Letter to the Editor: Converging Approaches to Autistic Online Discourse.
Blend BERTopic with DSM-5 coding to map how autistic Reddit themes cluster and interconnect.
01Research in Context
What this study did
Kakuszi et al. (2026) wrote a short letter, not a lab study. They argue we should mix two tools to read autistic Reddit posts. First tool: BERTopic, a robot that finds hot themes without labels. Second tool: DSM-5 codes, the checklist clinicians already use.
The team wants to lay the two maps on top of each other. The goal is to see which Reddit topics link to which official autism features.
What they found
The paper is pure theory, so there are no numbers yet. The authors simply claim the blend will ‘yield deeper insight’ into how autistic people talk online. In short, they pitch a recipe, not a cake.
How this fits with other research
Jackson et al. (2025) already tested a hybrid autism check-up for adults. They mixed open client stories with DSM-5 items and got rave reviews. Brigitta wants to do the same trick with Reddit words instead of clinic words.
Mathur et al. (2026) also shout for change, but they want ABA to sit and listen to autistic scholars first. Brigitta skips the listening step and jumps straight to coding speech. The two papers push reform in different gears—ears versus algorithms.
Graber et al. (2023) warn that forcing neurotypical goals can hurt. If BERTopic finds ‘stimming is fun’ themes, but DSM-5 tags them as ‘abnormal,’ the merge could still pathologize normal autistic chat. The idea is exciting, yet the ethical guardrails are missing.
Why it matters
If you assess teens or adults who live online, you need their real words. This method could give you a live feed of what matters to autistic clients before you write goals. Try running BERTopic on a subreddit they like, then pair the top five themes with DSM-5 items you already track. Share the merged list with your client and ask, ‘Does this feel like you?’ You might start the plan with their voice, not yours.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Pick one autistic-led subreddit, run BERTopic on last month’s posts, and match the top theme to one DSM-5 criterion you track.
02At a glance
03Original abstract
We read with great interest the study by Fong et al. (2025) which used BERTopic to analyze autism-related Reddit posts and map the thematic structure of online autistic discourse. Their data-driven, context-sensitive approach provides valuable insight into naturally emerging topics within this community. Notably, the most prominent themes—such as social relationships, stimming, and sensory sensitivities—align closely with core features of Autism Spectrum Disorder. This unsupervised approach provides an important baseline, but we believe future research would benefit from convergence with alternative, theory-driven frameworks. In our smaller-scale, clinical-perspective study, we employed a supervised approach (Kakuszi et al. 2025). We established predefined clinical categories derived from the DSM-5 symptoms and found that the focus points of the discourse-specifically emotion regulation, social communication, and sensory processing-showed notable methodological and thematic consistency with the dominant topics found by Fong and colleagues. This consistency suggests that both emergent and theory-based categorizations capture the most salient aspects of the autistic experience. Looking ahead, we propose that the next crucial step is to move beyond mere topic identification to explore how these themes organize into broader, complex thematic systems. Specifically, a major issue is that a single post can conceptually contribute to more than one topic by discussing many themes; however, in the BERTopic implementation in Fong et al.'s paper, each post is assigned to exactly one topic cluster (Grootendorst 2022). We used hierarchical cluster analysis to DSM-5-based categories to examine how they grouped together. Similarly, examining the co-occurrence and hierarchical organization of the identified topics could provide a richer understanding of how themes connect within the discourse. We believe that combining unsupervised topic modeling with supervised, clinically-grounded classification–validating emergent structures against established frameworks and exploring the interrelations among various topics–offers a powerful and complementary methodological direction for research on online autistic communities. This work was supported by the Hungarian Scientific Research Fund (NAP2022-I-4/2022). The authors declare no conflicts of interest. Data sharing not applicable to this article as no datasets were generated or analysed during the current study.
Autism research : official journal of the International Society for Autism Research, 2026 · doi:10.1002/aur.70183