In the case of large-scale epidemiological studies, there is evidence of substantial disagreement when lay diagnoses of schizophrenia based on structured interviews are compared with expert diagnoses of the same patients. Reasons for this level of disagreement are investigated in the current study, which made use of advances in text-mining techniques and associated structural representations of language expressions. Specifically, the current study examined whether content analyses of transcribed diagnostic interviews obtained from 150 persons with serious psychiatric disorders yielded any discernable patterns that correlated with diagnostic inconsistencies of schizophrenia. In summary, it was found that the patterning or structure of spontaneous self-reports of emotion states in the diagnostic interview was associated with diagnostic inconsistencies of schizophrenia, irrespective of confounders; i.e., age of patient, gender, or ethnicity. In particular, complex emotion patterns were associated with greater disagreement between experts and trained lay interviewers than were simpler patterns.