Psychologist logo
Twitter study
Emotion

Twitter study confirms the power of “affect labelling”

Emotions are calmed by putting them into words.

09 January 2019

By Christian Jarrett

You might imagine – as prior research suggests many people do – that putting your feelings into words will only intensify them. In fact, many laboratory studies have found the opposite to be true. Stating out loud, or writing down, what you are feeling – a process that psychologists call “affect labelling” – seems to down-regulate emotions, diminishing their intensity.

Now an intriguing study has explored this phenomenon outside of the lab, analysing over a billion tweets to find examples of when people used a tweet to put their emotional state into words. From analysing the emotional language used in preceding and subsequent tweets, Rui Fan and his colleagues were able to see how the act of affect labelling influenced the course of an emotional state. “We found that, for a majority of individuals, emotional intensity decreased rapidly after their explicit expression in an ‘I feel’ statement,” the researchers write in their paper in Nature Human Behaviour.

Fan and his team identified over 42,000 English language tweets in which an individual had stated they were feeling a positive emotion (for example, writing “I feel happy” or “I feel awesome”) and another 67,000-plus in which an individual had stated they were experiencing a negative emotion (for example, writing “I feel sad” or “I feel terrible”).

Next, they used an established algorithm to analyse the emotional language content of any tweets written in the 6 hours prior to the affect labelling tweet, and 6 hours afterwards (the algorithm uses over 7,000 commonly used English words previously scored by human raters in terms of their relative positivity or negativity). The emotional language scores provided by the algorithm allowed the researchers to chart the intensity of the Twitter users’ emotions prior to and after they had explicitly labelled how they were feeling.

The picture that emerged was that both positive and negative emotions began ramping up prior to an act of affect labelling (captured in the “I feel …” tweet) and then rapidly calmed afterwards. On average, the positive emotional experiences were longer lasting than negative (94 versus 85 minutes).

Also, the ramping up and subsequent calming of positive emotions was symmetrical around the peak that coincided with the “I feel …” affect labelling tweet. In contrast, the build-up of negative emotions was longer and more gradual prior to the relevant “I feel …” tweet, followed by a faster decline to baseline levels.

Comparing between the genders, there was a suggestion that the calming power of affect labelling was more striking for women than men, especially for negative emotions.

The researchers acknowledged their approach has its drawbacks – for instance, to what extent are people’s emotional expressions on social media a performance rather than a true reading of their emotional state? It’s possible too that the findings would vary in other languages or cultures.

Also, while analysing vast quantities of data gathered from social media offers many advantages (such as large sample sizes and real-life data), there is a loss of experimental control, and experts often disagree about the meaning of the data and how best to interpret it (see here for an example in a different context). It’s notable that in the current study, there was no comparison with how positive and negative emotional states unfold in the absence of affect labelling (i.e. what, in an experimental set-up, would be the control condition).

These issues aside, in an accompanying commentary on the new research, Matthew Lieberman – a leading researcher on affect labelling – describes the findings as “remarkable”. He adds: “The authors took a creative approach to studying affect labelling out in the real world and produced some of the strongest and most comprehensive data in support of the role of affect labelling in dampening affective intensity.”

All of which raises an obvious but relatively unexplored question – why does putting our feelings into words lead to a dampening of those feelings? Hopefully, future studies will address the relevant mechanisms, which Lieberman (in a recent review) suggests could include a distracting effect, a reduction of uncertainty, and/or something to do with “symbolic conversion” – the distance and closure created by translating emotions into words.