Whether melancholia (endogenous depression) is a distinct entity or simply a severe case of major depressive disorder (MDD) remains a topic of debate. In this paper, the author discusses how to prove melancholia empirically and the difficulty in doing so. To prove the existence of the disease, it is first necessary to identify the syndrome using factor or cluster analysis. Then, the validity of the diagnosis must be tested by examining the correlation between the syndrome and validators. However, it is difficult to show consistently a correlation between melancholia and validators using the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition criteria for melancholia (DSM-MEL). One of the reasons for this is that four of the eight DSM-MEL items overlap with the criteria for MDD. Another reason is that the symptoms of melancholia have a "distinct quality" in that the patient cannot clearly verbalize, which is difficult to define explicitly. These limitations may be overcome by defining melancholia based on a key characteristic, psychomotor disturbances, or by using a prototypical diagnostic measure. Taking the perspective that psychiatric symptoms do not exist as entities, but rather, are the final result of dialogical co-constructs arising from the clinical relationship, suggests that demonstrating the existence of mental disorders has certain limitations.
 Author's abstract