he Covid-19 pandemic has highlighted the value of the language services sector in a global crisis. It’s also highlighted the right of every person to receive clear, accurate and accessible information as quickly as possible. One way to do that is for the language services sector to incorporate plain-language principles into assessments of source-text quality. That in turn gives us a strong basis for supporting our clients and partners to communicate more clearly.

The coronavirus that causes Covid-19 has affected the entire world, and therefore has generated a global response. Developing and distributing information and countering misinformation has been a critical part of that response. Governments, health agencies and aid organizations have led those efforts, generating vast quantities of new information in an urgent, unpredictable and highly dynamic environment.

Since early in the crisis, linguists have made that information available in many languages. The World Health Organization, as always, published content in the six official languages of the UN. However, UN languages simply don’t cut it in the communities that many aid organizations support. To communicate effectively, aid organizations often need information in under-served local languages. And so the pandemic has highlighted, more than ever before, the value of quality language services.

Fortunately, international agencies and national governments have been supported by community-based organizations throughout the world. Those organizations have made information available to immigrant communities and marginalized language speakers who have limited access to information. For example, Doctors of the World translated NHS Covid-19 guidelines into more than 50 languages.

But effective communication requires more than providing information in multiple languages. When information addresses unfamiliar topics and encourages changed behavior, it must also be clear and unambiguous.

Aimee Ansari, the Translators without Borders (TWB) executive director, thinks that aid organizations will need to dramatically rethink how they interact with people in a crisis. “The pandemic has forced our sector to find new ways to engage with people remotely,” she said. “Opportunities for face-to-face communication will remain limited, so we’ll have to understand the elements of clear communication, and the best ways to achieve them.”

Our national leaders provide some unfortunate examples

Clarity is and will remain important for end-readers and translators. Yet I’ve seen many Covid-19 source texts that are unclear and ambiguous. I’ve been particularly surprised by the convoluted approach taken by some national leaders.

For example, the transcript of the national address that UK Prime Minister Boris Johnson delivered on March 23 includes the following sentence: “Shopping for basic necessities as infrequently as possible; one form of exercise a day, for example, a run, walk, or cycle alone or with members of your household; any medical need to provide care or to help a vulnerable person; and traveling to and from work, but only where this is absolutely necessary and cannot be done from home.”

US President Donald Trump offered some rather ambiguous hope in a national address delivered on March 11 when he said: “If we are vigilant — and we can reduce the chance of infection, which we will — we will significantly impede the transmission of the virus.”

Political pontificating aside, how can language service providers (LSPs) measure clarity in a source text, and how do we know when to encourage the writer to revise it? Such questions aren’t only relevant to linguists working on Covid-19 information. They’re also relevant to the language services sector more broadly and will remain so in the post-Covid-19 world, whatever form that takes.

As a plain-language editor, I think, perhaps unsurprisingly, that plain language is part of the answer.

Clarity and a network of dedicated linguists is key

TWB works at the intersection of language services and humanitarian aid. We coordinate a virtual network of more than 30,000 volunteer linguists, 10,000 of whom have joined this year. On the strength of that network, we support around 90 nonprofit humanitarian organizations to provide information in some 160 languages. We’ve translated more than 1.8 million words of Covid-19 source text since January, working in more than 90 languages. Our dashboard, https://translatorswithoutborders.org/translations-covid-19/, tracks our output as we update it each day.

Passing the million-words milestone offers an opportunity to reflect on our achievements and how we’ve managed them. It’s also an opportunity to consider how we might improve our support to further amplify the impact of our partners.

Many of our nonprofit Covid-19 partners need translations in marginalized target languages that are under-served by most LSPs and by existing language technology. They request challenging turnaround times. They lack the resources to engage professional writers, editors or proofreaders. Commercial LSPs are generally beyond their reach. In any case, few LSPs are agile enough to meet the huge and unpredictable language services requirements that emerge in a crisis. Fewer still can respond quickly to one that requires such a vast multilingual response.

So we’re proud of and grateful to the many linguists who responded to our call for support in the Covid-19 response. Many of those people have stood with us at the intersection between language services and humanitarian aid for many years. They’ve answered the call before. Others have recently joined our community, inspired by their belief that everyone has a right to information in a language they understand. All of the linguists that support us have to work fast, often in traumatic circumstances, and with distressing source texts. Many of those linguists are MultiLingual readers. If you’re one of them, thank you. You’re the unsung heroes in this response.

Well before this pandemic, we knew that providing clear, consistent and accurate information in marginalized languages saves lives. We also knew that to provide that information, linguists need clear, consistent and accurate source texts. Unfortunately, the urgency inherent in a crisis means that such source texts are often elusive.

Typically, the information that communities need in a crisis isn’t immediately obvious. Other times the need is so broad it’s hard to know where to start. Local politics can influence the messaging and the timing of its release. Educated guesses are sometimes the most impactful starting point from which to provide information. Eventually, some certainty emerges, but it shifts by the day, or sometimes by the hour. Any delay in translating and distributing information threatens lives, so careful crafting and editing of source texts is seldom a priority. It’s little wonder that source-text quality is less than linguists might like.

Our work in the Covid-19 crisis has highlighted this issue with source-text quality. It’s caused us to question the best way to address it, and who’s best placed to drive the change. Increasingly, we think that the answers are plain-language assessments driven by the whole language services sector.

Garbage in, garbage out

At a time when the world has desperately needed clear, accurate and accessible messaging, we’ve seen the opposite. Many of the Covid-19 source texts we’ve translated are unclear or ambiguous. Translators generally commit to producing target texts that reflect the style and tone of the source text. To query it or change it requires consultation with the author, generally leading to delays in the translation.

Quality standards and frameworks such as ISO 17100:2015 and TAUS’s Dynamic Quality Framework establish processes and metrics to measure target-text quality. But interestingly, these standards do not take into consideration the impact of poor source-text quality on target-text quality. That’s surprising, given the longstanding “garbage in, garbage out” principle that’s so beloved in the technology sector.

Of course, it’s not always appropriate or even possible to measure source-text quality. It would be inappropriate for translators to query the quality of creative texts such as poetry or novels. Or a peer-reviewed research paper.

In many cases, there’s little to gain from LSPs and translators assessing or querying the quality of a source text. But sometimes there’s a lot to gain. An unclear or ambiguous source text invariably reduces translation efficiency. It leads to an unclear and ambiguous target text and possibly a dissatisfied client. At best, it generates a protracted translation process, excessive costs and a blown budget. But how can we assess the clarity of a source document and interpret the results?

To measure clarity, measure plain-language elements

Use the active voice

The active voice is clearer than the passive voice because it specifies who is responsible for an action. That’s an important element of accountability in aid work. It also makes the verb more dominant, which focuses the reader’s attention on the relevant action. The active voice typically uses fewer words too, which can make a difference to both reading effort and translation effort.

Despite those obvious advantages, the passive voice was an issue in eight of the ten documents we assessed.

Use short sentences

Sentence length is one of the most significant influences on reader comprehension. A general plain-language recommendation is to limit sentences to 20 words or fewer. Even though most readers can comprehend longer sentences, it requires more reading effort.

Long sentences were an issue in eight of the ten documents we reviewed. They used multiple complex sentences with up to 50 words. The opening sentence of one document contains 42 words, several clauses, various tenses and multiple verbs. It also addresses several ideas: “In several countries and on global level, persons with disabilities have spoken out that government information is not being shared in accessible formats or that measures are not made to compensate for reduction in support services that persons with disabilities depend on.”

A plain-language alternative with 32 words in three simple sentences is: “Many persons with disabilities have said that they are not adequately supported by their government. Some governments don’t provide information in accessible formats. Others don’t compensate enough for reduced disability support services.”

Use key terms correctly

A clear source text prevents translators having to assume or infer any information. That’s important because incorrect assumptions and inferences can cause misunderstandings and rumors. Misused or unclear terms also undermine an organization’s credibility as a trusted information source.

Three of the ten documents that we reviewed use a key term incorrectly. For example, two of them assert that Covid-19 is a type of coronavirus. To avoid confusion, they need to convey that Covid-19 is the disease caused by a new type of coronavirus.

Several of the other documents use “coronavirus” and “Covid-19″ interchangeably. Errors or oversights like this confuse readers and require extra reading effort to untangle the true meaning.

Favor familiar terms and use them consistently

A key plain-language principle is to use one term to describe one concept. In the documents we reviewed, the majority used multiple terms to refer to similar concepts.

For example, some documents use both “separate” and “isolate” to refer to the same concept. Others use “the disease” and “the illness” to mean the same thing. One document uses “germ” and “virus” to mean the same thing.

All of these pairs are legitimate synonyms. In some contexts, readers might even view such variation as a sign of creativity. But in an urgent humanitarian response, we should encourage writers to always choose clarity over creativity.

Online corpora are valuable tools to determine the term that readers are likely to be most familiar with. Corpora like the Corpus of Contemporary American English, News on the Web and iWeb contain valuable word-frequency data. The data serves as a useful proxy for comparing the likely familiarity of alternative terms.

For example, we recently advised one of our partners to use “caring for” in preference to “attending to” when referring to Covid-19 patients. We based our recommendation on word-frequency research. It revealed that “caring for” occurred on average five times more frequently than “attending to,” as shown in Figure 1.

Use concise wording

Assessing source-text quality offers benefits to writers and translators