I have a job in which I must try to differentiate between persons who are actually psychotic and those who are faking psychosis, and it is often quite difficult. I would like to identify characteristics of language that have a high probability of accurate discrimination. I have included an example of language from (I believe) a genuinely psychotic individual.

He stated, "I’ve always been adjusted toward righteousness. My name is [Abraham Lincoln] so they know I’m a man of good qualities. They’ve always wanted me to be a leading assister." I asked how he has been getting along on the ward. He stated, "I don’t speak to the people in the dayroom, but whenever I get a chance to speak, I do. My house - my radiance and substance - my keeping it positive affects my neighbors." He mused, "There have been some expensive situations. They’re unexplainable, as far as the influence and the vibration of the system; the appreciation was hard for me to grasp. It was irresponsible behavior - the staff I was around on 102 - usually it’s the patients or people of lesser quality who make you feel uncomfortable. This time it was the staff being ceremonial around the medications. I don’t want to sound like a conjurer so I won’t try to explain it. It was expensive because I didn’t get a chance to appreciate things, so it cost me a few days of service."

  • 1
    It's the old problem of telling who's lying and who's not. It's not a linguistic problem; there are no magic formulae that one can depend on. Sorry.
    – jlawler
    Mar 10 '14 at 21:56
  • @jlawler: Still, there may be resources on this topic. It sounds interesting even if it probably wouldn't achieve what the OP expects. Mar 10 '14 at 23:21
  • Oh, there's millions being spent on it every day. The NSA, FBI, and CIA would all like to be able to tell who's lying from recorded voices alone (though they may not have to distinguish psychotics from fakes). However, it ain't that simple.
    – jlawler
    Mar 10 '14 at 23:24
  • 1
    I'm afraid this won't sound very helpful to you, but the only thing you can and should do is get those people a clinical evaluation by a psychiatrist. Any information that can be given within the length restrictions of an answer in this forum will give you very limited guidelines on which you should never make evaluations about someone's mental health.
    – robert
    Mar 13 '14 at 22:19
  • @Jlawler there are certainly some resources if you look into, say, appraisal theory. I've seen some theses on the subject, but I'm not sure whether they've since been published.
    – jimsug
    Mar 19 '14 at 13:42

This is certainly not something I'm an expert in, but it's interesting enough for me to take a poke around through google. There clearly is a literature of this stuff out there.

http://www.ncbi.nlm.nih.gov/pubmed/18768300 states that the linguistic features of schizophrenic language are mostly at the macro-linguistic level, which is generally not where computational linguistics techniques are at their best. My initial impression is that might point you toward running linguistic tests rather than looking at samples of arbitrary language. IS that viable in your context?

http://www.hindawi.com/journals/schizort/2012/825050/ looks interesting, but it doesn't address the question of lying. Still, for someone to immitate schizophrenia in this sort of testing would presumably require a very detailed knowledge of how schizophrenics behave.

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2441888/ is also interesting. It involves measuring patterns of brain activity using electrodes while conducting linguistic tasks. That would make lying much harder. Is that viable for you? (The well known difficulties associated with 'lie detectors' may or may not translate into this context, where it's not the lying per se that's being looked for).

I don't really know how on-target these references are, but I imagine that you'll quickly find more by following the citations and reverse citation links (eg through citeseer or google scholar)

  • The EEG/ERP study does not indicate an effect strong enough to be useful as a single-case diagnostic.
    – user3503
    Jun 12 '14 at 10:52

I'm not an expert in the field, but I believe "psychoses" is too broad a topic. I have come across studies that show a link between word-usage and autism or schizophrenia.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.