Topic modelling online depression forums: beyond narratives of self-objectification and self-blaming

Abstract

Background

Depression raises a double challenge: besides the negative mood and the intrusive thoughts, the relation to the self also becomes difficult. Online forums are analysed as communicative platforms enabling the interactive reconstruction of the self.

Aims

The discourses of online depression forums are explored. Firstly, narrative patterns are identified according to their thematic focus (e.g. dysfunctional body, challenges of intimacy) and discursive logic (e.g. information exchange, support). Secondly, narratives are analysed in order to describe various ways of grounding a depressed self.

Methods

∼70.000 depression-related posts from the biggest English-speaking online forums (e.g. www.reddit.com/r/depression, www.healthunlocked.com) were analysed. Quantitative (LDA topic modelling) and qualitative (deep reading) approaches were used simultaneously to determine the optimal number of topics and their interpretation.

Results

13 topics were identified and interpreted according to their content and communicative function. Based on the inter-topic distances four clusters were identified (medicalized, intimacy-oriented, critical and uninhabitable self-narratives).

Conclusions

The clusters of the 13 topics highlight various ways of narrating depression and the depressed self. Based on a comparison with a systematic review of mental illness recovery narratives, depression forums cover most narrative genres and emotional tones, thus create a unique opportunity for integrating the depressing experiences in the self.

Keywords:

• Depression
• narrative identity
• online forum
• text mining
• topic model

Acknowledgements

The authors would like to thank Fanni Máté for her contribution to data collection.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data used in this study is collected by the SentiOne social listening platform and are not publicly available due to confidentiality reasons but the pre-processed data are available from the corresponding author on reasonable request.

Notes

1 Researches about social media and depression should also be mentioned as a related, but distinct field (e.g. Moreno et al., 2011 64; Schwämmlein & Wodzicki, 2012). While social media platforms also enable sharing depressed mood, they differ from forums: the users are not anonymous, also the community is not organised around a certain mental health issue. In case of social media one is reaching out for help within their own network, in case of online forums a new network is established.

2 The selected forums included: www.reddit.com/r/depression, www.depression-understood.org, www.psychforums.com/depression, healthunlocked.com, www.mentalhealthforum.net, www.alonelylife.com, psychcentralforums.com, www.dealingwithdepression.co.uk, www.beyondblue.org.au, www.depressionforums.org, www.beatingthebeast.com.

3 For a recent overview see Maier, 2021 64.

4 https://www.nltk.org/book/ch02.html

5 http://www.nltk.org/howto/collocations.html

6 For a more detailed explanation of the methodology of similar topic modelling see Németh et al., 2021 64.

7 Relevance of a term (for the exact definition see Sievert & Shirley, 2014) is the sum of its topic-specific frequency and a penalty term that is an increasing function of the term’s overall frequency. The reason for introducing the penalty term is that terms that often occur in the given topic but are very common in the whole corpus are less relevant for the topic. The sum is a weighted sum, using weight of λ. The larger the λ, the less penalty is introduced. We set λ to be 0.6, which we found to give well interpretable results, and which was also found to be optimal by Sievert and Shirley (2014). As an illustration of relevant words per topic, see the visualization in the Appendix. Most relevant posts were defined as having a contribution of minimum 90% from the given topic, meaning these posts were mostly about the topic.

8 The remaining two topics (7 and 11) are distant from all the other ones, which expresses their marginal relevance for the mainstream discussions.

9 Even if participants of online depression forums are not necessarily “recovered” from depression, the comparison with such researches is not unjustified. Recovering from mental illness is not a finite process with a clear endpoint – therefore recovery narratives are never elaborated at the end of the process retrospectively. They are continuously experimented with. As online forums host such attempts, the forum posts are interpretable as “recovery narratives in the making”.

10 Analysing emotional tones with quantitative “sentiment analysis” is a promising direction of comparison. However, such analysis would have over-stretched the frames of the current article, so we decided to rely solely on a qualitative comparison of the most relevant expressions and posts.

11 A full comparison should include all of these dimensions – however, due to the lack of space, in this article only those two dimensions are analysed, which proved to be the most relevant. Also, in a more detailed analysis, each topic could be compared with these dimensions (not just their clusters) – this option was also given up due to lack of space. Taking these limitations into consideration, our comparison serves mostly a demonstrative purpose: it is supposed to establish links with the output of the related non-online research.

Additional information

Funding

This work was supported by the Higher Education Excellence Program of the Ministry of Human Capacities (ELTE–FKIP). The funding source had no involvement in conducting the research or in the preparation of the article.