P. CASTROGIOVANNI* - L. MORETTI - G. DE LISIO ~r 
VERBAL BEHAVIOUR. OF THE PSYCHOTIC 
AND PSYCHONEUROTIC PATIENTS: 
AN APPROACH ACCORDING TO THE METHODOLOGY 
OF COMPUTATIONAL LINGUISTICS 
The theoretical premise behind our investigation is based on the 
view, held by many authors, that not only the meaning but the structure 
of language itself is in some way the expression of the psychological 
characteristics of the person speaking. 
With this theory in mind we thought it would be interesting to 
check whether this supposed connection between verbal expression and 
personality is also confirmed by results of projective tests. We chose 
to use t~orschach test because the ten cards form a highly standardized 
stimulus and because, at the same time, the cards are indefinite and have 
little effect on the subject's verbal expression. 
The aim we set ourselves was to study the connection between 
linguistic expression and the characteristics of personality, in order 
to find out whether the morphological structure of the language used 
in the responses to the test might provide further information to add 
to that obtained by traditional signs. As is well known, the subjects' 
way of expressing themselves when confronted with each card often 
differ greatly even when their responses are given the same signs in 
the psychogram. 
METHOD 
We do not intend to go into details here about the method used, or 
the principles of selection of the subjects or the characteristics of the 
* Istituto di Psichiatria dell'Universit~t di Pisa (Dir. Prof. P. Sarteschi). 
72 1,. CASTROGIOVANNI- L. MORETTI- G. DE LISIO 
sample (already reported in a previous paper).1 The essential data is as 
follows. 
The sample is made up of 119 subjects, 32 females patients at the 
Psychiatric Clinic, University of Pisa (15 schizophrenics, 14 neurot- 
ics and 3 classified as "various ") and 87 relatives (41 relatives belong- 
ing to the schizophrenics, 36 to the neurotics and 10 to "various "). 
The responses to the Rorschach test were recorded on tape and 
then typed out, before being transcribed into perforated computer cards. 
The cards were then run through a computer which provided an alpha- 
betical list of all the words used by the 119 subjects. Below each word 
was given the context in which each word appeared. 
At this initial phase of our research we decided to limit our investi- 
gation to the analysis of traditional grammatical categories or parts 
of speech (substantives, adjectives, adverbs, verbs, pronouns, articles, 
prepositions, conjunctions, interjections). However in order to go deeper 
in our analysis of language, we further split these grammatical cate- 
gories into sub-divisions, taking gender (masculine and feminine) and 
number (singular and plural) of the parts of speech into consideration too. 
Therefore the lemma and the code number indicating the corres- 
ponding grammatical categories and sub-categories were transcribed 
opposite each word in the list drawn up by the computer. 
The computer calculated the number of words for each grammatical 
category and the respective percentages in relation to the total number 
of words spoken for each card, so that a summarizing table was obtai- 
ned for each subject. 
Then the total values for each category and sub-category for each 
subject were worked and finally for the various groups: schizophrenics, 
neurotics, various and the respective relatives. 
Being well aware of the arbitrariness of attempting to apply tradi- 
tional statistics tests to language, we deliberately decided not to try to 
make a statistical analysis at the initial stage of our examination of 
the data. We should be able to do this later, at the end of our analysis, 
when the results themselves might have suggested which of the many 
statistical tests and elaborations that might be borrowed from statisti- 
cal linguistics would prove most suitable. 
1 1 ). CASTROGIOVANNI, S. A. CERaI, G. MAI~PBI, P. J. PASQUINUCCI, G. TORRIGIANI, 
A. ZAMI'OLLI, Analisi llnguistica delle risposte al test di Rorschach di schizofrenici e neurotici 
e dei rispettivifamiliari. I: Metodologia e priml risultatl di analisi condotta mediante calcolatori 
elettronki, in <~ Neopsichiatria ~), XXXlV (1968), p. 810. 
VERBAL BEHAVIOUR OF PSYCHOTIC AND PSYCHONEUROTIC PATIENTS 73 
CONCLUSIONS 
The findings obtained so far in this first approach to the problem 
suggest some interesting conclusions: 
1. The grammatical profile is similar in the four groups consid- 
ered. Grammatical structure therefore represents the flame tmderlying 
the variability of language behaviour and as such has a certain static 
characteristic. 
2. The schizophrenic group shows certain specific aspects in 
respect to the other three groups: reduced number of words: higher 
percentages of substantives and articles; higher conceptual/articulatory, 
substantive/adjective, substantive/verb ratio, lower verb/adjective ratio. 
These linguistic characteristics of schizophrenic patients are correlated 
wit hthe autistic attitude of these subjects towards the test of Rorschach. 2 
3. The distribution of the grammatical categories for each card 
suggests the hypothesis that the type of stimulus (in this case the ten 
cards) does not alone determine substantial variations in the morpho- 
logical structure of speech. An exception is constituted by those ca- 
tegories (substantives, articles, adjectives) more closely dependent on 
the experimental situation in which the subject is asked to represent 
something. 
In a following phase of the study it was attempted to verify, at a 
micro and macrolinguistic level, whether, in the same patient, signif- 
icant linguistic variations during the evolution of a psychotic episode 
were present. 
In one case of a maniac-depressive psychosis, maniac type, 5 samples 
of spontaneous speech, corresponding to various stages in the evolution 
of the psychopathological picture, were taken. 
This material was examined at a phonematic, lexical, morpholog- 
gical, syntactic level in order to show whether there are variations of 
linguistic structures correlated with the variations of the symptomatol- 
ogy. 
This material is still in course of elaboration. 
The results are reported in detail on the following research: P. CASTROGIOVABINI, 
A. T~LAr~A, Pdmi risultatl dl un'anallsi statistica morfologica e lessicale delle risposte al test 
di Rorschach nelta prospettiva di uno studio dei rapporti fra psicopatologia e linguaggio, in A. 
ZAMPOLLI (ed.), Linguistica matematica e calcolatori, Firenze, 1973. 

