naitian

@naitian.org

1006 Followers 849 Following

NLP / CSS PhD at Berkeley I School. I develop computational methods to study culture as a social language.

Claude Code says that it has been Canoodling for 2 minutes and 46 seconds.
Screenshot of Apple weather app for Berkeley, CA with a warning for an “extreme cold watch”. Today shows currently mostly sunny 53 degrees with a high of 57 and low of 48.
Figure from the paper which shows how, for difficult tasks, directives and options tend to increase immediately before tasks are completed and drop afterwards, whereas the inverse is true for confirmations and expressives.
This video shows two players navigating a maze. The orange player is trying to direct the blue player using "left" and "right", but at first there is confusion on what those directions refer to. The orange player asks "which direction is right to you" and the blue player physically moves to demonstrate the direction. This allows them to successfully agree on terminology and progress forward in the level.
A figure demonstrating the different aspects of the corpus described in the tweet. There is a main isomorphic 3D view of a level in the Portal 2 co-op game, with some portals, lasers, and the blue and orange players. Inset, there are first-person captures of the blue and orange player views. There is also a box containing the transcribed dialogue with timestamps and labels for the discursive acts. Finally, there is a box containing a task and a list of subtasks. Some subtasks are already crossed out, with the time that they have been completed. The last subtask ("Player 2 places portal 4 on wall 4") is marked incomplete. The dialogue is as follows: Blue: Can you put your other portal up here? (tagged as directive) Orange: Where? (tagged as request for clarification) Blue: On uh, on this wall. (tagged as directive) Blue: So that it uh points at the circle. (tagged as directive) Orange: Okay. (tagged as commit) The full list of subtasks is: Task: Redirect lasers Subtask: Player 1 places portal 1 on wall 1. (completed) Subtask: Player 1 polaces portal 2 on wall 2 or 3. (completed) Subtask: Player 2 places portal 3 opposite of portal 2. (completed) Subtask: Player 2 places portal 4 on wall 4. (incomplete)
Centralisation of the means of production and socialisation of labour at last reach a point where they become incompatible with their capitalist integument. This integument is burst asunder. The knell of capitalist private property sounds. The expropriators are expropriated. The capitalist mode of appropriation, the result of the capitalist mode of production, produces capitalist private property. This is the first negation of individual private property, as founded on the labour of the proprietor. But capitalist production begets, with the inexorability of a law of Nature, its own negation. It is the negation of negation. This does not re-establish private property for the producer, but gives him individual property based on the acquisition of the capitalist era: i.e., on cooperation and the possession in common of the land and of the means of production.
Amir Goldberg on stage at IC2S2. The slide behind him highlights the difference between orthogonality (are they different but not opposed?) and oppositionality (are they opposed?) in evaluating the similarity of interpretations.
A poster for "Culture is not Trivia: sociocultural theory for cultural NLP" which takes the form of a flow-chart. The central question, and the starting point of the flow chart, is "What is culture in cultural NLP?" An arrow is labeled "wait, so what's cultural NLP?" This leads to a block explaining that the goals of cultural NLP are described in section 2 of the paper. They include inclusivity, depth, discerning, and adaptiveness. That leads to an arrow that says "that sounds great!". But there are recurring challenges in this kind of work! Section 3 surveys some of these: a discomfort around the proxies being chosen, a lack of coverage, and a lack of dynamicity. That in turn leads to an arrow labeled "Hm, sounds like we need to figure out..." and it leads back to the main question: "What is culture in cultural NLP?" A final arrow extends below this block: "Well, who's to say, really?" This points to sociocultural linguistics. Section 4 explores how other disciplines, like sociolinguistics, linguistic anthropology, and discourse analysis have faced similar challenges in the past. Section 4.2 gives an overview of sociocultural linguistics, which is a set of principles tying together some convergent themes: emergence, positionality, indexicality, relationality, and partialness. One arrow extends from this asking, "what's that have to do with cultural NLP?" Section 5 gives a case study of how indexicality clarifies how to think about stereotypies in the context of mining cultual knowledge from the web. Another arrow says "How can I build safe NLP systems?" Section 6.2 explores how localization can serve as a useful model from building culturally aware technologies because it forces developers to define culture explicitly and tractably. Finally, an arrow asks "how can I study culture with NLP methods?" Section 6.1 lays out theoretically motivated directions for future empirical and theoretical work in computationally modeling culture.
Still image from season two of Nathan Fielder's show "The Rehearsal", depicting a large hall meant to evoke imagery of Nazi Germany, with large Paramount banners hanging in the background.
A quote from the original paper: The elusiveness makes sense for money and politics, but up to now, it has not been reasonable - or even considered possible -- for art. GenAI images make it possible.
A screenshot of the terminal showing model training progress. The label for one of the progress bars says "Calidation Dataloader" instead of "Validation Dataloader".
Culture is not trivia: sociocultural theory for cultural NLP. By Naitian Zhou and David Bamman from the Berkeley School of Information and Isaac L. Bleaman from Berkeley Linguistics.
Ralph Wiggum of the Simpsons sitting on a bus: “I’m in danger,” he says while chuckling.
Sotheby's auctioneer Oliver Barker eliciting bids, living up to his surname and leaning forward on the rostrum in a pose that evokes the image of a bird dog at alert.
An excerpt from a conversation analysis paper describing a scene during a journalism hackathon where some group members (Kat, Mia, and Jen) are unsure whether a fourth member (Lee) is joining their group, so Kat semi-awkwardly (though this could just be a projection of my own social anxiety) broaches the subject directly. Illustrations next to the transcript describe Kat waving at Lee, and Kat and Jen celebrating after Lee affirms his participation.
Table showing some phrases with high and low emotional range, and their corresponding entropy values. The phrase with the lowest emotional range is “Could I ask you something?” and the phrase with the highest emotional range is “All rise.”
Line graph from the paper that shows the average emotionality of films has decreased since 1980, reaching its lowest point around 2010.
Diagram showing three concepts from the paper: First, dialogue phrase groups, which are groups of semantically equivalent phrases generated by clustering audio transcriptions. Second, utterances, which are pieces of audio extracted using speech segmentation. Finally, emotion labels for each utterance are classified with a speech emotion recognition model. Multiple utterances are associated with each dialogue phrase group in a many-to-one relationship.
Screenshot of the abstract for the paper, "Once More, With Feeling: Measuring Emotion of Acting Performances in Contemporary American Film". The abstract reads: "Narrative film is a composition of writing, cinematography, editing, and performance. While much computational work has focused on the writing or visual style in film, we conduct in this paper a computational exploration of acting performance. Applying speech emotion recognition modes and a variationist sociolinguistic analytical framework to a corpus of popular, contemporary American film, we find narrative structure, diachronic shifts, and genre- and dialogue-based constraints located in spoken performances.
screenshot of a jupyter notebook showing some code fetching vote count data from the washington post
A pile of library cards collected from all over the world
Screenshot from Google Docs where the autocompletion suggests I want to write about a "disciplinary discipline".
The listings for two papers about shibboleths in Google scholar, one of which is about enregisterment and identity in Michigan's Keweenaw Peninsula, and the other is about SAML and resource access management standards.