícia Amaral and Sandra Kübler. 2021. Word embeddings and semantic shifts in historical Spanish: Methodological considerations. Digital Scholarship in the Humanities 1-21. (print version: Volume 37, Issue 2, June 2022, Pages 441–461, doi.org/10.1093/llc/fqab050)
Zuoyu Tian, Dylan Jarrett, Juan Escalona Torres and Patrícia Amaral. 2021. BAHP: Benchmark of assessing word embeddings of Historical Portuguese. Proceedings of SIGHUM 2021. Available at: aclanthology.org/2021.latechclfl-1.13
Papers and poster presentations:
Patrícia Amaral, Hai Hu, Zuoyu Tian and Sandra Kübler. Upcoming. Model evaluation for diachronic semantics: A view from Portuguese and Spanish. Workshop “Computational models of diachronic language change", ICHL, University of Heidelberg, September 1-4, 2023.
Zuoyu Tian, Dylan Jarrett, Juan Escalona Torres and Patrícia Amaral. BAHP: Benchmark of assessing word embeddings in Historical Portuguese. Poster, LaTeCH-CLfl (The 5th Joint SIGHUM Workshop on Computational Linguistics for Cultural Heritage, Social Sciences, Humanities and Literature), EMNLP 2021.
Patrícia Amaral, Hai Hu and Sandra Kübler. Lexical and grammatical shifts: A distributional analysis of embargo. Paper presented at the Linguistics Symposium on the Romance Languages 51, April 29-May 1, 2021, Urbana-Champaign.
Amaral, Patrícia, Zuoyu Tian and Juan Escalona Torres. Tracing semantic change in Portuguese: A distributional approach to porém. Poster, presented at the Vienna Workshop on Portuguese Linguistics – Empirical Approaches to Portuguese, University of Vienna, Austria, Dec. 11-12, 2020.
Hu, Hai, Patrícia Amaral and Sandra Kübler. Methodological concerns in modeling language change for medieval texts. (Long paper) 2nd Workshop on Computational Detection of Language Change 2020, University of Gothenburg, Sweden, November 25, 2020. (held online due to COVID 19)
Invited talks:
Amaral, Patrícia. “The company words keep: Studying semantic change in Portuguese and Spanish”. Seminar Series in Linguistics, Australian National University, November 10, 2022.
Amaral, Patrícia. “Lexical and grammatical change in Spanish: Some insights from distributional semantics”, Department of Spanish and Portuguese Linguistics Colloquium, University of Illinois at Urbana-Champaign, October 28, 2021.
Breaking Your Research Into Chunks: Phasing, Targeting Funding and Working with Collaborators
Working in Phases
As you develop a DH project, you should be considering the work in phases. Phases can be thought of as smaller iterations in a larger project, defined by discrete goals, funding streams, and sub-projects from team members. Each phase of the project can help move forward over time, and work through difficulties and gaps in the initial proposal.
A segment from the original IDAH Faculty Fellow proposal:
‘There are several levels of collaboration that I require, involving both the exploration of the corpus data (either using existing resources for Spanish and Portuguese or adapting existing ones to these languages, since they have fewer resources available for data analysis) and higher-level exploration of methods that can be used to study semantic change.’
However, once starting this project, it became apparent that the historical texts (corpora) were not in a usable form for computational analysis. The texts would have to be converted into the .txt format, a process that is time-consuming.
From this problem, Patricia isolated 3 needs:
- Need 1: create a corpus that allows us to use distributional methods and/or clean existing ones.
- Need 2: Compare resources from different languages: challenges posed by resources for different languages
Thinking about a very large, multi-year project as having more than one “end” has a number of advantages. Primary among them is the consideration of several distinct project phases, each ramping up in scope and results, which lets you:
- Start with small sections of the project that are achievable
- move the project forward more consistently over time
- help all members of the team build their expertise
- make connections to other scholars
For the IDAH faculty fellowship Patricia realized that multiple steps were required:
- The creation of a corpus (or the collection of texts required for computational analysis)
- The analysis of different language corpora (what kinds of collections of English, French, Chinese and so on. . . are available? Are there materials in Spanish?)
- Experiment with the corpus using SMALL data sets. Do these experiments have reliable results? Are the findings replicable?
- How will these findings help determine the next funding stream? What will need to be done to apply for the internal, New Frontiers Grant?
Each phase should have layered outcomes, which are composed of small steppingstones and turn-ins. Team members can follow ideas or concepts that apply to their research or expertise, building side projects related to the central one. The team should think of these layered outcomes as separate deliverables from journal articles and poster sessions to internal corpora and literature reviews.
The layered outcomes
- let many people get experience
- let team members build new skills
- get feedback from reviewers along the way
- meet experts in highly specialized tasks (e.g. evaluation of computational models to study semantic change)
- refine the phased strategy
- prepare the next step of funding/publishing
Following and her team published their findings. Different collaborators led different paper projects:
- Hu, Hai, Patrícia Amaral and Sandra Kübler. Methodological concerns in modeling language change for medieval texts. (Long paper) 2nd Workshop on Computational Detection of Language Change 2020, University of Gothenburg, Sweden, November 25, 2020.
- Amaral, Patrícia, Zuoyu Tian and Juan Escalona Torres. Tracing semantic change in Portuguese: A distributional approach to porém. Poster, presented at the Vienna Workshop on Portuguese Linguistics—Empirical Approaches to Portuguese, University of Vienna, Austria, Dec. 11-12, 2020.
- In 2021: 3 conference presentations and 2 publications
- In total: 5 conference presentations and 4 publications (one more under review)
Targeted Funding
Acquiring funding requires the project to be seeded from something small and grown into something larger. It would be very difficult to apply for a NEH Digital Humanities grant in the earliest stages of a project, especially if you are unfamiliar with the terrain of the field. Instead, we recommend thinking small and developing the project through internal funding streams to learn the field, identify gaps in your needs, and learn how to address them. This phase can also include the development of relevant prototypes, surveys of the literature in a given field, and the learning of specific technologies for the project. This can include:
- IDAH Faculty Fellowship
- IDAH Summer Incubator
- HASTAC Scholars Program
As you develop the project, you can begin to apply to larger and more prestigious internal revenue streams, using the insights you gleaned, the prototypes you developed, and the findings you have documented as part of the previous round of funding. These internal grants can include:
- New Frontiers Fellowship (Now the IU Presidential Arts and Humanities Program)
- Collaborative Faculty Fellowships
- Seed Grants
You can also think about funding specific arms of the project, rather than the whole thing, by breaking the larger project into smaller chunks.
Finally, as you continue to scale the project up, you can apply for external funding using the publications you have produced through the previous resources.
The initial application was for a one-year IDAH faculty fellowship. As a consequence of the COVID-19 pandemic, the faculty fellowship was extended for an additional year. The smaller-scale prototypes and data-cleaning activities that resulted contributed to receipt of an IDAH senior fellowship, a smaller but more flexible model that supported several publications based on the early data models and prototype. These provided the foundation for a successful New Frontiers Fellowship application.
Collaborators
Collaborators can bring additional skills and perspectives to the team and afford the development of spin-off projects and publications. Collaborators contribute unique strengths and abilities and can make specific contributions to the project’s progress. Collaboration also means incorporating your collaborator’s priorities and their authorship goals. Identify portions of the project that are suited to the collaborator’s needs and skillset, and try to match funding sources to the partner’s needs, skills and available funding.
In the two phases of research, Patrícia worked with five fellow scholars: Hai Hu (graduate student in Computational Linguistics), Sandra Kübler (faculty, Linguistics), Juan Escalona Torres (graduate student in Spanish and Portuguese), Dylan Jarrett (graduate student in Spanish and Portuguese), Zuoyu Tian (graduate student in Computational Linguistics and Cognitive Science).
The IDAH Faculty Fellowship team was different than the New Frontiers Grant team because of how the project changed, and how different graduate student’s expertise could be applied to the main research goals.
Your collaborators will have their own interests and skills. Embrace their goals in the context of the larger project, because they can do work that will help expand the project beyond its initial scope.
Together with your collaborators, ask:
- What can become its own separate, publishable project?
- Can one of the projects become a Ph.D. dissertation?
- Would a team member profit from a specialized internship/incubator?
- Can a team member profit from doing a literature review related to a component of the project?
- Should the PI develop a book proposal? A graduate course?
- How can the PI create space for learning and mentoring opportunities?
One team member—Zuoyu Tian—brought his own research questions to the collaboration. His interest in change over time in Chinese addressed broader conceptual issues that the project team was trying to answer and led into his PhD dissertation research. Zuoyu also drew on the team’s work as first author on an article about evaluating word embedding models for studying semantic change, which became a new portion of the project. (See: Tian et al. 2021)
Building your research strategically: Layering Outcomes
As these processes overlap, each small component can inform and reinforce other aspects of the larger project. Peer review coming from poster sessions, journal submissions, conferences, and graduate seminars (as well as internally from the team), informs other projects as they are being developed, presented, or reported.
This feedback becomes one step in this iterative process, and is layered into the outcomes of the PI and the team:
- Layered outcomes help build in the broader-impact-criteria response from the beginning rather than having to think about it as an afterthought (not just specialized journals but also public impact)
- You develop new skills for each outcome
- You develop tools that can be shared with a wider community
- The team expands their expertise and visibility
- The outcomes help the growth and training of the group itself
- The outcomes become teaching tools for graduate seminars as well as new members to the team
It is easy to think on intellectual merit—the “research”—but it can be harder to see the “broader impact”—graduate student training and dissertations, the creation of resources for open-access platforms, the distribution of research and tools to specialized and lay audiences. The phased, layered approach helps see that every kind of knowledge practice from teaching team members, to producing research tools or collections are as significant to the research process as traditional scholarly outputs.
Moving Forward: How do I Apply a Phased, Layered Approach to My Own Research
Start by asking the following questions,
- How can you break down the problems or components of your project?
- Can you identify venues or outlets for the different components?
- Can you identify possible collaborators for each component?
- How does each step listed above “become” an end?
- A publication?
- An online resource?
- A teaching tool?
- If you think of intellectual merit and broader impacts, how does your project fit and how can it be improved?