2. Counting Words & Phrases - [E3] 3. Danescu-Niculescu-Mizil, Cristian, Lillian Lee, B. Pang, and John Kleinberg. 2012.

[JunsolKim]

Post questions here for this week's exemplary readings: Danescu-Niculescu-Mizil, Cristian, Lillian Lee, B. Pang, and John Kleinberg. 2012. “Echoes of Power: Language Effects and Power Differences in Social Interaction.” Proc. 21st Int. Conf. World Wide Web: 699–708.

[thisspider]

Authors claim to have found a domain-independent way of identifying status in interaction using a measure derived from coordination based on function words. There can be no doubt that they have succeeded; their measures have at least some success in identifying high-status speakers.

But what, in fact, has been measured?

(1) We might suppose that the higher status speaker will ask questions and the interlocutor answer those questions. This might yield high coordination, because the interlocutor might repeat part of the question in their answer. This would suggest, that the coordination measure identifies who asks questions and that the question-asking behavior is more fundamental. Is the coordination fundamental, or does it simply indicate some other feature of the interactions?

(2) The above is just a toy example and we can think of many others. The paper could alleviate our doubts if it gave us samples from the conversations analyzed. I would like to see what conversation patterns are driving the macroscopic statistical effect. How and when should computational social science present samples of the texts it is summarizing? What are the conventions and best practices here?

[Jasmine97Huang]

The article measures language coordinations in goal-oriented text. However, I wonder if the formula makes sense for conversational text among people with different power status and dependencies. Additionally, seems like the coordination measures implies one-on-one interpersonal communication exchange, would it work for mass communication such professor (high power) to students (low power)?

[Jiayu-Kang]

The research design of this article is strong in that it differentiates the more stable baseline characteristics and the power that changes abruptly. However, I'm still skeptical toward the consistency of measuring language coordination using function word classes. Although it is more generalizable to other contexts as discussed in the article, is it possible that the style markers is less accurate in capturing coordination? How should researchers convince the audience that it is good proxy?

[ValAlvernUChic]

It's really cool that the article manages to show some evidence that language coordination in text itself can be enough to establish power differentials between different parties. We often think about tone (to sift out condescension) or emotion (the most outwardly signal) when ascertaining power so the fact that it can be seen through just conversation is cool. I'm wondering what else beyond language coordination in text might be a plausible way of identifying power differentials. I can imagine that changes in words used (not necessarily to coordinate but to establish/differentiate) might also be possible! Also, the parties involved (wiki editors/admins and judges/lawyers) are from the same domains - they "speak the same language" - and thus coordination itself might be simpler. I'm imagining cases of migrant workers and their employers and whether we can expect language coordination there as well. Thanks all!

[Emily-fyeh]

The article claims to find some form of language coordination among power hierarchical relationships, which pioneers in the study of goal-oriented dialogues. The marriage of the sociological exchange theory and the linguistic style markers is definitely a strong combination holding potential to generalize into other dialogue occasions regardless of domain. The research design also focuses on capturing the 'abrupt' change of power (i.e. acquiring the admin position among Wikipedians) to discern the coordination effect from personal characteristics, which is persuasive for me. However, I personally would like to know more about how they count and label the linguistic labels and how these language usages spread interpersonally. For me maybe the timing is another important factor, for example, the frequent interaction/exposure to the contents from the upper-status people before the occurrence of language coordination.

[LuZhang0128]

This is an amazing article that shows us how people with low power would coordinate with people with high power. The knowledge bases of the participants, however, are similar to each other, meaning that they are able to coordinate fairly easily. I wonder if we are able to record the conversation between patients and doctors, or kids and teachers, are we going to get the same effect? Or if the powerful needs to adjust their language in this case. Moreover, the author mentions that the lawyers are more likely to coordinate towards chief justices compared to the associate justices. I wonder if the effect of power on coordination is linear, or if it will diminish as the power level goes up.

[mikepackard415]

This is definitely a really cool article. I have a friend in law school who used to listen to Supreme Court oral arguments and memorize some of the lawyer's key phrases.

I wonder whether similar methods could be applied to study teams or group projects. If we can extract information about the evolving power dynamics within groups that start out on equal footing, we could potentially learn about what makes a productive or happy team, or even what makes for good leadership in different situations.

[zixu12]

This is an interesting paper and it keeps reminding me of computational linguistics, and this framework quantitatively reveals the power structure in the wiki and the supreme court. As this paper is published in 2012, I am wondering whether there is some new research following or developing from this framework and what are the interesting findings?

[hshi420]

This is more related to psychology: this paper suggests that linguistic style coordination is positively correlated with power, and I was wondering about the mechanism. Is the linguistic style coordination caused by the power difference? Or is it caused by respect(or fearness) by the low power speaker in the discussion. In that case, linguistic style coordination does not necessarily show power difference.

[MengChenC]

So the paper observed and proposed some interesting social phenomena with language data, and compare those behaviors. However, these kinds of research usually do not examine the outcomes with deep psychological perspectives. Therefore, I am interested in what would be the explanations behind these findings, and how can we validate them? Are they related to biological psychology, evolutional psychology, or even behavioral neuroscience? I think knowing these can rather better understand human beings.

[kelseywu99]

Very interesting research on power relationships in linguistic styles in social interaction, the subject sort of reminds me of Goffman's the presentation of self in everyday life. I was wondering if the conclusion may be understood in a reverse way, such as that linguistic style coordination indicates a power difference between two parties? I think it applies to our everyday routine, i.e. we talk differently when we are addressing someone "higher up" yet we talk normally when we are having a casual conversation with friends. I was thinking if this research may be expanded into analyzing casual interaction online as well, such as internet influencers responding to fans' comments.

[hsinkengling]

One of the things I find curious is whether the style change is situational, I wonder if there are ways to track some kind of pseudo-longitudinal data, matching the same interlocutors in separate incidents, and measuring the variation between different occurrences?

[AllisonXiong]

It's definitely an inspiring research that combined sociology theories with computational linguistics. I was curious about and would love to know more technical details about the quantifying of marker m. Also, I think cross-domain comparison may be interesting. For instance, the definition and strength of 'power' seems to differ in the two samples, and i was wondering does that effect the level of corrdination?

[sudhamshow]

This was an interesting read! Finding evidence of power differences using language co-ordination seemed like a novel approach. But, is the framework developed generalisable for different languages, given the wide differences in language constructs? For e.g, A few Indian languages emphasise respect to elders by recommending the use of plural pronouns (they, them) to parents, elders unlike English. A few South Indian languages however use plurals only to male elders in the family (father, uncles, grandfathers). By using the language co-ordination framework, would it be possible to come to the same conclusions of power difference among people in different structurally equivalent classes, using these kinds of varied language setting?

[yuzhouw313]

Aligning one's linguistic coordination with the listener or target audience based on their status and the dynamics of dependence is an intuitive aspect of effective communication. This coordination is often measured by the usage of specific categories of function words within a communication and their alignment with a designated marker. However, a critical question arises when attempting to generalize these findings: how applicable are they to different communication contexts, specifically in distinguishing between professional or specific conversations and casual daily interactions? I would imagine that in professional conversations, the likelihood of using similar word utterances might be higher due to the specialized nature of the discourse within a particular field. While the authors acknowledge the domain independence of their designs and findings, an intriguing consideration emerges. Is there a mechanism in place to control for personal traits that could potentially influence linguistic coordination and, in turn, context setting?

[joylin0209]

This article talks about the power relations of language style in social interaction, which I found interesting and also generated some curiosity. First of all, I would like to know more about the strategic content of language style change. For example, under different power relationship interactions, is it common to use different words for the same meaning, and what are the examples? The second question is, can this analysis method even be used in other language systems, or even compared with English? For example, in Chinese, there are two different words to refer to "you", which are divided into honorific terms (usually used for people with greater age, rank, and power) and non-honorific terms. Therefore, a text containing the honorific "you" can easily convey the meaning of respecting the reader. Is it possible to express the difference in power relations through a single word in English? I think it would be interesting if we could compare the language style strategies of different language systems in social interaction.

[volt-1]

Exploring power relationships through social interaction texts is a fascinating subject, the context of the interaction influences the degree of language coordination. I am interested in how technological interventions (such as muting and turning off video) in online communications reflect the implicit power dynamics (such as the order of speaking, duration of speaking, etc.) through language coordination. Are these subtle expressions of power more difficult to perceive and analyze?