DEPTH @ LSHTM Research blog

Two members of the DEPTH team, Cicely Marston (supervisor) and Shelly Makleff (PhD student) discuss how best to present quotations from interview transcripts when writing up. We talk about how to present ‘untidy’ speech (e.g. ‘um’, ‘er’, repetition), how much to ‘tidy up’ quotes, and the implications of any ‘tidy up’.

Shelly’s interviews and analysis have been done in Spanish and the quotes she presents in the final write up are translated into English. Here we present a lightly edited version of a supervisory email interchange we thought might be useful to others. And we would love to hear your views in the comments – we certainly don’t have all the answers.

To leave a comment, make sure you’ve clicked on this article’s headline so you are reading the post itself, not the DEPTH blog homepage.

SM: How do you clean up a transcribed quote to present it in an article? Every time I cut some words, even just filler words, should I mark these omissions with an omission marker (such as […])? Or do I have the liberty to just cut those fillers without a […], in order to create a clean and readable quote?

CM: In my opinion all cuts should be marked with an omission marker (e.g. […]). I have argued about this with a journal before because newspapers use ellipses to indicate omissions (rather than a specific omission marker that only indicates omissions). The issue is that when you do this, there is no obvious way to mark pauses in someone speaking so you would need to find another pause marker that won’t be confused with an omission marker. You could do this by writing [pause] every time, but this also makes quotes hard to read if there are a lot of pauses. When you are using translated quotations, it is less clear what to do because for instance, you might keep the translation ‘clean’ by not including every single one of the filler words (though I would recommend you keep them as much as possible where there is a direct translation (e.g. in Mexican Spanish, hesitation where people say ‘este…’ can be translated as ‘um…’ in English), or at least if there is no direct equivalent, make sure you keep the spirit of the original which might have involved hesitations).

For translations, where it is good practice to provide the original language version in an appendix, one way to get around this is to present the original language quotations with all the pause markers etc included, and then present ‘tidied up’ translations in the body of the article. If you do this, you should mention it in the methods section so the reader knows they can refer to the original language quotations. Note that ‘tidying up’ is particularly challenging when you are working in your non-dominant language, which is all the more reason to present the original language transcript excerpts verbatim.

SM: Ok, so sounds like you’d always use […] to signify every piece of cut text in the article. For a conference poster, do you think it’s ok to leave out the […] for filler words so it’s smoother to read?

CM: I would keep it precise i.e. show where you have edited – I assume you won’t cut all the ums and errs. I get quite suspicious when I see a perfect quote because very very few people speak in complete sentences with no hesitations. if you genuinely think the hesitations are unimportant in any given instance, then you *can* edit them, but make sure there is a note that you have done this somewhere on the poster, for transparency.

SM: If I’m adding clarifying info in [], do I do that instead of or as well as the words that are being replaced? In other words, would it be “So for them [the students]” or “So for [the students]”?

CM: I would go with the longer version so that it is clear what they actually said versus what is your interpretation/explanation.

SM: Can I add punctuation and make sentences to create more clarity, when the speech was transcribed as a long run-on sentence?

CM: Yes, definitely improve the punctuation – transcriptions are almost always badly punctuated, especially when the narrative includes reported speech, in which case transcribers often give up on attempting to punctuate it altogether – and to be fair it can take a while to get it right even if it is quite obvious without punctuation what the speaker has said. It is worth doing because it does make it much harder to read when transcripts not properly punctuated. If you are not sure how to punctuate the sentence from the transcript alone (e.g. it is unclear where the emphasis in the sentence was), you will need to go back to the original audio to ensure your ‘new’ punctuation correctly represents what was said.

SM: I wish there were guidelines for this! In a quick internet search, I didn’t find any, at least not that are clear per discipline. While looking for guidelines I did see an article about the diversity of perspectives among academics about how they edit qualitative quotes. One perspective in favour of editing out the filler words pointed out that if participants saw their own quote with all the filler words, they’d feel embarrassed, and it isn’t an expectation that everyone speaks perfectly but that as researchers we should present their ideas as clearly as possible in a way they’d feel comfortable with. And actually, in Mexico when we shared the transcribed quotes with the health educators, they felt embarrassed about it, joking that they needed diction classes. They even made a meme of their horrified reactions when they read their words on paper and heard how they talked (see below). For the presentation of the data, we hadn’t really cleaned up the quotes, it was mainly verbatim, but the idea still stands- the way we represented them didn’t make them sound eloquent, and that embarrassed them.

CM. I agree that if the quotes have names attached to them, the person might prefer a ‘cleaned’ version, but your quotes are anonymised and so from an individual perspective I don’t think that is too much of a concern.

Having said that, it’s true that original, not tidied-up quotations might contribute to a discourse of the ‘other’ being inarticulate. People who are looking for ways to find others inferior will likely find them regardless and so I’m not sure that compromising the integrity of the transcript will help (although I’m open to arguments to the contrary).

Overall, though, going along with the idea that there is a ‘better’ way to speak brings its own problems. Should we all speak in perfect sentences? Who determines what is ‘perfect’ or ‘best’? This is especially difficult if you are trying to ‘tidy up’ sentences that were spoken in another language than our dominant language. As researchers we should commit to transparency. Interviewers who worry about their sentences should probably listen back to the interviews – they will hear that it sounds very normal, even if they hesitate, repeat words, use filler words, reframe questions and so on. It’s important to build rapport – if you don’t naturally speak in 100% full sentences in real life, why would you do so in an interview? Being inauthentic, or struggling to present a more perfect self, may well have a negative impact on the interview overall.

What do you think? How have you handled these issues? Let us know in the comment section below. If you can’t see where, click on this article headline and scroll to the bottom.

Despite widespread media attention, repeated public health alerts, and the US President’s declaration of the opioid overdose epidemic as a ‘public health emergency’, fatal overdoses continue to rise.

A key factor is pharmaceutical fentanyl – a powerful synthetic opioid commonly used in clinical settings to reduce moderate to severe pain. Since 2014 illegally manufactured fentanyl, and many of its analogues, have been used to adulterate street drugs, principally heroin. Fentanyl is considerably more potent by weight than heroin and has been responsible for a significant number of fatal drug overdoses in North America since 2015.

Preliminary data released by the Centres for Disease Control and Prevention suggest that in 2016 more  than 20,000 people died from unintentional overdoses involving synthetic opioids (not including methadone) in the United States. In Canada, 3,671 people died from unintentional overdose in 2017 – 72% of these deaths can be attributed to fentanyl or fentanyl analogues.

So what can be done?

Harm reduction agencies in North America have begun to distribute test-strips which, when dipped in a drug solution, are able to detect fentanyl and many of its analogues. Is this a potential answer to the overdose problem? Perhaps; but caution is required.

First, research suggests that while self-testing technologies may be accurate in a laboratory, they may be considerably less effective when used in a real-life setting. Contamination, for example with injecting equipment, is more likely to occur in a street setting than a laboratory.

Second, the test strips are only able to indicate whether or not fentanyl is present in a drug solution, they do not indicate how much fentanyl is present. In cases where a comparatively large proportion of a drug is fentanyl, ‘test hits’ taken by users to reduce their risk of overdose in response to a positive test may be ineffective in reducing the risk of overdose.

Third, while some test strips have demonstrated greater than 95% accuracy, this still leaves the possibility of a false negative in one in twenty tests of fentanyl-adulterated drugs. People may inject heroin up to five times a day rendering false-negative a fairly regular occurrence. Additionally, test strips have shown to have a low detection limit, a drug sample may test positive when a minimal amount of fentanyl is present. There is a risk that users may not consistently modify behaviour if drugs are routinely testing positive without producing the anticipated psychoactive effect.

Fourth, not all test strips are created equal. DanceSafe, a US-based public health organisation, in conjunction with the University of California San Francisco, tested five fentanyl test strips from three different manufacturers and found that four of the test strips did not detect carfentanil (a fentanyl analogue that is considerably more potent by weight than fentanyl); one test strip did not work at all.

The need to tackle the fentanyl crisis is clear. A recent drug-checking pilot carried out in Vancouver found that 90.6% of drugs submitted tested positive for fentanyl. Though fentanyl was most commonly found in heroin, it was identified in other street drugs as well, including: amphetamines, methamphetamines, crack, and cocaine.

And the problem is not confined to North America. Since 2014, six fentanyl variations have been identified in the UK, with several overdoses attributed to fentanyl occurring in the North East of England in 2017.

But should we be promoting test strips as a potential solution to the crisis?

In June 2018 Health Canada cautioned against relying on test strips alone, suggesting that other precautions were necessary to prevent overdose including: never consuming alone, consuming a lower dose – “start low and go slow”, and only using drugs in the presence of someone who is equipped with, and knowledgeable about the administration of, naloxone.

Providing appropriate messaging on how to interpret test results and their limitations is key. We must understand more about the accuracy and acceptability of fentanyl self-testing, and research should also aim to inform the development of an effective means of dispensing test strips – and associated risk-reduction measures – to those who do not regularly come into contact with harm reduction services. Education and appropriate harm reduction messaging to mitigate against the risk of false negatives alongside strategies to prevent or reverse overdose is crucial.

There is huge potential in self-testing technology but it must be used as part of a package of risk-reduction measures. The problem is clear but the answers are not.

Fentanyl self-testing outside supervised injection settings to prevent opioid overdose: Do we know enough to promote it?

You can also read Catherine’s research in the latest New York Times article, available here.

Our article, ‘Slow co-production’ for deeper patient involvement in health care’, has just been published in the Journal of Health Design. It’s open access so anyone can read it, download it and share it. The article argues that ‘slow co-production’, achieved by involving patients in in-depth research, can help deepen patient involvement in health care. We describe how slow co-production offers a mutually beneficial form of patient and public involvement. It promotes patient-centered knowledge and helps us to examine and reflect on the co-production processes themselves, rather than always rushing to evaluate the end product as if it came into being without a process and a series of human relationships.

The thinking behind this article was our different and shared experiences of time constraints in qualitative fieldwork. We discussed, for example, those times when we felt pressured to move on from fieldwork when we didn’t feel we had been able to fully examine the narratives that we were learning about, or the interviews we were conducting. This restriction to fieldwork research is usually required for budget and time constraints, but it can feel frustrating to have to withdraw from the work being conducted when it feels like there is more to be gained.

More positively, we recognise that we have also experienced opportunities for more thorough, in-depth ‘slow’ research, including the Sickle Cell project that this new article takes as its case study. We believe that this kind of ‘slow’ research, as previously explored in healthcare research by Vincanne Adams et. al (2013) and Heather Mendick (2014) amongst others, is an important antidote to the ever-building pressures that are heaped upon the researcher. To make things yet more complicated, these pressures come from a range of different external sources: for example, a rush to complete fieldwork within a limited timeframe to minimise costs for the funding body, or unexpected delays in participant recruitment that mean that tightly organised fieldwork cannot extend beyond the ethical clearance period granted by the host organisation.

This idea of ‘slow’ working isn’t new (see, for example, Honoré, 2005, or ‘A Call for Slow Scholarship’ by Hartman and Darab, 2012), but it is worth repeating: so much of academic research is subject to the restrictions of budget, outcome, time constraints and metrics that need ticking off that we forget how valuable time and space are for germinating our best ideas. The useful analytical frameworks, or new theories, or vital links we make between seemingly unrelated sets of results are often produced in the scarce time we get to think deeply about concepts or theories at length. It is this type of time which has become most compressed in the contemporary neoliberal university, where there are so many competing demands on the researcher’s time and attention, from emails to admin to marking to service.

In our discussions we came to a consensus that for us, ‘slow’ co-production is about having time to genuinely engage with others in dialogues that cover more ground than simply sharing ideas, although of course ideas come from dialogue so the two concepts aren’t separate. As well as sharing ideas, we can think of slow co-production in terms of humanity and human rights: specifically, reducing the gap between the researcher and the researched, and trying to find common ground in a qualitative methodology that productively disrupts some of the traditional boundaries that we have seen replicated over and over across social sciences and public health research. Therefore, as well as the point that taking more time in the research field can develop more positive, meaningful relations between the different parties in the research, we also want to stress that this in-depth, detailed, sensitive approach to interviewing is a specific form of co-production, and one that we want to pursue further across our DEPTH projects.

So, whilst our article is primarily arguing for the importance of involving patients and participants in qualitative research from the start and throughout the lifespan of a research project, we hope it also offers a small insight into the opportunities enabled by ‘slow’, thorough working between researchers and participants. This ‘slow’ process – scarce as it sometimes feels the opportunity is – can offer a really valuable way into mutually beneficial collaboration.

References

Adams V, Burke NJ, Whitmarsh I. (2014) Slow Research: Thoughts for a Movement in Global Health. Medical Anthropology, 33(3): 179-97. https://doi.org/10.1080/01459740.2013.858335 ​​​​​

Hartman Y, Darab, S. (2012) A Call for Slow Scholarship: A Case Study on the Intensification of Academic Life and Its Implications for Pedagogy. Review of Education, Pedagogy, and Cultural Studies, 34(1): 49-60. https://doi.org/10.1080/10714413.2012.643740

Honoré C. (2005) In Praise of Slowness: Challenging the Cult of Speed. New York: HarperCollins.

Mendick H. (2014) Social class, gender and the pace of academic life: What kind of solution is slow? Forum: Qualitative Social Research, 15(3):7. http://nbn-resolving.de/urn:nbn:de:0114-fqs140374.