Introduction
If any man comes to the gate of poetry without the madness of the Muses, persuaded that skill alone will make him a good poet, then shall he and his works of sanity with him he brought to naught by the poetry of madness. (Phaedrus, translated by Hackforth, Reference Hackforth1972)
The idea from antiquity onwards that poetry is closely allied to madness is as frequently stated today. Symptoms of psychotic illnesses have often been identified in creative individuals, perhaps particularly in writers (Jamison, Reference Jamison1989, Reference Jamison1993; Sass, Reference Sass2001). And among them, poets have been proposed as a highly vulnerable group (Post, Reference Post1994). Perhaps the most well-known proponent, Jamison (Reference Jamison1993), has proposed a particular link between creative writing (and ‘romanticism’ in particular) and bipolar disorder. Consistent with this, an early family study by Andreasen (Reference Andreasen1987) reported that writers presented higher rates of mental illnesses, and bipolar disorder in particular. Higher prevalence of creativity and affective disorder was also found in first-degree relatives of creative writers. With an explicit focus on bipolar disorder, Nowakowska et al. (Reference Nowakowska, Strong, Santosa, Wang and Ketter2005) reported both bipolar and creative groups to share greater cyclothymic, dysthymic and irritability traits. Patient-based studies have also supported superior creativity in bipolar disorder (Santosa et al. Reference Santosa, Strong, Nowakowska, Wang, Rennicke and Ketter2007) especially during manic phases (Soeiro-de-Souza et al. Reference Soeiro-de-Souza, Dias, Bio, Post and Moreno2011). More recently, a large-scale population study (Kyaga et al. Reference Kyaga, Lichtenstein, Boman, Hultman, Långström and Landén2011) found that authors in a creative profession possessed a greater likelihood of schizophrenia, bipolar disorder, depression, suicide and substance abuse. In an alternative vein, Sass (Reference Sass2001) has linked ‘modernism’ – its proponents and its products – with the symptoms of schizophrenia in particular, suggesting that this differs in kind from the historical links between ‘romanticism’ and affective disorder. Of course such broad sweeping artistic movements are not easily empirically studied in contemporary writers.
Turning from psychopathology to personality, elevated positive schizotypal traits (analogous to positive psychotic symptoms but more broadly distributed in the general population) have been found in a range of creative groups, including the sole study to date to focus on poets (Nettle, Reference Nettle2006). Ando et al.'s recent study (2014) of comedians identified elevated traits that the authors argued related to both schizophrenia and bipolar disorder: this study substantially contributed to the methods of our own. We predicted that poets would possess both greater self-reported schizotypy and bipolar symptoms. We were also interested to explore the relationships between style of poetry and traits given the theories of Jamison and Sass discussed earlier.
Method
Recruitment of poets was carried out by emails to online poetry networks, poetry societies and poets known to H.M. (herself a published poet) mainly in the UK. H.M. also contacted other poets via online social media. All were asked to complete an online survey based on that of Ando et al. (Reference Ando, Claridge and Clark2014) with the addition of the Mood Disorder Questionnaire (MDQ: Hirschfeld, Reference Hirschfeld2000). This diagnoses bipolar disorder by self-report if seven or more symptoms are endorsed as occurring at the same time, and as causing ‘moderate-to-severe’ problems. The online survey was anonymous but gave contact details of the authors. In addition to the shortened O-LIFE questionnaire (Mason et al. Reference Mason, Linney and Claridge2005), the survey contained basic demographic information, type of poetry, whether published or not, and length of time writing poetry. General population O-LIFE scores used for comparison formed part of the normative sample as used by Ando et al. (Reference Ando, Claridge and Clark2014) matched for age. The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
Results
The poets’ sample consisted of 294 individuals (111 men and 183 women) with a mean age of 44.31 years (s.d. = 13.33). The norms sample consisted of 808 individuals (238 men and 570 women) with a mean age of 45.29 years (s.d. = 14.48). The majority of poets (91.8%) had published poetry in a variety of forms, and over half (52%) had been writing poetry for at least 12 years. Of the range of self-reported types of poetry, over two thirds (70.4%) endorsed a ‘lyric’ style, over half (58.2%) a ‘narrative’ style, and around a third (31%) an ‘avant garde’ style (not mutually exclusive categories). In terms of self-reported diagnosed mental illness, two poets reported schizophrenia, 15 reported bipolar disorder (5.1%), 152 reported depression (51.7%) and 80 reported anxiety disorder (27.2%).
O-LIFE and MDQ scores
Table 1 shows the mean scores for the groups on the four O-LIFE subscales. Differences between the groups were explored using MANOVA. As gender was not found to be a significant factor, differences between the groups are reported for each of the subscales (see Table 1). There were highly significant differences favouring higher scores in poets for Unusual Experiences, Cognitive Disorganization and Impulsive Nonconformity. According to the MDQ criteria, 54 poets (18.37%) met diagnostic criteria for bipolar disorder at some point in the past. This can be compared to the 3.7% base rate identified the US general population on the same measure (Hirschfeld et al. Reference Hirschfeld, Calabrese, Weissman, Reed, Davies, Frye, Keck and Wagner2003).
Differences in O-LIFE subscales between poets and controls
n.s., Non-significant.
** p < 0.01.
To test the hypothesis that style of poetry could be predicted by O-LIFE subscale and MDQ scores, MANOVAs were conducted for the three commonest styles (‘lyric’, ‘narrative’ and ‘avant garde’). The endorsement of an ‘avant-garde’ style (F = 3.92, p < 0.01) was associated with higher scores on Unusual Experiences (F = 12.10, p < 0.01) and MDQ (F = 15.54, p < 0.01).
Conclusions
In line with our hypothesis, poets showed a high level of psychotic personality traits consistent with several other studies of creative groups (Nowakowska et al. Reference Nowakowska, Strong, Santosa, Wang and Ketter2005; Nettle, Reference Nettle2006; Kyaga et al. Reference Kyaga, Lichtenstein, Boman, Hultman, Långström and Landén2011). They did so both on indicators of the positive symptoms of psychosis, and, quite markedly, on self-reported symptoms of bipolar disorder. Clearly the latter is not a confirmed psychiatric diagnosis; however, it is around five times the rates using the same measure seen in the general population. Although completely different in methodology to Andreasen's longitudinal study (1987) of 30 prominent writers, the result is broadly consistent with her finding that 12 of these (43%) suffered a bipolar episode at some point. Perhaps the most interesting findings pertain to the style of poetry associated with proneness to psychosis. If ‘avant-garde’ poetry is analogous to ‘modernism’, the association with positive schizotypy is supportive of Sass's thesis (2001) concerning schizophrenia. However, the association was not restricted to schizotypal traits – it also extended to bipolar symptoms. It is plausible that this style of poetry allows freest expression, or has the capacity to push against the boundaries of what is possible. Although we did not explicitly measure ‘over-inclusive thinking’ it is frequently advanced as a process linking creativity with psychopathology, and schizophrenia in particular (e.g. Keefe & Magaro, Reference Keefe and Magaro1980). Whether this is the grounds for such a link, it seems that poets, and ‘avant-garde’ ones in particular, are characterized by the marked presence of psychotic traits (for both schizophrenia and bipolar disorder).
Limitations
The online nature of the methodology does not allow estimation of the response rate or representativeness of the sample, and differs from that of the normed data. The means of recruitment may have particularly attracted respondents with greater distress. These traits and symptoms are self-reported and not corroborated by independent assessment. While the vast majority of poets described a range of published formats, it is not possible to vouchsafe the actual nature of their poetic productions.
Acknowledgements
Thanks are due to Professor Gordon Claridge for details of the survey and to the many anonymous poets who helped complete the study.
Declaration of Interest
None.
