A computer program is now able to predict psychosis simply by observing language patterns, and with 100 percent success. Humans have been only 79 percent successful with that, says a report by RT News, an autonomous nonprofit by “TV-Novosti”. There is also news from America.

Read this in Polish

 

Predilection for “mental technology” has found a way not only into Russian columns. There are pages from the USA, too. The power of ELM to provide quick but near optimal solutions to the training of Single Layer Feedforward Networks (SLFN) allows extensive exploration of discriminative power of feature spaces in affordable time with off-the-shelf computational resources.

 

It was folk lore, some time ago, you scared children into good behavior. There had to be reasons to abandon the habit, however, it seems they might have been forgotten: would “pixies” be having a comeback, as computer algorithms to tell the future for grown-up people, too? Most people are familiar with the supernatural slasher. 😉

 

 

One good way to deal with a bogey is to have a look at it. Let us read the text. ELM is short for Extreme Learning Machines. Results by these machines can work in ensembles for Computer Aided Diagnosis. ELMs can help train artificial Single Layer Feedforward Networks. ELMs are pre-programmed, fast, affordable. There is yet nowhere in the text to read that human brains do not rely on “single-layer networks”, and really they do not.

 

Purportedly, it is semantic analysis to let the computer diagnose the human. For the first time in my life, I turn to machine translation, for poetry. We can try Emily Dickinson.

Lost

 

I lost a world the other day.
Has anybody found?
You’ll know it by the row of stars
Around its forehead bound.

 

Here, the pixie would lose the world, rather than a world.

 

LOST English to Polish GOOGLE

 

I am not going to elaborate on errors in Polish word endings, in which the pixie, with an Andersenian alas, does not make it for the fifth grade. Further, Polish does not have the articles, the words a, an, or the, but it does not neglect grammatical determinateness. In Polish, if we say “świat”, the sense is as for “the world”. If we want to say “a world”, we have to modify the noun.

 

Świat jaki zgubiłam, dni temu parę.
Czy odnalazł go kto?
Poznać po gwiazdach się daje
Na czele i w krąg.

 

With natural languages, we can have back-translation. The human translator receives the original and translated text, to back-translate into the language of the original, for verification. Here is the computer back-translating itself. A single-layer network would probably act similar. The Polish word “rząd” has a singular sense encoded, that of a government: the computer implies authorities, whereas supposed to leave it to people. 😉

 

LOST computer back translation Polish to English GOOGLE

 

The forehead in the verse is not literally a forehead. It is a metaphorical “forehead of a world”. The bogey evidently does not know about metaphors; more, makes a masculine world (“he-world”) in American, trying to tie around “his” forehead. Polish has collocations as “around the head” (wokół głowy) or “on the forehead” (na czole), also for headbands.

 

Feel welcome to my Emily Dickinson website.

Let us compare a few more lines. May it be The wind now.

 

Of all the sounds despatched abroad,
There’s not a charge to me
Like that old measure in the boughs,
That phraseless melody
The wind does, working like a hand
Whose fingers brush the sky,
Then quiver down, with tufts of tune
Permitted gods and me.

 

THE WIND English to Polish online GOOGLE

 

Here, the computer complains it is not paid. The Polish “opłata” can be a fee, charge, payment, et caetera. 😉

Back-translation lacks basic grammatical coherence, already in the first line.
THE WIND computer back translation Polish to English GOOGLE

 

All in all, I do not think computers would be good enough at language, to provide diagnoses or prognostics. They are not good enough for translation, and can work only as software to help format, or spare some of the typing. Further, language would not be a good diagnostic for a mental disturbance. Mental disorders are not primarily speech disorders, or consequences of language defects.

 

As Irena Styczek noted, within her holistic approach to speech therapy, schizophrenia may develop into schizophasia. We may view schizophasia as a phonologically driven discourse. The person would be “following speech sounds”, and not word sense. The example of word salad here would show a recourse to the palate (sadness-outage):

 

sadness cups coffee printer power outage.

 

Schizophasia, if it occurs, comes in late stages of the illness. Diagnosis most often would come sooner.

 

Brain fMRI would be probably better, in helping discern schizophrenia from other disorders. As Vander and others published in year 1985, neurological research had redefined schizophrenia as a somatoform disorder. Neuro-imaging detected differentiative biological structuring of altered function, in patient brains. Studies implied a developmental background, as a viral infection, malnutrition, or other, during fetal life or early after birth. The findings were not to encourage preemptive diagnoses. Importantly, the computer was to provide neuro-imaging. It was not to diagnose or predict illness.

 

Read this in Polish

 

Advertisements