Automated analysis of free speech predicts psychosis onset

in high-risk youths

Studies of mental functions, behaviors and the nervous system.

Moderators: Calilasseia, ADParker

Automated analysis of free speech predicts psychosis onset

#1  Postby Keep It Real » Sep 02, 2015 5:32 pm

Derived speech features included a Latent Semantic Analysis measure of semantic coherence and two syntactic markers of speech complexity: maximum phrase length and use of determiners (e.g., which). These speech features predicted later psychosis development with 100% accuracy, outperforming classification from clinical interviews. Speech features were significantly correlated with prodromal symptoms.


Full Journal Article
You're only conscious when you're thinking about consciousness.
User avatar
Keep It Real
Suspended User
THREAD STARTER
 
Posts: 9171
Age: 39

Print view this post

Ads by Google


Re: Automated analysis of free speech predicts psychosis onset

#2  Postby Keep It Real » Sep 02, 2015 6:03 pm

Image


The feature which seems to separate the majority of psychotic participants from those who did not develop psychosis seems to be semantic coherency. In other words, those who developed psychosis were less likely to repeat themselves in the interview. Am I reading that right? I find the journal article quite cerebrally challenging, I have to admit.
You're only conscious when you're thinking about consciousness.
User avatar
Keep It Real
Suspended User
THREAD STARTER
 
Posts: 9171
Age: 39

Print view this post

Re: Automated analysis of free speech predicts psychosis onset

#3  Postby Keep It Real » Sep 02, 2015 7:53 pm

Oh, come on you clever lot, I suffer from schizophrenia but really struggle to ascertain the differences in their speech in interview, from the journal article. Can anybody break it down into laymen's terms? I'm fascinated but can't interpret the journal article to my satisfaction.
You're only conscious when you're thinking about consciousness.
User avatar
Keep It Real
Suspended User
THREAD STARTER
 
Posts: 9171
Age: 39

Print view this post

Re: Automated analysis of free speech predicts psychosis onset

#4  Postby zoon » Sep 02, 2015 8:24 pm

I'm not one of the professionals or mathematicians here, so my guess comes with a pinch of salt. Semantic coherence seems to be not so much a matter of repeating oneself as of using words with related meanings in the same phrase (or group of phrases), where word relatedness has already been quantified by getting computers to break down many texts (some books mentioned were the Origin of Species, Moby Dick and Pride and Prejudice) to give a numerical rating to which words usually come close together, it's the kind of massive number crunching which would not have been possible without computers. It's just about which words usually appear close to each other, which isn't the usual meaning of "meaning" or "semantics", but it's something computers can measure. Quoting the paragraph in your link from which the above guess comes:
Bedi et al wrote:We employed a novel combination of semantic coherence and syntactic assays as predictors of psychosis transition. For the semantic analyses, we used a well-validated approach to automated text analysis previously used to analyze speech in schizophrenia,3 LSA17. LSA is a high-dimensional associative model that rests on the premise that word meaning is a function of the relationship of each word to every other word in the lexicon. If semantically similar words co-occur in texts with consistent topics more frequently than do unrelated words, then the semantic similarity of two words can be quantitatively indexed by the frequency of their co-occurrence in a sufficiently large corpus of texts.
User avatar
zoon
 
Posts: 3214

Print view this post

Re: Automated analysis of free speech predicts psychosis onset

#5  Postby Keep It Real » Sep 02, 2015 8:39 pm

zoon wrote:I'm not one of the professionals or mathematicians here, so my guess comes with a pinch of salt. Semantic coherence seems to be not so much a matter of repeating oneself as of using words with related meanings in the same phrase (or group of phrases), where word relatedness has already been quantified by getting computers to break down many texts (some books mentioned were the Origin of Species, Moby Dick and Pride and Prejudice) to give a numerical rating to which words usually come close together, it's the kind of massive number crunching which would not have been possible without computers. It's just about which words usually appear close to each other, which isn't the usual meaning of "meaning" or "semantics", but it's something computers can measure. Quoting the paragraph in your link from which the above guess comes:
Bedi et al wrote:We employed a novel combination of semantic coherence and syntactic assays as predictors of psychosis transition. For the semantic analyses, we used a well-validated approach to automated text analysis previously used to analyze speech in schizophrenia,3 LSA17. LSA is a high-dimensional associative model that rests on the premise that word meaning is a function of the relationship of each word to every other word in the lexicon. If semantically similar words co-occur in texts with consistent topics more frequently than do unrelated words, then the semantic similarity of two words can be quantitatively indexed by the frequency of their co-occurrence in a sufficiently large corpus of texts.


Fascinating, isn't it? I still don't understand the experiment fully but find it highly supportive of the language=thought hypothesis. It would be interesting to note if any of the authors of the studied texts showed psychotic symptoms. I'm guessing they assume they didn't. It's probably something of a spectrum IMO. It seems so very foreign to only be able to glimpse the precursors to illness with the use of number crunching computers...
You're only conscious when you're thinking about consciousness.
User avatar
Keep It Real
Suspended User
THREAD STARTER
 
Posts: 9171
Age: 39

Print view this post

Re: Automated analysis of free speech predicts psychosis onset

#6  Postby ScholasticSpastic » Oct 13, 2015 9:48 pm

Keep It Real wrote:<snip> language=thought </snip>

:shock:

Please don't do that. No it doesn't.
"You have to be a real asshole to quote yourself."
~ ScholasticSpastic
User avatar
ScholasticSpastic
 
Name: D-Money Sr.
Posts: 6354
Age: 45
Male

Country: Behind Zion's Curtain
United States (us)
Print view this post

Re: Automated analysis of free speech predicts psychosis onset

#7  Postby Keep It Real » Oct 22, 2015 1:00 am

Sure, a dog can think, but most waking human thought does take the shape of language imo....I think in terms of words at any rate,,,,speech to be more exact. An internal dialogue. Don't you?
You're only conscious when you're thinking about consciousness.
User avatar
Keep It Real
Suspended User
THREAD STARTER
 
Posts: 9171
Age: 39

Print view this post

Ads by Google


Re: Automated analysis of free speech predicts psychosis onset

#8  Postby igorfrankensteen » Oct 22, 2015 1:10 am

I suggest going with something like "use of language reflects or illustrates thought processes." That is, what the idea of automated analysis, or even "manual" analysis of speech based on.

If you use the "=" sign around mathematically oriented people too casually, then tend to get "jumpy."
User avatar
igorfrankensteen
 
Name: michael e munson
Posts: 2114
Age: 67
Male

Country: United States
United States (us)
Print view this post

Re: Automated analysis of free speech predicts psychosis onset

#9  Postby Keep It Real » Oct 22, 2015 1:16 am

I'm sure you're right; agree. If I'd have written "equals" instead do you think it would have elicited the same response?
You're only conscious when you're thinking about consciousness.
User avatar
Keep It Real
Suspended User
THREAD STARTER
 
Posts: 9171
Age: 39

Print view this post

Re: Automated analysis of free speech predicts psychosis onset

#10  Postby igorfrankensteen » Oct 22, 2015 2:08 am

Yes.
User avatar
igorfrankensteen
 
Name: michael e munson
Posts: 2114
Age: 67
Male

Country: United States
United States (us)
Print view this post

Re: Automated analysis of free speech predicts psychosis onset

#11  Postby ScholasticSpastic » Nov 02, 2015 8:15 pm

Keep It Real wrote:I'm sure you're right; agree. If I'd have written "equals" instead do you think it would have elicited the same response?

Yes. "Equals" and "=" mean the same thing. They denote an equivalency. Language and thought are not equivalent. If they were, we would be unable to think until we learned language- and how would we learn language without thinking? Our thoughts, once a language has been learned, are constrained by our language or structured by our language. But they are not equivalent things. Just as riverbeds are not rivers, atmospheric layers are not clouds, and mountain valleys are not glaciers.
"You have to be a real asshole to quote yourself."
~ ScholasticSpastic
User avatar
ScholasticSpastic
 
Name: D-Money Sr.
Posts: 6354
Age: 45
Male

Country: Behind Zion's Curtain
United States (us)
Print view this post


Return to Psychology & Neuroscience

Who is online

Users viewing this topic: No registered users and 1 guest

cron