Breaking down the barriers

x Sponsored content on Research Live and in Impact magazine is editorially independent.
Find out more about advertising and sponsorship.

Mind your language! This is not a warning about rudeness, although rudeness might be an unintended consequence; rather, it’s a pointer to the problems that arise from using sloppy, hard-to-understand or unengaging language when conducting research. Language that fails to resonate with panellists, or is wide of the mark in terms of research objectives, may equate to opportunities missed, spiralling costs and data of dubious quality. It may prove an insurmountable barrier to accessibility.

There are two major aspects to language-related problems that risk undermining research: clarity of communication and cultural adaptation. In the case of the latter, this is most obviously an issue for international/global research, although it is equally applicable when conducting research among minority groups in the UK who do not speak English as a first language. More on that later, but first let’s address the huge area that is international research.

International projects with data collection in one or more overseas country, but controlled and invoiced from the UK, have long been a particular strength of UK suppliers. In 2016, The Market Research Society (MRS) found projects of this kind represented 36% of UK research revenues, and the picture seems little changed since then.

As noted in the Research Live Industry Report 2021: “The UK retains its long-term status as a hub for management of international projects.” Given that around one third of projects emanating from the UK are international in nature, cultural adaptation is certainly far from a minor consideration. Yet many researchers believe there is still a long way to go in getting the language aspects of such projects right – even in English-speaking, or supposedly English-speaking, markets.

“In a global research context, people tend not to think about English-speaking markets,” says Ruth Partington, chief executive of Empower Translate, a specialist in market research localisation. “We do a lot of English adaptation – adapting UK English for markets such as the US, Australia, Singapore and India. We do global adaptation for accessibility. If you go into China and you speak UK English, it’s much less recognisable to the Chinese than American English, because they’ve all grown up watching Friends, or whatever American TV series it might be. So, there are the nuances and contextualisation.”

A good example of how layers of meaning may easily be missed by an English speaker from outside the UK is provided by the so-called ‘pork pie plot’ against Boris Johnson, in which Conservative backbenchers with concerns about ‘partygate’ met in Melton Mowbray. How likely is it that someone who has never lived in the UK would grasp that Melton Mowbray is the home of the pork pie, or the fact that ‘porky pie’ is also rhyming slang for a lie?

Clearly, one would not necessarily be dealing with such layers of complexity when doing market research. Nevertheless, it is important to understand – and, where necessary, explain – cultural context, so that people can really engage with what you ask them.

Mustard Research strategy director Andrew Wiseman says that, on balance, he would argue that local language is preferable for international research, but that the debate is complex. “For some audiences, especially in the business-to-business (B2B) space, using English internationally is the norm, on the premise that decision-makers in global organisations typically speak English as a secondary language. However, for many consumer audiences, the fluency of English is unlikely to be native, although admittedly at a higher level of fluency than our own foreign language skills may be.

“Of course, it’s helpful to have access to language skills in the immediate research team, but this mustn’t be taken as a replacement for native language speakers when it comes to translation. At Mustard, our translation partners are picked on the basis of their native language capability. This provides obvious speed and cost benefits, but also the cultural and linguistic nuances of how consumers in those markets talk about the issues that the survey is looking to explore.”

Wiseman notes that there are plenty of phrases that work well in English, but that don’t necessarily translate to other global languages – an issue that can be overcome by using native translators. He also recommends native proofreading, rather than back-translating, both for grammatical and contextual reasons. Finally, he asserts that a collaborative approach with your local market client contact is vital, to ensure that a balance is struck between keeping the feedback informal while delivering the data needed for decision-making.

“It’s also worth thinking about the context of phrases such as ‘quite’ and ‘very’ in research,” adds Wiseman. “There’s a famous picture that does the rounds from time to time: ‘What the English say, what the English mean, what anyone else takes away.’ In the context of something being ‘quite good’, typically this would mean it was OK but had room for improvement, whereas the global context could be ‘that was good enough’, meaning job done. So, again, ensuring that scale questions are interpreted in context is critical.”

The case for investment

Some agencies and clients may be disinclined to invest in professional translation because they see it as a way to keep costs under control. However, this is likely to be a false economy, and it should be noted that the cost of expert translation is far from high. Partington says translating an English survey into the languages of your study typically represents less than 5% of total project cost.

“Translation is such a small cost comparatively compared to the richness of what you’re going to receive from it as well,” she argues. “For global research, translation should be king. It should be the north star. Everybody should think, ‘I am doing this piece globally, I have to translate it and the translation has to be good’.”

Good translation will help researchers reach their quota quickly and meet the timelines projected to the client. Ultimately, it should lead to better value and higher-quality insights – great, actionable points, in other words.

Cultural differences must also be borne in mind. This calls for knowledge of local market habits and attitudes. “There are very few occasions when you’d want to use English over a native language,” says Opinium CEO James Endersby. “One of the biggest difficulties in international research, we find, is cultural differences in response patterns. For example, cultural acquiescence bias often means certain markets – for example, China – will almost always come out top for levels of agreement. This makes cross-market comparisons extremely difficult. It is also a known problem even within markets, because certain ethnic or cultural groups will be more likely to agree with any statement. Some techniques, such as bipolar scales, can mitigate this, but there’s no perfect solution.”

Researchers need to think about both client objectives and the respondent experience when designing surveys. All too often it is easy to focus too much on the former, and not enough on the latter. If the respondent experience is not a good one, then meeting client objectives will suffer in any case. It is vital that client language is translated into phrasing that the respondent audience will easily understand.

The aim should always be to use the language that allows respondents to fully understand the questions being asked. In some cases, points out Savanta executive vice-president Oliver Wright, assumptions should not be made about what that language is, and respondents should instead be given the choice of languages that are spoken in their country. When translating surveys, a native speaker should always proofread them to ensure accuracy and that any complex terms or colloquialisms are translated correctly.

“Cognitive interviews in local languages can help ensure that questions are worded in a way that ensures full understanding across multi-country projects,” remarks Wright. “In qualitative research, using native moderators or interviewers can help the research to be inclusive of different sub-groups of the target population.

“When the research is targeted at hard-to-reach or vulnerable groups, it can be beneficial to use participatory techniques so that research subjects are involved in the design. This helps ensure that the language used is appropriate to the target group and their experiences.” Savanta used such an approach for a project for the British Red Cross on the experiences of female asylum seekers.

The sensitivities and need for caution and thoroughness relating to work such as this, where the subjects may have difficult and even upsetting experiences to relate, are immediately apparent. Yet, in ostensibly less-sensitive areas – such as brand and consumer research – it is still important to tread with care. Researchers are continually under pressure to find ways to make research more cost-effective, quicker and easier, which may lead to shortcuts that are not necessarily helpful in the long run, and may, in fact, be detrimental.

This can, says Charlotte Sykes, research director at DJS Research, lead British researchers to conduct qualitative research in other English-speaking markets around the world for reasons such as: “Because I know the project”; “It’s cheaper for me to do it than get an agency involved”; and “It will be quicker to do it myself”. Such temptations should be resisted, asserts Sykes, because of the “dichotomy between speaking the same language and speaking the same culture”. In more than 10 years of researching internationally, she has learnt that the two are very different.

“As we all know, brand is complex,” says Sykes. “And researching brand is extremely complex. Add to that additional layers of linguistic and cultural nuance, and things can go very wrong, very quickly.”

We may think we speak the same language as our English-speaking international neighbours, but do we really? It is surely impossible to decipher explicit and implicit associations for a brand without in-depth cultural context for each market. Sykes provides the example of Americans not knowing what HP Sauce is – arguably making it the perfect accompaniment to pork pies in the rich buffet of international misunderstandings.

In terms of advice, Sykes advocates getting in-market support and localising methods. Localisation should extend to data-collection tools and analysis, on the basis that there really is no such thing as plain English.

A good example of tool localisation comes from MMR Research Worldwide. According to its consumer experience director, Caroline Withers, it adapts its tools for the advanced nature of ecommerce in parts of Asia, while ensuring it reaches target consumers through appropriate and relevant channels in less-technologically developed markets, such as Africa.

“For example, for studies in Africa, our South Africa team adapt their research by looking to connect with consumers through familiar services such as WhatsApp, rather than strict quantitative research tools,” says Withers. “This flexibility gains insights direct from consumers in a familiar context, without compromising on the research validity or reinventing the wheel, all at the same time.”

Market research needs to constantly evolve to capture actionable, appropriate learnings for businesses large and small. As globalisation continues apace, the industry must ensure regional and local customs are still researched, represented and understood. Withers explains that a dwindling in the “easy wins” of mass-market, international products means clients are increasingly looking to target specific consumer segments in specific regions, and throughout these focused campaigns, language is key to representing those target consumers’ needs. As a result, the entire industry must look to connect with consumers across borders, to develop brands and offerings to delight people wherever they live and whatever language they speak.

The pitfalls of automated translation

Artificial intelligence (AI), machine learning and robotic process automation (RPA) are commonplace today. Capabilities have come on in leaps and bounds and we often use such technology without a second thought. So, why not turn to Google Translate when working on international projects? What’s the worst that could happen? Unengaging research, potential embarrassment, reputational damage and maybe even legal problems all spring to mind.

The obvious pitfall with machine translation is that, if you are not a native speaker of the language you are trying to capture, you simply don’t know whether it’s appropriate. Dullness, nonsense and offence can easily slip through the cracks when translation is done via the internet.

Gender is a particular minefield that is tricky to navigate with machine translation. Pronoun errors can irritate or offend, and it is worth bearing in mind that, in some countries, it may be illegal to be anything other than male or female. Familiarity with, and sensitivity to, local linguistic and cultural norms should not be underestimated.

There is a lot to recommend when it comes to using professional translators with expertise in research, but it’s still possible to go wrong, even when following this route. Clients and research agencies sometimes expect mountains to be moved in an unfeasibly short span of time.

“Let’s say I’m an agency that invests three weeks going backwards and forwards to my client to build this beautiful survey, written in English, that my client signs off on and thinks is wonderful,” says Partington. “I spend hours programming it, making sure the routing is great, making sure it looks good on screen. I do all of that stuff and then give my translation agency two-and-a-half days to turn around 23 markets for a 20-minute interview. What’s that actually saying about your global market? It says: we don’t value you and the translation is a throwaway piece.”

Those new to international research may not realise that keyboard layout may also sometimes pose problems. Even in countries as geographically close to the UK as France and Belgium, for instance, the AZERTY keyboard (rather than the QWERTY one we use here) is the standard. This can be a factor in online surveys in which participants are asked to use specific keys to answer questions or respond to an image. Needless to say, keyboard challenges are magnified when it comes to languages such as Mandarin Chinese, which have a character set that differs totally from the English alphabet.

Another factor to consider is how to approach ‘right to left’ languages such as Arabic, Hebrew and Urdu – or, indeed, languages where visualisation is particularly challenging. An example of the latter is Thai, which has a space at the beginning and end of a sentence, but no spacing between words, resulting in huge wraps of text lines. Such differences need to be factored into survey planning and design to minimise the risk of confusing participants. Imagine a 1-10 sliding scale. In a right-to-left language, it should be designed so that one is on the right and 10 on the left – the reverse of what we are used to (although experiments have been done on flipping the scale for ‘left to right’ readers to see if that boosts engagement).

Unsurprisingly, what are known as ‘rare languages’ present bigger challenges than widely spoken ones. Papiamento, Creole languages of the Dutch Caribbean that, combined, have fewer than 350,000 native speakers, fall into this category. Delivering on briefs encompassing rare languages requires a robust quality management framework, according to Partington – hence, Empower’s ISO certification.

“We do a lot of validation in the supply chain. We need to assure ourselves that the external people we use are properly qualified, that they are who they say they are, and will be able to do what we need. For some languages, we don’t have any capability internally to validate, so that means the process is not only to find a translator, but also to find somebody else of equal quality to check the work that has been carried out by the initial translator. I think when you’re buying translation the question is not ‘Will my translation be perfect?’ The question is, ‘Is the way that it is being managed something I can rely on and trust?’”

The participant alienation problem

It’s understandably easy to offend, confuse or bore people through bad translation or a poor choice of words. However, participant alienation is an even greater worry when conducting an important piece of research into, let’s say, a rare disease or condition among minority groups. In such instances, the participant pool from which you can draw may be small to begin with. Speaking to them inappropriately almost inevitably means you won’t get the engagement you need to carry out the research. The project will fail and, significantly and detrimentally, the voice of that subgroup will not be heard.

The market research industry needs to think long and hard about participants and how to engage them when writing surveys. Priority is given to question numbers, length, opens and routing. In an ideal world, says Opinium’s Endersby, all surveys would be given a final human check by a “non-researcher, with no category knowledge or skin in the project”, who would answer questions such as: ‘Do you understand what it is that we are asking’?

Clients are understandably passionate about their brand, their sector and their objectives, and often assume the same of respondents. What’s more, the majority of people they communicate with on a day-to-day basis will be similarly engaged. At best, this can lead to ambiguous shorthand and, at worst, incomprehensible and mind-numbing jargon. Endersby defines the mark of a good researcher as the ability to combine a strong understanding of formal logic with genuine empathy – two very different skills. The goal is to be the voice of consumers, not only by reporting their views, but by understanding how to engage them with research in the first place.

“Our job is to translate our clients’ objectives and needs into questions that are easily understandable by their audience,” says Endersby. “This can often be a tricky balance, and when we are talking to participants about particularly technical things, we introduce a phase into our research where we cognitively test our questionnaires with a small number of the target audience before we launch fieldwork. This helps us sense-check questions, language and flow with people who are not research savvy and, potentially, not familiar with the client’s internal language.”

At the other end of the scale from being overly technical or jargon-heavy is the danger of patronising your audience by trying to mimic them – for instance, adopting slang terms when speaking to young people, which can be cringeworthy and counterproductive. “At best, you are dad dancing in a nightclub; at worst belittling the respondent and prejudicing their likely responses,” says Endersby.

Thinking about participants not only entails talking to them in a language they understand and won’t recoil from, but also aiding the analysis process by appreciating the context behind the responses they give. For example, customer satisfaction research typically uses the term ‘satisfied’ in some way. What does this truly mean? Does it mean the same to everyone – and what about the context?

“If I’m answering a survey about water, does ‘satisfied’ simply mean that ‘I turn my tap on and water comes out’, given that I might not have had any other contact with the water company,” ponders Mustard’s Wiseman. “Understanding the term ‘satisfied’ without the context of their broader experience, or their expectations, can make results more difficult to implement. After all, would consumers talk to their friends in such terms: ‘I’m really satisfied with the meal I’ve just eaten’, or would they be more likely to say ‘That was brilliant/terrible/awesome/rubbish/not as good as last time’.”

It should go without saying that an appreciation of the way people actually speak is a handy skill for a researcher. All the more so in an era of survey fatigue. In the early days of the pandemic, when working from home was a novelty, some people were happy to complete surveys as a way of filling their time. But just as we have seen Zoom fatigue set in for those who feel their lives have become an endless series of video calls, many consumers have become fed up with completing surveys. For some, the disaffection may be sparked by a sense that surveys are an invasion of their privacy or spare time in a period when it has been a struggle to prevent work from encroaching too much on home life. Others, perhaps, have the impression that surveys are a bit lazy. After all, why should they, the consumer, make the effort to share their precious information and opinions?

Using snappier, more engaging and more relevant language certainly has a role to play in overcoming participant alienation. As a stakeholder group, participants are sometimes overlooked compared with the end audience for the results, and there’s certainly an opportunity for that balance to be addressed. Whether by pushing back on ‘adding just a couple more questions’ requests from clients, challenging the use of internal language, or asking ‘what are you going to do with the results from that question?’, there are ways in which survey length and flow can be improved for participants.

“As consumers are digesting media in multiple formats today, with clinically proven decreasing levels of concentration, so market research must adapt to this new normal,” says Withers. “Communicating clearly, flexibly and effectively is key to consumer engagement. We are constantly learning and adapting our survey language to be more personable and more like a normal conversation.

“We look to provide the human touch with video questions and feedback, as well as integrating AI technology and gaining more conversational responses via chatbots. All of these elements are important to boost engagement, as well as increase the depth and insight we learn from consumers by reaching out in a language and format that can truly connect with them.”

Language can be a barrier to accessibility, and the onus is on the research industry to prevent that. Keeping cost per interview to a level that is reasonable and accessible is a big issue, with huge implications for those working in this field, and achieving it depends on engaging your participants right from the start. This, in turn, depends on using language that is clear, compelling and appropriate. In short, there is a very strong commercial argument for getting your language right.