Linguistic analysis is of great potential benefit to psychiatry as a research and assessment tool, but the skill and time it demands means that it has not been widely used. This paper describes a much simplified form of syntactic analysis.

A detailed protocol for the Brief Syntactic Analysis (BSA) was written, based on earlier work by Morice and Ingram. Three psychiatrists were trained in its use, and inter-rater reliability established through independent ratings of 12 transcripts taken from a mixed group of psychiatric patients and a group of non-psychiatric controls. Concurrent reliability of the BSA against the Morice and Ingram analysis was established by comparing measures from the two methods on 16 transcripts of mixed patients.

There were high levels of agreement between the three psychiatrists and between the BSA and the Morice and Ingram analysis, although one-way ANOVA indicated that for some variables there were small but statistically significant absolute differences between the two. The reasons for this were discussed. A principal components analysis confirmed the presence of three factors corresponding closely to the three families of linguistic variables.

The results indicate that psychiatrists can be trained to use a syntactic analysis with high levels of agreement. The BSA, which takes much less time to complete, produces measures that are comparable with the original analysis from which it was derived.

Several studies have revealed linguistic differences between diagnostic groups. This study investigates the extent to which these differences are accounted for by factors such as chronicity, or disturbances in cognition associated with acute psychosis.

Transcripts of interviews with patients suffering from RDC schizophrenia (n=38), mania (n=11) and controls (n=16) were examined using the Brief Syntactic Analysis (BSA). Patients were within two years of first onset of psychotic symptoms, and received tests of working memory and attention.

The speech of patients with schizophrenia was syntactically less complex than that of controls. Patients with schizophrenia and mania made more errors than controls. These differences were, to some extent, related to group differences in social class, working memory and attention, although significant group differences in language persisted after the effects of covariates were removed.

The study confirms the existence of differences in the speech of psychiatric patients. Low complexity appears to be a particular feature of speech in schizophrenia, even in the earliest stages of the condition. The importance of this finding is discussed in relation to two recent theories of schizophrenia: Crow's evolutionary model, and Friths neuro-psychological model.