Our aim was to test the findings of a study which claimed that, if the syntactic structure of schizophrenic speech were subjected to a detailed linguistic analysis, clear differences would be demonstrated between schizophrenic, manic and control populations. It was confirmed that schizophrenics do have less syntactically complex speech which contains more errors. Using linguistic variables in a discriminant function analysis, it was possible to predict diagnoses correctly in 79% of cases.

A computer-assisted analysis of samples of free speech from acute schizophrenics (n = 50), chronic schizophrenics (n = 27) and normal subjects (n = 50) enabled a comparison of the linguistic profiles of the three groups. The chronic group consistently emerged as the most impaired, on measures of complexity, integrity (error) and fluency of speech, with the acute patients performing less well than normal speakers but better than chronic patients. Demographic differences could account for only a small number of the linguistic differences. A comparison of chronic schizophrenics from the community and those from long-stay wards suggested that their poor linguistic performance was in some way related to the illness process and not to institutionalisation. Three possible explanations for these results were considered, particularly the possibility that low complexity of speech, negative symptoms and poor outcome are in some way related.

To investigate whether language in schizophrenia deteriorated progressively, 11 schizophrenic subjects, 9 manic subjects and 9 controls were re-tested after an interval of three years using the computer-assisted syntactical analysis technique of Morice. In 13 of the 16 linguistic variables described as hallmarks of schizophrenic speech decline, deterioration was noted in schizophrenics in the direction predicted and relative to the manic and control groups. The deterioration was most pronounced in complexity and integrity of speech. One variable remained unchanged and two (semantic variables) showed marginal improvement. It was concluded that language, and in particular syntax, does deteriorate in the schizophrenic process.

Selected subjects from a group of first onset schizophrenics (aged under 30 years) were taken from a previous study and placed in one of two groups depending upon whether they had exclusively positive symptoms (n = 9) or a mixture of positive and negative symptoms (n = 9). Their linguistic profiles were compared with those of a group of controls (n = 10) matched for educational attainment and parental social class. Both groups of schizophrenics had significantly lower integrity scores, suggesting that they made more syntactic and semantic errors. Those patients who presented with negative symptoms tended to have speech of lower syntactic complexity than the other two groups, although the difference just failed to reach statistical significance. It is suggested that syntactic and semantic errors are state dependent features associated with positive symptoms, whereas low syntactic complexity may be a more enduring feature associated with the presence of negative symptoms.

No abstract