research articles bibliotherapy

An official website of the United States government

Official websites use .gov A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

• Publications
• Account settings
• Advanced Search
• Journal List

The impact of school-based creative bibliotherapy interventions on child and adolescent mental health: a systematic review and realist synthesis protocol

Hayley redman, g j melendez-torres, alison bethel, judith green.

• Author information
• Article notes
• Copyright and License information

Corresponding author.

Received 2023 May 9; Accepted 2024 Feb 12; Collection date 2024.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

There is a need to identify evidence-based interventions to be delivered in schools that can be used to improve child and adolescent mental health and wellbeing. Creative bibliotherapy is one proposed intervention. However, there has been, to date, no comprehensive assessment of the evidence for its impact on mental health and wellbeing. To fill this gap, we will conduct a systematic review and realist synthesis.

A systematic search of the bibliographic databases APA PsycINFO, Medline (via Ovid), CINAHL, ERIC, Education Research Complete (via EBSCOhost) and Web of Science (SCI, SSCI, AHCI, ESCI) for school-based creative bibliotherapy interventions on child and adolescent mental health. Types of study to be included: cohort studies, non-randomised comparative evaluations, randomised controlled trials. The data from all included studies will be summarised descriptively and strength of evidence appraised. This is a potentially large field of practice, with heterogeneous interventions; we will use methods from intervention components analysis to describe and categorise the range of components and approaches used in included interventions. To understand how interventions work and in which contexts, we will use methods from realist synthesis to develop an exploratory account of mechanisms in different settings and for different young people (contexts).

Findings will assess the range of evidence for the impact of creative bibliotherapy on child and adolescent mental health and wellbeing, the strength of evidence for the impact identified, and describe potential mechanisms. This review will be useful for a wide range of stakeholders considering implementing or developing interventions using creative bibliotherapy in school-based settings.

Systematic review registration

This protocol was registered at the International Prospective Register of Systematic Reviews ( https://www.crd.york.ac.uk/prospero/ ), registration number CRD42023410333. This review is funded by Wellcome Trust (221457/Z/20/Z).

Supplementary Information

The online version contains supplementary material available at 10.1186/s13643-024-02482-8.

Keywords: Bibliotherapy, Child and adolescent mental health, Schools

At least one in four of the UK population experience a mental health issue at some point in their lives [ 1 ]. Future in Mind , the report from the UK’s Children and Young People’s Mental Health and Wellbeing Taskforce, estimates that over half of mental health problems in adult life start by the age of 14 and 75% by age 18 ([ 1 ], p. 9). The report emphasises the need for evidence-based early intervention and multi-sectoral action. The report also stresses the importance of promoting mental health and wellbeing to everyone—not just focusing on mental illness and diagnosis. Universal services, such as schools, can play a key role in preventing mental health problems and promoting mental wellbeing. Furthermore, the effectiveness of taking a whole school approach to well-being has been shown for both physical and mental health and well-being outcomes, for example, body mass index, tobacco use, and being bullied ([ 1 ], p. 36).

To improve child and adolescent mental health and wellbeing, the UK Government has begun to implement strategies outlined in the 2019 NHS Long Term Plan [ 2 ] and the 2017 Green Paper on Transforming Children and Young People’s Mental Health Provision [ 3 ]. Several of these plans are relevant to promoting mental health and well-being in all children in schools, including the integration of mental well-being learning into the curriculum and ensuring a whole school approach to wellbeing. The National Institute for Health Care Excellence (NICE) guidance on Social, emotional, and mental well-being in primary and secondary education [ 4 ] recommendations include ensuring ‘that the curriculum for all pupils includes evidence-based, culturally appropriate information about social, emotional and mental wellbeing to develop children and young people’s knowledge and skills as part of the whole-school approach’ including the integration of ‘relevant activities into all aspects of education to reinforce the curriculum offer about social, emotional and mental wellbeing and skills’ and ‘universal interventions’ ([ 4 ], p. 9–10). Creative bibliotherapy could be an efficient and effective tool in a school’s arsenal to promote mental health and wellbeing. Suvilehto [ 5 ] argues that many teachers already practice bibliotherapy in some manner, without giving their practice a formal name. This systematic review will assess the range and quality of evidence for the effectiveness of creative bibliotherapy in school settings.

Creative bibliotherapy

Bibliotherapy practice is a multifaceted and complex mixture of approaches and interventions operating under the broad banner of using books to heal ([ 6 ], p. 18). An analysis of the 100 most cited papers on bibliotherapy identified depression, anxiety, panic disorder, insomnia, and aphasia as key areas for bibliotherapy’s application [ 7 ].

There is no universally agreed definition of bibliotherapy–Hicks argues there is almost as much diversity in the definition of the term bibliotherapy as there is in its practice ([ 6 ], p. 13). Bibliotherapy spans a continuum,stretching from the use of creative literature to promote health and well-being at one end to clinical intervention and psychological therapy at the other, with considerable variation in between ([ 6 ], p. 13). Sitting on either end of this continuum, Brewster [ 8 ] presents two distinct models of bibliotherapy: self-help and creative, representing a synthesis of models from the literature that also reflect current practice in the UK. These terms have since been widely used in the literature (for example [ 9 , 10 ]):

Self-help bibliotherapy: the use of nonfiction self-help books, often recommended by medical practitioners, to provide practical help to people with mental health problems.

Creative bibliotherapy: the use of fiction and poetry to work with individuals and groups to promote better mental health.

Creative bibliotherapy can be delivered in a number of ways. There are two key models of creative bibliotherapy currently practiced in settings such as the UK. The first stresses the individual and individual reading—the right book must be found for the right person at the right time. This model is practiced by The Reading Agency, a UK-based charitable organisation that works with partners in both the health and education sectors to promote using the ‘proven power of reading’ to help people tackle ‘life’s big challenges’ [ 11 ]. The second concentrates on reading literary works from the canon, that is, classic texts in English, such as the works of Jane Austen, Charles Dickens, and William Shakespeare, aloud in a group setting to facilitate discussion. This model is practiced by The Reader, another UK-based charity that works to bring people together, including schools and families, to ‘experience and enjoy great literature, which [they] believe is a tool for helping humans survive and live well’ [ 12 ]. Troscianko and colleagues highlight an interesting divergence in the literature: whilst the majority of existing bibliotherapy theory entails individual reading, most empirical work has assessed group reading [ 13 ].

This systematic review will focus on creative bibliotherapy. C reative bibliotherapy is defined as: the reading and discussion of creative texts including, but not limited to, fiction books, short stories, picture books, and poetry, including alternative formats (e.g. e-books). In comparison to the literature on self-help bibliotherapy, the evidence base for creative bibliotherapy is much smaller and more eclectic. Although the practice has been dated back to ancient Greek medicine [ 14 ], the UK National Association of Primary Care (NAPC) notes the word was rarely used in medical practice until 2004, when the UK National Institute for Health and Care Excellence (NICE) produced new guidelines for depression [ 15 ]. These guidelines suggested that for patients with mild depression, healthcare professionals should consider recommending a guided self-help programme based on cognitive behavioural therapy (CBT). Hicks [ 6 ] notes that this was driven in part by the UK’s socio-political context of ‘a health sector charged with becoming more productive, using resources more effectively, building capacity and engaging people in taking responsibility for their own health’ in the early 2000s ([ 6 ], p. 13).

There is an established evidence base for the effectiveness of self-help bibliotherapy for a range of mental health and wellbeing outcomes at all ages (e.g. [ 16 , 17 ]). However, the evidence base for creative bibliotherapy is far less well-developed. Furthermore, Troscianko [ 9 ] has argued that current research and practice of creative bibliotherapy is underdeveloped in its understanding of the mechanisms of change. Existing theories are based on minimal empirical evidence and are largely based on the individual reader paradigm [ 13 ].

Preliminary searches located two systematic reviews that have examined evidence of the impact of creative bibliotherapy on mental health and wellbeing. The first, from Montgomery and Maunders, investigates the effectiveness of creative bibliotherapy for internalising, externalising, and pro-social behaviours in children [ 10 ]; the second, from Glavin and Montgomery, reviewed creative bibliotherapy for post-traumatic stress disorder [ 18 ]. The eight randomised controlled trials included in Montgomery and Maunders’ review suggest creative bibliotherapy has a small to moderate positive effect on child behaviour [ 10 ]. No studies met the inclusion criteria for the second review. The authors note that whilst excluded studies ( N  = 13) provided valuable qualitative information regarding bibliotherapy’s acceptability and utility, they lacked a robust study design [ 18 ]. Neither review aimed to describe the mechanisms of change. However, Montgomery and Maunders note that ‘Although no definitive model of creative bibliotherapy emerges from the included studies, all interventions reflected to some extent the evidence-based steps of CBT’ [ 10 ]. They suggest that each creative text used ‘provided opportunity for identification of unhelpful beliefs and behaviours, challenging of their meaning, and the development of new beliefs and behaviours’ [ 10 ]. In the later systematic review, Glavin and Montgomery propose that this transporting effect of literary reading may also explain a ‘possible causal linkage between reading and PTSD [post-traumatic stress disorder] treatment through the lens of prolonged exposure techniques’ in which phenomena that would be threatening in the real world can be safely engaged within the fictional one [ 18 ]. These theories are discussed below.

Mechanisms of change

The realist synthesis will develop a provisional programme theory for creative bibliotherapy, drawing on evidence from the studies included in the systematic review, and also on the existing literature on mechanisms of change. For self-help bibliotherapy, Troscianko [ 9 ] argues there is an underlying assumption that (if) it works, it works because the therapeutic model (cognitive behavioural therapy [CBT]) it is based on works. Some suggest similar mechanisms for creative bibliotherapy. Dwivedi and Gardner argue that experiencing stories through fiction, poetry, and film could act on these same CBT mechanisms, teaching ‘new attitudes and belief systems’ [ 19 ]. During reading, both cognitive processes and emotional processes occur [ 20 , 21 ]. Cognitive processes such as recognition and reframing are key to the recognition of unhelpful cognitions and, as such, elicit more realistic thoughts and assumptions. Emotional processes, such as empathy and identification, allow for previously unconsidered and unhelpful cognitions to surface, allowing the reader to be challenged with new ways of interpreting these through insight into a fictional world. The reader begins ‘to understand others and their plights from perspectives other than [their] own’ ([ 20 ], p. 62).

Creative texts can emotionally transport the reader into a story that is both pleasurable and rewarding, with certain stories providing an opportunity to engage safely with emotional difficulties while the characters the reader connects with deal with their own [ 20 ]. Empirical research has shown that fiction is processed differently from non-fiction, with a respective difference in brain activation; for more on neural mechanisms of change see, e.g. Tribe et al. [ 22 ]. It is proposed that fiction improves our ability to understand other people’s perspectives, due to the way our brains process and comprehend narratives [ 23 ]. This is theorised as the ‘transportation effect’ whereby stories have the power to transport readers from the real to the narrative world [ 24 ]. Green argues that transportation into the narrative world ‘can lead to real-world belief (and behaviour) change’ [ 24 ]. Similarities, both demographic and between the reader’s life and the character’s story, may lead to a stronger sense of transportation ([ 25 ], p. 27). Sharing the experience with others in the real world (as in the group bibliotherapy model outlined above) allows the reader to form connections and community, considered building blocks in mental health recovery [ 26 ].

Shrodes [ 27 ] and Hynes and Hynes-Berry [ 28 ] present models of the conceptual effects and processes involved in successful bibliotherapy [ 8 , 29 ]. These models are presented in Fig.  1 .

Models of bibliotherapy [ 27 , 28 ]

Both of these models are based on the individual reader paradigm, which Troscianko and colleagues [ 13 ] argue suggest as emphasising the similarity between the reader, their ‘problem’ and the arc of the protagonist’s story. This similarity prompts a connection between the reader and the protagonist, provoking the subsequent stages of these models. Troscianko and colleagues critique the shortcomings of these models based on the individual reader paradigm, which faces both ‘empirical and theoretical obstacles’ for explaining ‘how and to what extent ‘similarity’ is therapeutically beneficial’ [ 13 ].

Jones [ 30 ] proposed core processes across the creative art therapies (CATs) including artistic projection, perspective and distance, embodiment, non-verbal experience as detailed by others, but also the playful space and the informed player, the participating art therapist, the active witness, and the triangular relationship. This theory places significant weight on the participant-therapist relationship and interaction. Research from the self-help model has tested the efficacy of various degrees of therapist involvement in bibliotherapy. This is often in the context of drivers to find low-cost alternatives to traditional therapy or to reduce contact hours. A meta-analysis of self-help interventions in the management of depressive symptoms, however, found ‘pure’ (as opposed to ‘guided’ self-help) had a minimal effect size (0.06) on depression [ 31 ]. There are concerns that this may have led to an exaggeration of the effectiveness of bibliotherapy without facilitation [ 32 ]. There is little research on the extent to which therapeutic contact drives the effectiveness of creative bibliotherapy. Furthermore, although noted by Billington et al. [ 33 ] as an essential ‘mechanism of action’, the extent to which the group setting drives change has not come under scrutiny in the literature. Any impact of these interactions (either with a therapist or within a group) could be attributed to Lazarus and Folkman’s ‘transactional model of stress and coping’ which contends a person’s capacity to cope and adjust to challenges and problems is a consequence of transactions (or interactions) that occur between a person and their environment [ 34 ].

Across the literature, for both self-help and creative bibliotherapy, little attention is paid to the actual content of the literature used for bibliotherapy. Brewster notes in most studies the only information provided about the text is ‘the number of pages in the book and a quantified reading age’ making it difficult to assess the importance of linguistic style and for self-help texts the ‘therapeutic approach and balance of instruction and reflection that facilitates effective treatment’ ([ 35 ], p. 9–10). Other than brief explanations of what the canon includes and a presumption that these texts contain narratives to which all can relate, there is a lack of information available on what books are used for creative bibliotherapy in studies. However, the individual reader paradigm would argue against such a universalised approach. Research by McNicol [ 36 ] concerning the reception of Arthur Frank’s illness narratives, found that restitution and chaos narratives were not well received: they were deemed unsatisfying, unsettling, and unhelpful. Narratives describing how to accept, live with and better understand health conditions were much better received, described by Frank as stories that ‘tell of searching for alternative ways of being ill’ ([ 37 ], p. 117). Again, this assumes an ‘identification’ based connection between the reader and protagonist which we neither know how or to what extent this is therapeutically beneficial [ 13 ].

Furthermore, a consideration of the book as a symbol or material object is not covered in the existing literature. Does the act of receiving a physical book specifically chosen for you or a personalised recommendation offer on some level the feeling of being cared for? If a patient was prescribed a book by a GP would this feel more personal and caring than a ‘regular’ prescription of pills? Although unconfirmed by the data generated, Lundmark hypothesised the Bible, for example, serves two functions as a coping tool: a collection of stories that bear meaning to the reader; and as an artifact, with the focus on the Bible as a ‘tangible physical object’ ([ 38 ], p. 142). This suggests there is scope for testing the significance of the book as a material object to the outcome of bibliotherapy interventions. As such, we will include interventions that use both physical books and alternative formats (e.g. e-books), and the realist synthesis will allow consideration of the effects of the book as material object.

Troscianko and colleagues highlight how the theoretical models from the individual reading paradigm tend to follow a common pattern: they emphasise the similarity between the reader’s problematic experience and the arc of the protagonist’s story [ 13 ]. Billington et al. [ 33 ], propose a set of four significant ‘mechanisms of action’ based on an evaluative study of group creative bibliotherapy. The first three were deemed essential to its success, while the fourth was considered influential: (1) ‘A rich, varied, non-prescriptive diet of serious literature’ including a mix of fiction and poetry. Both literary forms allowed participants to discover (or rediscover) modes of thought, feeling, and experience. (2) The group facilitator’s role ‘in expert choice of literature, in making the literature ‘live’ in the room and become accessible to participants through skillful reading aloud, and in sensitively eliciting and guiding the discussion of the literature’. The facilitator’s alert presence in relation to literature, the individual and the dynamics of the group is a complex and crucial element of the intervention. (3) The group’s role ‘in offering support and a sense of community’, discussions elicited in response to the texts allowed personal ideas, feelings, opinions and experiences to be shared, which was ‘demonstrably critical in ‘knitting’ the group together’. (4) The environment in contributing to the ‘atmosphere, group dynamic and expectation of the utility of the reading group’. This study identified that a group located in a mental health drop-in centre were ‘much more willing to engage with the literature for its own sake from the very outset of the study’ in comparison to a group located in a GP surgery, who tended to view the literature as something ‘prescribed’ [ 33 ]. These four mechanisms, however, are challenging to evaluate, as it is difficult to determine if the observed effects are due to all, none, or any subset of these four factors. Furthermore, the importance of the four factors in different configurations may be highly variable in different contexts.

By drawing on realist thinking of causation, we will identify indicative causal processes (i.e. mechanisms) that lead to the impacts of bibliotherapy. Where possible we will compare the causal processes we identify to those theorised in the literature, and by doing so potentially refine our understanding of bibliotherapy.

Methods/design

This protocol was registered at the International Prospective Register of Systematic Reviews ( https://www.crd.york.ac.uk/prospero/ ), registration number CRD42023410333. The protocol is being reported in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) statement, attached in Additional file  1 .

The systematic review and realist synthesis will answer the following questions:

What are the impacts of school-based creative bibliotherapy interventions on child and adolescent mental health?

What mechanisms can be identified through which impacts are achieved, and in which contexts?

Eligibility criteria

(p) population.

5–16-year-old children and adolescents attending mainstream schools—the equivalent to primary and secondary school ages in the UK.

(I) Intervention

School-based interventions that use creative bibliotherapy to improve children and adolescents’ mental health and wellbeing. Creative bibliotherapy involves the reading and discussion of creative texts including, but not limited to, fiction books, short stories, picture books, and poetry, including alternative formats (e.g. e-books). Non-fiction, didactic and self-help texts will be excluded.

Interventions must include literature-related components, as defined above. No restrictions will be placed on the discussion format (facilitated group discussion, peer-to-peer, individual with teacher/librarian) or setting within the school (class-based, whole school, small group). Interventions will be excluded if they are designed to be delivered by a clinical practitioner/health professional.

(C) Comparator

Studies will be included that include any valid comparator e.g. other intervention, a waitlist control, do nothing, or treatment as usual.

(O) Outcome

All mental health and wellbeing outcomes will be included, including, but not limited to depressive symptomatology, anxious symptomatology, internalising problems, externalising problems, conduct disorders, disruptive behaviour. Additional outcomes will include mediators of mental health and mental wellbeing, including but not limited to self-concept, self-efficacy, and mindfulness.

Information sources

The following bibliographic databases will be searched: APA PsycINFO, Medline (via Ovid), CINAHL, ERIC, Education Research Complete (via EBSCOhost) and Web of Science (SCI, SSCI, AHCI, ESCI).

Search strategy

The searches will include both free text terms and controlled vocabulary terms when available and appropriate, these will be developed by the information specialist. No date or language restrictions will be applied at the searching stage. Forwards and backward citation searching will also be undertaken in Scopus using the final included articles from the database searches. A draft search strategy is available in Additional file  2 .

Study records

Data management.

Details of all searches will be recorded. Search results will be downloaded to EndNote desktop software. Studies sourced through supplemental hand searching will be recorded and imported into EndNote.

Selection and data collection process

In the first screening, two reviewers will independently screen titles, abstracts and keywords of all the studies yielded by the search against the inclusion/exclusion criteria, displayed in Table  1 , using Rayyan [ 39 ].

Inclusion and exclusion criteria

Studies that fail to meet the inclusion criteria and any duplicates will be excluded, and full texts will be obtained for studies that appear to meet the inclusion criteria or for which there is any uncertainty. Full texts will then be screened independently by two reviewers to determine whether they meet the inclusion criteria, with reasons for exclusion noted. Disagreements will be resolved by a third reviewer.

Data for the systematic review will be extracted from included studies to map the range of evidence, assess quality in terms of risk of bias, and identify evidence for a realist synthesis. The latter will include evidence relating to a provisional programme theory for creative bibliotherapy. Data extraction will be performed by one reviewer using a data extraction form developed by the researchers for the purposes of this review. This form will be refined by the reviewer until the data extraction is complete, to ensure the appropriateness and usefulness of all fields.

Risk of bias in individual studies

Quality appraisal for the systematic review will be carried out by two members of the review team using the Cochrane tools for risk of bias in randomised and non-randomised studies. However, as this systematic review is aiming to map the range of evidence for school-based creative bibliotherapy no studies will be excluded on the grounds of quality.

Data synthesis: mapping the evidence

The data from all included studies will be summarised descriptively. Tables and text will provide key study characteristics which will be summarised and appraised. Details will include study characteristics (first author, publication year, origin), study design (sample size, population characteristics, risk of bias assessment criteria), intervention features (duration, mode of delivery), drop-out rates and assessment tools for primary and secondary outcomes.

Because these interventions are heterogeneous, we will use methods from intervention components analysis [ 40 ] to describe and categorise the range of components and approaches used in the included intervention.

Data synthesis: identifying potential mechanisms and contexts

Subsequently, to understand how interventions work and in which contexts, we will undertake a realist synthesis [ 41 ] to develop an exploratory account of how these interventions work (mechanisms) in different settings and for different young people (contexts). Capacity precludes a full realist synthesis: this will be restricted to the data from studies included in the systematic review, and theoretical evidence on potential mechanisms of change from selected theoretical studies (see Mechanisms of Change, above). We will develop this account using methods of constant comparison, working in pairs to consider the included evidence, and relate these to different intervention strategies identified in the components analysis. Reviewers will seek out the contextual (C) influences that are hypothesised to have triggered the relevant mechanism(s) (M) to generate the outcome(s) (O) of interest [ 41 ]. Synthesis will consist of developing an initial programme theory, informed by our narrative review of literature and an iterative process of refining this from explicit accounts from studies in the systematic review. We will then compare ‘how the programme was supposed to operate’ to the ‘empirical evidence on the actuality in different situations’—all along C-M-O lines. Analytic purchase comes from the ability to describe and understand the many contingencies that affect the likelihood of such interventions generating their intended outcomes [ 41 ]. In turn, this will provide exploratory indications about what schools might need to put into place to ensure the intervention is most likely to trigger the right mechanism(s) to produce the desired outcomes. This will contribute to a refined programme theory, which can be tested in future evaluations.

This protocol describes a planned systematic review of comparative studies of creative bibliotherapy interventions to understand the impact of school-based creative bibliotherapy interventions on child and adolescent mental health, with a realist synthesis. We are aware that this draws on two rather different paradigms of evaluation. Whereas the comparative studies included in the systematic review draw on probabilistic methods for identifying effect sizes, a full realist review would draw on a wider range of evidence, selected for its capacity to address and then refine the programme theory [ 42 ], and underpinned with a generative model of causality. There are good epistemological grounds for rejecting claims of ‘realist’ perspectives in reviews in which the underlying evidence is derived from probabilistic causal designs [ 43 ]. However, we believe that our synthesis can draw on some of the insights of a more realist perspective, using a narrower set of evidence. This combination has been used effectively in studies of similar topics, such as therapeutic writing [ 44 ]. Our synthesis will generate indicative insights on the mechanisms and contexts that are important to consider when designing future interventions of creative bibliotherapy. By undertaking a realist synthesis, even of a somewhat selective body of evidence, this review will provide an initial exploratory account of how these interventions work in different settings and for different people—the context-mechanism-outcome (CMO) configurations—to help us understand how, why, and for whom an intervention produced the desired and undesired outcomes. Capacity prevents a full realist review, and we recognise that by restricting the documents eligible for inclusion, we will be providing exploratory causal explanations. These can be used in future empirical studies to confirm, refute, or refine theorised CMOs. When completed, the findings of the systematic review may be of interest to educational professionals, health and social care practitioners, commissioners and providers, as well as professionals who work in the voluntary and community sectors. These will also provide an initial programme theory for testing in future evaluations of creative bibliotherapy.

The systematic review will identify evidence of impact from studies with a counterfactual. A realist synthesis will shed light on the mechanisms of change for creative bibliotherapy, which is currently underdeveloped in the literature.

Limitations

Given the lack of clarity in distinguishing between creative and self-help bibliotherapy in the literature, it can be difficult to deduce which bibliotherapy is being referred to in some texts.

Inclusion criteria restricted to studies with a comparator will potentially exclude more holistic appraisals of creative bibliotherapy. Capacity precludes a full realist synthesis, which might furnish better developed theory about the interaction of context and mechanisms.

Acknowledgements

The review is funded by Wellcome Trust. It is part of a larger study on Reading as Therapy conducted by the Wellcome Centre for Cultures & Environments of Health in collaboration with Exeter City of Literature. We are grateful to the wider study team for comments and input, in particular Anna Cohn Orchard and Liv Cooper from Exeter City of Literature, Des Fitzgerald and Gillian Partington.

Authors’ contributions

All authors contributed to the protocol development. AB developed the search strategy. GJMT developed the data synthesis. HR and JG drafted the protocol. All authors refined the protocol. All authors read and approved the final version of this protocol. JG is the guarantor.

This research was funded by the Wellcome Trust [Grant number 221457/Z/20/Z]. For the purpose of open access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission.

Availability of data and materials

Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate.

Not applicable.

Consent for publication

Competing interests.

The authors declare that they have no competing interests.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

• 1. Department of Health. Future in mind: promoting, protecting and improving our children and young people’s mental health and wellbeing. London: Department of Health; 2015. Available at: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/414024/Childrens_Mental_Health.pdf .
• 2. NHS. The NHS long term plan. London: NHS; 2019. Available at: https://www.longtermplan.nhs.uk/publication/nhs-long-term-plan/ .
• 3. Greening J, Hunt J. Transforming children and young people’s mental health provision: a green paper. London: Department of Health and Social Care and Department for Education; 2017. Available at https://www.gov.uk/government/consultations/transforming-children-and-young-peoplesmental-health-provision-a-green-paper .
• 4. NICE. Social, emotional and mental wellbeing in primary and secondary education . London: National Institute for Health and Care Excellence. Available at: https://www.nice.org.uk/guidance/ng223 . [ PubMed ]
• 5. Suvilehto P. We need stories and bibliotherapy offers one solution to developmental issues. Online J Complement Altern Med. 2019;1(5):1–4. doi: 10.33552/OJCAM.2019.01.000523. [ DOI ] [ Google Scholar ]
• 6. Hicks D. An audit of bibliotherapy/books on prescription activity in England. London: MLA; 2006. [ Google Scholar ]
• 7. Xu Z, Liu R, Guo L, Gao Z, Gao Z, Liu X, Li J, Li B, Yang K. The 100 most-cited articles on bibliotherapy: a bibliometric analysis. Psychol Health Med. 2022;28(9):1–7. [ DOI ] [ PubMed ]
• 8. Brewster L. Books on prescription: bibliotherapy in the United Kingdom. J Hosp Librariansh. 2009;9(4):399–407. doi: 10.1080/15323260903253456. [ DOI ] [ Google Scholar ]
• 9. Troscianko ET. Fiction-reading for good or ill: eating disorders, interpretation and the case for creative bibliotherapy research. Med Humanit. 2018;44(3):201–211. doi: 10.1136/medhum-2017-011375. [ DOI ] [ PubMed ] [ Google Scholar ]
• 10. Montgomery P, Maunders K. The effectiveness of creative bibliotherapy for internalizing, externalizing, and prosocial behaviors in children: a systematic review. Child Youth Serv Rev. 2015;55:37–47. doi: 10.1016/j.childyouth.2015.05.010. [ DOI ] [ Google Scholar ]
• 11. Readingagency.org. UK: the reading agency: about. 2023. Available at: https://readingagency.org.uk/about/ . [cited 20 Apr 2023].
• 12. Thereader.org. UK: The reader: what we do. 2023. Available at: https://www.thereader.org.uk/what-we-do/ . [cited 20 Apr 2023].
• 13. Troscianko ET, Holman E, Carney J. Quantitative methods for group bibliotherapy research: a pilot study. Wellcome Open Res. 2022;7:79. doi: 10.12688/wellcomeopenres.17469.1. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• 14. Brewster L. Medicine for the soul: bibliotherapy. Austral Public Libr Inform Serv. 2008;21(3):115–119. [ Google Scholar ]
• 15. NAPC. Reading Well: books on prescription: how bibliotherapy can help your patients and save your practice time and money. London: National Association of Primary Care; 2018. Available at: https://napc.co.uk/wp-content/uploads/2017/09/Reading-well.pdf .
• 16. Moldovan R, Cobeanu O, David D. Cognitive bibliotherapy for mild depressive symptomatology: randomized clinical trial of efficacy and mechanisms of change. Clin Psychol Psychother. 2013;20(6):482–493. doi: 10.1002/cpp.1814. [ DOI ] [ PubMed ] [ Google Scholar ]
• 17. Lewis KM, Amatya K, Coffman MF, Ollendick TH. Treating nighttime fears in young children with bibliotherapy: evaluating anxiety symptoms and monitoring behavior change. J Anxiety Disord. 2015;1(30):103–112. doi: 10.1016/j.janxdis.2014.12.004. [ DOI ] [ PubMed ] [ Google Scholar ]
• 18. Glavin CE, Montgomery P. Creative bibliotherapy for post-traumatic stress disorder (PTSD): a systematic review. J Poet Ther. 2017;30(2):95–107. doi: 10.1080/08893675.2017.1266190. [ DOI ] [ Google Scholar ]
• 19. Dwivedi K, Gardner D. ‘Theoretical perspectives and clinical approaches’ in Dwivedi, K. The Therapeutic Use of Stories. London: Routledge; 1997. [ Google Scholar ]
• 20. Oatley K. A taxonomy of the emotions of literary response and a theory of identification in fictional narrative. Poetics. 1995;23(1–2):53–74. doi: 10.1016/0304-422X(94)P4296-S. [ DOI ] [ Google Scholar ]
• 21. Oatley K. Meetings of minds: dialogue, sympathy, and identification, in reading fiction. Poetics. 1999;26(5–6):439–454. doi: 10.1016/S0304-422X(99)00011-X. [ DOI ] [ Google Scholar ]
• 22. Tribe KV, Papps FA, Calvert F. “It just gives people hope”: a qualitative inquiry into the lived experience of the Harry Potter world in mental health recovery. Arts Psychother. 2021;74:101802. doi: 10.1016/j.aip.2021.101802. [ DOI ] [ Google Scholar ]
• 23. Altmann U, Bohrn IC, Lubrich O, Menninghaus W, Jacobs AM. Fact vs fiction—how paratextual information shapes our reading processes. Soc Cogn Affect Neurosci. 2014;9(1):22–29. doi: 10.1093/scan/nss098. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• 24. Green MC. Narratives and cancer communication. J Commun. 2006;56:S163–S183. doi: 10.1111/j.1460-2466.2006.00288.x. [ DOI ] [ Google Scholar ]
• 25. McNicol S. ‘Theories of bibliotherapy’ in Brewster L, McNicol S. Bibliotherapy. London: Facet Publishing; 2018. [ Google Scholar ]
• 26. Leamy M, Bird V, Le Boutillier C, Williams J, Slade M. Conceptual framework for personal recovery in mental health: systematic review and narrative synthesis. Br J Psychiatry. 2011;199(6):445–452. doi: 10.1192/bjp.bp.110.083733. [ DOI ] [ PubMed ] [ Google Scholar ]
• 27. Shrodes C. Bibliotherapy: a theoretical and clinical-experimental study. Berkeley: University of California; 1949. [ Google Scholar ]
• 28. Hynes A, Hynes-Berry M. Bibliotherapy: the interactive process a handbook. New York: Routledge; 1986. [ Google Scholar ]
• 29. McCulliss D. Bibliotherapy: historical and research perspectives. J Poet Ther. 2012;25(1):23–38. doi: 10.1080/08893675.2012.654944. [ DOI ] [ Google Scholar ]
• 30. Jones P. The arts therapies: a revolution in healthcare. Abingdon, New York: Routledge; 2020.
• 31. Gellatly J, Bower P, Hennessy SU, Richards D, Gilbody S, Lovell K. What makes self-help interventions effective in the management of depressive symptoms? Meta-analysis and meta-regression. Psychol Med. 2007;37(9):1217–1228. doi: 10.1017/S0033291707000062. [ DOI ] [ PubMed ] [ Google Scholar ]
• 32. Febbraro GA. An investigation into the effectiveness of bibliotherapy and minimal contact interventions in the treatment of panic attacks. J Clin Psychol. 2005;61(6):763–779. doi: 10.1002/jclp.20097. [ DOI ] [ PubMed ] [ Google Scholar ]
• 33. Billington J, Dowrick C, Hamer A, Robinson J, Williams C. An investigation into the therapeutic benefits of reading in relation to depression and well-being. Liverpool: The Reader Organization, Liverpool Health Inequalities Research Centre; 2010. [ Google Scholar ]
• 34. Lazarus RS, Folkman S. Stress, appraisal, and coping. New York; Springer Publishing Company; 1984.
• 35. Brewster L. ‘Bibliotherapy: a critical history’ Brewster L, McNicol S. Bibliotherapy. London: Facet Publishing; 2018. [ Google Scholar ]
• 36. McNicol S. The impact of educational comics on feelings and attitudes towards health conditions. Manchester Metropolitan University [9 July 2020]. 2015.
• 37. Frank AW. The wounded storyteller: body, illness, and ethics. Chicago: University of Chicago Press; 2013.
• 38. Lundmark M. The Bible as coping tool: Its use and psychological functions in a sample of practicing Christians living with cancer. Arch Psychol Relig. 2019;41(2):141–158. doi: 10.1177/0084672419871116. [ DOI ] [ Google Scholar ]
• 39. Ouzzani M, Hammady H, Fedorowicz Z, Elmagarmid A. Rayyan—a web and mobile app for systematic reviews. Syst Rev. 2016;5:1. doi: 10.1186/s13643-016-0384-4. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• 40. Sutcliffe K, Thomas J, Stokes G, Hinds K, Bangpan M. Intervention Component Analysis (ICA): a pragmatic approach for identifying the critical features of complex interventions. Syst Rev. 2015;4(1):1–3. doi: 10.1186/s13643-015-0126-z. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• 41. Wong G, Westhorp G, Pawson R, Greenhalgh T. Realist synthesis. RAMESES training materials. London: The RAMESES Project; 2013. [ Google Scholar ]
• 42. Pawson R, Greenhalgh T, Harvey G, Walshe K. Realist synthesis-an introduction. ESRC Res Methods Prog. 2004;2:55. [ Google Scholar ]
• 43. Marchal B, Westhorp G, Wong G, Van Belle S, Greenhalgh T, Kegels G, Pawson R. Realist RCTs of complex interventions–an oxymoron. Soc Sci Med. 2013;94:124–128. doi: 10.1016/j.socscimed.2013.06.025. [ DOI ] [ PubMed ] [ Google Scholar ]
• 44. Nyssen OP, Taylor SJ, Wong G, Steed E, Bourke L, Lord J, Ross CA, Hayman S, Field V, Higgins A, Greenhalgh T. Does therapeutic writing help people with long-term conditions? Systematic review, realist synthesis and economic considerations. Health Technol Assess. 2016;20(27):1–368. doi: 10.3310/hta20270. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Data availability statement.

• View on publisher site
• PDF (1.2 MB)
• Collections

Similar articles

Cited by other articles, links to ncbi databases.

• Download .nbib .nbib
• Format: AMA APA MLA NLM

Add to Collections

SYSTEMATIC REVIEW article

Bibliotherapy as a non-pharmaceutical intervention to enhance mental health in response to the covid-19 pandemic: a mixed-methods systematic review and bioethical meta-analysis.

• 1 Centro de Investigaciones Económicas, Administrativas y Sociales, Instituto Politécnico Nacional, Mexico, Mexico
• 2 Cross-Functional Group in Clinical Ethics, XXI Century National Medical Center, Mexican Social Security Institute, Mexico, Mexico
• 3 Facultad de Medicina, Universidad Nacional Autónoma de México, Mexico, Mexico
• 4 Departamento de Traducción y Ciencias del lenguaje, Pompeu Fabra University, Barcelona, Spain
• 5 Servicio de Endocrinología, National Institute of Pediatrics, Mexico, Mexico
• 6 Metabolic Diseases Research Unit, XXI Century National Medical Center, Mexican Social Security Institute, Mexico, Mexico

Background: A non-pharmaceutical treatment offered as psychological support is bibliotherapy, which can be described as the process of reading, reflecting, and discussing literature to further a cognitive shift. The coronavirus disease 2019 (COVID-19) pandemic demands a response to prevent a peak in the prevalence of mental health problems and to avoid the collapse of mental health services, which are scarce and inaccessible due to the pandemic. Thus, this study aimed to review articles on the effectiveness of bibliotherapy on different mental health problems.

Methods: A systematic review was conducted to examine relevant studies that assess the effectiveness of bibliotherapy in different clinical settings as a treatment capable of enhancing a sense of purpose and its surrounding values. To achieve this, a systematic review, including a bioethical meta-analysis, was performed. A variant of the PICO (Participants, Intervention, Comparison, and Outcome) model was used for the search strategy, and the systematic review was conducted in three databases: PubMed, Bireme, and OVID. Inclusion criteria were relevant studies that included the keywords, excluding documents with irrelevant topics, studies on subjects 15 years or younger, and in languages besides Spanish or English. Starting with 707 studies, after three rounds of different quality criteria, 13 articles were selected for analysis, including a hermeneutic analysis, which was followed by a fourth and final recovery round assessing bibliotherapy articles concerning healthcare workers.

Results: Our findings showed that through bibliotherapy, patients developed several capacities, including the re-signification of their own activities through a new outlook of their moral horizon. There are no research road maps serving as guides to conduct research on the use of bibliotherapy to enhance mental health. Additionally, values such as autonomy and justice were closely linked with positive results in bibliotherapy. This implies that bibliotherapy has the potential to have a positive impact in different settings.

Conclusions: Our contribution is to offer a road map that presents state-of-the-art bibliotherapy research, which will assist institutions and healthcare professionals to plan clinical and specific interventions with positive outcomes.

Introduction

Amid the coronavirus disease 2019 (COVID-19) pandemic, a pressing problem faced by different health ministries is the mental health of the population, this includes both those who have been social distancing and staying indoors for long periods at a time and those considered essential workers who have continued laboring despite the considerable risk—among them are healthcare workers (HCWs). This has exposed the population to a variety of psychological diseases such as sleep disorders, depression, anxiety, and burnout. These disorders affect a broad spectrum of individuals from different backgrounds and across ages ( 1 – 3 ). If not addressed, the prevalence and financial burden of mental health disorders in communities affected by COVID-19 will grow exponentially. Furthermore, both the healthcare and socioeconomic systems will collapse if a significant improvement is not made in the diagnostic approach, prevention, and non-pharmacological treatment of psychological disorders.

Mental health problems can be addressed through a plethora of available treatments, such as psychosocial therapies or cognitive behavioral therapy (CBT), which are provided by trained psychologists ( 4 ). However, regardless of the availability of different treatments, few people with mental health problems have sought help during the COVID-19 pandemic ( 5 ). This setting, along with the need for social distancing, poses a challenge when considering available treatments for improving mental health.

One of the available non-pharmacological treatments in psychological literature is bibliotherapy ( 6 ). This is defined as reading as a guide to therapeutic change; bibliotherapy has been studied by mental health scientists in recent years as a tool, different from traditional interventions, that improves the readers' lives ( 7 , 8 ). Though several definitions have been created to conceptualize bibliotherapy , they all focus on three essential elements: reading material for inside- or outside-session use, a therapeutic and achievable goal, and personal improvement.

Bibliotherapy is better understood as the process of reading, reflecting upon, and discussing literature (personal narratives and stories). This discussion of curated literature promotes cognitive shifts within the reader ( 9 ). It is crucial to note that bibliotherapy differs from self-help strategies as the reflection and discussions of literature take place in a structured setting ( 10 ). The reading material is also subjected to scrutiny and has a specific purpose or problem that it addresses.

The use of books in a systematic clinical setting offers the possibility of improving mental health at a low cost. In addition, it represents an alternative for those who are hesitant to receive treatment for mental health problems ( 11 ). For essential workers, including healthcare professionals, it is essential to be treated and to get help. Not only do mental health problems lead to moral distress, but they can be associated with loss of values when treating patients, which in turn reduces the quality of care.

Bibliotherapy has shown positive results for various mental disorders in different trials, which justify the rational and empirical evaluation of this approach. However, in the existing literature, it remains to be studied whether bibliotherapy can enhance values that contribute to obtaining a sense of purpose. Therefore, the main objective of the present work is to review the principal studies that assess the effectiveness of bibliotherapy as a treatment to enhance a sense of purpose and values in those with different mental health problems. Based on these results, healthcare professionals and institutions can plan clinical and specific interventions that are well-tested, assessed, and valued and show clinical effectiveness in improving mental health and the work environment. As a result, different mental health issues, such as anxiety, depression, sleep disorders, and burnout, can be addressed through bibliotherapy.

A diagnostic screening was performed on five databases: PubMed, Bireme, OVID, Philosopher's Index, and JSTOR, searching for the most recent articles on the use of bibliotherapy as a non-pharmacological intervention to help mental patients and HCWs. Subsequently, a systematic review was conducted up to February 2018 to obtain original articles about available literature-based non-pharmacological treatments (bibliotherapy). This search was complemented with a screening up to 2020, incorporating the COVID-19 pandemic. The search strategy was based on the PICO (Participants, Intervention, Comparison, and Outcome) approach coupled with the PRISMA (Preferred Reporting Items for Systemic Reviews and Meta-Analyses) checklist ( 12 ). However, the comparison variable was removed due to the research being focused only on bibliotherapy as an intervention. The PIO (Participants, Intervention, and Outcome) strategy, which includes participants or problems, the intervention or exposure, and outcomes, was used to systematically search all databases. This type of modified search derived from the PICO model has also been used in other systematic reviews ( 13 – 15 ).

Search Strategy/Literature Search

The search strategy was carried out on five computerized databases: PubMed, Bireme, Philosopher's Index, JSTOR, and OVID. The first database used was PubMed, produced by the National Library of Medicine (a public body that depends on the National Institutes of Health of the United States), which, according to Cochrane ( 16 ), contains approximately 16 million references to journal articles from the year 1950 onward, as well as 5,200 indexed journals. This makes PubMed/Medline the most widely used database in the health sciences field globally. Subsequently, a search was carried out in the Virtual Health Library (VHL) database, produced by Bireme (Latin American and Caribbean Center for Information in Health Sciences), which is a specialized center of the Pan-American Health Organization (PAHO). There is a considerable importance to this database because it contains (indexed) the most relevant scientific literature journals from Latin America and the Caribbean. We performed the search as well in Philosopher's Index, which is a premier database designed to find publications of interest in the field of philosophy. The axiological analysis is the great interest for philosophers worldwide. Following is JSTOR that is a cross-functional database.

Finally, a search was carried out in OVID, the world's most trusted medical research platform, which has been a vital part of healthcare for over 20 years. OVID's flagship platform is the leading choice, globally, among clinicians, researchers, educators, and students in the medical, scientific, and academic fields.

Mesh terms were used to review this phenomenon. The following keywords were used for Participants: “healthcare personnel,” “healthcare professional,” “healthcare manpower,” “physician,” “doctor,” “nurse,” “social worker”; Intervention: “bibliotherapy”; and Outcome: “liberty,” “empowerment,” “tolerance,” “justice,” “benevolence,” “equity,” “respect,” “charity,” “beauty,” “autonomy,” “purity,” “ethical values,” “axiology,” “personal identity,” and “dignity.” The Boolean operator “AND” was used to link PIO variables, while the operator “OR” was used to combine keywords from the same variables. This search strategy was used to obtain relevant articles from each database, as shown in Figure 1 . All references were stored in Mendeley Desktop. Due to the scarce results on health workers, a redirection to the general population was done using the screened papers that were already in our curated database.

Figure 1 . Flowchart. It shows the selection process used to retrieve the final 13 articles. In the first selection round, 707 articles were obtained from three computerized databases (PubMed, Bireme, and OVID). Works with double references, irrelevant topic, or not written in English or Spanish were excluded during the second round of screening, leaving 25 articles. The third round of selection included a quality criterion in which the full text was read. Finally, 13 studies with over 80% were selected for a hermeneutic analysis.

Eligibility Criteria

In the first round, articles were selected from every electronic database and were screened for relevance based on their title and abstract. Works with a double reference, on participants 15 years old or younger, irrelevant topic, related to letters or books, and not written in English or Spanish were excluded. Before the second round, three authors performed an iterative review to confirm the actual relevance of each paper.

In the second round of selection, the full text of the articles was considered. A quality criterion, as seen in Supplementary Table 1 , was used to assess the methodological value of each reference. The articles were read and classified according to the following criteria: (a) clear investigation objectives, (b) inclusion of a research question according to the objectives, (c) adequate and solid methodology, (d) terminology definition, and (e) results according to the objectives. Each criterion was worth 20%, with a maximum quality score of 100%.

In the third round, 13 articles met the optimal quality criteria (higher than 80%). Each article was analyzed and codified using Atlas. ti software to highlight the backgrounds of the bibliotherapy interventions, methodological elements, and results. All articles were codified to develop a deeper qualitative analysis based on the different codes, their relations, co-occurrences, and their networks. The bioethical meta-analysis searched for networks that could link bibliotherapy as an intervention to enhance social, ethical, and professional values.

Finally, a fourth round was executed as a recovery round to retrieve any articles directly linked to the intervention of bibliotherapy for HCWs.

Relevance Assessment of the State of the Art of Bibliotherapy

The first round yielded 707 initial articles, after which 25 articles were retrieved in the second round of selection; lastly, 13 articles met the quality criteria and were retained for further analysis ( Figure 1 , Supplementary Figure 1 ).

Studies that had clear research questions and objectives, definitions of the measured concept, valid measuring instruments, detailed description of the methods, information of the targeted population, characteristics of the participants, addressed missing values, and used appropriate statistical analysis were considered. The results are shown in Table 1 .

Table 1 . Quality criteria for the second round of selection.

These 13 studies were analyzed in depth to highlight the strengths of each work. The following considerations were added to the analysis: if it was treatment or prevention, targeted disorder, duration of intervention, sessions per week, instrument used, and main results ( Table 2 ).

Table 2 . Description of the articles in the third round of selection.

One of the main findings showed that all studies addressed a treatment perspective rather than a prevention perspective. Three of these studies did not specify the length of the intervention, and two out of these 13 studies did not mention the kind of literature they used ( 17 , 18 ).

Considering the main results of these studies, three of them found no differences between the bibliotherapy group and control group ( 17 , 19 , 20 ). However, results from four other studies indicated that bibliotherapy may facilitate self-concept and an internal locus of control ( 20 – 23 ).

Another study showed that with bibliotherapy, though there was a significant change, it did not result in a better intervention than traditional treatment ( 19 ). Another study pointed out that bibliotherapy was better than being on the waiting list ( 24 ). Finally, the remaining three studies found that bibliotherapy was a potential self-help resource ( 10 , 24 , 25 ).

Of the 13 studies included, two used bibliotherapy as an additional treatment ( 19 , 23 ), while the remaining 11 studies tested bibliotherapy as the main treatment. One paper described pretreatment evaluation ( 17 ), while 12 studies evaluated pretreatment and posttreatment. None of the studies showed any adverse effects of using bibliotherapy as a treatment; however, two studies described the effect as not adverse, although this finding was not significant ( 19 , 26 ).

Regarding improvement, out of the 13 studies, six did not specify the percentage of patients who improved ( 11 , 20 – 22 , 24 , 27 ). Bilich et al. ( 10 ) describes that 31% of the total sample showed clinically significant changes ( 10 ), Hodgings et al. ( 26 ) reported a 23% improvement, and Kaldo et al. ( 18 ) reported that 68% of the progress was made in the bibliotherapy group ( 26 ). Furthermore, Macdonald et al. ( 25 ) mentioned a 100% advance in participants receiving bibliotherapy, and Wright et al. ( 23 ) specified an 89% recovery of the participants in the experimental manipulation condition.

The three studies found in the recovery round, which assessed articles directly linked with bibliotherapy in HCWs, showed the impact that different literary works can have. While two studies focused on the nursing population, Amar ( 28 ) examined the impact of personal stories on nursing students' education and Harrison ( 29 ) studied the use of imaginative literature in scholarly inquiry, Andersonet al. ( 30 ) assessed the importance of empathy in both physicians and patients through the use of graphic stories. Taken together, all three articles highlight the positive impact of bibliotherapy for healthcare personnel.

Semantic Networks as an Initial Compass

The database stored in Mendeley was analyzed for the frequency of the terms used in the title, keywords, and abstract of each work according to the search parameters specified in the Methods section. The results of the first search yielded 488 articles (without double references) ( Figure 2A ), and five studies were discarded because no abstract was found. Based on this volume of articles, the word count was modified to consider only terms with 75 coincidences, meaning those with a significant frequency ( Figure 2 ).

Figure 2 . Keyword frequency. Analysis of database stored in Mendeley regarding the words in the title, keywords, and abstract. (A) Four hundred eighty-eight articles were retrieved in the first round, (B) 35 in the second, (C) 25 in the third, and (D) 13 in the fourth. Results showed a coincidence in the most frequent words.

The five most common terms in the literature records were treatment (560), care (386), health (378), depression (316), and interventions (271). At this point in the research, bibliotherapy did not appear as a frequent word, since there were only 126 coincidences, placing it at the 38th position. As bibliotherapy was the key concept of the present study, another round was conducted to obtain relevant studies for the analysis.

In the second round, which yielded 35 documents, three were discarded due to the lack of an abstract. The most frequent terms in those articles were bibliotherapy (62), treatment (40), group (35), patients (31), and health (28), where the terms treatment and health coincided with the previous selection ( Figure 2B ).

In the crossed iteration of the second round, 25 articles were retrieved, where two were again discarded because they did not have an abstract. The same list of terms, as in the previous round, was found; nevertheless, based on the significantly reduced volume of articles, the coincidences were not limited. The most frequent words in these documents were bibliotherapy (40), treatment (30), group (24), health (22), and patients (22) ( Figure 2C ).

These new results showed a coincidence in the most frequent words except in the order of health and patients, since these terms were inverted in the present third round. It is worth noting that bibliotherapy was the most frequent word in the 25 articles that point out the proper delimitation of each selection round.

In the third round of selection, 13 final articles were retrieved. One of them did not have an abstract. The same list of terms as in the previous search was found, and there was no restriction on the number of elements in the word count. The most frequent words were bibliotherapy (23), group (23), treatment (21), patients (17), and participants (15) ( Figure 2D ).

There was a coincidence in four out of the five terms compared to the previous round of selection. This change shows that the articles had a strong methodological component in the implementation of bibliotherapy, which is consistent with the present systematic review.

In the last round, we also researched the complete text of the 13 articles; the most frequent words were group (427), treatment (402), bibliotherapy (295), participants (288), study (258), depression (244), health (191), mental (160), patients (160), and clinical (156).

Findings From Bibliotherapy and Values

Hereafter, a table was created to show values that were considered in each paper ( Supplementary Tables 1 , 2 ). The different values that a study impacted are shown, with a cognitive shift on the part of the participants or considered relevant, on how they view their own life, their treatment, or everyday activities. By doing this, it was found that all articles spoke about gaining autonomy from bibliotherapy, 10 works addressed liberty as a central value in the intervention, and five articles regarded being proactive toward treatment. No articles that highlighted honesty, veracity, justice, or beauty were found (as shown in Table 3 ). To specify what is understood for each value, examples of quotes can be seen in Table 4 .

Table 3 . Values considered in the last round of review.

Table 4 . Instances of values.

To develop the hermeneutic analysis, each paper was read in depth and codified using Atlas. ti software to create networks of terms. One example is shown in the next image ( Figure 3 ), where values such as liberty, autonomy, and justice are closely linked and associated with positive results in bibliotherapy.

Figure 3 . Values linked to bibliotherapy. Network of values performed in Atlas. ti. It illustrates how values such as autonomy, liberty, and justice are associated with positive results regarding bibliotherapy treatment. The figure depicts how autonomy is a crucial value found in all 13 texts of the final round as well as a network between values. The identification numbers refer to each of the 13 texts and the line in that text.

In the third round of selection, the analysis showed that autonomy, justice, and freedom were the values most frequently discussed in these papers. Autonomy was a value closely linked with positive results in the case of bibliotherapy because this is one of the ways to promote a self-help treatment that could promote empowerment and allow for the control of an increasing number of situations in patients' lives, enable them to solve their own problems, and acquire the skills necessary to do it ( 19 ). These studies showed improvements related to self-concept and locus of control, which also reflect improvements in the autonomy of a patient. However, bibliotherapy was also found to promote autonomy in the participation of a patient in their own treatment by being proactive when given access and benefiting from the information about their illness or condition and lowering the caregiver's burden by improving compliance and reducing anxiety episodes ( 27 ). Values such as autonomy and proactiveness also showed an impact on values, such as liberty, due to the enhanced self-efficacy and self-concept, an improvement in life possibilities, options open to patients with fewer symptoms, and better control of their day-to-day lives.

Meanwhile, justice may also be considered a value linked to bibliotherapy in the view of these works because it allows for a great deal of access to this form of intervention. Wright et al. ( 23 ) states that bibliotherapy is accessible to individuals who may be geographically or otherwise isolated; it is also a valuable form of treatment for those with limited economic resources and helps caregiving institutions to pay attention to larger groups with limited personnel. Bibliotherapy also allows for a more private way to address these health issues, without having to deal with negative perceptions or the reticence of those not willing to share their concerns with others.

These considerations may also be applied to a specific population, that is, HCWs who, due to the COVID-19 pandemic, have been under significant stress and their mental health has been threatened due to the conditions they face, moral burdens, and workload. In the literature reporting the specific case of mental health among HCWs, we found that all the previous stress factors signaled to widespread anxiety- and depression-related disorders ( 31 – 33 ). There was no report on bibliotherapy used to help HCWs; however, it stands as a viable option due to the logistic difficulties of offering standard treatments amid the pandemic.

A Hermeneutic Analysis on How Bibliotherapy Works

The analysis of these texts stemmed from phenomenological and hermeneutic approaches. The latter's, insofar as it concerns a naive reading, results are shown in Table 5 .

Table 5 . Hermeneutic analysis.

Phenomenologically, each paper was read in accordance with real, sensible experiences described. Thus, the reading focused on real-life scenarios behind research subjects. Compared with a wait-list control group, individuals receiving relapse prevention (RP) exhibited significant reductions in the frequency of panic attacks, panic cognitions, anticipatory anxiety, avoidance, and depression. In addition, individuals in the RP group were more likely to attain a “clinically significant change” in status on both panic-free status and level of avoidance more frequently than individuals in the control group ( 23 ). For example, in an article focusing on panic attacks, subjects are written in as individuals, far beyond being mere topics of interest due to panic attacks.

Although the table may reflect the subjects as mere numbers or components in the study, for the purposes of the bioethical and value-based medicine research that is behind this, the subjects are taken as single phenomena, independent of numbers or statistics. The real element of value then is the final discussion and conclusions that each paper arrived at and how they reflected on the effectiveness and ineffectiveness—that bibliotherapy may have as treatment or how it could aid in treatment. Research on bibliotherapy yields benefits when teaching people about the value of literature and how it may impact their daily lives and their day-to-day practices. As was seen in the 13 articles reviewed for this study, bibliotherapy's results and effects vary across the board; however, the general consensus seems to be positive.

A Road Map and Compass to Bibliotherapy as a Non-pharmaceutical Intervention

The results of prior studies illustrate some of the best practices that should be implemented in offering bibliotherapy ( Figure 4 ). Several studies were designed to compare control and experimental groups, with relatively small groups (most of them with fewer than 50 participants). Most of the studies' length was between 3 and 6 months, and these were often the ones that had positive results. The aim of the studies was to treat disorders such as depression, stress, and anxiety, as well as functional psychosis, among others; it is an option that can be considered as far reaching for a large population, including HCWs, that has mental health problems during the COVID-19 pandemic ( Figure 4 ).

Figure 4 . A road map to the systematic review. On the road map, we can appreciate the following: (1) That most of the studies on the last round of analysis were comparative experimental studies (9) that used a control group. (2) The studies most often had sample sizes smaller than 50 participants (6), but larger studies with less than 200 participants were also frequent. (3, 4) These studies tried to measure the efficacy of bibliotherapy mostly on patients with depression, anxiety disorders, and functional psychosis. (5) A variety of standardized tests, scales, and questionnaires were used along with interviews to measure the degree of change achieved through treatment. (6) Autonomy and liberty were the values most often related with positive results in these studies. (7) Bibliotherapy was offered in-person and through telephone sessions, although many of the studies did not pay enough attention to this aspect of the study. (8) A wide array of literature options was offered for these treatments, most frequently the clinician-supported problem-solving literature was used. (9) Nine out of the 13 studies reported positive results of bibliotherapy, which was considered a cost-efficient therapy suitable for mild to moderate disorders.

It is important to note that there are still several features of bibliotherapy that need further research. For example, most of the studies found were not specific about the type of sessions in which the therapy was offered; the recurring types of sessions were in-person and through telephonic contact. The last of these options, as well as online contact, is especially important if we wish to offer mental healthcare during times in which health services may be needed to be reserved for those who are critically ill. Another feature of bibliotherapy, which needs further research, is the materials, books, and booklets used in this type of therapy. Out of 13 studies, six used problem-specific books to offer therapy; however, the structure of such materials is not clear and cannot easily be reproduced. Moreover, one should consider those who cannot easily access hospitals; thus, studies on the wider fiction literature that may help are needed.

The articles present themselves as sturdily scientific; however, their motivations are to aid patients who suffer from problems such as mental disorders, addiction, and imprisonment. The most valuable piece of information for the purposes of this study is the reports of the individual studies' subjects, as more patients reported feeling an improvement with the kind of bibliotherapy provided by the conductors of the investigations. Although none of the 13 articles directly discussed bibliotherapy in treating service workers, doctors, medical personnel, or even workers who suffer from burnout, some of the articles may be pointing in this direction, aiding us in our own implementation of bibliotherapy as an alternative intervention in mental healthcare. Studies that address topics such as anxiety, depression, caregivers, and stress-related issues may have a direct impact on the way we understand the approach to be taken for our own subjects.

This study specifically addresses the issue of values—or lack thereof—and their possible furtherance through literature while simultaneously encompassing the axiological works of several philosophers that shed light on this topic. The selection of literary works has been carefully curated to each reflect a distinct value, which will thereafter be applied in clinical practice through value-based medicine, which reflects and grapples with these absences in evidence-based medicine. One of the main objectives is to take the correct approach when trying to understand and, in a way, treat our subjects with the implementation of literature as a manner of therapy.

The methodological approach was based on these previous studies on the effectiveness of bibliotherapy. It is considered that, given the literature's effectiveness as a reflection medium, it could impact the attendees in their everyday practice ( 10 ). The most common therapeutic intervention used for bibliotherapy is CBT. CBT helps clients identify their distorted and depressogenic thinking and learn more realistic ways to frame their experiences by reading and conducting exercises that are completed at home, with minimal or no supervision from a therapist ( 10 ).

The results showed autonomy and justice linked with positive results in bibliotherapy often because this type of therapy could promote empowerment, decision-making, and problem-solving. Enhancement of clinical autonomy was also often reported where patients were prone to participate in their own treatment, reducing the caregiver's burden by improving compliance and reducing anxiety episodes. Though the role that reflection plays in reviewing the experience of others and how it strengthens self-control and decision-making is not clear, its effects are noticeable.

Bibliotherapy is a complementary resource to the clinical treatment of a disease. It is a strategy that helps patients, through literature, to cope with their situation by identifying with the experiences lived by the characters, and from reading, to develop their own tools to make better decisions about their health and exercise control over their lives and their illness. It is well-known that literature, as a reflection of human existence, leads those involved to reflect on themselves and their environment, and that, in addition to its esthetic character, it possesses the richness of confronting individuals with their emotions, values, feelings, and conflicts. It is also a way of helping individuals express, live, and solve these. It is an intrinsic character of literature to serve as therapy, catharsis, and cure for any conflict that disrupts our existence, and that is why human beings have always resorted to it (and, of course, also to the arts) in some way as the best medicine for life.

The process involves three phases: identification, catharsis, and insight ( 34 ). First, the reader creates a bond with the character with whom they identify most; then, this character encounters a conflict and resolves it; and finally, the reader, having experienced the conflict of the character through the text, reflects on personal circumstances and internalizes some behaviors represented in the book that will serve as tools to resolve their own conflicts. Nussbaum ( 35 ) for the same reason points out that “the novel's capacity to explore the length and breadth of a life, but the combination of this exploratory power with the presence of a character who will count as a high case of the human response to value, that creates the telling argument.”

The key to moral behavior not only implies theoretical understanding, but it must be connected with practice, unleashing a clear consciousness in the reader in such a way that unpacks their moral objectives, values, and hierarchy of values, creating moral abilities involved in reading and interpreting it ( 35 , 36 ). Hence, patients can re-signify their own matters, being able to think about their lives and conflicts from a broader moral horizon.

In this study, we explored the heterogeneity of the outcome of bibliotherapy and its value network relationship. The results implied that the patient uses and develops several capacities in an indispensable way such as emotion, creativity, values, moral horizon, and imaginative capacity. This means that, as readers, we assume the challenge of unpacking our imperfections (such as physical, ethical, and axiological). Nussbaum ( 35 ) expresses it this way, “We notice the way we are inclined to miss things, to pass over things, to leave out certain interpretative possibilities while pursuing others.” In brief, to teach us “how we should live.” However, as Pellegrino [( 37 ), p. 16] states: “She or he can enmesh us in the variegated particulars of an imagined life, but that cannot replace the hard work of normative ethics. In the end, the reader must choose whether to accept, reject, or modify his or her own way of life in light of the experience gained by the evocations of affect and thought in a work of fiction.”

Patients with psychosis improved in their clinical symptomatology and cognitive and psychosocial functioning after having attended a reading group program compared to patients who did not attend such structured activities. Patients who attended the group also reported that reading activity had a positive impact on group cooperation dynamics and that it was perceived as highly pleasant, useful, and interesting ( 22 ). Although not to be taken as a single means to treat a patient, it is an aid to other kinds of therapy such as CBT.

The scientific literature reports demonstrate certain benefits from bibliotherapy, maybe not surpassing those of other psychological treatments, but since one of the advantages of the treatment is that it can be widely available, bibliotherapy can and should be considered when developing public policies to help take care of the mental health of those affected by the COVID-19 pandemic, and for physicians, nurses, and other healthcare professionals to cope with the saturation of healthcare services during the COVID-19 pandemic. There is still plentiful details of the treatment and the phenomenon to discover and be systematically assessed to expand the benefits for health personnel and prevent diseases such as sleep disorders, anxiety, depression, and burnout, which greatly decrease the quality of life of communities and healthcare professionals. However, some thought must be given to the mechanisms of implementation of such therapies, where the most common instruments of bibliotherapy are books, which are currently difficult to share. In this sense, electronic books and materials would probably be a better option for implementation.

After this systematic review, we respond to our main research question, building up the road map, and many conclusions regarding bibliotherapy can be drawn. First, when the methodology of a bibliotherapy treatment is conducted cautiously, positive effects can be seen, regardless of the diseases. It can be noted that bibliotherapy treatments promote values as supplementary profit. One of the main positive aspects of bibliotherapy is that it is a low-cost alternative that can reach those unable to access treatment during the COVID-19 pandemic; it is an integrative and multidisciplinary treatment that links psychology, medicine, humanities, and literature. Hence, this means that bibliotherapy could potentially be applied to a larger population and healthcare personnel and, when implemented in a structured way, could have a positive impact on enhancing mental health amid the COVID-19 pandemic.

Data Availability Statement

The original contributions presented in the study are included in the article/ Supplementary Material , further inquiries can be directed to the corresponding author/s.

Author Contributions

MA-B, NA-B, PS, and AH-B conceived and designed the experiments. DM-F, IM-C, MA-B, AH-B, PS, and NA-B performed the systematic research and/or bioethical meta-analysis, and analyzed the data. DM-F, IM-C, MMA-B, SR, AH-B, PS, MA-M, and NA-B wrote the paper and contributed to helpful discussions. All authors contributed to the article and approved the submitted version.

The authors declare that this study received funding from the Instituto Politécnico Nacional SIP-IPN 20200228 and 20210515, as well as from COFAA-IPN. The funder was not involved in the study design, collection, analysis, interpretation of data, the writing of this article, or the decision to submit it for publication.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments

We are indebted to Federica Porcu for performing some of the preliminary searches. The authors would like to acknowledge the experimental support and fruitful discussions provided by MSc Ana Beatriz Serrano-Zumago from the Cross-functional group of clinical ethics at Centro Médico Nacional Siglo XXI, IMSS. We also wish to thank Dr. Cristina Revilla-Monsalve, Joaquín González, and Dr. César González for their support. DM-F would like to thank Mexico's National Council on Science and Technology (CONACYT) for the scholarship number 1000198, which allowed her to further this research. The contributions of the assigned pre-graduate research fellows at the Universidad Iberoamericana are greatly appreciated. We are also thankful for the contributions of Rogelio Ezequiel and Alonso Loyo for the artwork.

Supplementary Material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpubh.2021.629872/full#supplementary-material

1. Palgi Y, Shrira A, Ring L, Bodner E, Avidor S, Bergman Y, et al. The loneliness pandemic: loneliness and other concomitants of depression, anxiety and their comorbidity during the COVID-19 outbreak. J Affect Disord. (2020) 275:109–11. doi: 10.1016/j.jad.2020.06.036

PubMed Abstract | CrossRef Full Text | Google Scholar

2. de Lima CVC, Cândido EL, da Silva JA, Albuquerque LV, Soares L, de M, et al. Effects of quarantine on mental health of populations affected by Covid-19. J Affect Disord. (2020) 275:253–4. doi: 10.1016/j.jad.2020.06.063

3. Peng M, Mo B, Liu Y, Xu M, Song X, Liu L, et al. Prevalence, risk factors and clinical correlates of depression in quarantined population during the COVID-19 outbreak. J Affect Disord. (2020) 275:119–24. doi: 10.1016/j.jad.2020.06.035

4. Suyi Y, Meredith P, Khan A. Effectiveness of mindfulness intervention in reducing stress and burnout for mental health professionals in Singapore. EXPLORE. (2017) 13:319–26. doi: 10.1016/j.explore.2017.06.001

5. Dissanaike S. How to prevent burnout (maybe). Am J Surg. (2016) 212:1251–5. doi: 10.1016/j.amjsurg.2016.08.022

CrossRef Full Text | Google Scholar

6. Cohen LJ. The experience of therapeutic reading. West J Nurs Res. (1994) 16:426–37. doi: 10.1177/019394599401600407

7. Castro Santana A, Altamirano Bustamante N, Castro Santana A, Altamirano Bustamante N. ? ‘Leer para estar bien?: prácticas actuales y perspectivas sobre la biblioterapia como estrategia educativo-terapéutica. Investig Bibl Arch Bibl Inf. (2018) 32:171. doi: 10.22201/iibi.24488321xe.2018.74.57918

8. Cohen LJ. Bibliotherapy. The therapeutic use of books for women. J Nurs Midwifery. (1992) 37:91–5. doi: 10.1016/0091-2182(92)90143-Q

9. Lanza ML. Literature: a vehicle for emotional connection between clinician and client. Arch Psychiatr Nurs. (1991) 5:313–8. doi: 10.1016/0883-9417(91)90030-9

10. Bilich LL, Deane FP, Phipps AB, Barisic M, Gould G. Effectiveness of bibliotherapy self-help for depression with varying levels of telephone helpline support. Clin Psychol Psychother. (2008) 15:61–74. doi: 10.1002/cpp.562

11. Chien W, Thompson D, Lubman D, McCann T. A randomized controlled trial of clinician-supported problem-solving bibliotherapy for family caregivers of people with first-episode psychosis. Schizophr Bull. (2016) 42:1457–66. doi: 10.1093/schbul/sbw054

12. Mamédio C, Roberto M, Nobre C. The Pico strategy for the research question. Rev Latinoam Enferm. (2007) 15:1–4. doi: 10.1590/S0104-11692007000300023

13. Roman P, Carrillo-Trabalón F, Sánchez-Labraca N, Cañadas F, Estévez AF, Cardona D. Are probiotic treatments useful on fibromyalgia syndrome or chronic fatigue syndrome patients? A systematic review. Benef Microbes. (2018) 9:603–11. doi: 10.3920/BM2017.0125

14. Machado De Lima DV, Aparecida Lacerda R. Hemodynamic oxygenation effects during the bathing of hospitalized adult patients critically ill: systematic review. ACTA Paul Enferm. (2010) 23:278–85. doi: 10.1590/s0103-21002010000200020

15. Mlenzana NB, Frantz JM, Rhoda AJ, Eide AH. Barriers to and facilitators of rehabilitation services for people with physical disabilities: a systematic review. Afr J Disabil. (2013) 2:22. doi: 10.4102/ajod.v2i1.22

16. Centro Cochrane Iberoamericano. Manual Cochrane de Revisiones Sistemáticas de Invervenciones . Barcelona: Centro Cochrane Iberoamericano (2012).

Google Scholar

17. van Lankveld JJ, Grotjohann Y, van Lokven BM, Everaerd W. Characteristics of couples applying for bibliotherapy via different recruitment strategies: a multivariate comparison. J Sex Marital Ther. (1999) 25:197–209.

PubMed Abstract | Google Scholar

18. Kaldo V, Ramnerö J, Jernelöv S. Involving clients in treatment methods: a neglected interaction in the therapeutic relationship. J Consult Clin Psychol. (2015) 83:1136–41. doi: 10.1037/ccp0000039

19. Joling KJ, van Hout HPJ, van't Veer-Tazelaar PJ, van der Horst HE, Cuijpers P, van de Ven PM, et al. How effective is bibliotherapy for very old adults with subthreshold depression? A randomized controlled trial. Am J Geriatr Psychiatry. (2011) 19:256–65. doi: 10.1097/JGP.0b013e3181ec8859

20. Kohutek KJ. Bibliotherapy within a correctional setting. J Clin Psychol. (1983) 39:920–4. doi: 10.1002/1097-4679(198311)39:6<920::AID-JCLP2270390616>3.0.CO;2-N

21. Evans K, Tyrer P, Catalan J, Schmidt U, Davidson K, Dent J, et al. Manual-assisted cognitive-behaviour therapy (MACT): a randomized controlled trial of a brief intervention with bibliotherapy in the treatment of recurrent deliberate self-harm. Psychol Med. (1999) 29:19–25. doi: 10.1017/S003329179800765X

22. Volpe U, Torre F, De Santis V, Perris F, Catapano F. Reading group rehabilitation for patients with psychosis: a randomized controlled study. Clin Psychol Psychother. (2015) 22:15–21. doi: 10.1002/cpp.1867

23. Wright J, Clum GA, Roodman A, Febbraro GA. A bibliotherapy approach to relapse prevention in individuals with panic attacks. J Anxiety Disord. (2000) 14:483–99. doi: 10.1016/S0887-6185(00)00035-9

24. Reeves T. A controlled study of assisted bibliotherapy: an assisted self-help treatment for mild to moderate stress and anxiety. J Psychiatr Ment Health Nurs. (2010) 17:184–90. doi: 10.1111/j.1365-2850.2009.01544.x

25. Macdonald J, Vallance D, McGrath M. An evaluation of a collaborative bibliotherapy scheme delivered via a library service. J Psychiatr Ment Health Nurs. (2013) 20:857–65. doi: 10.1111/j.1365-2850.2012.01962.x

26. Hodgins DC, Currie SR, El-Guebaly N, Diskin KM. Does providing extended relapse prevention bibliotherapy to problem gamblers improve outcome? J Gambl Stud. (2007) 23:41–54. doi: 10.1007/s10899-006-9045-1

27. Buwalda FM, Bouman TK. Cognitive-behavioural bibliotherapy for hypochondriasis: a pilot study. Behav Cogn Psychother. (2009) 37:335–40. doi: 10.1017/S1352465809005293

28. Amar AF. Violence education in nursing: critical reflection on victims' stories. J. Forensic Nurs. (2008) 4:12–8. doi: 10.1111/j.1939-3938.2008.00002.x

29. Harrison E. Advancing nursing scholarship through the interpretation of imaginative literature: ancestral connectedness and the survival of the sufferer. Adv Nurs Sci. (2001) 24:65–80. doi: 10.1097/00012272-200112000-00007

30. Anderson PF, Wescom E, Carlos RC. Difficult doctors, difficult patients: building empathy. J Am Coll Radiol. (2016) 13:1590–8. doi: 10.1016/j.jacr.2016.09.015

31. Civantos AM, Byrnes Y, Chang C, Prasad A, Chorath K, Poonia SK, et al. Mental health among otolaryngology resident and attending physicians during the COVID-19 pandemic: national study. Head Neck. (2020) 1597–609. doi: 10.1002/hed.26292

32. Zerbini G, Ebigbo A, Reicherts P, Kunz M, Messman H. Psychosocial burden of healthcare professionals in times of covid-19 – a survey conducted at the university hospital augsburg. GMS Ger Med Sci. (2020) 18:1–9. doi: 10.3205/000281

33. Rodriguez RM, Medak AJ, Baumann BM, Lim S, Chinnock B, Frazier R, et al. Academic emergency medicine physicians' anxiety levels, stressors, and potential stress mitigation measures during the acceleration phase of the COVID-19 pandemic. Acad Emerg Med. (2020) 27:700–7. doi: 10.1111/acem.14065

34. Shrodes C. Bibliotherapy: A Theoretical and Clinical-Experimental Study. (1950). Available online at: https://openlibrary.org/books/OL14596117M/Bibliotherapy_a_theoretical_and_clinical-experimental_study (accessed April 25, 2019).

35. Nussbaum MC. Flawed crystals: james's the golden bowl and literature as moral philosophy. New Lit Hist. (1983) 15:25. doi: 10.2307/468992

36. Jalongo MR. Bibliotherapy: literature to promote socioemotional growth. Read Teach. (1983) 36:796–803.

37. Pellegrino ED. Professionalism, profession and the virtues of the good physician. Mt Sinai J Med. (2002) 69:378–84.

Keywords: bibliotherapy, litherapy, mental health, coronavirus disease 2019, pandemic, values, bioethics, systematic review

Citation: Monroy-Fraustro D, Maldonado-Castellanos I, Aboites-Molina M, Rodríguez S, Sueiras P, Altamirano-Bustamante NF, de Hoyos-Bermea A and Altamirano-Bustamante MM (2021) Bibliotherapy as a Non-pharmaceutical Intervention to Enhance Mental Health in Response to the COVID-19 Pandemic: A Mixed-Methods Systematic Review and Bioethical Meta-Analysis. Front. Public Health 9:629872. doi: 10.3389/fpubh.2021.629872

Received: 20 November 2020; Accepted: 12 January 2021; Published: 15 March 2021.

Reviewed by:

Copyright © 2021 Monroy-Fraustro, Maldonado-Castellanos, Aboites-Molina, Rodríguez, Sueiras, Altamirano-Bustamante, de Hoyos-Bermea and Altamirano-Bustamante. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Adalberto de Hoyos-Bermea, adehoyos@ipn.mx ; Myriam M. Altamirano-Bustamante, myriamab@unam.mx

† These authors have contributed equally to this work

Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.

An official website of the United States government

Official websites use .gov A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

• Publications
• Account settings
• Advanced Search
• Journal List

Quantitative methods for group bibliotherapy research: a pilot study

Emily t troscianko, emily holman, james carney.

• Author information
• Article notes
• Copyright and License information

Email: [email protected]

No competing interests were disclosed.

Accepted 2023 Nov 20; Collection date 2022.

This is an open access article distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Version Changes

Revised. amendments from version 1.

In the revised version of this article, changes were made in response to reviewers’ comments. The most significant of these were: 1) further elaboration on some of our reading-group procedures and clarification of the differences between these and the protocols of Shared Reading; 2) more discussion of bibliotherapy, including the tension between bibliotherapy and group reading practices; 3) additional explanation of some of our choices regarding data analysis; and 4) adjustments to Table 1 and correction of Figure 5. Please see our responses to the reviewers for more details.

Bibliotherapy is under-theorized and under-tested: Its purposes and implementations vary widely, and the idea that ‘reading is good for you’ is often more assumed than demonstrated. One obstacle to developing robust empirical and theoretical foundations for bibliotherapy is the absence of analytical methods capable of providing sensitive yet replicable insights into complex textual material. This pilot study offers a proof-of-concept for new quantitative methods including VAD (valence–arousal–dominance) modelling of emotional variance and doc2vec modelling of linguistic similarity.

VAD and doc2vec modelling were used on conjunction with qualitative coding to analyse transcripts of reading-group discussions plus the literary texts being discussed, from two reading groups each meeting weekly for six weeks (including 9 participants [5 researchers (3 authors, 2 collaborators), 4 others] in Group 1, and 8 participants [2 authors, 6 others] in Group 2).

In-text–discussion similarity was inversely correlated with emotional volatility in the group discussions (arousal: r = -0.25; p = ns; dominance: r = 0.21; p = ns; valence: r = -0.28; p = ns). Enjoyment or otherwise of the texts was less significant than other factors in shaping the significance and potential benefits of participation. (Texts with unpleasant or disturbing content that strongly shaped subsequent discussions of these texts were still able to sponsor ‘healthy’ discussions of this content.)

Conclusions

Our methods and findings offer for the field of bibliotherapy research both new possibilities for hypotheses to test, and viable ways of testing them. In particular, the use of natural language processing methods and word norm data offer valuable complements to intuitive human judgement and self-report when assessing the impact of literary materials. We also share observations on facilitation protocols, interpretative practices, and how our group reading model differs from other trials of group reading for wellbeing.

Keywords: bibliotherapy, evaluation, group reading, narrative, literature, linguistic analysis

Introduction

It is intuitively plausible that reading ‘literature’ might have effects relevant to mental health and wellbeing, but is it also true? If so, what effects, and by what mechanisms do they arise? Is it possible to generalize, given the vast scope for variation in texts, readers, and contexts of reading?

Research on ‘creative bibliotherapy’ has begun to address these questions. Creative bibliotherapy, the reading of literary texts (which may include prose fiction, poetry, and/or drama) for health benefits, is distinct from ‘poetry therapy’, which tends to use poetry rather than narrative or dramatic forms of literature, and which often includes writing as well as reading poetry. Beyond this, however, bibliotherapy is a contentious term, one that encompasses a wide range of practices, contexts, and rationales—from 1-1 encounters in which a ‘bibliotherapist’ analyses a client’s problem and suggests appropriate reading matter to reading groups in which a book is read and discussed during regular meetings. Shared Reading, the most common form of organized reading for wellbeing in the UK and Europe, does not call itself bibliotherapy because any therapeutic effects were initially seen as secondary, with the main goal being to expand access to ‘great writing’: to ‘give immediate access to complex writing that might otherwise be at least daunting and at worst unavailable to a large section of the population’ 1 . Longden et al. 2 propose a framing of Shared Reading as ‘implicit psychotherapy’ in which usefulness remains implicit and potential because the activity remains squarely literary (p. 118), concluding ‘We believe that recovery , restoration or realisation may be more appropriate terms than therapy’ (p. 119). Some studies have, however, made the directly therapeutic potential more explicit, as in Longden and colleagues’ 3 investigation of Shared Reading versus waitlist on quality of life for individuals with dementia. Even here, though, ‘wellbeing’ is the reference point and ‘bibliotherapy’ is not mentioned by name.

As in literary studies more generally, only relatively recently has empirical research begun to inquire into the health/wellbeing-relevant mechanisms and effects to individual and shared reading. This research has typically taken a bottom-up data-driven approach rather than adopting the theoretical constructs relied on in much bibliotherapy theory, which consist largely of concepts adapted from psychoanalysis, such as identification and catharsis. Further, when assessing research conducted so far in this broad area, an interesting divergence arises: The majority of existing bibliotherapy theory concerns individual reading, while most empirical work has involved group reading. The theory is based on minimal empirical evidence, and the empirical work has not yet been used to derive a more evidence-based theoretical account, although it has generated many hypotheses for efficacy and mechanisms of change.

Drawing on existing theoretical and empirical research on bibliotherapy, and using relevant tools from other areas (experimental psychology, natural language processing, cognitive literary studies), this study aimed to contribute to the project of mapping out testable hypotheses of bibliotherapeutic change. We adopted a group-reading methodology to connect the theoretically and empirically driven traditions via investigation of both text-centred and broader social aspects of how reading exerts change. In the rest of this introduction, we consider 1) the purported and observed therapeutically relevant effects of creative bibliotherapy in group settings and 2) the hypothesised mechanisms of therapeutically relevant change (in group and individual settings).

Therapeutically relevant effects

Many wide-ranging claims are made for the therapeutic value of reading, taking in purported benefits to self-understanding, self-expression, and self-esteem; interpersonal and communication skills; and creativity, change, and coping and adaptive functions, amongst others. We acknowledge that we equivocate in our discussion here between ‘therapeutically relevant’ and ‘cognitive and emotional engagement’. However, part of the project of creative bibliotherapy is to purposely challenge any easy distinction between effects of literature that are ‘healing’ and those which foster ‘wellbeing’. It is becoming increasingly the case, for instance, that the literature and practices associated with fostering population-level mental health emphasise wellbeing and self-care as part of health care interventions. Take the ‘Every Mind Matters’ NHS social media campaign. This initiative aims to forestall larger mental health problems by providing audiences with tools that help encourage healthy sleep patterns, avoidance of anxiety triggers, and positive mood. These are not “therapeutically” relevant in the narrow sense but are nevertheless part of the UK government’s focus on improving mental health before it becomes a biomedical issue. Insisting too strongly on a distinction between therapeutic and the salutary effects in reading would, we suggest, close down one of the more progressive innovations in thinking about mental health and wellbeing.

When it comes to putting such claims to the test, the most extensive empirical studies of group bibliotherapy have been carried out by Josie Billington and colleagues in the Centre for Research into Reading, Literature and Society at the University of Liverpool, in collaboration with The Reader. Their interventions typically involve reading a mixture of fiction and poetry, and have included:

Groups at a GP practice, run by a trained facilitator for patients and local residents 4 .

Groups in healthcare settings led by a project worker, for adults with depression 5 .

Groups organized for individuals with or vulnerable to mental illness, isolation, or unemployment who volunteer for The Reader plus other local volunteers, led by the founder of The Reader 2 .

Groups in a range of community and healthcare settings led by English undergraduates 6 .

Groups in prisons led by a trained Reader in Residence 7 , 8 .

Groups led by a project member in healthcare environments, for people with dementia (and sometimes the staff who care for them), using mostly poetry 3 , 9 .

Groups for Mersey Care NHS Trust service users, led by a trained Reader in Residence, and also training NHS staff to grow a lasting reading-group culture 10 .

Billington and colleagues hypothesised that group reading should bring improvement in the areas of social, mental/educational, and emotional/psychological wellbeing, and found qualitative evidence of improvements across all areas, including: enhanced concentration, interest in learning, self-awareness, and capacity for self-expression; increased confidence; reduced sense of isolation 5 . Similarly, the Mersey Care initiative, assessed in a Merseyside Service User Evaluation, documented ‘improvements in confidence, self-esteem, self-expression, memory, concentration, creativity, social engagement, listening skills and overall health and well-being’ 10 . Robinson 4 reported positive effects on mood, loss of self (being ‘taken out of oneself’), concentration, confidence and self-esteem, pride and achievement, and communication skills, as well as appreciation of the opportunity to reflect on experiences in a supportive environment, and of a common purpose and shared ‘journey’.

Where quantitative measures have been used, reduction in dementia symptom severity 9 and improvement on depression markers on the PHQ-9 5 , 11 have been observed, though numbers were small and causality cannot be established because neither study included a control group. Other quantitative measures of change are rare, but in a 12-week crossover design, Longden and colleagues 2 found a substantial effect size for an increase on the ‘purpose in life’ subscale of the Ryff Scale of Psychological Well-being after 6 weeks of Shared Reading versus a social activity focused on the built environment, where no change or a reduction on this scale was found. No significant differences between conditions were found for other scales administered, including positive and negative affect; depression, anxiety, and stress; mastery; and mental well-being, although some effect sizes suggested trends meriting further investigation. A later study by Longden and colleagues 3 found improved quality of life for individuals with mild to moderate dementia with three months of Shared Reading versus waiting control condition, and no change to very low levels of psychopathological symptoms. A systematic review 12 found a small to moderate effect on internalizing, externalizing, and prosocial behaviours amongst children from studies with a range of procedures involving stories, poems, or films, plus various forms of interpretive support. A later review of creative bibliotherapy for post-traumatic stress disorder (PTSD) 13 found no high-quality studies but did yield some suggestions that understanding and communication may be enhanced by group interventions involving reading. Meanwhile, a study using Persian poetry found evidence of quantitative improvement in mood, specifically a reduction in depression and an increase in hope amongst women with breast cancer receiving chemotherapy 14 .

In other existing qualitative work, mood improvement is a common focus of inquiry. Mood was treated as the central dimension of change in a qualitative study using poetry about disability, which draws distinctions between nervous (i.e. emotional) arousal , energetic arousal (action readiness), and hedonistic tone (valence) 15 . Pettersson’s user-focused study using poetry and short stories found six main categories of reading function reported by the four reading group participants who completed subsequent interviews: informational, escapist, social, perspective-creational, aesthetic, and therapeutic 16 . Pettersson points out that the first three of these align with Brewster’s outline of four user-centred models of bibliotherapeutic outcome for mental health problems: informational, escapist, social, and emotional (including empathic and cathartic) 17 . The absence of the emotional function raises the possibility that, for her participants, the emotional is subsumed in the therapeutic. That said, however, the primary observed benefits were interpersonal and pragmatic, including improved self-confidence and ability to perform simple daily activities (including reading, willingness to engage in social activities, and capacity to complete daily chores). The interpersonal strand of these findings aligns with the characterization of the Shared Reading group as an ‘affordance nest’ for socially distributed meaning-making (and tolerance of the absence of ready meanings) in Skjerdingstad and Tangerås’s 18 case study of a single group session.

In sum, then, changes on a wide range of social, cognitive-emotional, and behavioural dimensions are sought and observed in existing literature. Beyond the possibilities that in the absence of controls, randomization, and blinding, researchers are observing what they want to observe and participants are telling researchers what they know they want to hear or what they themselves want to believe, other questions arise. In particular, the breadth of documented effects raises the question of the extent to which they can be attributed to the group meetings or the reading of the text(s). Would a regular group meeting with no literary object of focus, or reading the same texts on one’s own, have similar effects? Is engaging with both text and group contributing something specific? If so, what does each element offer; by what means? Here questions arise regarding mechanisms of change, and there is less evidence to draw on.

Elicitors and mechanisms of change

What ‘active ingredients’ are responsible for bibliotherapeutic change?

Some researchers draw on existing theoretical frameworks like Vygotsky’s model of deep understanding 6 , 19 or reader-response models of creative and participatory reading 5 . Billington and colleagues 5 propose ‘four significant components or “mechanisms of action”’: reading material, facilitator, group dynamics, and physical environment. They elaborate as follows:

‘A rich, varied, non-prescriptive diet of serious literature, including a mix of fiction and poetry (the former fostering “relaxation” and “calm”, the latter encouraging focused concentration). Both literary forms allowed participants at once to discover new, and rediscover old and/or forgotten, modes of thought, feeling and experience.’ (p. 6)

‘The role of the group facilitator in expert choice of literature, in making the literature “live” in the room and become accessible to participants through skilful reading aloud, and in sensitively eliciting and guiding discussion of the literature. The facilitator’s social awareness and communicative skills were critical in creating individual confidence and group trust and in putting the group’s needs above those of the individual where necessary. The facilitator’s alert presence in relation to literature, the individual and the dynamics of the group is a complex and crucial element of the intervention.’ (p. 6)

‘The role of the group in offering support and a sense of community.’ (p. 6) (Evidenced by increased ‘reflective mirroring’ of others’ ‘thought and speech habits’, and increased cooperation and personal confidence.)

‘The environment in contributing to atmosphere, group dynamic and expectation of the utility of the reading group.’ (p. 7)

The researchers described the first three as ‘essential in its success’ and the last as ‘influential’. At present, however, these are not falsifiable hypotheses. The observed effects may be due to all, none, or any subset of these factors. The relative importance of the four factors in different iterations may (or may not) also be highly variable between contexts. The individual mechanisms could be to some extent isolated by adjusting reading-group procedures in order to assess their relative contributions—for example, by selecting ‘unserious’ literature, by democratizing the facilitation (as in the present study), or by altering group size and composition or environmental setting.

In the dementia study cited earlier 9 , brevity and variety of texts, length of meetings (one hour), an informal setting (a lounge), and the presence of a staff member are highlighted as crucial. Robinson 4 stresses the importance of reading aloud (as a means of building confidence and sharing encouragement) and of the expert facilitator’s contributions, including in deciding how long to spend informally chatting before reading, judging when to stop reading for discussion, and helping people start with texts accessible and enjoyable enough to stay motivated for the longer term. Focusing more on textual form and content, Daboui and colleagues 14 suggest that the spiritual aspects of Persian poetry help in increasing hope, and that poetry as a form generally helps with communication about taboo subjects like death.

These observations of factors contributing to efficacy suggest ways of narrowing down the wide range of possible contributing factors, but they do not lead directly to accounts of the mechanisms by which efficacy is achieved. Several attempts have been made to set out a multistage cognitive process to account for observed changes. Gorelick, for example, sets out four phases of reading-stimulated therapeutic change: recognition, examination, juxtaposition, and application to self 20 . Billington, Longden, and Robinson 8 single out ‘memory and continuities’ and ‘mentalisation’ (exercise of Theory of Mind) as mechanisms by which shared reading may have a protective or therapeutic function for problems such as depression, self-harm, and personality disorders. Montgomery and Maunders propose that the mechanisms of creative bibliotherapy might be roughly equivalent to those of cognitive behavioural therapy: They categorize these into ‘cognitive reading processes’ (recognition and reframing) and ‘emotional reading processes’ (empathy, emotional memories, identification) and suggest that parallel forms of identification, challenging, and replacing of negative thoughts occur in bibliotherapy, resulting in ‘new attitudes and belief systems’ 12 .

In their later paper, Glavin and Montgomery 13 propose that the transporting effect of literary reading may permit a form of exposure therapy in which things that would be threatening in the real world can be safely engaged with in the fictional one. This application of the literary-theoretical concept of transportation brings their paper into contact with theories developed beyond the realm of bibliotherapy specifically, by cognitive literary scholars studying the psychological effects of reading in other contexts of psychological difficulty, including bereavement or post-traumatic distress. Kuiken and Sharma 21 , for instance, identify the complex phenomenon of ‘sublime disquietude’, resulting from a mixture of perceived emotional discord, self-perceptual depth, and inexpressible realizations, all of which literary reading can induce. Sikora, Kuiken, and Miall 22 explore the interplay of presence and absence that occurs during literary reading after bereavement in relation to a gradual acceptance of poignant memories of the dead person. Kuiken, Miall, and Sikora 23 investigate how different forms of self-implication affect the emergence of this type of readerly response, especially via the blurring of boundaries between reader and narrator. This in turn relates to broader recent work on the many varieties of ‘personal relevance’ that a text may prompt, with a range of emotional and interpretive consequences 24 .

Theoretical models from the individual reading paradigm tend to follow a common pattern: They emphasise similarity between the reader’s problematic experience and the arc of the protagonist’s story. This similarity is believed to prompt an identification-based connection between reader and protagonist that generates (possibly via a catharsis-like reaction) insight into the nature of a problem, in turn eliciting a problem-solving phase in which the reader learns from the protagonist and makes personal changes 25 – 29 . This model faces both empirical and theoretical obstacles 30 – 32 , not least a paucity of supporting evidence and a failure to pin down how and to what extent ‘similarity’ is therapeutically beneficial. The limit case here would be reading about one’s own experiences—for example, via diary-writing—and although therapeutic writing has a growing evidence base for many conditions and situations (on poetry therapy, see Ramsey-Wade and Devine’s 33 review), it seems unlikely that reading literature written by others should exert its effects via the same mechanisms, merely diluted.

Building on survey data 32 suggesting that for at least one type of illness (eating disorders), heightened reader–protagonist similarity can generate reader perceptions of significantly harmful, rather than helpful, effects, in this study we were open to the possibility that reading might generate uncomfortable, distressing, and even apparently anti-therapeutic experiences for readers. We were also open to the possibility that short-term negative experiences, and individuals’ perceptions of them as unhelpful, might not be the whole story; such difficult experiences may contribute to positive longer-term effects. Longden and colleagues 2 found that Shared Reading increased negative affect and suggested that rather than being a drawback of the procedure, the effect is ‘consistent with some of its intrinsic value (ie, literature’s power to open individuals up to a range of emotional states)’ (p. 118). This type of hypothesis is compatible with both catharsis and exposure-therapy hypotheses of bibliotherapeutic efficacy, and with a more general view that the value of the experience of literary reading derives from complexity and multidimensionality rather than a simple feel-good effect.

Other perspectives are provided in research showing that readers who score high on a ‘search for meaning’ scale—a metric correlating with propensity for depression—are more receptive to literary (over non-literary) versions of a text 34 . Similarly, several papers give theoretical and empirical grounds for thinking that fiction can reduce anxiety by offering predictive schemes for thinking about social interactions 35 – 37 . Carney 38 also explores the role of predictability in culture more generally, suggesting how entropy (a measure of unpredictability) might differentially impact on anxiety and depression.

The primary objectives of this study were as follows.

To record and transcribe a full set of reading-group discussions to generate detailed data on group–text interactions.

To trial new computational methods for sensitive analysis of the resulting complex textual material.

To generate new hypotheses on potential therapeutic efficacy and its mechanisms to guide future research in group bibliotherapy.

Ethical approval

Ethical approval for this study was provided by the University of Oxford Social Sciences and Humanities Interdivisional Research Ethics Committee, ref. MS-IDREC-C1-2015-155. All researchers involved in this project completed the NIH online training course ‘Protecting Human Research Participants’ and familiarized themselves with the BPS code of conduct and the University’s data protection and academic integrity guidelines.

Reading group procedures and participants

Our study took the form of two closed ‘Books, Minds, and Bodies’ reading groups in two consecutive university terms (October to December 2015 and January to March 2016), the first group meeting for seven weekly sessions, the second group for eight weekly sessions, as required to complete the selected texts. Two groups were run in order to generate a wider range of text–participant interactions than would have arisen in a single iteration. All 3 authors participated in the first group; EH and ET participated in the second group. The authors took responsibility for welcoming participants, covering housekeeping announcements, and establishing basic guidelines for conduct at the start of the first session, and thereafter participated in the reading and discussion in the same way as all other participants.

Other participants were recruited via an advert posted on noticeboards and local online events listings, offering a chance to ‘explore connections between reading and mental health and wellbeing' with a group of professional researchers interested in these topics. Group 1 included 9 participants (3 authors, 2 colleagues, 4 others); Group 2 included 8 participants (2 authors, 6 others, reducing to 5 others after session 3 after one participant dropped out due to other commitments). Group 1 included 6 females; Group 2 included 7 females, and the male participant was the one who dropped out after session 3, leaving an all-female group. Group 1 consisted solely of students and researchers (ranging in career stage from undergraduate to postdoc), while the four non-author participants in Group 1 were neither students nor researchers/academics. Other demographic information (e.g. age, education, occupation) was not gathered. We sought a minimum of 4 and a maximum of 6 recruited participants per group, to allow for a range of perspectives without compromising the intimacy and trust that can more easily be created in a smaller group. We required participants to be aged 18 years or over, to have an interest in narrative and/or mental health, and to be available for weekly two-hour sessions during the term. Other than age, the only exclusion criterion was that participants be able to confirm that ‘I do not currently suffer from a mental health disorder’. This criterion was included to mitigate the risk of harmful effects being generated by participation. ET interviewed participants beforehand to increase the probability of sustained and positive commitment by providing information to prospective participants about the role they would play in the group. ET provided the information sheet before the meeting, and asking the following questions in person:

What drew you to this project?

Are you able to commit to weekly sessions until [date]?

What are your expectations for the project?

Have you ever been part of a fiction reading group?

Are you comfortable with a rotating facilitator role?

After the roughly 20-minute meeting, potential participants were invited to take their time to decide whether they wished to take part, and after confirming their interest by email were sent further information on text selection (see below). Paper consent forms were provided for participants to check and sign at the first meeting of each group, with the opportunity to ask further questions. All but one of the interviewed candidates participated; the exception had to pull out due to scheduling difficulties.

Group 1 (henceforth MT , for ‘Michaelmas Term’, Oxford University’s autumn term) met at 6 pm on Mondays; Group 2 (henceforth HT , for ‘Hilary Term’, the spring term) met at 2:50 pm on Fridays. Meetings took place in one of two similar meeting rooms at Balliol College, Oxford. Texts were selected democratically: We circulated a list of five possible books suitable in length for a term’s meetings, providing either a short (one-page) excerpt or a link to an Amazon.co.uk preview. Text length was the primary factor in shortlist selection; beyond this, we drew on our collective knowledge about texts likely to reward the close attention involved in being read aloud and discussed at length. Each participant ranked the full selection in order of preference and vetoed any book they had read before. Fyodor Dostoevsky’s Notes from the Underground was selected for MT (in the Oxford World Classics translation by Jane Kentish) and Ted Chiang’s Story of Your Life and Other Stories for HT. We chose not to read the same text in both groups to ensure that all participants shared in the experience of discovering a new text together and also to capture as much variance as efficiently as possible (rather than attempting to control for variance as will be appropriate in a follow-up study).

We also note in passing that issues of ‘literariness’ and ‘quality’ emerge as an item of concern here, given that creative bibliotherapy centres on the idea that literary works are the drivers of therapeutic and cognitive change. On this question, we had to strike a balance between identifying texts that no one had read and honouring individual preferences. The two resulting texts are works of recognised quality, in that one is by a major European novelist (Dostoevsky) and one is by a leading contemporary practitioner of science fiction (Chiang). We suggest that these choices withstand critique on grounds of literary achievement, and that the distinction is, in any event, an unhelpful one: What counts as literature has always been contested, and what counts as a diversion in one historical period can become the literature of a subsequent one.

Participants were presented with their own copy of the selected book, provided with a pencil, and encouraged to make markings in the book if they wished. Copies were collected between sessions to ensure participants did not read ahead on their own; these were given to participants to keep at the end of the term. In the final MT meeting, having finished reading the main text, we asked each participant to bring a poem of their choice to read and introduce to the group. In the final two HT meetings, having read four Chiang stories, we read two short stories by Franz Kafka in English translation: ‘A Hunger Artist’ and ‘Jackals and Arabs’. These additional sessions are excluded from the linguistic analysis of the literary texts and the transcripts (six for each term) covered in this article.

During the sessions, we opted to invite all participants to contribute equally from the outset to reading the text aloud. In some Shared Reading studies, the authors observe that participants grow more confident and co-operative over time: Billington and colleagues (2010) 5 , for example, note that over the 12 months for which the group met, participants increasingly ‘took the initiative in supporting one another’s comments, in guiding the direction for discussion and in offering to read aloud from the text themselves’ (p. 7; see also Robinson, 2008, pp. 6–7 4 ). Here we invited such co-operative contributions from the outset. Books were read aloud by participants, switching with each paragraph (or, where these ran for longer than a page, with each new page). Sessions lasted 2 hours, the first hour spent reading, followed by an hour for discussion with refreshments (wine / soft drinks and nibbles in MT, tea and biscuits in HT).

Audio recording began at the start of the second hour, and no notes were taken in addition to the recording. In another contrast with the typical protocols for Shared Reading, we chose to adopt a shared facilitation model in which each session was facilitated by a different participant, usually self-nominated. This decision was taken for two reasons: 1) to allow the groups to be conducted and participated in by ourselves as a research team with no training in formal facilitation in Shared Reading or any other model; 2) to reduce any perceived hierarchy between the researcher participants and others (whether academics themselves or otherwise).

Facilitation was described to participants in the facilitator information sheet as follows:

The facilitator role, which will rotate at each session between participants, is broadly to keep the conversation going. This involves making sure that all participants feel able to contribute and perhaps occasionally asking someone who has been speaking for a long time to open the conversation to others. No more than one person should be speaking at once—and no private comments or conversations. This is a group discussion.

We have put together a list of possible questions that might be useful when you are in this role.

The questions included 5–7 questions in each of the following 4 categories:

Emotional response (e.g. ‘Do you care what happens to the main character(s)?’)

Interpretive response (e.g. ‘Did you ever feel that what was being described wasn’t what the passage was really “about”?’)

Mental imagery (e.g. ‘Do you find you’re imagining more or less or differently than you do when you read alone?’)

Drawing connections between real life and the narrative (e.g. ‘Do you think anything in this passage could help you think through or deal with difficulties in your life?’)

The questions were designed to cover a wide range of potential types of response to the texts, going beyond interpretive, meaning-focused responses to encompass emotional and sensory dimensions, as well as opening up possibilities for connections with real-world experience that might be relevant to health and wellbeing. In practice, the questions were barely used, since conversational prompts from facilitators were rarely needed, and facilitators in general preferred to generate them ad lib when required.

Collectively, the decisions to democratize the choice of text to be read, the reading-aloud of the text, and the group facilitation meant that our procedures deviated significantly from the Shared Reading tradition, in which a trained ‘reader leader’ selects the texts, reads it (though contributions from others may be invited), and facilitates the discussion of it. Together they maximized the self-contained simplicity of the group-reading setup: there was no prior training based on complex pre-existing theory or tradition. They also maximized the democratic nature of the experience, given that all participants played comparable roles in the selection, the reading, and the guiding of discussion.

A final difference from the Shared Reading protocol was that the reading of the text proceeded uninterrupted (aside from brief clarifications of vocabulary or similar) for a full hour. This format allowed us to record the entirety of the discussions without needing to record the reading as well; this was useful given that some participants expressed a degree of nervousness or inhibition about reading aloud to begin with. The clear demarcation also allowed the direct and indirect experiences of textual engagement to unfold with their own distinct dynamics, rather than the reading being repeatedly interrupted by discursive reflection. This allowed for greater analytical clarity. Overall, our priority was to create a simple set of group-reading procedures in which to trial quantitative methods for assessment of the group’s practices and effects, rather than to either replicate or challenge Shared Reading or any other model.

After both groups had concluded, EH and ET transcribed the audio recordings of the discussions, EH taking MT and ET transcribing HT. One HT session was transcribed by a group participant, who was paid for her contribution; her transcript was checked and edited. Participants’ contributions were pseudonymized using colour codes to introduce their discussion interventions, and the codes were stored separately from the transcripts in a password-protected file.

While aware of the potential for researcher bias in a participatory framework of this kind, we considered that it could not be directly minimized at the group participation stage given the naturalistic setting, but that variations in individual perspectives would be manifest by all participants (some of whom were, in MT, also academics in other fields, i.e. there was no simple dissociation between ‘researchers’ and ‘participants’). In contrast to many prior studies, all forms of linguistically manifested bias were made explicit in the discussion recordings and transcripts, and were thus treated as an integral part of the complex dynamics under investigation. At the stage of analysis as opposed to participation, meanwhile, our methods were expressly designed to reduce the problematic forms of bias evidenced in qualitative content analysis, as we go on to document below.

Participant feedback . After the final reading-group session, participants completed an online survey designed to complement our analysis of their discussion contributions with direct self-report on the experience of taking part. The survey included questions about the enjoyment and significance of different elements of the reading-group experience, and forms of learning and change participants perceived to have resulted from taking part. To test for differences between the two groups, feedback from participants was analysed using an independent samples t-test that compared means between the two groups.

Comparison with other studies’ procedures

The procedures used in our study both resemble and differ from other studies along several dimensions. Table 1 summarises the comparisons.

Table 1. Comparison of procedures across cognate studies.

Analytical methods.

All analysis as outlined below was conducted on the full dataset resulting from 17 total individual participation instances (this included 2 repeat participations by researchers EH and ET) minus 1 participant who dropped out for non-study-related reasons after session 3 of HT.

Rationale . The guiding principle of our data analysis was that the discussion transcripts, including their linguistic relationship to the texts under discussion, stand at the centre. This analytical principle derives from the hypothesis that if shared literary reading is having wellbeing-relevant effects, they will be visible in the linguistic patterns of the text-prompted discussion. We acknowledge that the inclusion of control and experimental groups would represent the optimal way to evaluate the effects of interest here. However, our aim was not to engage in formal hypothesis testing, but to gain pilot data from which hypotheses could constructed. With respect to existing research in this area, some previous studies have recorded and transcribed discussions 2 , 4 , but transcripts have been relatively underused as sources of analytical insight. Our methods focused on two dimensions of participant response: emotional reaction and cognitive elaboration. Reading is a complex amalgam of emotional and cognitive responsiveness, and any thorough appraisal of its processes and impacts needs to attend to both 42 . As our ambitions included quantifying the reactions of our reader groups, this meant finding ways to measure the cognitive and emotional variation implicit in our data. We resolved this problem by making use of word norm data and unsupervised machine learning in the following ways.

Emotional response . Traditionally, an impediment to supplementing qualitative assessments of the impact of reading has been the difficulty of measuring subjective responses. This is especially so with respect to emotional response, given that there is no universally attested taxonomy of emotion. When this emotional response is expressed verbally, the facility of language for expressing the same emotions in different ways compounds the problem. It follows that quantitative analysis of our transcript data has to try to solve the problem of extracting nuanced emotional information from text.

Our response to this challenge was to make use of word norm data. Essentially, word norms are corpora of language that have been rated by human participants along a specific dimension or set of dimensions 43 – 45 . Warriner and colleagues 44 present 13,914 common English lemmas (words stripped of morphological variation) that have been rated by human participants for valence , arousal , and dominance . According to dimensional models of emotion, each discrete emotion can be represented in terms of an underlying set of finite components 46 . The VAD model of affect identifies valence, arousal, and dominance as the underlying factors responsible for emotional variation 47 , 48 . That is, each discrete emotion can be thought of as a specific combination of valence (how pleasant or unpleasant it is), arousal (how stimulating or sedating it is), and dominance (how in-control or controlled it makes someone feel). Thus, anger is a low-valence, high-arousal, low-dominance emotion, while contentment is high-valence, low-arousal, and high-dominance. These extensive word-norm data provide an empirically validated way of assessing the overall emotional impact of a word, making mean VAD easy to calculate. Though valence is often understood in polar terms (positive or negative), we followed Warriner and colleagues’ (2013) 44 approach of treating it as a scalar metric (0 = low valence, 1 = high valence) to allow for comparison with arousal and dominance, which are not polar in nature. We could have performed a median split to capture polarity, but we felt it best to retain the original formulation in Warriner et al. ’s data for the same reasons as they offer. The great value of the VAD norms is that they provide a low-dimensional proxy for emotional variation that is not restricted to words that are ostensibly emotional in their character (mood words like ‘happy’, ‘sad’, ‘angry’, etc.). They therefore provide a versatile means of establishing emotional variation on the basis of word use in linguistic documents. To this extent, they have an obvious value when it comes to capturing how our participants responded to texts and to each other on a per-session basis. Necessarily, they are also limited: emotion is conveyed not just by lexical choice, but also by phrasing, tone, and body language; equally, as averages of rater responses, VAD word norms are not sensitive to homonymy and other subtleties of usage. Moreover, competing dimensional models exists, with different dimensions as well different numbers of dimensions 46 . No doubt, these other dimensional models of emotion could have been used if word-norm data existed on a similar scale to VAD, and this could have affected our results. However, we are confident that that the alignment of the VAD dimensions with physiologically and socially fundamental features of human emotional cognition means that any differences would be in detail and terminology, rather than in our broad conclusions. In any event, certainty on this point could only be secured only by re-running the analysis using word-norm data that do not presently exist.

The emotion of each text (both the discussion transcripts and the literary texts under discussion) was calculated by taking the mean of valence, arousal, and dominance across all the words of that text. Although this has the advantage of computational simplicity, within-text random variation means that the longer the text, the more they will tend towards the background mean for these values for English. As our texts were relatively short and of roughly equal length, we felt that any effects of reversion to the mean would be small and spread equally across all texts. This analysis was performed by JC using bespoke scripts written in python 3; the script ascertained the value for each word on the transcript for each session for valence, arousal, and dominance using the Warriner 44 data and returned an average for the entire session. With respect to taking the mean of responses as our variable of interest, we acknowledge that a case could also be made for using the variance: Literary scholarship is, after all, at least as interested in the role of texts in stimulating variation in responses as in coordinating them. However, as our study was guided by the assumption that there would be coordinating effects of reading texts in groups, any subsequent statistical testing would reveal if there wasn’t by not allowing us to disqualify the null hypothesis that texts have no coordinating effects on responses. As the use of variances would not have been of value in generating testable hypotheses consistent with our aims, we therefore did not use them as point estimates of our data.

Cognitive elaboration . Perhaps a greater challenge than measuring emotional response is measuring cognitive response. Allowing that the difference between ‘emotional’ and ‘cognitive’ is somewhat artificial, the fact remains that the space of possible concepts is wider than the set of possible emotions—it is, in fact, at least the size of a language’s vocabulary. Given that concepts can also be recursively combined to produce new concepts, this means that the set of possible concepts is combinatorically large. Until recently, this has meant that the task of extracting topics and themes from linguistic documents has been the preserve of the best pattern-matcher we know: the human brain. However, advances in unsupervised machine learning mean that it is now possible to automatically identify the specific ways in which words are combined with others to produce recurring items of content. In particular, word-embedding algorithms like word2vec, doc2vec, GloVe, and BERT provide empirically robust methods for capturing semantic variation at the level of word, document, and context that can be deployed at scale 49 – 53 . Like all machine-learning methods, these algorithms are sensitive to initial parameter selection, so there is no sense in which they provide objective measures of semantic variation. Nevertheless, they inject a useful amount of statistical rigour into a practice that is often a hostage to ideological agendas.

For our purposes, the algorithm of most value was doc2vec, a document-level analogue to word2vec. Where word2vec represents the behaviour of a word across a corpus of documents by training a shallow neural network to predict its association patterns, doc2vec is trained by associating document tags with the word vectors comprising the document. (These word vectors are mathematical descriptions of how a word behaves in a corpus; the tags are the document name used to group the word vectors associated with a specific document.) Doc2vec thus represents high-level semantic variation across documents in a precise way, thereby making the discrete documents of a corpus comparable. In our case, the relevant corpora consisted of the transcripts of each session and the relevant portions of each text read in that session. The specific implementation of the doc2vec algorithm we used was that associated with the Gensim natural language processing library for python 54 . We note here that the doc2vec algorithm operates by representing a document as a mathematical vector and using this representation to (amongst other things) compute document similarity; it does not capture specific forms of textual identification. If readers, for example, identify with characters in the discussion and talk about it, then this may impact on the doc2vec metrics. However, there is no way of establishing this, and it was not our intention to use doc2vec to do this.

When using the doc2vec algorithm, the key parameters involve 1) choosing the size of the moving window of words amongst which semantic relations are assumed to obtain and 2) specifying the number of training epochs used by the algorithm. The first parameter is essentially a discourse specifier: In genres like poetry, this window should be large, given the dense interdependence of semantic elements; in technical writing it should be small, in view of the emphasis on strict denotation. In our case, we resolved on using the average sentence length per group ( Figure 1 ), given that sentence lengths in discursive conversation can often be quite long. The number of training epochs was 30k, which trial and error show is the point at which results stabilized.

Average sentence length per group for discussions in a ) MT and b ) HT. MT=Michaelmas Term; HT=Hilary Term.

Data cleaning . Both emotional and cognitive measures required texts to be cleaned in a way that made them amenable to automated analysis. In practice, this meant tokenizing the text into words and phrases and eliminating redundant variation across these words and phrases. Tokens were extracted by splitting character strings on whitespace; these were regularized using parsers from the spaCy natural language processing library for python 55 . In practice, this involved lemmatizing each token, removing case variation, eliminating punctuation markers, and dropping stopwords (‘it’, ‘a’, ‘an’, ‘on’, ‘the’, etc.) that conveyed no semantic information. This reduced each text to a list of words that captured its semantic content in a consistent way.

Qualitative coding . EH and ET read the full set of transcripts and conducted a manual coding process for relevant features. Given our interest in the potential of group reading to have impacts on life beyond reading itself, and given existing theoretical work founded on notions of ‘similarity’, we adopted the categories of ‘personal relevance’ outlined by Kuzmičová and Balint 24 as our starting point. These included: personal relevance , perceived similarity / simile-like identification , wishful identification , and metaphor-like identification . Other categories identified in the first 2 transcripts from each term as necessary to the coding process included: expression of emotional engagement , expression of no emotional engagement , engagement with character , liking (of text), dislike (of text), self-qualification , and human condition . The resulting 11 categories were applied to coding the remaining transcripts.

Emotion by session

VAD analysis of the transcripts showed clear differences in emotional profile both between individual sessions and between the two groups ( Figure 2 ) 56 . On the whole, HT exhibited higher levels of valence and arousal and lower levels of dominance across sessions, though these differences were not statistically significant ( V : t = -1.52, p = .16; A : t = -1.3, p = .21; D : t = -1.38, p = .19). This may point to greater emotional febrility in HT, but it may also be an artefact of discussing very different texts. Although the lack of significance indicates that the differences may be the result of random chance, linguistic data exhibit high variation, meaning that significance is unlikely to be found in a sample size this small whether an effect is present or not. We therefore proceeded to analyse the emotional profile of the texts under discussion.

Figure 2. Change in levels of arousal, dominance, and valence by group and session.

MT=Michaelmas Term; HT=Hilary Term. The numbers on the y axis are scaled between 0 (min.) and 1 (max.).

We found that text–discussion similarity was inversely correlated with emotional volatility in the group discussions (arousal: r = -0.25; p = ns; dominance: r = 0.21; p = ns; valence: r = -0.28; p = ns). In practice this means the greater the level of similarity between texts and discussions as measured by the doc2vec algorithm, the lower the arousal in the discussion of the text. That is, people used less energizing or stimulating language––a quality associated with emotionally volatile language. This possibly arises from there being less disagreement amongst individuals concerning their reception of the text. (See additional data in the online repository: text_sim.)

It should be noted that MT-S6 had no text as such; instead, participants were invited to reflect on their experiences of participation and their relevance for further investigation of bibliotherapy. This meta-reflective structure presumably accounts for the outlier low arousal value for this session relative to the others. In Table 2 we offer a brief indication of some possible reasons for the outlier status of the HT-S6 VAD values.

Table 2. HT, Session 6 versus other HT sessions.

Emotion by literary text.

Since one possible driver of emotional responsiveness in participants is the emotional profile of the texts they are discussing, we took VAD measures of the discussed texts. Considered in aggregate, these measures positioned Chiang as on average lower in arousal and higher in valence and dominance than Dostoevsky ( Figure 3 ). These differences were not statistically significant; nor did we expect them to be, given likely reversion to the mean for both authors over longer stretches of text. Within each author, there were relatively small differences between specific sections or stories ( Figure 4 ).

Average a ) valence, b ) arousal, and c ) dominance values for Ted Chiang’s Story of Your Life and Other Stories and Fyodor Dostoevsky’s Notes from the Underground .

VAD (valence, arousal, dominance) values in a ) Dostoevsky and b ) Chiang by session selection.

Emotion by word norm data

An important element of establishing emotional variation involves identifying the actual words used. We did this by concatenating all transcripts for each group so as to create two corpora. After establishing VAD values for each word, we performed a decile split for each of valence, arousal, and dominance so as to split the data into ranked tiers. By comparing the top and the bottom deciles, this allowed us to gain insight into the kinds of word driving the emotional profile of each group (see an example in Figure 5 ).

Figure 5. Sample word clouds for valence in MT and HT.

The top row shows the words most responsible for negative valence in each term, the bottom row those most responsible for positive valence. Because most words have some positive valence (the scale goes from 0 to 1), words common to both lists were excluded. MT=Michaelmas Term; HT=Hilary Term.

Cognitive elaboration

Our expectation was that measuring doc2vec document similarity between transcripts and texts for each group would show up a pair-wise relationship, such that the transcript of a session would be semantically closest to the text read in that session. Surprisingly, this was not the case: there was no obvious pattern linking a text to a session, and the method was, in any event, highly sensitive to parameter selection. What did emerge were striking between-group differences with respect to whether the group reproduced literary content in general, relative to non-literary content (i.e. the degree to which transcripts resembled the texts being read or resembled other transcripts) ( Figure 6 ). As can be seen, the MT discussions reproduced far more of the semantic content of the Dostoevsky selections considered in aggregate than the HT sessions with the Chiang stories. In other words, the MT group was more ‘on topic’ than the HT group, if being ‘on topic’ is taken to mean discussing the texts. An additional finding was that in sessions where the doc2vec similarity was low, VAD ratings varied more, suggesting that these sessions were more emotionally febrile.

Figure 6. Semantic similarity between transcripts and texts in MT and HT.

Note that a value of 0 means purely random association and a value of 1 means identity. Thus, lighter hue means more similar semantic content. Here, this is visible in the bottom half of the left-hand figure, which shows that transcripts were more similar to Dostoevsky selections than to each other. Note that, unlike in MT, HT transcripts in aggregate semantically resemble each other more than they resemble the Chiang texts. MT=Michaelmas Term; HT=Hilary Term.

Participant feedback

In both groups, most responses to the post-participation questionnaire followed a similar pattern ( Figure 7 ). In assessing the significance of different elements of their reading and reflection during the sessions, the participants rated highest their emotional responses to the text and discussion, plus the perceived relevance of the texts to their lives. Engagement with the language of the text was rated lowest; engagement with textual meaning was rated of intermediate significance. Participants rated enjoyment of all elements of participation (the group, the additional short texts, reading aloud, listening to others read) significantly higher than enjoyment of the main text. Learning (about the text, about reading, and about oneself) was rated relatively low, and change (in reading habits, self-esteem, interactions with others, and in general) was rated as low (with general change higher than more specific types).

Figure 7. Differences in post-participation feedback, by group and theme.

Intergroup analysis of participant feedback revealed three statistically significant differences amongst the 32 questions posed ( Figure 8 ):

Figure 8. Significant intergroup differences in post-participation feedback: enjoyment of the principal texts, enjoyment of the short texts, and enjoyment of hearing others read aloud.

Scales range from 0 to 5, where a higher score indicates greater agreement with the statement. MT=Michaelmas Term; HT=Hilary Term.

How much did you enjoy Notes from the Underground / Story of Your Life and Others ? ( p = .02)

How much did you enjoy the short texts in the final session? ( p = .03)

Did you enjoy listening to other people read sections of the text aloud to the group? ( p = .04)

Qualitative coding

For the full coding results, please see the online repository (qualitative_coding_results). Five categories (expression of emotional engagement with the text; self-qualification [no. of instances]; self-qualification [no. of words]; and liking and dislike of the text) were considered relevant and coded by both EH and ET; 3 categories (engagement with a character, expression of complexity or difficulty, and personal revelation) were coded by EH only; and 6 (explicit absence of emotional engagement, personal relevance, perceived similarity, metaphor-like identification, wishful identification, and human condition) by ET only. Differences were manifest between our coding outputs for the 5 common categories; we felt that these were in theory resolvable, but not relevant to the purposes of the present analysis. ET’s use of the categories derived from Kuzmičová and Balint 24 resulted in near-zero outputs for ‘metaphor-like identification’ and ‘wishful identification’ (1 instance across the 2 categories in all 12 transcripts), low levels of perceived similarity (21 total instances), and medium levels of personal relevance (68 instances). ET’s self-generated category ‘human condition’ (expressions of commonalities in human experience) yielded a far higher total (120 instances), with significantly more in MT than HT (82 v 38).

Liking and disliking the text

We found that enjoyment and dislike of the texts read did not impact on how valuable participants felt the sessions to be. Statistical analysis of post-participation feedback indicated that reported enjoyment of the main texts being read was low, and significantly lower for HT than MT. Even when texts were explicitly disliked, participants enjoyed the discussions they prompted, and all other aspects of their participation. One HT participant remarked in response to the question ‘How much did you enjoy the stories by Ted Chiang?’, ‘ It did not matter. The stories, even when not enjoyable, still triggered discussions, interactions and people expressed their opinions .’ Other aspects of enjoyment may play an important role: for instance, reported enjoyment levels for listening to the texts being read aloud by others were higher overall, and significantly higher for MT than for HT, supporting the relative insignificance of enjoyment of the text itself in overall experience and effects of participation.

Liking and disliking the discussion

Many discussions of emotional register focus on whether the register is positive or negative—that is, on valence. We found that valence alone was insufficient to capture the range of emotional variation in how participants reacted to the sessions. Valence levels were higher in HT discussions. However, HT also manifested higher levels of arousal (associated with stress) and generated reports of finding the space unsafe. Conversely, MT participants evinced lower valence and arousal but higher dominance. In both terms, however, participants reported low levels of change in self-esteem or social interaction.

In this study, we designed a novel variant on group reading for wellbeing, differentiated from commonly used Shared Reading protocols in that all participants contributed equally to choosing which text to read, to reading the text aloud, and to facilitating the discussion. These design choices aimed to democratize the group reading experience and to provide a test intervention not associated from the outset with a complex set of assumptions and training protocols as in the Shared Reading methodology. In these senses, the procedures capitalize on the group setting as an opportunity for exertion of personal agency in a social context—aligning with Skjerdingstad and Tangerås’s 18 emphasis on the group’s potential to drive distinctive forms of distributed cognition. These choices increase the contrast with the directed nature of the individual bibliotherapy model, where the bibliotherapist selects the text, the patient/client reads the text alone, and the discussion is guided by the therapist.

Our analysis used new quantitative methods in the attempt to provide a combination of richness and replicability needed to answer questions about human responses to complex aesthetic phenomena. Using VAD (valence–arousal–dominance) modelling of emotional variance and doc2vec modelling of linguistic similarity to analyse the discussion transcripts and texts under discussion from two reading groups, we found an inverse correlation between text–discussion similarity and emotional volatility in the group discussions. Specifically, doc2vec analysis demonstrated that in verbal similarity taken as a whole, MT manifested significantly higher levels than HT, but that high arousal plus low dominance were also present. We also found no link between discussion valence and therapeutically relevant outcomes: The higher-valence group discussion (in HT) involved higher arousal and lower dominance, and neither group reported appreciable change in self-esteem or social interaction attitudes/habits. This suggests that higher valence does not necessarily translate into outcomes that reflect autonomy and self-directed action—traits that are often absent in mental health conditions.

Post-participation feedback also suggested that enjoyment or otherwise of both the texts and the discussion was less significant than other factors in shaping how participants perceived the significance and potential benefits of their participation. This suggests that any therapeutic use of fiction need not be restricted to straightforwardly enjoyable or accessible texts. Given that all participants contributed to text selection, and that this selection was presumably guided at least in part by anticipated liking or enjoyment (though perhaps also by emotional responses like identification), the low levels of enjoyment are striking. Anecdotally, it seemed that the opening pages of the chosen texts were considerably more straightforwardly enjoyable than later material. However, other potential selection aids (book blurbs, excerpts from later in the text) would come with their own drawbacks. The finding that enjoyment of the text is not paramount aligns with Shared Reading’s common emphasis on elements of reading that are not dependent on or reduced to enjoyment. Skjerdingstad and Tangerås 18 , for instance, suggest that the group dynamic may make enjoyment less relevant than it is in individual reading by allowing for intrinsically shared emotional-interpretive experiences such as being moved by another group member’s emotional reactions or personal disclosures in response to the text. It contrasts, however, with other suggestions that a minimum level of enjoyment is important to achieve: ‘However it was important that the groups enjoy what they are reading, or at least want to continue with the book’ (p. 9) 4 .

Our findings are limited by the absence of a direct control group in which either the same participants read a different book or new participants read the same book. We decided against this option in order to allow everyone to experience a new text together and to maximize rather than control for variance. Any follow-up research should involve a control condition to establish causal links between textual features and discussion variables. Further limitations emerge with respect to our not having captured factors of interpersonal variation to do with reading history, education levels, or personality type. As these data may have mediated the impact of the texts read, including them in future research may provide more material for hypothesis generation and testing.

Analytically, our ability to draw robust conclusions from the linguistic data was limited by the absence of both a strong signal and a large sample size (of discussion material plus texts under discussion). This limitation could be addressed by rolling out such group meetings on a larger scale and pooling the textual materials, although automated transcription tools would in this case ideally be trialled to reduce the time investment of manual transcription. Other expansions such as testing effects of texts in non-narrative genres, or involving participants with specific demographic characteristics or (with appropriate safeguards) with current mental health conditions, could serve to evaluate the generalizability of the current findings.

Our results are based on a small sample of individuals in an idiosyncratic study setup. As such, caution should be used when generalizing to larger samples of readers. However, almost all real-world reading occurs in idiosyncratic ways, and by running the two groups we sought to control for some of the relevant variation between individuals. Moreover, it is hard to see how any experimental setup concerning group reading can capture the large number of factors that attend group reading. For these reasons, we suggest that some aspects of our results may generalize, but further research will be needed to establish which.

In the remainder of this section, we offer some starting points for broader interpretation of our findings with respect to the types of text–discussion interactions that may be therapeutically positive, as well as timescales of potential change. We begin by linking our work with other researchers’ observations on group facilitation procedures.

Reading group facilitation

The role of the reading group facilitator has been the subject of frequent discussion in existing group bibliotherapy research, and the standard approach is to use a trained ‘reader leader’ rather than to share facilitator responsibilities amongst participants. The experiences of ET and EH in this study suggest there may be benefits to rotating facilitation to involve all participants actively in steering the group dynamics; the key, however, is that facilitation occur, and that it be active. Active facilitation is important in any scenario involving structured discussion of literature, for several reasons. In the first instance, it keeps discussion focused on the text. This aligns with Billington’s reflections on the facilitator’s role, which emphasise the specifically literary guidance facilitators give: one ‘essential’ component for ‘success’ is ‘The role of the group facilitator in expert choice of literature, in making the literature “live” in the room and become accessible to participants through skilful reading aloud, and in sensitively eliciting and guiding discussion of the literature’ (p. 6) 5 . In the second, the facilitator role regulates conversational dynamics between participants. This supports Robinson’s observation that facilitator duties include ‘bring[ing] people in as much as possible into the general discussion’, being ‘willing to give everyone space to talk and read and reflect, and being ‘able to make them feel that their contributions to any discussions was [sic] valuable and interesting to others in the group’ (p. 5) 4 . Finally, there is a value in having an individual present who can respond to unanticipated or problematic issues emerging over the course of the discussions, and minimize any potential damage—always a possibility when reading complex texts that elicit a wide range of experiences. All of this becomes more important when the group contains researcher–participants, because they may be inclined to take a more passive role in in their capacity as observers. ET’s personal reports after HT sessions, for instance, highlight the difficulties of participating, facilitating, and researching all at once: ‘ I tried to “facilitate” and thereby prevented myself participating. ’ The researcher perspective is encapsulated in ET’s habit of ‘ reminding myself that this is all part of the experiment, and that it doesn’t matter how it goes; it matters that we learn something from how it goes. ’ Sometimes more active intervention is needed, whether by a pre-trained facilitator or from a group member who assumes this role with lighter-touch guidance—perhaps emphasising the importance of encouraging proximity to the text. The choice of facilitation protocols will depend on who is taking part and for what reasons.

Open and closed interpretation

What types of text encourage constructive interpretive and discursive patterns? Our experience was that texts that were interpretively ‘open’ did a better job of sustaining discussion and promoting a sense of shared purpose than texts that were interpretively ‘closed’. The former are texts that allow readers freedom to speculate as to their meaning without having an obviously ‘right’ answer; the latter are more like puzzles that can be solved in a singular, unequivocal way. The distinction may be aligned with Barthes’ 57 distinction between lisible (readerly, or literally readable) and scriptible (writerly, or literally writable) texts: the former directing interpretation down well-worn paths, the latter drawing attention to themselves and inviting interpretive elaboration. Our result converges with the less specific suggestion made by Billington and colleagues 5 that one of the four mechanisms of bibliotherapeutic action is that the literature being read (a mixture of fiction and poetry) be ‘serious’ and ‘rich’, although it may contradict the suggestion that the fiction ought to foster ‘relaxation’ and ‘calm’. Responses to open (or closed) complexity may be beneficial without being relaxing or calming.

Another discovery was that interpretive openness in a text facilitated deeper social interactions between participants than are typically had between strangers. Participants reported that hearing interpretive contributions from others generated curiosity and that later this was rewarded by discovering the personal origins of the contribution:

they’d come up and have an opinion on some part of the chapter that we read, and then I’d think ooh why did they think that, you know, that’s really mysterious [laughter] And then over the next couple of sessions I’d learn more about them, and that was really interesting. (MT-S6)

However, this positive effect is lost when discussion strays too far from the text. We should also note that change in broader social interactions was not widely reported in the post-participation feedback, perhaps because of the nature or duration of this intervention relative to the patterns of participants’ everyday interactions.

Text–discussion proximity

Direct correspondences between text and discussion seem not to be manifested in the emotional sphere in either group. However, our direct probing of text–discussion similarity via the doc2vec analysis demonstrated a clear difference between the two groups: In verbal similarity taken as a whole, MT manifested significantly higher levels than HT. For HT, the highest level of text–discussion similarity occurred when ‘Liking what you see’ was under discussion, a story about physical appearance, social appraisal, and other topics that have a bearing on day-to-day social life. This was the only story that participants said (in discussion and in the post-participation feedback) they had thought and talked about outside the session, and one participant described it (near the start of the subsequent session) as having ‘ probably resonated with me more than the others, just I guess cos it’s kind of more immediately applicable to everyday experience ’ (HT-S6). Correspondingly, participants in both groups considered the act of drawing connections between the text and their own lives to be of relatively high significance to their experience of reading and reflecting during the sessions. One mechanism of its significance may simply be that by definition it encourages text–discussion proximity, and therefore a greater experience of control during the discussion.

Our results for both groups (an association between low doc2vec similarity and high VAD variance) suggest that discussion needs to be grounded in the interpretive possibilities of the text for it to be therapeutically positive. There is, however, no direct connection on VAD grounds between the emotional profile of the texts and the discussions. Our results therefore challenge the similarity hypothesis concerning bibliotherapeutic mechanisms, in which benefits are derived via a close pedagogical relationship between the protagonist’s psychological situation and progression and the reader’s. We found that value for personal insight and wellbeing was sometimes derived despite the lack of obvious markers of similarity between reader and protagonist. For instance, the Underground Man’s lack of growth was perceived by this MT participant as a spur to personal growth:

Particularly after the first few sessions in terms of how people occupy such different headspaces, and also discussions about him wanting to control social situations and set up a moment where he is in a certain position of power/standing shoulder to shoulder. It made me think more about not controlling relationships around me, which was helpful! (post-participation feedback)

When asking how text–discussion similarity is maintained or lost, the difference between discussions centred on author intention versus character motivation seems instructive. The former tended to deflate the world of the text and have the effect of closing down interpretive activity by assuming that ‘definitive’ answers exist. The latter kept the text interpretively open and thereby stimulated ongoing discussion, perhaps because there is clearly no singular ‘fact of the matter’ when it comes to a fictional character’s motivations. One linguistic manifestation of this difference across the two groups was that the pronoun ‘he’ referred more often to the author in HT (along with ‘they’, also for the author) and more often to the protagonist in MT; engagement with characters was also correspondingly higher in MT (see online data file qualitative_coding_results). This challenges the suggestion made by Robinson that author and character are equivalent as objects of interpretive engagement: that ‘participants appeared to become enmeshed in the plot, as they developed their own theories around characters’ actions and motivations, and the author’s intent in using particular words, or including descriptions of particular contexts’ 4 . In our experience, speculation about the author may be more likely to lead to closed discursive forms: statements of liking/dislike, or statements aimed at establishing biographical facts.

Human condition, self-qualification, and VAD-guided word frequency as indicators or mediators of positive group dynamics

One important mediator of positive discussion dynamics may be a category that emerged in the qualitative coding of the discussion transcripts: reflections on the human condition. This emerged in a bottom-up fashion out of the attempt to employ the four major categories of ‘personal relevance’ set out by Kuzmičová and Balint, and the realization that none of those categories accommodated an important and frequently recurring phenomenon: the act of seeing something in the text or character(s) as resonating with a general or universal human tendency, or of seeing a character as an ‘everyman’ figure, etc. For instance:

GOLD: I really loved the um, the bit about memory. [pause] Again because I think it’s true. There are things in everybody’s memory that he doesn’t divulge to everyone but only to friends and so on. I thought that was… [pause] And that writing is often the way of processing those things in a preparatory way, to revealing something. [pause] There’s no self apart from an autobiography in some sense. (MT-S2)

GREEN: He’s also got that typical outsider syndrome, of feeling that you’re superior to everybody else, and looking down on them, while also feeling extremely insecure when he’s actually in their presence, so that you’re living constantly in this world of your own making. (MT-S3)

Frequency of human condition mentionings was significantly higher in MT than in HT (though low in the outlier S6 for both groups), and sessions that included most mentions also tended to include higher numbers of expressions of emotional engagement. It is possible that this type of personal relevance-drawing serves as a happy medium between making and expressing a direct personal connection and keeping discussion at an unthreatening but perhaps also ineffectual level of generality. The group reading context in particular may encourage this form of less individualized relevance attribution, as a way of speaking for oneself but also for a collective. Such comments may have a beneficially inclusive effect, and may also involve cause-and/or-effect links, or ambiguous overlap, with more directly personal connection-making. In this exchange, for instance, ORANGE’s human condition mention generates BLUE’s personal relevance mention:

ORANGE: Because um—let me have a look. [pause] Sorry, I just have to just check back to the things I underlined. [pause] Because he’s obsessed with his social footing and the status, but he tries to achieve it by bumping into people or—the way people look at him, and there’s this constant sense of shame and—he talks about things like physicality and physical size, and I just found it really sort of—a bit like the primate rank, fighting. In a way that I don’t feel—or I don’t—I haven’t experienced or thought about as much in the sense of like the female gender. I don’t know. I think that’s why.

BLUE: I think I was probably just instantly translating from kind of jostling with shoulders to I don’t know, idiots not getting out of the way of me when I’m on my bike or something, and then size of body to like shape of body, or... I think I was probably doing the switch very automatically and that was why it didn’t feel alien. (MT-S3)

The danger also exists that expressing a personal view on a universal phenomenon may make others feel excluded or misunderstood by unjustified generalization. Specifically, the capacious pronoun you may sometimes serve as a linguistic veil for a contribution arising directly out of personal experience not shared more widely. As a form of distinctly discursive relevance-seeking, we suggest that ‘human condition’ statements are worth further investigation.

A second, related phenomenon observed in the discussions was the conspicuous number of self-qualifying statements. These included statements like ‘I don’t know whether anyone else felt that’ or ‘maybe that’s just me’, and other indications of not knowing, not remembering, or otherwise emphasising one’s subjectivity or bias. These statements, occurring significantly more often in MT, were a frequent feature in both groups’ discussions. Although self-qualification could be seen as a trivial indicator of default false modesty, it may also serve a helpful function for social cohesion by moderating the strength of claims being made (including softening human-condition observations). Pragmatically, it often also seemed to provide an easy entry point for the next speaker:

YELLOW: Um, yeah, but again it’s not in the same way, he’s not hating himself anymore there. He’s just indifferent to himself. It’s a different kind of stance, you know. I mean he doesn’t seem petty, whereas I think up to this point, nearly everything he did is petty, you know like the little breakdown he has, and you just you know, you wanna give him a slap and say snap out of it. But here, I felt that was — again, does anybody else have any feelings on that, cos I’m not sure…

GOLD: I might agree, if it wasn’t for the fact that he carried on writing.

YELLOW: Yeah, fair point. (MT-S5)

Self-qualification may also, rather than being a causal driver in its own right, be an effect or a correlate of other ways in which discussion is made more inclusive, for example a general awareness of the importance of maintaining conversational flow amongst participants. In this sense it may relate to the various form of semantic and syntactic echoing identified as a significant contributor to positive group dynamics by other researchers 5 , 11 .

Finally, VAD-structured word clouds provide a different type of clue to textual manifestations of constructive conversational dynamics. For example, the word solution was the most frequently spoken in the high-valence and high-dominance selections for HT ( Figure 5 ), having arisen 13 times in S4 (and almost never in any other session), which was the highest-dominance and highest-valence of all 6 sessions. Instances of its use make clear that participants were grappling with the difficulties of trying to make sense of the plot and authorial intention (in this instance, examples of closed complexity), as well as its wider relevance to problems humankind are currently facing (e.g. overpopulation), with more open-ended scope. Throughout the discussion, the word solution operates as a fulcrum between exploration of closed and more open forms of complexity. Its usage patterns add detail to the suggestion that such transitional dynamics may be associated with positive valence and feelings of control over the discussion. Thus VAD-structured word frequency mapping may, alongside human-condition observations and self-qualifying statements, be a useful indicator as to where to embark on close reading as a source of more sensitive insights regarding the direction taken by a specific text-prompted discussion.

Transcending the qual-quant distinction

Though our study was motivated primarily by the desire to observe the effects of group reading, we feel that some concluding comments are in order on the methods that we used to conduct our analysis. In particular, we made a conscious effort to transcend the too-easy distinction between qualitative and quantitative methods. For good reason, work on literary texts has traditionally been qualitative in nature; texts, after all, are vehicles for creating meaning, and meaning can be experienced but not enumerated. Moreover, for much of the history of literary studies, computational methods simply did not exist (or exist with sufficient availability) to be used in the course of routine scholarship. This situation has changed markedly in the last decade or so, and the emergence of text embeddings and large language models has conclusively demonstrated that semantic content can be captured numerically. This means that the insights generated by contextually informed close reading can now be complemented by reliable and accessible methods from NLP and computational linguistics.

As with all interdisciplinary undertakings, the results may not cohere in any absolute way, or they may cohere at levels other than those expected. Nevertheless, only a curiously shortsighted view of textual scholarship would reject the potential offered by the triangulation of qualitative and quantitative approaches to language. Our aim here has been to demonstrate one way in which such complementarity might play out; there are many others. But whichever one is chosen, we believe that the project of understanding how texts, culture, and cognition interact will be furthered by using all the methodological tools available––and that rejecting qualitative tools in favour of quantitative ones, or vice versa, is merely to dress up subjective prejudice as intellectual conviction.

Taking the longer view

A general question underlying the analyses in this study is a question about timescales. In particular, is it possible (or indeed likely) that a reading and discussion experience that has negative qualities (feels unpleasant, uncomfortable, or even unsafe) may elicit positive change (increased understanding, constructive action, etc.) at some point following the reading and discussion? Conversely, are the most positive experiences (on whichever metrics we select) more or less likely to generate positive change (of whichever type is valued)? This question of short- versus longer-term good versus ill is one that we addressed through in-depth analysis of the verbal dynamics of the discussion sessions and consideration of the participant feedback at the end of each group’s series of meetings. The free-response feedback from both groups included observations on a) learning about oneself and others, b) mood/relaxation benefits, and c) changed habits or attitudes around reading that cannot necessarily be gleaned from analysis of what was said during the discussions, and particularly not from the levels of reported enjoyment of the texts being read. Here is a selection of the concrete changes reported:

I learned that it is hard for me to clearly articulate certain types of emotional or experiential responses and maybe that is something I need to work to improve!

I’m actively looking for a book to read in its entirety. How long this motivation will linger... I’m not sure.

I emerged with a sense that people are nicer than I thought they were, and more inclined to be charitable in my interpretation of motives more generally.

I became more relaxed with respect to other problems after the reading group.

                                                                                                                                                                                          (MT)

it has thankfully increased willingness to see others points of view.

I came back in a much better mood. I would listen to my daughter when we were reading together and discuss what was happening.

I built a momentum as it were, continuing with the state of free flowing self expression even after the session.

You realise who you may look and sound alike, have similarities with and it provided a new talking point(s), idea, concept to recommend to others (without discussing content of discussions).

It’s been a busy and sometimes difficult term for me since recovering from surgery- it’s one of several activities that I’m pleased I took part in and proud I completed.

Yes wanted to be able to have a new routine and read around topics not necessarily thought of or enjoyed before. Also to further number of folk we know.

                                                                                                                                                                                          (HT)

It may or may not ever be possible to predict on the basis of textual and/or discussion content which of these benefits is mostly likely to accrue, let alone to tailor text, select participants, or guide discussion to maximize its likelihood. We suggest the next steps for group bibliotherapy research and practice should be open to several possibilities:

that negative effects are possible, in the short and/or longer term

In two personal reports which ET made after HT-S5 and HT-S6, she reports feeling ‘ excluded ’, feeling that ‘ there was no space and no time for me in the conversation ’, feeling resentment of other participants for never allowing silence between contributions, being keen to get away after the end of the session, finding it hard to re-engage with people after the end of the session, and subsequently feeling ‘ distanced from everything and everyone, and angry, and very fragile ’ for a number of hours, including responding to mental illness-related online material in a ‘ more viscerally defensive/aggressive ’ mode than usual. She concludes: ‘ I’m left feeling that in the wrong hands, or with the ‘wrong’ text, or just through bad luck, this thing really could be dangerous for people. I’m generally tired at the moment, but I’m healthy and generally strong and fairly well-balanced. If I weren’t, I imagine that getting past this might have taken much longer. And who knows, perhaps with more time still, I’ll feel that I’ve learned something about myself that needed to be learned, and that the negative short-term reaction was a fine price to pay for insight that took longer to come. But right now, I feel dislike and resentment and a lingering sense of unsettlement that I can’t quite see turning into something good. ’

that enjoyment of the discussion may be minimally related to liking of the text

I’ve learnt that I can still enjoy reading a text I don’t like when it’s in this kind of context. (HT post-participation feedback)

that positive and negative effects of participation may be minimally related to liking of the text

My relationship with reading has definitely changed. I see the value that it has to create a new and sometimes uncomfortable experience. Previously, I only considered reading to be strictly for the purpose of gaining knowledge. I also feel confident in sharing my opinions about a book. (MT post-participation feedback)

that positive effects of participation may be related to the text-proximity of the discussion

I also learned how reading a work of fiction together can bond people and allow for the development of group identity. I felt that reading in a group fostered a sense of collaborative journeying through the text, with the acts of reading and discussion marking a unique space and time. (MT post-participation feedback)

The last two quotations also indicate the value of experiencing a text for the first time in the bibliotherapeutic group, as well as the significant felt impact of discovering it together. Reading aloud, as a process that heightens the experience of a shared and ‘live’ sensory-cognitive journey, is likely to be relevant here. In other words, the way a text is encountered matters, possibly more than the nature of the text itself. As for the group discussion, having begun with a sceptical attitude as to the value of actually talking about a book as opposed to simply convening for discussion about something else, we find ourselves concluding that the book really does make a huge difference. In particular, our results suggest that talking about the book is importantly different from not talking about the book, and that emotional unpredictability is greater when the book is left behind. Literary scholars persuaded that literature has effects is not the world’s most attention-grabbing headline, but we hope that this pilot study has demonstrated how novel quantitative methods for sensitive textual analysis can flesh out the claim and generate useful hypotheses for future testing.

Acknowledgements

We are grateful to Nela Brockington for her involvement in designing and organizing the reading groups, for taking part in the MT group, and for stimulating conversations on methods, analysis, and the bigger picture. We thank Thor Magnus Tangerås and Moniek Kuijpers for the care they took in providing detailed peer review comments to help us strengthen this article. Finally, we would like to thank everyone who took part in our reading groups for giving their time and their enthusiasm to the project.

Funding Statement

This work was supported by Wellcome [205493; a fellowship awarded to James Carney]; and a grant from the Balliol Interdisciplinary Institute awarded to Emily Holman.

The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

[version 2; peer review: 2 approved]

Data availability

Underlying data.

Oxford University Research Archive: Books, Minds, and Bodies dataset. https://doi.org/10.5287/bodleian:gJZz9KDE0 56 .

This project contains the following underlying data:

Qualitative_coding_results.xlsx

raw_text_data.xlsx

text_sim.xlsx

qualitative_coding_results.xlsx

participant_feedback_results.xlsx

Data are available under the terms of the Creative Commons Zero "No rights reserved" data waiver (CC0 1.0 Public domain dedication).

• 1. Davis J: Enjoying and enduring: groups reading aloud for wellbeing. Lancet. 2009;373(9665):714–5. 10.1016/s0140-6736(09)60426-8 [ DOI ] [ PubMed ] [ Google Scholar ]
• 2. Longden E, Davis P, Billington J, et al. : Shared Reading: assessing the intrinsic value of a literature-based health intervention. Med Humanit. 2015;41(2):113–120. 10.1136/medhum-2015-010704 [ DOI ] [ PubMed ] [ Google Scholar ]
• 3. Longden E, Davis P, Carroll J, et al. : An evaluation of shared reading groups for adults living with dementia: preliminary findings. J Public Ment Health. 2016;15(2):75–82. 10.1108/JPMH-06-2015-0023 [ DOI ] [ Google Scholar ]
• 4. Robinson J: Reading and talking: Exploring the experience of taking part in reading groups at the Vauxhall Health Care Centre. Liverpool. 2008; (115). Report No.: 8. Reference Source
• 5. Billington J, Dowrick C, Hamer A, et al. : An investigation into the therapeutic benefits of reading in relation to depression and well-being. Liverpool. 2010. Reference Source [ Google Scholar ]
• 6. Billington J, Sperlinger T: Where does literary study happen? Two case studies. Teach High Educ. 2011;16(5):505–16. 10.1080/13562517.2011.570439 [ DOI ] [ Google Scholar ]
• 7. Billington J: “Reading for life”: Prison reading groups in practice and theory. Crit Surv. 2011;23(3):67–85. Reference Source [ Google Scholar ]
• 8. Billington J, Longden E, Robinson J: A literature-based intervention for women prisoners: preliminary findings. Int J Prison Health. 2016;12(4):230–43. 10.1108/IJPH-09-2015-0031 [ DOI ] [ PubMed ] [ Google Scholar ]
• 9. Billington J, Carroll J, Davis P, et al. : A literature-based intervention for older people living with dementia. Perspect Public Health. 2013;133(3):165–73. 10.1177/1757913912470052 [ DOI ] [ PubMed ] [ Google Scholar ]
• 10. Billington J, Davis P, Farrington G: Reading as participatory art: an alternative mental health therapy. Journal of Arts & Communities. 2013;5(1):25–40. 10.1386/jaac.5.1.25_1 [ DOI ] [ Google Scholar ]
• 11. Dowrick C, Billington J, Robinson J, et al. : Get into Reading as an intervention for common mental health problems: exploring catalysts for change. Med Humanit. 2012;38(1):15–20. 10.1136/medhum-2011-010083 [ DOI ] [ PubMed ] [ Google Scholar ]
• 12. Montgomery P, Maunders K: The effectiveness of creative bibliotherapy for internalizing, externalizing, and prosocial behaviors in children: a systematic review. Child Youth Serv Rev. 2015;55:37–47. 10.1016/j.childyouth.2015.05.010 [ DOI ] [ Google Scholar ]
• 13. Glavin CEY, Montgomery P: Creative bibliotherapy for post-traumatic stress disorder (PTSD): a systematic review. J Poet Ther. 2017;30(2):95–107. 10.1080/08893675.2017.1266190 [ DOI ] [ Google Scholar ]
• 14. Daboui P, Janbabai G, Moradi S: Hope and mood improvement in women with breast cancer using group poetry therapy: a questionnaire-based before-after study. J Poet Ther. 2018;31(3):165–72. 10.1080/08893675.2018.1467822 [ DOI ] [ Google Scholar ]
• 15. Czernianin W: Poetry as a therapeutic medium in shaping mood. J Poet Ther. 2016;29(3):135–45. 10.1080/08893675.2016.1199513 [ DOI ] [ Google Scholar ]
• 16. Pettersson C: Psychological well-being, improved self-confidence, and social capacity: bibliotherapy from a user perspective. J Poet Ther. 2018;31(2):124–34. 10.1080/08893675.2018.1448955 [ DOI ] [ Google Scholar ]
• 17. Brewster E: An investigation of experiences of reading for mental health and well-being and their relation to models of bibliotherapy.University of Sheffield.2011. Reference Source [ Google Scholar ]
• 18. Skjerdingstad KI, Tangerås TM: Shared reading as an affordance-nest for developing kinesic engagement with poetry: a case study. Cogent Arts & Humanities. 2019;6(1): 1688631. 10.1080/23311983.2019.1688631 [ DOI ] [ Google Scholar ]
• 19. Soter AO: Reading and writing poetically for well-being: language as a field of energy in practice. J Poet Ther. 2016;29(3):161–74. 10.1080/08893675.2016.1199510 [ DOI ] [ Google Scholar ]
• 20. Gorelick K: Poetry therapy. In: Malchiodi C editor. Expressive Therapies. New York: The Guilford Press.2005;128–9. [ Google Scholar ]
• 21. Kuiken D, Sharma R: Effects of loss and trauma on sublime disquietude during literary reading. Sci Study Lit. 2013;3(2):240–65. 10.1075/ssol.3.2.05kui [ DOI ] [ Google Scholar ]
• 22. Sikora S, Kuiken D, Miall DS: An uncommon resonance: the influence of loss on expressive reading. Empir Stud Arts. 2010;28(2):135–53. 10.2190/EM.28.2.b [ DOI ] [ Google Scholar ]
• 23. Kuiken D, Miall D, Sikora S: Forms of self-implication in literary reading. Poet Today. 2004;25(2):171–203. 10.1215/03335372-25-2-171 [ DOI ] [ Google Scholar ]
• 24. Kuzmičová A, Balint K: Personal relevance in story reading: a research review. Poet Today. 2019;40:429–451. 10.1215/03335372-7558066 [ DOI ] [ Google Scholar ]
• 25. Shrodes C: Bibliotherapy: a theoretical and clinical-experimental study.University of California, Berkeley.1950. Reference Source [ Google Scholar ]
• 26. Russell DH, Shrodes C: Contributions of research in bibliotherapy to the Language-Arts Program. I. Sch Rev. 1950;58(6):335–42. Reference Source [ Google Scholar ]
• 27. Pardeck JT: Using literature to help adolescents cope with problems. Adolescence. 1994;29(114):421–7. [ PubMed ] [ Google Scholar ]
• 28. Pardeck JT, Pardeck JA: Treating abused children through bibliotherapy. Early Child Dev Care. 1984;16(3-4):195–203. 10.1080/0300443840160304 [ DOI ] [ Google Scholar ]
• 29. Shechtman Z: The contribution of bibliotherapy to the counseling of aggressive boys. Psychother Res. 2006;16(5):645–51. 10.1080/10503300600591312 [ DOI ] [ Google Scholar ]
• 30. Detrixhe JJ: Souls in jeopardy: Questions and innovations for bibliotherapy with fiction. J Humanist Couns Educ Dev. 2010;49(1):58–72. 10.1002/j.2161-1939.2010.tb00087.x [ DOI ] [ Google Scholar ]
• 31. Troscianko ET: Fiction-reading for good or ill: eating disorders, interpretation and the case for creative bibliotherapy research. Med Humanit. 2018;44(3):201–11. 10.1136/medhum-2017-011375 [ DOI ] [ PubMed ] [ Google Scholar ]
• 32. Troscianko ET: Literary reading and eating disorders: survey evidence of therapeutic help and harm. J Eat Disord. 2018;6(1): 8. 10.1186/s40337-018-0191-5 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• 33. Ramsey-Wade CE, Devine E: Is poetry therapy an appropriate intervention for clients recovering from anorexia? A critical review of the literature and client report. Br J Guid Counc. 2018;46(3):282–92. 10.1080/03069885.2017.1379595 [ DOI ] [ Google Scholar ]
• 34. Carney J, Robertson C: People searching for meaning in their lives find literature more engaging. Rev Gen Psychol. 2018;22(2):199–209. 10.1037/gpr0000134 [ DOI ] [ Google Scholar ]
• 35. Carney J, Wlodarski R, Dunbar R: Inference or enaction? The impact of genre on the narrative processing of other minds. PLoS One. 2014;9(12): e114172. 10.1371/journal.pone.0114172 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• 36. Carney J, Robertson C, Dávid-Barrett T: Fictional narrative as a variational Bayesian method for estimating social dispositions in large groups. J Math Psychol. 2019;93: 102279. 10.1016/j.jmp.2019.102279 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• 37. Carney J, MacCarron P: Comic-book superheroes and prosocial agency: a large-scale quantitative analysis of the effects of cognitive factors on popular representations. J Cogn Cult. 2017;17(3–4):306–30. 10.1163/15685373-12340009 [ DOI ] [ Google Scholar ]
• 38. Carney J: Culture and mood disorders: the effect of abstraction in image, narrative and film on depression and anxiety. Med Humanit. 2020;46(4):430–443. 10.1136/medhum-2018-011459 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• 39. Robinson J, Billington J: An evaluation of a pilot study of a literature-based intervention with women in prison: short report. 2012. Reference Source
• 40. Tukhareli N: Bibliotherapy in a library setting: reaching out to vulnerable youth. Can J Libr Inf Pract Res. 2011;6:1–18. 10.21083/partnership.v6i1.1402 [ DOI ] [ Google Scholar ]
• 41. Dubrasky D, Sorensen S, Donovan A, et al. : “Discovering inner strengths”: a co-facilitative poetry therapy curriculum for groups. J Poet Ther. 2019;32(1):1–10. 10.1080/08893675.2019.1548924 [ DOI ] [ Google Scholar ]
• 42. Mar RA, Oatley K, Djikic M, et al. : Emotion and narrative fiction: interactive influences before, during, and after reading. Cogn Emot. 2011;25(5):818–33. 10.1080/02699931.2010.515151 [ DOI ] [ PubMed ] [ Google Scholar ]
• 43. Brysbaert M, Warriner AB, Kuperman V: Concreteness ratings for 40 thousand generally known English word lemmas. Behav Res Methods. 2014;46(3):904–11. 10.3758/s13428-013-0403-5 [ DOI ] [ PubMed ] [ Google Scholar ]
• 44. Warriner AB, Kuperman V, Brysbaert M: Norms of valence, arousal, and dominance for 13,915 English lemmas. Behav Res Methods. 2013;45(4):1191–207. 10.3758/s13428-012-0314-x [ DOI ] [ PubMed ] [ Google Scholar ]
• 45. Lynott D, Connell L, Brysbaert M, et al. : The Lancaster Sensorimotor Norms: multidimensional measures of perceptual and action strength for 40,000 English words. Behav Res Methods. 2020;52(3):1271–1291. 10.3758/s13428-019-01316-z [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• 46. Rubin DC, Talarico JM: A comparison of dimensional models of emotion: evidence from emotions, prototypical events, autobiographical memories, and words. Memory. 2009;17(8):802–8. 10.1080/09658210903130764 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ]
• 47. Mehrabian A: Basic dimensions for a general psychological theory: implications for personality, social, environmental, and developmental studies. Cambridge MA: Oelgeschlager, Gunn & Hain,1980;381. Reference Source [ Google Scholar ]
• 48. Bakker I, van der Voordt T, Vink P, et al. : Pleasure, arousal, dominance: Mehrabian and Russell revisited. Curr Psychol. 2014;33(3):405–21. 10.1007/s12144-014-9219-4 [ DOI ] [ Google Scholar ]
• 49. Rong X: word2vec parameter learning explained.2014;1–21. Reference Source
• 50. Mikolov T, Chen K, Corrado G, et al. : Efficient estimation of word representations in vector space.2013;1–12. Reference Source
• 51. Pennington J, Socher R, Manning C: Glove: global vectors for word representation. Proc 2014 Conf Empir Methods Nat Lang Process. 2014;1532–43. 10.3115/v1/D14-1162 [ DOI ] [ Google Scholar ]
• 52. Campr M, Ježek K: Comparing semantic models for evaluating automatic document summarization. Lect Notes Comput Sci. 2015;9302:252–60. 10.1007/978-3-319-24033-6_29 [ DOI ] [ Google Scholar ]
• 53. Devlin J, Chang MW, Lee K, et al. : BERT: pre-training of deep bidirectional transformers for language understanding. Palo Aalto,2018. Reference Source
• 54. Řehůřek R, Sojka P: Software framework for topic modelling with large corpora. Proceedings of the LREC 2010 Workshop on New Challenges for NLP Frameworks. Valletta, Malta: ELRA,2010;45–50. 10.13140/2.1.2393.1847 [ DOI ] [ Google Scholar ]
• 55. Honnibal M, Johnson M: An improved non-monotonic transition system for dependency parsing. Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing. Lisbon, Portugal: Association for Computational Linguistics;2015;1373–8. 10.18653/v1/D15-1162 [ DOI ] [ Google Scholar ]
• 56. Troscianko E, Carney J, Holman E: Books, Minds, and Bodies dataset. University of Oxford.2022. http://www.ora.ox.ac.uk/objects/uuid:bd0ada56-58d2-4832-9c8f-2c064faa4e99
• 57. Barthes R: The pleasure of the text. New York: Hill and Wang,1975. Reference Source [ Google Scholar ]

Reviewer response for version 2

Moniek m kuijpers.

Competing interests: No competing interests were disclosed.

This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

The authors have thoroughly revised the paper based on my feedback and that of the other reviewer. I am happy to say I approve of this version of the paper as it is without any reservations. It was a pleasure reading it.

Is the work clearly and accurately presented and does it cite the current literature?

If applicable, is the statistical analysis and its interpretation appropriate?

Are all the source data underlying the results available to ensure full reproducibility?

Is the study design appropriate and is the work technically sound?

Are the conclusions drawn adequately supported by the results?

Are sufficient details of methods and analysis provided to allow replication by others?

Reviewer Expertise:

Empirical literary studies; narrative absorption; shared reading and well-being; digital social reading; psychometrics

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

Thor Magnus Tangerås

I have now carefully read the article. I am quite impressed with the authors rigorous and extensive revisions, they are both appropriate and sufficient. Thus I find that the article can now be unconditionally approved.

I cannot comment. A qualified statistician is required.

Transformative reading experiences, shared reading, bibliotherapy

Reviewer response for version 1

This article introduces newly developed quantitative measures that can be used in the context of group or shared reading to evaluate the fluctuation and variation of emotional and cognitive aspects both in the texts that are read, and the discussions that follow these readings. These tools could be used to investigate whether and how textual features in the group discussion match textual features in the text that is read, allowing for more clearly pinpointing the mechanisms underlying group or shared reading that lead to various effects (such as therapeutic outcomes).

The article is written well, provides a clear and valuable contribution to the field and the discussion of the benefits of shared or group reading. I especially appreciate the critical reflection on bibliotherapy research, as this is something that is rarely touched upon in the literature, but is something that deserves our attention. We need more research in this area that is (as) free (as possible) of researcher bias. 

I do have a couple of concerns I share with the first reviewer. Addressing these would certainly strengthen the article. Most notably, I agree with the first reviewer that a clearer distinction should be made between The Reader's intervention Shared Reading and the group reading that was performed in this study. As the other reviewer suggests, you have basically invented and tested your own intervention, which I do not see as problematic in itself. However, I do see the need for reflecting more on the differences between your intervention and the Shared Reading intervention (mainly in the discussion section of your paper). Especially, with regard to facilitation and text selection.

In my understanding, Shared Reading does not involve reading out loud by all participants, just by a trained Reader Leader, which I think changes the directions in which the discussions are flowing, as compared to your intervention where the facilitator role is being shared between participants. With respect to the reflection at the end of the article about "if being left in the wrong hands, or with the wrong text, this thing could really be dangerous for people", I think this is a crucial difference to be discussed.

With regard to text selection, I found it interesting that you found that the text selection did not seem to matter much to your participants in terms of enjoyment of discussion or overall positive or negative effects of participation, as the text selection is a major tenant of the Shared Reading intervention. Additionally, you found that engagement with the language of the text was deemed of low significance for your participants. I was hoping to see more of a reflection on these results in the discussion, especially in relation to the Shared Reading intervention's emphasis on using literary texts of "high quality".

I also agree with the first reviewer that you will have to distinguish better between the two different methodological layers of your study. I would advise you also to make these layers clearer in your title and abstract, as the observational and qualitative coding you performed seem to me to be at least just as important and relevant (with respect to the main findings in your discussion) as the quantitative methods you employed.

Where I disagree with the first reviewer is the section on the tried VAD and doc2vec methods. To me the rationale for developing these methods was sound, as was much of the explication. Where I lost the "plot", was in how the method was applied.

I think the idea of mapping the emotional and cognitive variability in both the literary texts read and the transcripts of the discussions they inspired is a great idea, which is why I did not understand concatenating the data from the different sessions, in which you read different texts. And generally I had a hard time understanding why you would like to work with means, rather than with the variance your data shows per session (and use that as a way of generating hypotheses about differences between why certain texts lead to certain discussions). For example, the two wordclouds in Figure 5 almost look identical, which made me question what the usefulness of the method is, when collating data over several sessions.

Furthermore, I think you described the main purpose of introducing such methods into this field of study very well in your discussion, when you say: "VAD-structured word frequency mapping may, alongside human-condition observations and self-qualifying statements, be a useful indicator as to where to embark on close reading as a source of more sensitive insights regarding the direction taken by specific text-prompted discussion". What I take from this is that it is one method that should be combined with other - more qualitative - methods to help researchers determine where in there large amount of textual data they should look for insights into "how group reading works". I think it would improve the legibility of the paper, as well as the contextualization of this method in the larger research area, if you mention something like this earlier in the paper. And generally emphasize the importance of "method triangulation" in this field of research: yes, it is important to develop quantitative methods to investigate shared reading, but they should still be combined with qualitative data to make sense of what is happening during shared reading. The methods complement each other. 

Could you add a laymen terms explanation to the result that text-discussion similarity was inversely correlated to emotional volatility in group discussions? Does that mean that when there was more similarity between the words used in the text and the words used in the discussion of that text, there was less volatility in those discussions? And what exactly does volatility refer to (how should we contextualize it in the VAD context)?

I am personally more used to seeing valence  operationalized as two distinct categories: positive versus negative valence. How should we interpret valence in your study? When valence is high, does that mean that words are more pleasurable or more unpleasant?

When you discuss doc2vec similarity on page 16, I get the impression that you consider it the same thing as perceived similarity between reader and character. Am I right in assuming that, based on this paragraph, and if yes, could you elaborate on why you think it is similar? I get the sense that the doc2vec method can tell us something about the semantic similarity in terms of what words are used in the texts that are compared, but whether it could really help us reflect on the perceived similarity that readers felt between them and the characters in the story, I am unsure. Of course, you can use your observations and coding here to reflect on the importance of perceived similarity.

In general I would really like to see more of a reflection on the usefulness of the newly developed quantitative methods in the discussion. Right now, the discussion is mostly dedicated to - really important and relevant - insights gleamed from the observational and qualitative coding methods used, whereas the title of the paper would make the reader assume it is mostly about the quantitative methods.

Minor points

These are some smaller suggestions to improve the overall legibility of the paper.

I would consider changing the abbreviations HT and MT into something more general like ST (spring term) and FT (fall term), so it is easier for other researchers to interpret (those of us that are not used to the specific terminology used at Oxford University)

Related to that, sometimes when you are drawing conclusions about differences between the MT and the HT group it is unclear whether you are referring to differences between the groups themselves or differences between gathering during spring term versus fall term (e.g., greater emotional febrility in HT)

Could you elaborate what the column "Commentary" refers to in Table 2.

The first part of the note under Figure 5 does not seem to correspond to the actual figure (Figure 5 is just about the HT group, whereas the note implies that it is about both groups)

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

Competing interests: None to disclose.

Response to Reviewer 2 Introduction/general Thanks for these comments on the overall thrust of the paper, particularly as regards the distinction with Shared Reading. We have included more detail on the specific differences between our procedures and SR (with respect to facilitation, reading aloud, and text selection) and have also flagged more clearly that this is in fact a new set of procedures. We added a note near the end on the link between facilitation procedures and harm reduction. With respect to text selection, we added further arguments to justify our approach: specifically, that our texts needed to respect the previous reading of the participants, and were by any reasonable metric of a high literary quality. With that said, we also added the point that judgements of literary quality are not stable over time and individuals, and there will inevitably be a large amount of subjectivity in any text selection. Methods We clarified some points where our methods were not clear with respect to concatenating sessions together (which we didn’t actually do) and justified our decision to use point estimate measures (means) rather than measures of spread (variances). We also corrected the incorrect word cloud figure that you helpfully pointed out.   We appreciate your comment on our mixed methods, and we’re glad that you consider the qualitative coding to add value. On balance we feel that the computational methods trialled here represent the methodological innovation and thus are appropriate to foreground in the title. We have added mentions of the qualitative analysis and other contributions from the Discussion section to the abstract to give them more salience. And see below for an addition to the Discussion section itself regarding the important principle of quant/qual triangulation. Results We added some text that clarified what doc2vec, which we used to measure semantic similarity, can and cannot measure with respect to reader response, and briefly clarified our use of the term “emotional volatility”. Our use of a scalar (rather than polar) measure of valence was also justified. Discussion Expanding on our elucidations of the doc2vec method noted above, we also added clarification of its relation (or lack of) to reader/character similarity. We chose to dedicate a good proportion of the Discussion section to unpacking the insights gleaned from the qualitative analysis, as a way to conclude the paper on a more exploratory note and give the qual elements meaningful space. But in the effort to connect the two sets of methods more satisfyingly with each other, we have added a short new section on the significance of the quantitative methods and on the triangulation attempted here between the two. Minor points We considered various alternatives to the (admittedly arcane!) Michaelmas and Hilary, but after reflecting that no alternatives are problem-free (autumn versus fall; spring versus winter term; etc.) we opted to let MT and HT stand.   In all cases, mentions of intergroup differences refer to differences observed between the two groups as run in the two terms. We were unable to identify simple ways to clarify this further in the main text, but we hope this note provides adequate clarification. The somewhat cryptic Table 2 column label has been clarified.  And thanks again for spotting the incorrect file inclusion in Fig. 5. Thanks for your thorough and helpful feedback!

The article is interesting and provides a valuable contribution to research on literary reading in groups and the development of quantitative methods for analysing emotional and cognitive aspects of text-discussion relationships and complex reader responses.

The article is clearly structured, well written and engaging. The introduction covers most, but not all (see further down), relevant research on the effects and mechanisms of various bibliotherapies.

The rationale and objective of the study is clearly laid out:

There is an absence of analytical methods capable of providing sensitive yet replicable insights into complex textual material.

This pilot study offers a proof-of-concept for new quantitative methods including VAD (valence–arousal–dominance) modelling of emotional variance and doc2vec modelling of linguistic similarity.

And the sections on method, results and discussion are thorough and transparent. The analytic method is replicable.

In sum, I recommend that the article be indexed, but there are issues that should be addressed in each section of the paper:

Introduction:

The article would profit from a sharper discussion of what is implied by “bibliotherapy”.

At least two things should be highlighted here:

That “bibliotherapy” is a contentious term, and covers a highly diverse set of practices, contexts and rationales. For instance, some forms are centered around an interview with a prospective reader with a problem, the problem is analysed and the “bibliotherapist” selects and recommends apt reading material. What constitutes “apt” literature is problematic. E.g. should choice of text be based on “similarity of problems”? And the current study provides empirical findings of great value here.

Shared Reading, the group reading practice that is now most widely disseminated (not just in the UK but across Scandinavia and Europe) does not call itself a “bibliotherapy”. The purpose of Shared Reading was to allow everyone regardless of background the access to and enjoyment of “great reading” (cf Jane Davis, The Lancet). Therapeutic effects were seen as secondary gains, and not the effect aimed for. I think the current article would do well to recognize that a central premise of SR is that the chosen literature may not be “likeable” – challenging, threatening, etc. As such, this study lends empirical support to this central premise.

Also, a little bit of historical context would be useful. The “theorization” of bibliotherapy consists largely of constructs borrowed from psychoanalysis (identification, catharsis, etc). It is only in recent years that empirical approaches to bibliotherapy and literary reading as such have gained currency. And the most relevant of this research – that carried out by Kuiken et al.  on the one hand, and Phil Davis et al. – is primarily bottom-up and not theory driven.

Given that the article briefly discusses Billington’s account of four “mechanisms”, and the problems associated with establishing the impact of each, it would be useful to discuss how

One could isolate variables, given that this study in fact revolves around discussing the role of two of these: choice of reading material and group discussion, but has altered one premise of SR – that the facilitator be trained and highly skilled.

What is referred to as “method” here (and occasionally also “methodology”) is problematic.

On the one hand, it comprises reading group procedures and participants, on the other the methods of analysing data.

What should be stated clearly is that the authors have in fact invented their own group reading method. Whether that should be called Bibliotherapy or something else is another matter. Table 1, designed to show the differences in variables among various approaches, does not cover all relevant aspects here.

First of all, why have the authors decided to design their own unique blend? Given that Shared Reading forms the central touch point in terms of empirically-based group methods, perhaps it would be better to gather data from SR sessions? If not, why not?

Very little is said about the participants. Are they all academics/students at Oxford? Are they all avid readers? They are interviewed beforehand and primed as to their role, so they are highly motivated to participate. This is relevant. They get to choose the reading material, but from which options? And based on what? Perhaps the issue of identification or likeability or plot interest is already integral to that choice.

Why not discuss the relevant differences to Shared Reading? I think it is significant that the session is divided in two here: first read, then discuss. But the most problematic aspect is that facilitation is rotated among participants. The list of possible questions they are given beforehand are mostly evaluative, asking for a “head” response rather than immediate emotional reactions as they happen. This is a very important objection given that so much of the discussion of the results is about emotional responses.

Method of analysis:

This part is the central part of the method. How can one find and develop quantifiable and replicable ways of determining emotional and cognitive responses and interactions?

I find the explication of rationale  for selecting VAD and doc2vec, and the procedures and documentation of their implementation, solid and convincing. This part of the article needs no alterations. What is worth discussing, however, given that this is a pilot study, is: what would the difference be if a different emotion model was used?

Results and Discussion:

Results are clearly laid out and analysed. Emotions by session, by literary text, by word norm data – and cognitive elaboration.

The discussion is relevant and thorough.

My main objective here is:

There is a significant problem in that the article does not clearly distinguish between reader responses, interactions and reported worth of participation on the one hand, and therapeutic effects on the other. Cognitive and emotional engagement with text and with group can be established, and also how the participants self-report the value of participating. However, there is no bridge to documentation of therapeutic benefits/improved psychological well-being over time.

As stated in the article: Our analysis used new quantitative methods in the attempt to provide a combination of richness and replicability needed to answer questions about human responses to complex aes- thetic phenomena.

There is a big difference between the question of emotional-cognitive responses to aesthetic phenomena and the question of the therapeutic benefits of these responses. As such, the findings of the study may be more relevant to the discussion of reader response theory than bibliotherapy theory.

In the literature review, some highly relevant studies are omitted:

Longden, E., Davis, P., Billington, J., Lampropoulou, S., Farrington, G., Magee, F., ... Rhiannon, C. (2015). Shared reading: Assessing the intrinsic value of a literature-based health intervention. Med Humanities, 41, 113–120. doi: 10.1136/medhum- 2015-010704 . 1

Longden, E., Davis, P., Carroll, J., & Billington, J. (2016). An evaluation of shared reading groups for adults living with dementia: Preliminary findings. Journal of Public Health, 15(2), 75–82. doi: 10.1108/JPMH-06-2015-0023 . 2

One of these studies does in fact have a control group where a different group activity is used. And in one of these studies, pre-/post testing is used. If objective/quantifiable measures of improved mental health/wellbeing are to be found, there must be a concept of wellbeing and a way of measuring it systematically. Thus, if it were shown that participants did improve mental health, then hypothesizing that this improvement would be reflected in the transcripts would be of great importance.

(Another article not cited, (Shared reading as an affordance-nest for developing kinesic engagement with poetry: A case study, Kjell Ivar Skjerdingstad and Thor Magnus Tangerås) Discusses how participation can be of benefit even when one does not like the text.) 3

In sum, the article would profit from:

Clearer context for study.

Mention Longden et al. s articles.

Discussion of rationale and choice of procedure of group method.

In the discussion section, draw a distinction between therapeutic effects and reader responses. And elaborate on how the method can be developed further, and which hypotheses can be tested and which variable must be controlled for

• 1. : Shared Reading: assessing the intrinsic value of a literature-based health intervention. Med Humanit .2015;41(2) : 10.1136/medhum-2015-010704 113-20 10.1136/medhum-2015-010704 [ DOI ] [ PubMed ] [ Google Scholar ]
• 2. : An evaluation of shared reading groups for adults living with dementia: preliminary findings. Journal of Public Mental Health .2016;15(2) : 10.1108/JPMH-06-2015-0023 75-82 10.1108/JPMH-06-2015-0023 [ DOI ] [ Google Scholar ]
• 3. : Shared reading as an affordance-nest for developing kinesic engagement with poetry: A case study. Cogent Arts & Humanities .2019;6(1) : 10.1080/23311983.2019.1688631 10.1080/23311983.2019.1688631 [ DOI ] [ Google Scholar ]

Response to Reviewer 1 Introduction   Thank you for your helpful pointers to fleshing out our account of bibliotherapy, emphasising the breadth and contentiousness of the term and in particular its uneasy relationship with the Shared Reading paradigm. The Davis reference is very helpful there. We’ve made some additions to the Introduction to reflect these tensions, and have added something more explicit on why we do choose to use the term in the group reading context here. We appreciate the point that SR’s take on “likeability” is neatly compatible with our conclusions here about liking and enjoyment, and have drawn out that link explicitly. And we’ve made a few small additions on the history of bibliotherapy as theory and practice; thanks for the helpful sketch of the main contours. It’s a nice point that our study works directly with two of Billington’s four hypothesised mechanisms while testing the counterhypothesis with the trained facilitation. We added a brief note on testability, including in relation to the structure of our study.  Method We understand that foregrounding “methods” in the study title makes any terminological slippage here particularly salient. Overall, the scientific convention in which “Methods” covers all practices, both analytical and procedural, has contributed to some fuzzy boundaries here, but we have replaced “methods” with “procedures” where appropriate, to distinguish more clearly between the reading-group setup and the data analysis.  Thanks also for the suggestion to demarcate our reading-group procedures much more explicitly from the SR procedures. We now have a section enumerating the major differences and offering a rationale for adopting these rather than the typical SR methods—allowing that there is already variation amongst SR implementations, as shown in Table 1 (which has also been adjusted slightly to offer a more informative overview). We have now included a little more detail on the participants’ backgrounds, the initial interview, and the text selection process (touching also on the likeability question). On the last of these, the shortlist choices were inevitably arbitrary to some extent, but we did ensure that no single person’s taste or judgement predominated.  As for facilitation procedures: We have attempted to clarify in the main text that the facilitation questions offer, in our view, a reasonable balance of interpretive, emotional, sensory, and personal-relevance aspects, while being constrained by the fact of being precisely a retrospective discussion rather than attempting to tap in-the-moment immediacy. The latter would require significantly different methods (more along the lines of ecological momentary assessment, for example—which would be tricky to implement in a group setting). Never having taken part in a SR session ourselves, we don’t know how the decisions about when to pause to discuss—or indeed how to guide discussion in one direction or another—are in practice taken, in the training protocols or in their translation to specific groups. More investigation of what difference these small differences make would be an interesting direction for future research. Method of analysis We have clarified and expanded on our methods in several places where we were not sufficiently clear in our original text. We explained why the VAD model was used instead of other models of dimensions of emotion, and why we expect (though cannot prove) that there would be no substantive differences in our results if we had VAD-analogous linguistic data for other models of emotion.  Results and Discussion We went to greater lengths to clarify why the distinction between “therapeutic” and “cognitive/emotional” effects is less relevant than it first appears. We respect the reviewer’s observation that these are distinct phenomena, but we also referenced innovations in thinking about mental health that intentionally set out to collapse these distinctions in a way that is both productive and relevant to our ambitions. Thank you for flagging these three really interesting and valuable studies, all of which we have woven into our literature review, and mentioned elsewhere as appropriate. With respect to using a control-group methodology, we made it clearer that we are not engaged in hypothesis testing; our study was observational. This means that using a control vs experimental group design would have been premature. Many thanks for your constructive comments.

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Citations

• Troscianko E, Carney J, Holman E: Books, Minds, and Bodies dataset. University of Oxford.2022. http://www.ora.ox.ac.uk/objects/uuid:bd0ada56-58d2-4832-9c8f-2c064faa4e99

Data Availability Statement

• View on publisher site
• PDF (3.4 MB)
• Collections

Similar articles

Cited by other articles, links to ncbi databases.

• Download .nbib .nbib
• Format: AMA APA MLA NLM

Add to Collections