Company

Email tone detector: helpful tool or cultural imperialism?

This Sums Her Up PerfectlyI’m always interested to see how automation and artificial intelligence are being deployed to help people write more clearly. The latest innovation to catch my eye is a ‘tone detector’ from the popular writing assistant software developer Grammarly.

At the time of writing, the beta version of the tone detector is available as part of Grammarly’s Chrome browser extension and can be used on email sites such as Gmail and Yahoo. According to their website, the tone detector will text you how your message is likely to sound to the reader ‘by analyzing your word choice, phrasing, punctuation, and even capitalization’.

Grammarly says it can detect 40 different tones, such as ‘confident’, ‘optimistic’, ‘worried’ or ‘sceptical’. If the tone it detects isn’t the one you are aiming for (for example, your idea of confidence is its idea of aggression) you can choose different terms based on the software’s suggestions.

I love using technology in my work as a professional communicator, whether it’s the spell checker in Word to eliminate basic typos, or more sophisticated tools like PerfectIt to review the internal consistency of a piece of writing. And while I probably wouldn’t use a tone detector for my own emails, I can see how it might provide useful feedback in certain circumstances, particularly for an important work email to a prospective client or manager where striking the right tone can be crucial to the outcome.

Technology has its limitations of course, particularly when it comes to more nuanced human interactions. Sometimes the choice of tone can be deliberate, while sometimes it might relate a range of different cultural and sociological factors.

Robotics engineer and tech fellow at the AI Now Institute, Deborah Raji, worries that the concept of a ‘dominant tone’ calls into question the validity of less common methods of communication. Guardian Australia quotes Raji as saying that ‘whatever they’re using to train their models … automatically kind of supersedes any other dialect or smaller group that speaks differently, and that dialect ends up being branded as the wrong thing.’

Indeed, if everyone was to rewrite their emails according to a set of algorithms developed by small team of linguists and software engineers, then much of the richness of our communication would be lost. Language has a way of resisting homogenisation though, and new technologies often result in more not less diversity in our communications in ways that are impossible to predict.

So while tools like a tone detector can be useful, nothing beats a final read through of your email before you hit sent to check that it says what you want it to, and the tone fits the purpose.

Image credit: A. Strakey This Sums Her Up Perfectly via photopin (licence)

See also: