Lightnews — Scholar-powered news

Theresa Heyd

@theresaheyd.bsky.social

2.3K followers 990 following 7K posts

Je préfère ceux qui n'y arrivent pas pour la bonne et simple raison que je n'y arrive pas très bien, moi-même.

Posts Media Videos Starter Packs

Pinned

Theresa Heyd @theresaheyd.bsky.social · May 12

"The jokes that mean the most to you are the ones you don't quite get"
(Sedgwick 1996, Queer Performativity)

3 1 30

Theresa Heyd @theresaheyd.bsky.social · 4h

dad joke aside, what a time to be teaching this

two iconic images of summer 2025: left, the ICE bike chase in Chicago, right, the Coldplay kiss cam

Theresa Heyd @theresaheyd.bsky.social · 4h

preparing my class on the sociolinguistics of surveillance, but getting the feeling that I know Foucault about it

2 1 25

Reposted by Theresa Heyd

Ingebot Bachmann @ingebotbachmann.bsky.social · 4h

Honig und Nüsse den Kindern,

1 3

Reposted by Theresa Heyd

Tien @wtftien.bsky.social · 4h

I don't want smart cities, I want dumb, beautiful and affordable ones.

13 48

Theresa Heyd @theresaheyd.bsky.social · 6h

Was ich sagen möchte ist: hier ist mein Verständnis von gutem, respektvollem Zusammenarbeiten an der Uni. Wenn ihr gen AI verwenden möchtet, erwartet keine Zustimmung oder Absolution von mir.)

1 9

Theresa Heyd @theresaheyd.bsky.social · 6h

(wobei "dürfen" nicht ganz das Framing ist, um das es mir geht. Ich kann Studierenden hier schlecht Verbote erteilen, nicht nur wegen den Hürden der Kontrolle, sondern auch, weil ich sie als mündige Akteur*innen verstehe.

3 7

Theresa Heyd @theresaheyd.bsky.social · 6h

ich weiß nicht genau, was mit "KI nutzen, um (...) sich einen ersten Text zu erstellen" heißt. Den Text von gen AI erstellen lassen? Dann nein.

3 3

Theresa Heyd @theresaheyd.bsky.social · 6h

das Zerlegen des wissenschaftlichen Arbeitens in kleine und kleinste Iterationen ist etwas, das ich z.B. in Methodenkursen explizit thematisiere und übe. Eine Literaturempfehlung, die sehr spezifisch für mein Fach ist, ist da Heller et al. 2017, fantastische Ressource.

Critical Sociolinguistic Research Methods: Studying Language Issues That Matter

Critical Sociolinguistic Research Methods is a guide to conducting concrete ethnographic and discourse analytic research projects, written by top scholars for students and researchers in social scienc...

www.routledge.com

2 7

Theresa Heyd @theresaheyd.bsky.social · 6h

of course

Reposted by Theresa Heyd

Anna Neumaier @annaneumaier.bsky.social · 8h

Sehr sehr gute Darlegung!

Theresa Heyd @theresaheyd.bsky.social · 22h

my gen AI teaching statement

Summary:
If we work together, in the context of my teaching, or as part of prospective or ongoing supervision of term papers, projects, BA and MA theses, predoctoral and postdoctoral projects, I ask you and expect you to not make use of generative AI tools. In the context of writing and linguistic research, this concerns first and foremost the textual output of Large Language Models, but also extends to generative tools for image and video synthesis. I do not use generative AI in my teaching, research, preparation, assessing, grading, or interaction with you, and I expect the same of you. The only reasonable use case I currently see for generative AI tools is doing critical sociolinguistic research on them – not with them – and if this is something you are interested in pursuing, please feel free to reach out.
Please do not use generative AI to produce text that you submit to me for supervision or grading. This includes portions of text (“only the abstract”, “only the literature review”), draft versions, and related textual work such as shortening, expanding, summarizing, editing your own work.
Please do not use generative AI for the development of research questions, hypotheses, data and method outlines, literature reviews, or bibliographies.
Please do not use generative AI for the generation of data, or for the interpretation of data.
Please do not use generative AI to circumvent reading, e.g. by generating digests of summarized research literature.
Please do not use generative AI for your processes of brainstorming, pondering, reasoning, doubting, rejecting, and generally thinking and wondering about the work you are engaged in.

2 13

Theresa Heyd @theresaheyd.bsky.social · 8h

stars, they're just like us

Theresa Heyd @theresaheyd.bsky.social · 8h

langauge spotted in Brown (1980) ✅️

journal article "research on langauge and sex"

1 9

Reposted by Theresa Heyd

Ingebot Bachmann @ingebotbachmann.bsky.social · 16h

unbeantworteten Briefe an das Gestern vergessen!

2 6

Reposted by Theresa Heyd

pepperhouse.bsky.social @pepperhouse.bsky.social · 15h

Hell hole San Francisco at night.
#photography

6 30

Theresa Heyd @theresaheyd.bsky.social · 19h

das tl;dr ist: don't.

1 11

Reposted by Theresa Heyd

Ingebot Bachmann @ingebotbachmann.bsky.social · 20h

Bald mußt du den Schuh schnüren

1 4 12

Theresa Heyd @theresaheyd.bsky.social · 22h

Consequences for supervision, grading and collaboration:
If you work with me as part of teaching, supervision, grading, or a related collaboration, I ask you to follow the guidelines summarized above. If you disagree on principle with my approach, I suggest that you reconsider your choice of working with me if possible.
I do not ask you to document use of generative AI use in your work, e.g. in the data and methods section, in the sense of creating implied acquiescence.
If I find your work to contain explicit or implicit traces of generative AI (e.g. hallucinated citations), this will have a direct impact on my grading. Principles of plagiarism and responsible intellectual work apply, e.g. in the form of the English Department’s Honor Pledge.
These principles are not meant to intimidate your or create apprehensions around your work. If you feel unsure about viable methodologies and good academic practice, you are always welcome to get in touch and to discuss your ideas and challenges. I care deeply about data and research ethics and wish to treat them as a collaborative goal. If you feel overwhelmed by a particular exercise, task or part of your project, you are invited to seek support, from friends and fellow students, but also your lecturers, tutors, counselors and examiners.
Finally, and importantly, the goal of academic teaching, training, research and practice is not seamless perfection. Members of the academic community, students and senior researchers alike, tend to hold themselves to high intellectual standards, and rightfully so. However, making mistakes, coming to tentative answers only, or ending up with more questions rather than answers, is a normal part of doing academic work. As lecturers, we would rather see a documentation of your academic progress than a fabricated result generated with AI tools.

1 1 25

Theresa Heyd @theresaheyd.bsky.social · 22h

Individual reasons:
Using generative AI is not research. The bad quality of generative AI output is well-documented. Generative AI is not a resource for culling information, “knowledge”, or data; the output of LLMs constitutes statistically likely strings of words, not coherent and informed perceptions of the social world. Generative AI platforms are not tools for research (linguistic or otherwise) in the way that dictionaries and databases, corpora and repositories are parts of the linguistic toolkit. The outcomes of generative AI use are not reliable, not reproducible, not empirically grounded or anchored to any notion of data that we have in linguistics.
Writing is thinking, and thinking is the important part. The very point of research in the humanities and social sciences (and probably all other disciplines) is to grapple with concepts and theories, to make imperfect observations about the world, and to try and derive social meaning and insight from them. Whenever this feels challenging and messy, you are usually doing it right. Sitting down with your ideas, data and interpretations and trying to make sense of them is not a burdensome task on the to-do-list that can be outsourced where possible, but constitutes the essence of becoming and being a researcher. Writing up and/or visualizing your data and ideas is the path that leads to the creation of new knowledge – even and especially when it is tentative, iterative, and less-than-perfect.
Your voice as a researcher is what matters. As a supervisor, colleague, and fellow linguist, I am not interested in reading machine-generated material; it is your interpretation of data, concepts and theories that I am interested in. In short, it is your voice – your perspective, your take, your positionality – that makes your research output socially meaningful and of relevance to me.

1 1 27

Theresa Heyd @theresaheyd.bsky.social · 22h

Reasons:
There is an entire bundle of reasons for prohibiting generative AI from academic work. They fall into two broad categories. The first category is societal and concerns issues of ecological and human exploitation. The second category is more individual and concerns the underpinnings of intellectual work and academic interaction.
Societal reasons:
Generative AI is exploitative of natural resources. Its data centers consume massive amounts of water and energy, leading to substantial negative impact for the regions they are based in, and for the global climate context at large. (https://news.mit.edu/2025/explained-generative-ai-environmental-impact-0117)
Generative AI is exploitative of intellectual property. Generative AI is dependent on massive datasets of textual output for training purposes. The specifics of these datasets are often not made publicly available, but given their size, they conventionally include data that is copyrighted or otherwise protected, for example by the European General Data Protection Regulation. Datasets have been proven to include material such as literary and academic writing, newspaper material, social media discourse, and even digital interaction such as email and direct messages. Whereas the discipline of linguistics has been diligently working towards equitable and ethical forms of data use for decades, these forms of data consumption are unmitigated by ethical considerations. (https://towardsdatascience.com/the-coming-copyright-reckoning-for-generative-ai-b7fe0963c58f/)
Generative AI is exploitative of human labor. While generative AI is marketed and discursivized as automated and machine-based, a lot of the underlying work has been described as “fauxtomation” – it is being produced by humans, working often under precarious, underpaid and unfair conditions, to perform clickwork and related manual labor. This concerns in particular workers in the Global South, leading to the neocolonial reproduction of exploitative labor. (…

2 1 24

Theresa Heyd @theresaheyd.bsky.social · 22h

my gen AI teaching statement

6 7 74

Theresa Heyd @theresaheyd.bsky.social · 23h

Spotted in a colleague's office: International Gothic Association Conference 2011, English Department, Heidelberg. If that isn't the spookiest, prettiest conference poster of all times 💀

International Gothic Association Conference

English Department-Kettengasse 1-2= 69117 Heidelberg=

2-5 August 2011

Gothic Limits / Gothic Ltd

Conference poster in black white and Red. An ominous sunset over Heidelberg castle.

2 1 26

Theresa Heyd @theresaheyd.bsky.social · 1d

"wie wenig groovy diese Neunmalklugheit klingt" oh mein Gott

3 26

Reposted by Theresa Heyd

Magda Birkmann @magdarine.bsky.social · 1d

Schon lustig, dass das, was die Hilmars & co. letztes Jahr über Meyer/Hefter geschrieben haben, dem, was dieses Jahr die Olivers über Melle/Elmiger schreiben, diametral entgegengesetzt ist. Was als Konstante bleibt: Egal wie man's als Frau macht, macht man's falsch, und bestiehlt damit die Männer.

Die Sprache in "Die Holländerinnen" ist von kühler Klarheit, der Intellekt fühlt sich geschmeichelt, auch die innere Schwärze des Werks beeindruckt. Es gibt viele Preise, die dieses Buch verdient hätte. Der Buchpreis aber sollte Türen aufstoßen. In den gepeinigten Kopf des Helden von "Haus zur Sonne" etwa. Der Roman "Die Holländerinnen" aber wirkt abweisend, ein klassischer Braunruck. Glückwunsch an die Autorin Dorothee Elmiger, dass sie damit gewonnen hat.

13 46 170

Reposted by Theresa Heyd

Sylvia Plath Bot @sylvia-plath-bot.bsky.social · 1d

And apparently indestructible.

1 7

Reposted by Theresa Heyd

Jay Blackwood @jayblkwd.bsky.social · 1d

A composite #Moonrise over #Toronto last Wednesday. The Moon is shown over about a 15 minute span with almost 3 minutes between each copy of the Moon. I found it interesting for the color shifted during the rise.

#photography
#landscapephotography

Nightscape of Toronto and the CN Tower. It's a composite image showing the Moon at 6 different elevations, including one aligned with the CN Tower. The buildings were photographed blue hour and still have the golden glow of the Sun reflecting of them.

2 20