This page is intended to provide some context to potential students on what I work on and what I expect from students. You can also find information on my lab website, which may be more up to date. If you are a prospective PhD student interested in working with me, I encourage you to apply to JHU (information here) and include my name on your application.

While things may change, I am unlikely to take new PhD students for Fall 2026.
If you are currently a JHU graduate or undergraduate student interested in working with my lab, feel free to reach out to me!
I generally do not work with high school students or undergrads from other universities except through structured programs.

You certainly can email me, but I have found that these emails have effectively no impact on admissions decisions. Thus, as a rule I generally do not reply, because I believe that replying is more likely to mislead applicants into over-estimating the impact of their email than to accomplish anything else.

My primary field of research is natural language processing, and some of my work could be considered data science and AI. Within this space, I focus on social-oriented topics that broadly fall under three categories:


Can we use automated text processing tools to help us identify and understand social phenomena? Research questions in this space might include: what motivates successful social movements? What are large-scale propaganda strategies and are they effective? How do conceptualizations of "stereotypes" and "bias" differ across cultures and domains? This line of work sometimes involves developing new technology that facilitates open-ended text analysis and sometimes involves actually conducting those analyses, often in partnership with social scientists. The long-term impact goals of these projects often focus on informing policy decisions. Some papers in this space include:

• Chan Young Park*, Julia Mendelsohn*, Anjalie Field*, and Yulia Tsvetkov. "Challenges in Opinion Manipulation Detection: An Examination of Wartime Russian Media" Findings of EMNLP (2022). (text)
• Anjalie Field*, Chan Young Park*, Antonio Theophilo*, Jamelle Watson-Daniels, and Yulia Tsvetkov. "An Analysis of Emotions and the Prominence of Positivity in #BlackLivesMatter Tweets" PNAS (2022). (text, code)
• Anjalie Field, Gayatri Bhat, and Yulia Tsvetkov. "Contextual Affective Analysis: A Case Study of People Portrayals in Online #MeToo Stories" ICWSM (2019). (text, code)
• Anjalie Field, Doron Kliger, Shuly Wintner, Jennifer Pan, Dan Jurafsky, and Yulia Tsvetkov. "Framing and Agenda-setting in Russian News: a Computational Analysis of Intricate Political Strategies." EMNLP (2018). (text, code)

This line of work focuses on developing NLP models that have the potential to address a broad range of societal issues. I am currently particularly interested in public services, such as social services, policing, healthcare, and public health. In the past I have also conducted projects focused on propaganda, misinformation, and stereotypes and bias. The long-term impact goals of these projects focus on deployable technology. Some relevant papers include:

• Anjalie Field, Chan Young Park, Kevin Z. Lin, and Yulia Tsvetkov. "Controlled Analyses of Social Biases in Wikipedia Bios" WebConf (2022). (text, code)
• Nupoor Gandhi, Anjalie Field, and Yulia Tsvetkov. "Improving Span Representation for Domain-adapted Coreference Resolution" Workshop on Computational Models of Reference, Anaphora and Coreference at EMNLP (2021). (text)
• Anjalie Field and Yulia Tsvetkov. "Unsupervised Discovery of Implicit Gender Bias" EMNLP (2020). (text, code)

We cannot seek to develop models for "social good" and understand societal issues without self-reflecting on potential harms and issues in our own field. This line of research focuses on measuring and mitigating potential harms in NLP pipelines and their development. The long-term impact goals of these projects focus on methods to improve ethical NLP development as well as shaping policy, specifically around NLP use and development. In the past year, my primary interests in this space have focused on privacy and transparency. Sample papers include:

• Krithika Ramesh, Nupoor Gandhi, Pulkit Madaan, Lisa Bauer, Charith Peris, and Anjalie Field "Evaluating Differentially Private Synthetic Data Generation in High-Stakes Domains" Findings of EMNLP (2024) (text)
• Inna Wanyin Lin*, Lucille Njoo*, Anjalie Field, Ashish Sharma, Katharina Reinecke, Tim Althoff, and Yulia Tsvetkov "Gendered Mental Health Stigma in Masked Language Models" EMNLP (2022). (text)
• Anjalie Field, Su Lin Blodgett, Zeerak Waseem, and Yulia Tsvetkov. "A Survey of Race, Racism, and Anti-Racism in NLP" ACL (2021). (text)

These categories are not independent: if we want NLP to be for "social good", it needs to be ethical. If we want to identify and mitigate "bias" in our models, we need social science perspectives on what "bias" is. For a more formal overview, including how these concepts fit together and high-level goals, you can also take a look at my research statement from a few years ago.

Most of my research focuses on identifying abstract concepts in text, e.g. "bias", "propaganda", that can be subtle and implicit. This type of content is difficult to identify using established methods or supervised approaches, and projects typically involve developing new frameworks that incorporate relevant context. Some of the common challenges include accounting for confounding variables, reducing the need for hand-annotated data, and handling domain-specific concepts and terminology. Sometimes this involves deep-learning, sometimes it involves counting word frequencies, and sometimes it involves manual analysis.

Additionally, most of my work involves exploring new data that is digitized, but can be messy and unstructured (newspaper articles, social media, social workers' notes). A lot of projects have started with a great data set, where we need to figure out how to process it and what to do with it. Unlike working with standarized NLP datasets, which have been carefully curated, this process can involve a fair amount data wrangling. While I most commonly work with text data, some of my recent projects have focused on speech.

I am primarily looking for students who are highly motivated to work on social issues. While I do not expect incoming students to have extensive experience in NLP, students should be comfortable writing code. Students from interdisciplinary backgrounds are welcome. I have had lots of very productive collaborations with people in a variety of fields and institutions, including political science, economics, statistics, etc.

I will strive to maintain a lab that values inclusiveness and is a safe environment for all members. As a new professor, I expect to work closely with students, including regular meetings (e.g. weekly) and active involvement in projects, such as helping with paper writing. Students who are looking for hands-on involvement and close collaboration are a good fit. While I do expect frequent communication and regular progress, I have no intention of tracking students' work hours or vacation time, and students will be encouraged to set schedules that best support their mental health and productivity.