An accessible audio tour for non-native speakers: how to write the script

An accessible audio tour for non-native speakers starts with the source text, not the translation. Someone with limited language skills hears a long sentence full of subordinate clauses and loses the thread, and a translation of that same sentence is just as hard. Write your source in plain language and every language you make from it will be plain too. Here’s how to do that: how to write, how to translate and which voice to choose.

Why the language level of your source text decides everything

Automatic translation does exactly what you ask. Put in a forty-word sentence with three subordinate clauses and a technical term, and out comes a forty-word sentence with three subordinate clauses and a technical term in Polish, Ukrainian or Arabic. The technology makes your story available in more than 70 languages, but the readability comes from your own text.

It works the other way round too. Short sentences, everyday words and one idea per clip give you a text in every language that a visitor with limited language skills can follow. So you write it well once and benefit from it in every language.

There’s another reason to start simple. The same text works for visitors with low literacy in your own language, for children and for anyone walking through a busy gallery with headphones on and only half their attention to spare. Plain language isn’t a separate version for a separate audience; it’s the version that works for most people.

Which language level should you use?

For visitors with a limited command of your language or the target language, A2 to B1 is a workable starting point. A2 means everyday words, short sentences and no figurative language. B1 allows a little more room, including more abstract concepts, as long as you explain them.

Choose one level and stick to it throughout the tour. A clip at A2 followed by one at C1 is more confusing than a tour that’s slightly too simple everywhere.

If you want to be sure where you stand, have someone from the target group read the text. A language-level check on paper tells you less than a visitor who asks halfway through your clip what you mean.

How to write a clip in plain language

Keep clips short: forty to ninety seconds of spoken text, roughly a hundred to two hundred words. Anyone who loses the thread can pick it up again at the next object.

A few practical rules:

• One idea per sentence. Split every sentence with "because", "while" or "which" into two sentences.
• Sentences of ten to twelve words on average. Vary the length, but don’t let a sentence run to thirty.
• Active sentences with a clear person doing the action. "The painter made this in his studio" instead of "this work was produced in the artist’s studio".
• Everyday words over technical terms. If you do use a technical term, explain it in plain words in the next sentence and then stick to that same word.
• No figurative language, irony or puns. For some of your visitors, "the heart of the city" literally means an organ.
• Name the object the visitor is looking at, straight away in the first sentence. "You are now standing in front of the large map on the left-hand wall."
• Go easy on numbers and dates. One date sticks; five don’t.

What you give up is nuance. What you get back is a visitor who follows the story to the end. Save the extra depth for a second, optional clip or for the text on screen.

Which languages do you choose, and what do you do with the translation?

Start with your own visitors. Which languages do you hear at the front desk? Which groups come through a school, a community organisation or a civic integration programme? Four languages that people actually use are worth more than twenty that nobody taps. After that, your tour statistics show you which languages to add.

An automatic translation is a good starting point and rarely the finished version. There are three things you’ll always want to check:

1. Proper names and technical terms. Names of rooms, objects, people and places should either stay untranslated or get one fixed translation. Wandrz stays Wandrz in every language.
2. Form of address and tone. Every language has its own conventions for formal and informal address. What sounds friendly in Dutch may come across as too familiar in German.
3. The language level after translation. Machine translation sometimes picks a more formal or longer word than your source. Have a native speaker, a volunteer or a teacher from a language school read the translation, then change it in the platform so the correction stays.

Don’t write a separate text for each language. One clear source that you polish lightly in each language stays consistent and is easy to maintain.

Which voice should you choose for non-native speakers?

For most tours, a natural AI voice in the target language is the practical choice: you have it straight away in every language you’ve chosen, without a recording studio or scheduling. What matters for accessibility is not so much which voice, but how you set it up.

• Choose a native voice for each language. A Dutch voice reading Polish costs the listener extra effort with every word.
• Go a little slower than you’re used to. Someone listening in a second language needs more time to take it in. A calm pace is the cheapest accessibility gain there is.
• Build pauses into the text. A full stop and a blank line between two thoughts give the listener a moment to catch up. Write those pauses into your clip on purpose.
• Neutral, clear articulation over drama. A theatrical delivery with many changes in intonation is harder for a beginner to follow.
• Listen back to every clip yourself, in every language. Numbers, abbreviations and proper names are where things go wrong. If something sounds wrong, adjust the spelling in the text until it’s right.

A human narrator has a warmth that an AI voice doesn’t always match. For a tour in six or ten languages that you also want to keep up to date, that usually doesn’t outweigh the time and cost. A middle way: have your main language recorded by a narrator and use AI voices for the rest.

What else you need besides the audio

Audio helps visitors who find reading hard. Text on screen helps visitors who read a language better than they hear it. Offer both, with the same content, so visitors can read along as they listen.

That also means:

• An image with every clip, so visitors can see what it’s about even when a word escapes them.
• A language choice that’s clear right at the start, with each language shown in its own script and under its own name.
• GPS triggers or clear numbers, so nobody has to read to know which clip plays when.
• A short introduction in plain language on how the tour works. That’s another clip you write at A2.

In short

This is how you make an accessible audio tour for non-native speakers: write your source text at A2 to B1 with short, active sentences and one idea per clip, choose the languages your visitors really speak, have the automatic translation checked for names, tone and language level, and use a native voice at a calm pace. The technology takes care of the languages; the language level is up to you.

With Wandrz you build your tour right in the browser, translate it into more than 70 languages and adjust any translation where needed, all in your own branding. Want to know how to make your existing texts work for visitors who speak another language? Get in touch for a free consultation.