WhatsApp voice notes and sign language: why text is still the bridge

Articles about WhatsApp accessibility, ours included, tend to treat deafness as a volume problem: the sound does not get through, so let us turn sound into text. That holds for a large proportion of the people concerned. It does not hold for all of them.

For a Deaf signer whose first language is British Sign Language, a voice note transcribed into English does not arrive in their language. It arrives in a second language. That is a distinction worth a transcription service saying out loud rather than quietly skipping.

BSL is a language, not English with hands

This is the most widespread misunderstanding among hearing people: imagining sign language as English transposed into gestures, word for word.

BSL has its own grammar, its own syntax and its own vocabulary. The order of elements follows a spatial logic with no equivalent in English. It was recognised as a language in its own right by the British Sign Language Act 2022.

The consequence is direct: for a Deaf signer, reading written English is not reading their own language. It is reading a learned language, often acquired in difficult educational circumstances, and fluency varies enormously from person to person.

What that changes about transcription

It would be dishonest to conclude that transcription is pointless. It helps enormously. But it does not help the same way for everyone:

• For someone who became deaf or hard of hearing and whose first language is English, text removes most of the barrier. This is the most common case by number.
• For a Deaf signer, text is a bridge. A real, crossable bridge, but one that takes effort written English does not ask of a native reader.

What follows from that distinction is very practical: the quality of the writing matters more than people assume. A three-minute voice note returned as a dense block of prose does not have the same usefulness as a short, structured summary with the important information first.

That is one of the reasons Abrège 🍿 returns a summary by default rather than a word-for-word transcript: short structured text reads better than a transcribed stream of speech, for almost everyone, and particularly for anyone reading in a second language.

Writing to be read

The rules that make a message readable for a signer are the same rules that make it readable for everyone. There is no trade-off to make.

• One idea per sentence. Nested clauses are expensive for every reader.
• Important information first. The appointment, the time, the decision. Not at the end of a paragraph.
• Few figures of speech. Idioms are what hold out longest against a second language.
• Nothing left implied. What is not written will not be guessed, and that is not a comprehension failure.
• Lists when there are several items. Text that can be scanned beats text that has to be held in memory.

These rules apply to written messages, and they apply to how a transcription tool ought to return a voice note.

Sign language video has the same flaws as a voice note

An honest observation, which applies to us as much as anyone: when a signer sends a BSL video, that video raises exactly the same practical problems as the voice notes we spend our time writing about.

It cannot be skimmed. It takes the time it takes. It cannot be read discreetly in a meeting. It cannot be searched by keyword three weeks later.

The symmetry is worth noting, because it is a reminder that the problem with voice notes is not audio as such. It is the linear format: unskimmable, unsearchable. BSL video shares it.

In practice, many mixed conversations work by combining the two: video for the live exchange, text for whatever needs keeping, finding again or re-reading.

In a mixed group

Groups bringing together hearing people, hard of hearing people and signers are the trickiest case, and the most common one in families.

A few habits that cost little and change a lot:

• Follow an important voice note with two lines of text. Not the whole transcript, just the essentials.
• Never answer a written question with a voice note. It is the single most reliably excluding move.
• Accept that people reply in their own format. A group that tolerates text, voice and video without ranking them works better than one that imposes the majority's format.

Our piece on the accessible family WhatsApp group works through these rules in detail.

What a transcription tool can and cannot do

A service like ours converts audio into written English. It does not translate into BSL, and it would be dubious to imply otherwise: translating into a sign language is a profession, practised by interpreters, and no consumer tool stands in for that today.

What a transcription tool does do, and it has real value: it removes the dependence on a hearing third party to find out what a voice note contains. No longer having to ask someone to summarise a message addressed to you is a form of independence that counts, even when the destination language is not your first.

It is a bridge. Saying it that way is more accurate than claiming total accessibility.

Going further

Deaf-led organisations are the best source on these questions, well ahead of software companies. The British Deaf Association and RNID publish on access to information and on Deaf people's rights.

For the wider WhatsApp picture, our guide to WhatsApp for deaf and hard of hearing people covers the whole subject.