Preamble

Quite some time ago, we looked at a the representation of South Korea and the K-Wave in news articles published in The Guardian. This examination of the topics for which articles were tagged revealed a large-scale increase in articles on South Korea tagged for K-Wave topics and a reduction in articles tagged for North Korea.

This month, we re-examine an updated corpus of Guardian articles to get some impression of how the content of articles which touch upon the K-Wave differs from all other coverage of South Korea over the first twenty five years of the twenty first century.

Data

The data comes from The Guardian Open Platform and was accessed using the convenient R wrapper for its API. In its raw form, it consists of all articles tagged “world/south-korea”. These were then filtered for articles with an online publication date between the years 2000 and 2024, inclusive. This was a total of 2244 articles.

A K-Wave sub-corpus was identified as those which also included the top-level tags (i.e., those that come before an oblique like the “world” in “world/south-korea”) “tv-and-radio”, “film”, “music”, “food”, and “fashion”. This was a total of 316 articles, representing roughly 14% of the total corpus.

Keyness

We first compare these corpora in terms of characteristic keywords, that is those words that appear significantly more frequently in one sub-corpus compared to the other as defined by the the statistical measure of keyness.

This reveals the fairly unsurprising fact that words denoting types of media as well as individual pieces of media and artists are characteristic of the articles tagged for the K-Wave, whereas words relating to international and regional politics, most especially security on the Korean peninsula, are more characteristic of articles tagged for South Korea that are not tagged for the K-Wave. More interesting is the composition of the top twenty key terms for the K-Wave sub-corpus as well as the magnitude of the keyness scores assigned to them. Previous reports in this series have revealed that Korean cinema is not consumed at a volume comparable to K-Dramas on international streaming services (most recently here), that Korean cinema constitutes only a very small fraction of all K-Wave motivated exports (here), and that it is Korean food and K-Pop which are most widely perceived across the Anglosphere as well-known Korean cultural products (here). Despite this, the word with by far the highest keyness score is “film” and five other words unambiguously related exclusively to cinema also appear in the top twenty keywords (i.e., “films”, “bong”, ““parasite,”cinema”, and “movie”). While there are important caveats to bear in mind when interpreting keyness scores (especially significance measures, such as the chi-squared score used here, rather than effect size measures as outlined in this paper), these results nevertheless strongly suggest that Korean cinema occupies a position in the UK press discourse around the K-Wave that does not align with its actual, relatively low rate of consumption or the public interest.

Feature Co-Occurence in the K-Wave Collection

We now move on to consider exclusively the K-Wave sub-corpus. Specifically, we will look at words which frequently appear in the same articles. A commonly cited aphorism in quantitative linguistics is that “you shall know a word by the company it keeps” (Frith 1957: 11). Identifying and visualising words that co-occur frequently in the same text allows us to do precisely that.

Starting from an understanding of topics as recurring sets of co-occuring words across documents, from an impressionistic reading of this static visualisation it is possible to identify two primary topics: K-pop and Korean cinema. Both the nodes ‘film’ and ‘k-pop’ are connected by thick edges (representing a higher frequency of collocation) to a number of other nodes which have connections largely among themselves. Furthermore, they lie on the shortest path between many of these peripheral nodes and the others in the network, that is, they appear to have high betweenness centrality.

Also notable is a third group of nodes which connect both among themselves as well as to both the nodes ‘k-pop’ and ‘film’. Whether these words constitute a third topic, perhaps articles devoted to the K-Wave as a global cultural phenomenon, judging from the connotations of the words, or whether they simply appear in articles on both the topics is K-pop and Korean cinema cannot be established in this analysis.

Interactive Network Visualisation

Examination and interpretation of the static visualisation in the preceding section provides a good deal of insight. It is, however, possible to leverage the affordances of online, interactive visualisation to communicate these more explicitly. This we do in the following reproduction the network diagram. While its layout may differ significantly from the static visualisation, it nodes and their connections do not.

The categories above are analytically imposed from a qualitative consideration of the structure of the network and the meanings of the words attached to the nodes. By clicking on individual nodes, you can highlight them along with the other nodes to which they are directly connected. The drop down menu in the top left corner highlights nodes by analytically imposed group. Hovering over an edge provides additional information about the number of co-occurrences between the words it connects across the sub-corpus. Examining the network in this way reinforces the interpretation given for the static visualisation above.

Conclusion

Reporting on the K-Wave makes up a significant proportion of reporting on South Korea. Its focus is largely on K-Pop and Korean cinema, while other reporting on South Korea is largely devoted to international relations and security. As interesting as what features is here is what does not. None of the keywords for the K-Wave sub-corpus were related to Korean cosmetics (a large economic contributor to K-Wave motivated exports) or the wider phenomenon of K-Beauty. The factors underlying this, editorial and otherwise, might be worth considering. Also worth revisiting is keyness. Not only could the terms identified here be analysed using an effect size measure, but it may be instructive to compare the entire South Korean corpus, including the K-Wave sub-corpus, with a North Korea to determine what if any role the K-Wave plays in distinguishing reporting on the ROK from reporting on the DPRK.

Acknowledgement
This work was supported by the Core University Program for Korean Studies of the Ministry of Education of the Republic of Korea and Korean Studies Promotion Service at the Academy of Korean Studies (AKS-2021-OLU-2250004)