‘Are We There Yet?’: An Outsider’s View of Psychiatric Diagnosis
I recently read a fine piece by Awais Aftab on validity in psychiatric classification, and it prompted me to reflect on how I think about diagnostic systems and mental illness as someone outside psychiatry: a linguist who also happens to live with schizophrenia.
I had my work cut out for me when I began my PhD. I had no prior research experience. My BA and MA were in creative writing, so everything I did during my doctoral work had to be built from the ground up. I learnt research methods while designing experiments, practised critical appraisal while reviewing the literature, and gradually found my footing as a researcher. One of the earliest challenges was identifying my epistemology (how we know) and ontology (what can be known). In the end, I described myself as a positive pragmatist: someone who believes that reality can be observed and measured, while also taking the view that the best method is the one that solves the problem at hand.
Given that positionality, I tend to view arguments between the critical and mainstream branches of psychiatry—particularly debates about how "valid" DSM categories are—with a degree of indifference.
My own research required me to trace the history of formal thought disorder back to its earliest descriptions. Those early psychiatrists relied on descriptive psychopathology. They observed recurring patterns in patient populations and developed theories and rating scales that helped clinicians describe what they were seeing. Their overriding concern was practical: helping people.
In the case of thought disorder, however, this approach inevitably produced rigid descriptive frameworks. Language actively resists neat categorisation, something clinical linguists understand well, so it is unsurprising that these frameworks sometimes overfit the rich and varied linguistic behaviours they sought to describe. In the process, much of what patients continue to do well has been overlooked. Across more than a century of thought disorder research, only a handful of researchers have seriously documented preserved linguistic abilities—and I am fortunate to count myself among them.
At the same time, my pragmatic side fully understands why these categories were developed and why they proved useful. We simply were not "there" yet. Corpus linguistics, the approach I applied to naturalistic speech data, did not exist in Kraepelin's day, nor even during much of Andreasen's work. We now possess methods capable of far greater precision, but descriptive psychopathology solved the immediate problem of its time: providing clinicians with a systematic way to describe what they were observing and to support patients accordingly.
I view the DSM in much the same way.
My research ultimately suggested that formal thought disorder is both transdiagnostic and linguistically fingerprinted. We observe thought disorder in autism, mania, schizophrenia, and even in people with no identifiable psychiatric condition. However, the linguistic profile differs across these groups. The formal thought disorder associated with autism is measurably different from that associated with schizophrenia.
I also found something that much of the existing literature—with the notable exception of Peter Liddle's work—had not led me to expect. Thought disorder is not characterised by speech errors that are absent from healthy populations. Rather, the linguistic behaviours associated with formal thought disorder occur in everyone as natural artefacts of speech production. They simply occur more frequently, and become more noticeable, in schizophrenia and other neurodevelopmental and psychiatric conditions.
For that reason, I recommended moving away from a categorical conception of formal thought disorder and towards a dimensional one.
What I did not conclude, however, was that the concept of thought disorder, or the rating scales built around it, are therefore meaningless. I criticise aspects of their methodology and argue that developments in computational psychiatry can move us towards a far more precise understanding. But I do not regard the previous century of work as a mistake.
This is where I find myself parting company with much of critical psychiatry. It often treats the imprecision of our current methods as evidence that the entire enterprise is fundamentally flawed. I disagree.
My own experience, and the literature I have read, suggest something rather different. Psychiatric diagnoses are useful approximations. They represent the best conceptual tools available at a given point in scientific history. As our methods improve, those categories will almost certainly become more refined, more dimensional, and perhaps eventually replaced altogether. But that does not mean they have failed.
Science advances through approximation. Each generation inherits imperfect models, improves them, and passes better ones to the next. There is little value in discarding useful frameworks simply because they are incomplete. Better tools do not invalidate the work that came before them; they build upon it.
We will arrive at more precise understandings when the evidence and methods allow us to do so. Until then, researchers and clinicians are doing what science has always done: using the best tools available while working to create better ones.
