Class meeting #7 – Online music and sound communities – Monday 10/15

Readings:

Thelwall, Mike. 2018. “Social Media Analytics for YouTube Comments: Potential and Limitations.” International Journal of Social Research Methodology 21 (3): 303–16. https://doi.org/10.1080/13645579.2017.1381821.
Born, Georgina, and Christopher Haworth. 2017. “From Microsound to Vaporwave: Internet-Mediated Musics, Online Methods, and Genre.” Music and Letters 98 (4): 601–47. https://muse-jhu-edu.ezproxy.cul.columbia.edu/article/689473.

Optional

Andersen, Joceline. 2015. “Now You’ve Got the Shiveries: Affect, Intimacy, and the ASMR Whisper Community.” Television & New Media 16 (8): 683–700. https://doi.org/10.1177/1527476414556184.

Response:

How can the methodology described in Thelwall (2018) be used to supplement the investigation of Born and Haworth (2017)?

In your short response you may consider some of the following questions:

  • Do you think that Thelwall would approve of the use of their method in this context?
  • If not, what problems can you imagine that they would find with such a technique?
  • Do you think that Born and Haworth would accept conclusions drawn from the use of CFTC?
  • If not, what problems can you imagine that they would find with such a technique?

 

6 thoughts on “Class meeting #7 – Online music and sound communities – Monday 10/15

  1. yh2825

    I think both of the articles could use each other’s methodology to improve their analysis.

    For Born and Haworth, the article extensively uses Issue Crawler as the supporting technique to identify correlations and connections between genres based on the “hyperlinks” – however, this approach would only provide descriptions of the genre at a metadata level as if the hyperlinks serve as labels for the genre. It would be interesting to apply the CTFC framework to the research so that we can see the actual connections between the sonic data across genres. Instead of using “comment term” as the element of analysis, we can first extract features from the sonic data and apply the framework towards these features. Then, for example, a time series analysis or frequency analysis on these features would give a historical and structural view of the music genre.

    For Thelwall, although the article uses node graph to identify the clustering of comments between comments of different dance genre which is similar to the approach of Born and Haworth, it is possible to apply the same methodology within one single dance genre. By identifying the clusters within one genre, it is possible to then cross compare the clusters between different genres to see the correlation of the demographic information of these clusters – i.e. what is the correlation between identified groups of people and the clustering of comment terms across all genres?

  2. lnl2110

    Born and Haworth’s CFTC methodology, although it involves a “network” step, is far less interested in investigating an artistic community through the directionality of relationships within and outside of it, than Thelwall. The difference between these two investigations is the CFTC method can only derive conclusions about a genre by hypotheses about divisions within a community (ex. gender difference), whereas the IC method can be used to generate patterns within and outside of a community without needing that same degree of initial division. Co-link analysis allows for Thelwall to investigate the “types of creativity, discursivity, sociality” by means of visual analysis and identifying significant actors. CFTC analyses patterns in language with initial assumptions about creativity, discursivity, and sociality. Thelwall should not necessarily supplement the Born and Haworth’s investigation, but possibly precede it, and inform some of the CFTC method.

  3. erc2175

    I could imagine the CTFC methodology being used in a context like analysis of discussion on Reddit, the .microsound mailing lists, or other centralized forums. As such, Born and Halworth’s analysis, relying primarily on establishing connections via hyperlinks between entities rather than analysis of single entities, seems like it would have limited use for much of the CTFC methodology, especially analysis of individual users’ comments in correlation with their identities. One could, however, imagine using the CTFC methodology to explore centralized forums to determine sentiments regarding forces established by Issue Crawler as important or central. These sentiments could conceivably lead to greater understanding of issues like aspirational hyperlinking, and the degree to which these strategies work in various different communities.

  4. ijg2112

    The CTFC methodology that Thelwall discussed could be useful in analyzing the formation of the various musical genres that Born and Haworth explored. If the same analysis was run on internet texts relating to these genres at incremental dates in the past, the change in the results could help model the changes in the characteristics and community perceptions of each genre. This could also help provide a better understanding of how, and possibly why, the values behind a genre have evolved. While this method may not provide in-depth detail, it could help provide points of interest for further research by humans.

    I am uncertain that Thelwall would be completely confident of the CTFC methodology being used in this context. Born and Haworth were more focused on the keywords, subtopics, and themes within each genre. However, Thelwall concluded that CTFC method was not very successful in producing insightful subtopics about the subjects explored. This methodology might not produce the results that Born and Haworth are looking for. Additionally, the algorithms required to search for bigger phrases or idea are significantly more time consuming, so this might not be a realistic tool to use.

  5. clj2142

    Born and Haworth’s study would benefit from the qualitative data gathered using Thelwall’s method. While the quantitative data Born and Haworth collected using Issue Crawler is beneficial to understanding the structure of these microgenres, the nuances of a genre cannot be completely represented by quantitative data alone. Thelwall’s methods investigate genre at the level of the individuals involved, which allows one to look at word choice which then can be further examined by looking at what words are closely associated with different demographics (in Thelwall’s study he looked at gender differences). I think that because their approaches were so different, each author (or group of authors) would question the validity of the others’ work: Born and Haworth because Thelwall’s work is so subjective, Thelwall because Born and Haworth’s work relies heavily on the quality of the Issue Crawler algorithm and presents raw data only. Ideally, these two methods should be combined to study genre on both a structural and individual level.

  6. spn2120

    In my opinion, Born and Haworth could definitely take some cues from Thelwall’s methodology. Born and Haworth discuss “medium specific” research that includes both the quantitative and qualitative, yet the language of their analysis appears to stem more from western conceptions of modernism/post-modernism, genre, genre theory, etc. than the data of comments itself. While they note the importance of non-sonic data, their conclusions seem more theoretical than actually representative of the social interactions between online users. Studying inputs and outputs of hyperlink articles, they argue that the internet itself becomes embedded in the “assemblage” of mediations. However, it’s hard for them to convince me that their study pushes the discussion in any way. Thellwall poses questions to study the dynamics of gender or emotional response, and acknowledges the limitations of their own data collection (youtube names, the fact that comments on old videos weren’t as well saved), as well as the intersection between particular forms of art such as Bollywood and a certain “rare” dance that culturally intersect with other topics. Additionally, Born and Haworth use physical location (both virtually though certain hubs and domain name suffix) as a way to trace scope. I think that their analysis is too broad, and could benefit from some of the clustering methods proposed by Thellwall. Furthermore, as much as they mention the “global,” their personal observations seem to constitute the majority of their analysis of the four genres, rather than the language of commenting or other stats. While I’m intrigued by the theory behind their study, and I agree that factors such as globalization, genre, and mediation are important as a *framework* from which to test hypotheses, I am not at all convinced by their mode of research.

Leave a Reply