Danescu-Niculescu-Mizil, C., Sudhof, M., Jurafsky, D., Leskovec, J., & Potts, C. (2013). A computational approach to politeness with application to social factors. arXiv preprint arXiv:1306.6078.
Summary:
The paper focuses on exploring the content politeness on social platforms like Wikipedia, Stack Exchange. They conduct linguistic analysis exhibiting politeness on the corpus of requests and extract positive and negative politeness strategies from the context. In the experiment, two classifiers, a bag of words classifier (BOW) and a linguistically informed classifier (Ling.), are compared to illustrate the effectiveness of their theory-inspired features. They also show the relationship between politeness levels and social power/reputation. In general, we see a negative correlation between politeness and power on Stack Exchange or Wikipedia. They also add that there are differences in politeness levels across factors of interest, such as communities, geographical regions, and gender.
Reflection:
In the paper, they use a bag of words classifier as a strong baseline for the new classification task. However, it didn’t mention whether the BOW classifier is the state-of-the-art, thus the advancement of BOW needs to be discussed or some other classifiers are expected to be listed in the experiment. With a more effective classifier, where can we take advantage of it to benefit users or virtual communities? For instance, we may attach users in the community with a ‘politeness’ feature, hostile users will be automatically muted or banned if their politeness is lower than the floor value. We can also use it to evaluate the politeness of online customer service extensively used in the electronic business, which would benefit the environment of online shopping.
Voigt, R., Camp, N. P., Prabhakaran, V., Hamilton, W. L., Hetey, R. C., Griffiths, C. M., … & Eberhardt, J. L. (2017). Language from police body camera footage shows racial disparities in officer respect. Proceedings of the National Academy of Sciences, 201702413.
Summary:
Similar to the politeness-level difference analyzed in the above paper, one of the most non-negligible topics is the respect-level between different races(even gender, nationality). The paper collects footage from body-worn cameras, we extract the respectfulness of police officer language toward white and black community members by applying computational linguistic methods on transcripts. Disparities are shown in the speaking way of officers toward black versus white community members.
Reflection:
For the data collection, the source of datasets is not geographically diverse. Given that the issue of racial discrimination in different regions may not be of different levels of severity, I think it might be a better choice to collect footage in more than one place in the United States, not just Oakland, which would provide a holistic point of view of how officer treat white and black drivers. On the other hand, the race of officer is a key control factor since I’m curious about how black officer treats black and white drivers in their routine traffic stops, which is not discussed in detail though.
I noticed that the data they extracted were footage from body-worn cameras, transcripts were mainly used in the research. With the development of computer vision technology, extracting footage for the frame detection as an additional feature may help us to better understand the relation pattern of between police officers and drivers. For example, the frequency of conflicts between officers and drivers is also an important measure of respect degree.
Questions:
- For these topic-similar paper, is there a clear difference between being polite and being respectful?
- Since it is not easy for social scientists to measure how police officers communicate with the public, researchers use body-worn cameras to collect data. However, with body-worn cameras capturing interactions every day, how do we guarantee police officers would act exactly the same as the way when they don’t wear cameras? Based on this, keeping data collection procedure agnostic to police officers might be an alternative choice for researchers?
