Tolkie makes the complex texts of the government a lot easier to understand

By Youri van Heumen

While civil services love complex language and literary sentences, low-literate people are quite the opposite. In total, more than two and a half million people in the Netherlands have difficulty understanding a government text. Such an official text often contains necessary information that is crucial for the citizen.

“It’s a great shame when the government does not clearly convey the information to the people for whom it was originally intended”, say Laurens van den Berg and Jeroen van de Nieuwenhof, founders of Tolkie. They devised a total package of solutions to make texts much more understandable. In this episode of start-up of the day, Laurens and Jeroen tell more about their tool.

How did you come up with this idea?

Laurens: “The idea initially originated from journalism. We saw that the number of people who understand a journalistic text is very low. We are talking about at least 2.5 million people. These people can read, but not well enough to be able to understand a lot of texts. Journalists do try to explain complicated matters in an understandable way. That is our profession. But for people who have difficulty reading and writing, that is not enough.

The longer we wrestled with this problem, the more we got the idea that this problem is not limited to journalistic texts. Also think of texts from housing corporations, municipal news texts, and important information from the government. We want to change that by offering a digital version of the language buddy. This makes texts easier to understand because difficult words are explained differently. You can think of our system providing synonyms or example sentences, showing images or reading difficult words aloud or offering them for translation.”

How did you identify the need for such help?

Laurens: “We wanted to make sure that our solution would work as intended. That’s why we immediately started a market survey to map out the needs. The people who pay us are the organizations with a website, while the users are the low-literate people themselves. For example, we found out what the low-literate people had difficulty with, in particular difficult words and long and complex sentences in texts of, for example, a municipality. We came to the conclusion that there is a great demand for an automated solution from both organizations and the low-literate themselves to make texts more accessible and understandable.”

Towards a digitally literate society: elderly use Google for the first time in their lives – Innovation Origins

Only 49 percent of the Hungarian population has basic digital skills, compared to 56 percent in the EU. The Digital Success Programme is about to change that.

Why should organizations pay for this?

Jeroen: “Because they can also get a lot of value from our tool themselves. Organizations are presented with statistics and analyses about its use. This makes it clear which words are considered complex. Or certain sentences that are exceptionally often translated into another language, which says a lot about the users of the tool.”

What challenges have you encountered?

Jeroen: “One of the challenges is that our target group, the low-literate people, are quite difficult to reach. It is not the case that you can see from the outside whether someone is low-literate or not. In the end, we found several language groups with which we have been in touch a lot to test the tool. As time went by, we built a network of people who also like to stay in touch with us to improve our tool constantly.”

“Companies need to use their data more intelligently” – Innovation Origins

Every day, Innovation Origins puts a start-up in the spotlight. Today it is the Belgian company Agilytic.

Laurens: “Another challenge is making the current version even smarter. The definitions and synonyms we serve are still handwritten or checked. We did this because too often we still see that definitions from the dictionary are even more complex than the word itself. We are now investigating the suitability of new language models to speed up this process. Maintaining quality is essential in this.”

What do you do differently from the competitors?

Jeroen: “The answer is short: there are no people or organizations that do the same as we do. The only thing that comes close is the read-aloud function on websites. But that is also only a small part of our total package. There are a lot of drawbacks to that read-aloud function, for example, we don’t know if people are waiting for a voice to read the entire web page instead of the sentences that are an obstacle for people, and who is to say that people actually understand it? For some words, an image or an example sentence might simply be better.”

What are your goals for the future?

Laurens: “Over the past six months, we have been busy optimizing the tool. We want to bring it to the market as soon as possible. It is not that we are not active anywhere yet, because we are already running on the website of Omroep Tilburg. Nevertheless, cooperation with other organizations remains a key point within our company. Only in this way can we provide better and better help to the people who need it.”