A few weeks ago I got a very interesting email from Felix Peckitt, a student, poet, programmer and fellow ticcer. He got in touch to tell me about some investigations he’d been carrying out using the thousands of my vocal tics. I was fascinated by his explorations and invited him to write about his work in a guest post. So it’s with great pleasure that I now hand over to Felix to tell you more:
How happy or sad is the language in vocal tics? Is it always negative, inappropriate and coprolalic (swearing), or are there sometimes positive and funny moments? We often remember the most negative tics as they have a bigger social impact, but using the Touretteshero tic database of around 6500 tics and some statistics we can get a better picture of what’s going on
It would take a long time (1) to rate all the tics as being positive or negative by hand, and would be subjective for each person doing the rating. Luckily for us, there’s a technique called “sentiment analysis” which can do the heavy lifting. Firstly, each word in a dictionary is rated by volunteers in varying degrees of positivity or negativity. For instance, “amazing” would score highly, and “disgust” would receive a low score. Then, a program analyses sentence structures that affect the rating, for instance, “not good” would receive a low score.
Sentiment analysis was originally developed to look at movie reviews to find out if the film was rated positively or negatively without having to read the article – useful for websites like Rotten Tomatoes and IMDB (2). Since then, sentiment analysis has been used to look at social media, plot structure in novels, (3) and in many other areas
The most positive tic with a score of 8 was: “Amazing grace, how sweet the sound of little baby Jesus eating fish fingers.” The most negative tic with a score of -13 was: “Fuck it, shit it, block it, piss it.” However, over 60% of tics were not so extreme, scoring between -1 and 0. Notice how the distribution is symmetrical, and that there are two smaller peaks at around -4 and 2, before tailing off again. This goes to show that for Touretteshero at least, vocal tics aren’t so negative all of the time.
Michael Eckbert, for his help scraping the Touretteshero website.
Matthew Jockers, for the use and creation of the Syuzhet package.
Touretteshero, for the use of her tics.
My brother Oscar, who is an inspiration to me always.
How do tics compare to spoken language? How does the sentiment of tics change over time? How can we predict the sentiment of the next tic, i.e. what is the probability distribution of tic sentiment? Watch this space!
1) at 5 seconds per rating, about 9 hours
2) Turney, Peter (2002). “Thumbs Up or Thumbs Down? Semantic Orientation Applied to Unsupervised Classification of Reviews”. Proceedings of the Association for Computational Linguistics. pp. 417–424. Pang, Bo; Lee, Lillian; Vaithyanathan, Shivakumar (2002). “Thumbs up? Sentiment Classification using Machine Learning Techniques“. Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP). pp. 79–86.
I love reflecting on the language of tics, but analysing the sentiment isn’t something I’d ever considered before. It’s offered me a brand new perspective on something I experience every day. Thanks Felix for this wonderful, amazing, glorious, awesome, miraculous, splendid, breath-taking post.