nlp

How Far Can We Go With Harmony? Testing On Kufungisisa, A Cultural Concept Of Distress From Zimbabwe

Please select all the ways you would like to hear from Harmony project:

How Far Can We Go With Harmony? Testing On Kufungisisa, A Cultural Concept Of Distress From Zimbabwe

Many psychologists believe that mental illnesses can vary across cultures. In 1904, Emil Kraepelin initiated the field of comparative psychiatry after studying mental health disorders in Java, writing that “Die Eigenart eines Volkes wird auch in der Häufigkeit und klinischen Gestaltung seiner Geistesstörungen zum Ausdruck kommen,” meaning “The peculiarity of a people[ethnic group] will also be expressed in the frequency and clinical form of its mental disorders.”[1]

More than a century later, the emergence of global mental health research projects has opened a number of debates about the applicability of psychiatric categories to different cultural settings, such as those in the Diagnostic and Statistical Manual of Mental Disorders (DSM) series[2].

In 2013, the publication of DSM-5 included for the first time Cultural Concepts of Distress (CCD), referring to “ways that cultural groups experience, understand, and communicate suffering, behavioral problems, or troubling thoughts and emotions”[2].

Examples of cultural concepts of distress include:

  • “Los nervios” or “ataque de nervios” (Spanish) – literally “attack of nerves”, a concept of severe distress, often linked to interpersonal disputes in Latino and Spanish-speaking cultures[3]. “Los nervios” are associated with physical and functional impairments.
  • “Kufungisisa” (Shona) – literally “thinking too much”, this is a term used in Zimbabwe to describe both the cause and the symptoms of mental health problems such as anxiety and depression. Patel et al[4] suggested describing this as “feeling stressed” or as a “non-specific neurotic mental illness”.
  • “Hikikomori” (Japanese) – a form of severe social withdrawal whereby an adolescent or young adult becomes a recluse in their parents’ home and does not work or attend school for long periods of time[5].

Can we use Harmony to harmonise mental health instruments designed for different cultural settings where some disorders may have no direct equivalent in one culture?

Using Harmony in Google Colab

Zvaita sei kuti chembere yorasika, bere rorutsa imvi? (How is it that the old woman is missing and the hyena is vomiting grey hairs?)

Shona proverb (similar to English “there’s no smoke without fire”)

If you are a Python user, you can follow my experiment in this Jupyter Notebook, which you can open in Google Colab: https://github.com/harmonydata/

Shona (chiShona) is spoken in Zimbabwe and belongs to the Bantu language family, which also includes Zulu, Xhosa, and Swahili.

ShonaEnglish
fungathink
kufungato think
ndofungaI think
-isa(causative suffix: “to cause to do”)
-isisa(intensive suffix: “to do quickly”)
kufungisisathink deeply, think too much; a Shona idiom for non-psychotic mental illness

In Shona, derived verbs can be created from simple verbs by attaching suffixes to the verb stem.

I tried using Harmony to see how it would harmonise “kufungisisa” (thinking too much) to a Western instrument such as GHQ-12.

Although English is the best-resource language for natural language processing, multilingual NLP techniques are catching up even for lower-resourced languages. There exist some NLP models for Shona. I used the sentence transformer model Davlan/xlm-roberta-base-finetuned-shona which is a modification of ROBERTA trained on Shona texts[7]. I plugged one into Harmony and tried to match the Shona symptom questionnaire for the detection of depression and anxiety, which is used in Zimbabwe[6].

img

Above: the text of the Shona symptom questionnaire for the detection of depression and anxiety.

A problem I encountered was that the transformer model didn’t work for both Shona and English (it’s not multilingual, like Harmony’s default transformer model). I Google translated GHQ-12 into Shona as a temporary measure.

Also, the transformer model did not operate as a sentence transformer, but rather as a token-level transformer, so my sentence vectors were made by averaging token vectors across an input.

My model’s output is below:

img

Harmony and the Shona transformer model matched the question about “kufungisisa” to GHQ-12 question 1 “been able to concentrate on whatever you’re doing?” which seems approximately OK. However, I would need a Shona native speaker to validate my results.

Also, when we are using English and Portuguese texts, which has until now been our go-to bilingual combination, we have had the advantage that some multilingual models cover both languages and so it’s possible to directly compare in both source languages. In the absence of an LLM (large language model) which can handle both Shona and English, it’s not possible to directly compare English and Shona text, but my experiment above shows that Harmony can work on monolingual Shona text.

References

  1. E. Kraepelin: Vergleichende Psychiatrie. Zentralblatt für Nervenheilk. u. Psychiat. Nr. 174, 433-437. 1904 (39)
  2. American Psychiatric Association (2013) Diagnostic and statistical manual of mental disorders, 5th Edition. Washington, D.C.: American Psychiatric Publishing, Inc.
  3. Kohrt, Brandon A., et al. “Cultural concepts of distress and psychiatric disorders: literature review and research recommendations for global mental health epidemiology.” International journal of epidemiology 43.2 (2014): 365-406.
  4. Patel, Vikram, Essie Simunyu, and Fungisai Gwanzura. “Kufungisisa (thinking too much): a Shona idiom for non-psychotic mental illness.” The Central African journal of medicine 41.7 (1995): 209-215.
  5. Teo, Alan Robert, and Albert C. Gaw. “Hikikomori, a Japanese culture-bound syndrome of social withdrawal? A proposal for DSM-V.” The Journal of nervous and mental disease 198.6 (2010): 444.
  6. Shona symptom questionnaire, https://depts.washington.edu/edgh/zw/hit/web/project-resources/shona_symptom_questionnaire.pdf, retrieved 13 July 2023
  7. Alabi, Jesujoba O., et al. “Adapting pre-trained language models to African languages via multilingual adaptive fine-tuning.” arXiv preprint arXiv:2204.06487 (2022).

Related Posts

Harmony issue tracker

Harmony issue tracker

Tracking Harmony issues and raising feedback Have you found a problem with Harmony? You can use the links on the Harmony homepage at harmonydata.ac.uk - you can see there are buttons titled Raise NLP issue and Raise UI issue, which take you to the process for logging issues in Github. The issues remain in the issue trackers and can be assigned to people either under “Assignee” or in the comments:

Developer guide

Developer guide

Git and GitHub workflow The preferred workflow for contributing to Harmony’s repository is to fork the main repository on GitHub, clone, and develop on a new branch. Please read our general guide about contributing to Harmony. Fork the main project repository by clicking on the ‘Fork’ button near the top right of the page. This creates a copy of the code under your GitHub user account. For more details on how to fork a repository see this guide.

Signup to our newsletter

The latest news on data harmonisation project.

Please select all the ways you would like to hear from Harmony project:

You can unsubscribe at any time by clicking the link in the footer of our emails. For information about our privacy practices, please visit our website. We use Mailchimp as our marketing platform. By clicking below to subscribe, you acknowledge that your information will be transferred to Mailchimp for processing. Learn more about Mailchimp's privacy practices.