This is the third part in The Future of Digital Marketing series, which explores the impact of the COVID-19 pandemic on digital transformation and digital marketing as companies strive to deliver a consistent multimarket, multichannel experience. You may recall the backlash faced by clothing company H&M a few years ago when the retailer used a black child to model a hoodie with the words “coolest monkey in the jungle” printed across the chest. Accusations of racism ensued, along with bad press. A celebrity that partnered with the brand unceremoniously ditched the retailer. The company pulled the product from the market and issued an apology. It’s the type of scenario marketers go to great lengths to avoid. While this example demonstrates the undesirable consequences companies can face for offensive language, it’s hardly an isolated case. The public accused retailer Zara of being insensitive to people who have gluten intolerances based on its shirt that asked, “Are You Gluten Free?” Others have criticized luxury brands for insensitive imagery in their designs. Words are powerful. They can be used to promote harmony and goodwill or create divisiveness among people. Marketers have both ethical and financial reasons to get their language right. To achieve welcoming copy, marketers must avoid blatantly offensive and insensitive language. They must also eliminate signs of less obvious, implicit bias and foster the inclusion of people who have diverse backgrounds. Brands can achieve these goals through the thoughtful usage of language in public-facing product and marketing materials. The public has increasing awareness of such efforts or lack thereof. It is challenging to execute these initiatives when dealing with one language. It becomes even more difficult when multiple languages are involved. Lionbridge can help you to deliver inviting content that resonates with all your audiences. What is Bias and Inclusive Language? Bias is a judgement and leads people towards one option, nationality, or idea that is usually negative or prejudiced. It becomes codified in the language and expressions we use, both consciously as explicit bias and unconsciously as implicit bias. While we have become increasingly aware of bias, its deep roots in our education and language make it difficult to detect. As such, it may be impossible to eliminate unconscious bias entirely despite our best efforts. Still, we must try. Bias and inclusivity are increasingly playing a central role for brands as we continue to address the COVID-19 pandemic and its lasting impact. As companies move their customer and workforce experiences online, the content they create has become the primary medium for interactions. For instance, Statista highlighted a survey from August 2020 that showed online purchases of over-the-counter medicine and household goods grew over 45% compared with pre-COVID trends. Consumers, more than ever, are being exposed to companies’ online content. Inclusive language fosters a sense of belonging. It addresses prejudice and bias by reducing the weight and importance of a description from a person's identification. For instance, we can achieve inclusiveness by referring to the person first and the person’s disability or difference second. Conveying that there is “a person with a learning disability” focuses on the person first, whereas identifying someone as “a slow learner” equates the person with a condition. The same approach applies to people who belong to a religious, national, political, or social group. The emphasis on the human aspect allows for an environment where everyone can feel included and participate freely. Being aware of the existence of bias is an important first step to deal with the issue. Companies can address bias on multiple levels. The use of inclusive language when creating content is one important strategy. Why Should Marketers Focus on Implicit Bias and Inclusive Language? While it’s challenging to create culturally inclusive language and eliminate explicit and implicit bias in translation, it is clearly in a company’s best interest to try. In addition to being the respectable and responsible thing to do, companies can expect such efforts to help them expand their customer base, build greater brand trust and loyalty, elevate their reputation, and ultimately bolster their bottom line. We can point to recent social movements for playing a large role in consumer expectations around inclusion. For instance, rallies and marches led by the Black Lives Matter movement greatly influenced societal mores. Even those who do not actively participate in these types of demonstrations still expect to see and hear advertising messages aimed at a broader demographic range. This expectation was prevalent even before Black Lives Matter protests hit their peak. According to a 2019 Adobe report, 61% of Americans find diversity in advertising important and 38% show a stronger trust for brands that portray diversity. Want more evidence that consumers are paying attention? In 2020, U.S. adults recognized Nike as the top brand for advertising diversity, followed by Coca-Cola, Google, Apple, and Dove, according to Adobe research that was reported on by eMarketer. Consumers outside of the U.S. are also watching brands’ diversity efforts. A 2019 report on retail luxury goods by Mintel found more than half of the buyers from Germany, Italy, France , Spain, China, and the UK felt that luxury brands weren’t portraying enough diversity in advertisements. Diverse markets have huge spending power. Removing barriers between you and your customers will enable them to see themselves in your product and increase the likelihood that they will make a purchase. How Have Ad Campaigns and Other Initiatives That Feature Diversity of Voice, Inclusion and Image Grown? We can look at the actions of some of the biggest global retailers and select service providers to see a conscious shift towards inclusivity that is clearly gaining momentum: Apple and Google are replacing terms like “blacklist” and “whitelist” with more neutral terms like “allow list” and “deny list” to be more inclusive. The Houston Association of REALTORS® and some builders replaced the terms “master bedroom” and “master bathroom” with the terms “primary bedroom” and “primary bathroom." The word “master” has a connection to slavery. Japan Airlines was the first Asian airline to use gender-neutral language on flights and in airports. Instead of addressing passengers as “ladies and gentlemen,” the airline asks for the attention of all passengers. Other international airlines have previously taken similar steps. ASOS, a London-based clothing brand, is implementing nine new initiatives to offset racism. Among their efforts, they are launching a diversity and inclusion strategy, adding Black-owned brands to their offerings, and providing dedicated training (including bias training) for managers and hiring panels. The issue has also captured the attention of ADWEEK, which is encouraging marketers to create more inclusive copy. How Do Some Languages Lend Themselves to Inclusivity More So Than Others? Gender neutrality is one way to make people feel included. The goal is not to remove gender altogether but to reduce the negative impact of some gendered terms and expressions. It is easier to achieve gender neutrality in some languages than others. Non-gendered languages like Finnish, Turkish, Japanese, and some other Asian languages are very easy to neutralize because there are no grammatical genders to contend with. Natural gender languages like English and Chinese are easy to neutralize. Although these languages contain gendered pronouns, nouns are non-gendered. Gendered languages like French, Portuguese, Spanish, Arabic, and Hebrew are difficult to neutralize because of gendered pronouns and nouns. Sentences in these languages will often read awkwardly when translators make efforts to neutralize content. It’s important to take these factors into consideration when preparing content for translation. This will help to prevent issues from emerging during the localization process. What Strategies Can Marketers Put in Place to Avoid Bias and Promote Inclusivity in Their Multilingual Content? When working with global audiences, it’s imperative to create source copy that is inclusive, takes cultural differences into account, and doesn’t contain biased text. This will prevent flawed copy from being translated into other languages and being seen and scrutinized all over social media. Once a transgression happens, brands can face long-term consequences. The many lists of marketing faux pas that are in perpetual existence show how difficult it is to rebound from a mistake. The damage can be prevented when close attention is placed on getting the source content right and effectively localizing that copy. Furthermore, making a mistake in your source—and then having to correct it for all your other markets—is an avoidable expense when digital marketers focus on perfecting the content at the beginning of the process. Nonetheless, it is challenging to avoid bias. That’s because it can often be very nuanced. Content creators might not even be aware of their biases. Detecting bias can be particularly hard since the same words can be regarded as appropriate in one context or non-inclusive in another context. For instance, referring to grown women as “girls” in your ad copy could draw criticism, but saying “hey girl!” to a friend would not likely raise an eyebrow. It is paramount for a content creator to understand these subtleties. Marketers can train their content writers to become more aware of the existence of bias. At Lionbridge, efforts are made to promote inclusivity during translation by adding relevant guidelines in the style guide of each project. Linguists are then tested on the guidelines during onboarding to ensure instructions will be followed. While it is challenging to detect bias, you don’t have to rely solely on only humans to do it. Technology is another tool to turn to. Inclusiveness and bias detectors help to ensure content is compliant, respectful, and equitable. Until recently, it was not possible to rely on automations that detect bias because they were tough to build. However, advancements in Artificial Intelligence (AI) and natural language processing technologies have allowed for the creation of multiple tools that effectively help detect biased language that might be missed by humans because of their nuances. These tools often use machine learning and large corpora of data to assess the intent of the text and enable companies to identify both inappropriate and non-inclusive language. These solutions typically work in one of two ways: Real time suggestions are displayed as the content is written and the person working on the text must decide whether to accept the suggestion. Content governance gate checks allow companies to detect pieces of content that are not compliant with their guidelines. In June 2020, Microsoft Word added a new feature to its grammar checker that is available with a Microsoft 365 subscription. This new feature detects exclusionary language and suggests different wording for it. Google is focused on offering inclusive-language prompts within its G-Suite platform that will suggest alternatives to terms that are identified as ableist or unnecessarily gendered. And Lionbridge now offers an automated solution to detect source content that does not meet guidelines and other standards. We’ll tell you more about this Smairt™ Content offering later in this piece. Bias automation tools are only as good as the data being used to train the tools. However, these tools will become increasingly more important as the focus on inclusivity continues to rise and the technology becomes more sophisticated. How Does Bias and Non-Inclusive Language Detection Work? Tools that detect bias and non-inclusive language leverage numerous technologies. The simplest ones use a list of terms and topics that should not be included in the content. The more sophisticated tools, which use AI and machine learning technologies, infer meaning of the content and determine whether it is inappropriate in the given context. This is accomplished by using neural networks and large language models that help machines understand complex and subtle relationships between different words and phrases in text. Importantly, these detectors identify issues early on to help marketers get content out faster and reduce rework costs downstream. How Does Lionbridge Help Companies Ensure Content is Inclusive and Free of Bias? Lionbridge has always sought to remove explicit bias from content and has tools and solutions in place to help companies eliminate implicit bias as well. Lionbridge helps companies plan and create content with inclusivity at its core. Even before companies start to create content, Lionbridge can assist them in setting up a process that helps prevent bias. Preparing language for localization with the least amount of bias involves careful consideration from the start. Lionbridge’s experts are culturally aware, which results in translated or localized content that resonates with the people of any target market. Not all countries agree on what is considered offensive language. For instance, Italian linguists assert that the localized version of blacklist—a word on Apple’s forbidden list—would not resonate as offensive to an Italian audience because Italy and the United States do not share the same history with respect to slavery. Moreover, the linguists suggest that imposing cultural mores from the U.S. would only serve to alienate an Italian audience. Countries have varying levels of tolerance for inclusive language that must be considered when localizing content. For instance, Scandinavian languages have extensive legal guidelines that support the use and widespread adoption of inclusive language at all levels, whereas other countries, such as Portugal or certain nations in South America, resist such efforts. France’s education ministry announced earlier this year that the use of gender-neutral language will not be permitted in schools. The government agency asserted that gender-neutral language undermines the understanding of French. Lionbridge is an expert at addressing the needs of each target audience to prevent content creation that will come across as inauthentic. Lionbridge continuously strives to reduce implicit bias. As a leader in localization technology, we leverage Machine Translation and AI. We rely on our large corpus of curated data to make AI smart and teach it to use inclusive language. Our use of glossaries and style guides address blatantly offensive words, such as profanities, as well as more subtly coded terms. And our new Smairt™ Content proprietary algorithms will prevent unsuitable language from finding its way into translated content at all. Lionbridge’s Smairt™ Content automation checks content for 120 different language aspects. It highlights issues in your source content that may require corrections before moving to the next stage. If the algorithms detect any flaws in the source content, the text can be stopped from proceeding to localization. That way, companies can fix the problems once—in the source—and avoid spending time and money to correct errors multiple times in all their languages. Companies still have the option of moving flagged content to localization. In some cases, it may make sense to note the issue and analyze it later. These cutting-edge algorithms are now part of the Lionbridge Locaⁱlization Platform™. It addresses each step of the content journey and helps us achieve great accuracy during localization. We devote resources towards research and development to continuously make technological advancements that promote an optimal outcome. Lionbridge’s Language Quality Services (LQS) provide quality assurance evaluations into translated and localized content to ensure products and services resonate in each locale. LQS involves an exacting check of translated material measured against quality benchmarks that include inclusive language when inclusive language is desired. LQS assesses, annotates, and validates content so it can be used to make intelligent systems even smarter. LQS linguists use standard Multidimensional Quality Metrics (MQM), a rigorous assessment framework used to evaluate translated content. The results of the translation are then mapped to data analytics that offer a benchmark for the team to raise the quality of the output. Finalized content is then fed into the Translation Memory (TM), which continues to refine and improve the overall content. How to Spot Non-Inclusive Language Do you really know how to identify bias and non-inclusive language? Do you just “know it when you hear it”? The next time you have a doubt, ask yourself these questions to check your choice of words. ASK: Is this the right word? HOW TO DETERMINE APPROPRIATENESS: Replace the word in question with a simile using the same phrase. EXAMPLE: So easy, even your grandma can use it! Compared with: So easy, even a novice can use it! CONCLUSION: “Grandma” is an offensive word choice in this context. Novice is not an offensive word choice. ASK: Is this the right audience? HOW TO DETERMINE APPROPRIATENESS: Use the phrase in front of a different audience. EXAMPLE: My manager is a “slave driver.” Do you feel comfortable saying this in front of your Caucasian co-workers? What about in front of your Black colleagues Not likely. CONCLUSION: “Slave driver” is an offensive word choice. ASK: Is this a racial/national stereotype? HOW TO DETERMINE APPROPRIATENESS: Replace the race/nationality with another race/nationality in the phrase to test whether or not you should include a race/nationality in your text. EXAMPLE: “Asians” are good at math. Would you say “North Americans” are good at math? Not likely. CONCLUSION: The phrase, “Asians are good at math,” is a racial stereotype. Lionbridge leaders have a lot more to say about implicit bias and inclusivity. They’ve identified trends among Lionbridge’s client base and offer ideas to overcome the barriers that obstruct the creation of inclusive content. Read more in our blog, Reducing Implicit Bias. Get in touch Interested in ensuring that your content is inclusive and free from biased language? Reach out to us. How to Spot Non-Inclusive Language Do you really know how to identify bias and non-inclusive language? Do you just “know it when you hear it”? The next time you have a doubt, ask yourself these questions to check your choice of words. Lionbridge leaders have a lot more to say about implicit bias and inclusivity. They’ve identified trends among Lionbridge’s client base and offer ideas to overcome the barriers that obstruct the creation of inclusive content. Read more in our blog, Reducing Implicit Bias. Get in touch Interested in ensuring that your content is inclusive and free from biased language? Reach out to us.