ANALYSIS
How a pandemic changed our writing tone?
Understanding Grammarly’s tone predictor.
A while ago, I saw a meme where it was depicted how a boy spends the better part of their relationship explaining to the girl, “I didn’t mean that”. I don’t know about the stereotype, but that meme was for me. Humans are prone to miscommunication and can misinterpret if you fail to use the appropriate phrasing.
Chatting is worse. Hitting a perfect note is both necessary and hard if you want your message to get across effectively. Especially, in the absence of facial expressions and the pitch of the voice — while writing — messages pose a threat of miscommunicating your intentions. That’s when the “tone” of the written text comes into play.
To sound friendly (but not informal) or concerned (but not enquiring), choosing your words wisely is key to delivering your message effectively.
People with effective communication distinguish themselves from others with
- proper word usage,
- punctuation usage,
- conciseness of sentences,
- delivery of the message, etc.
This has become a more prominent part of the corporate culture. During my Masters in Business, we had a dedicated subject on Business Communication in which we learned the basic nuances of effective communication in corporate environment.
Fortunately, Grammarly’s tone detector feature helps you exactly with that — suggesting the correct tone of your writing. This feature of the grammar checking platform claims to identify the subtle contextual hints reflecting a range of tempers.
How Tone Predictor works
As per Grammarly's website, the tone predictor feature analyzes the following aspects to analyze the tone of your written text:
- Word Choice
- Phrasing
- Capitalization
- Punctuation
Taking these factors into account, it classifies the message in forty distinct tones, for instance — informative, concerned, formal, informal, regretful, excited, and many more.
For instance, a curious message would be one with the usage of words like “I would like to know”, a regretful message would be the one with a usage of apologetic words, an angry message would be with… well I guess, you got the gist.
Note: The tone detector needs at least 120 characters to be activated and it is available only in premium versions and doesn’t work on some sites like Zoho, Salesforce, etc.
When there’s tension in the Air
With this Machine Learning-based technology, they can analyze the tone of the mass or the tension in the air, so to speak. That’s what they did when the Covid-19 pandemic struck (depicted in the graph above).
They analyzed a striking (but obvious) deviation in two of the sentiments: Informative and Optimism. While within America, Grammarly observed a 75% drop in the usage of an optimistic tone in the message. At the same time, a rise of about 70% in the informative tone of the message was observed in early April when there was little information on the pandemic.
Though they have only revealed data until the outbreak converted into a pandemic. Though it is easy to anticipate that there must have been a more marked deviation in the tone of people when the imposed lockdown started getting to them — especially with the rising death tolls and riots following George Floyd’s death.
For instance, a disapproving and angry tone must have observed a high rise (among a certain group) with the rise of conspiracy theories and a concerned and disheartening tone must have dominated the tone of people losing jobs in the last couple of months.
Opportunities:
Talking about the technology alone, these data insights wherein poses a threat of privacy breach for some, yet present a lucrative opportunity for others. For instance, you can predict riots when they are still at a nascent stage. Even coupling the tone predictor with some other product-specific keywords can give real-time insights into how the product is being perceived.
But in its current form, the Tone Predictors can ease our communication, establish a universal understanding of tone, and make us more civil. Personally, I would give anything not to be slaughtered by a grammar-nazi.
The phrase, “I think we will be able to help you” is less confident than “we can help you.” This is what Grammarly’s tone detector does, more or less. And if you are still struggling to sow your foot on the grammatical ground, these little nuances would equip you with the necessary sophistication.
Though, these tone predictors explain that there is still a lot of room to be covered in terms of Big Data and machine learning opportunities. But with advancement, there could be some probable collateral damages as well.
For instance, such technology in the hands of a political party can provide them an unprecedented insight into the lives of voters. Another criticism is this technology cannot take into account a specific context or who is sending it. People do sound different when they say things verbally. So do we need such uniformity in our day-to-day conversations?
I am talking in the thin air, but from where we are currently standing I feel grateful for such technologies to exist and make our communication, and understanding of human reactions, better.
Since you are here:
Thanks for reading!
