The Great British Bake-Off Conversation
This year, we’ve gone even more technical than a technical challenge for The Great British Bake Off. Editors at OxfordDictionaries.com have been analysing the positive and negative language used by the judges in each episode so far, and we’re going to use our results to try to make our own predictions about what will happen next. Full details of the scoring are below, so let’s see: how have the judges been trying to sweeten their blows and spice things up with their language? Does what the judges say reflect the results of the show, or are they just trying to take the biscuit? The results of our analysis are visible below.
Our predictions didn’t work out for Week 7, but we were spot on for Week 8… have we got the right recipe for predicting the eventual winner?
We devised a scoring system that would allow us to give numerical values to the language the judges use with each contestant. If the judge makes a flatly positive statement—such as Mary Berry’s “It’s got a nice dome” of Marie’s Madeira cake in Episode 1—we give 2 points. For a qualified positive statement—as when Paul Hollywood describes Mat’s Spanish Windtorte as “quite neat” in Episode 4—we give 1 point. If a positive statement is emphasized, we give one extra point for each emphasizer used, so Ian gets 3 points for his “truly magnificent” bread sculpture in Episode 3 (2 for magnificent, then 1 for the emphasizer truly) while Paul gets 5 for his bread sculpture which is “very, very good indeed” (2 for good, and 1 each for the first very, second very, and indeed).
On the flipside of things, for a baldly negative statement, a contestant loses 2—so Ian lost two points when the base of his frangipane tart was described as “not crispy” in Episode 6. For a qualified negative statement—as when Mary said that Alvin’s biscotti were “a little bit soft in the middle” in Episode 2—we take away 1 point. Each emphasizer on a negative statement causes 1 extra point to be lost, so Ugné lost a total of 4 points when Mary called her sugar-free cake “very, very close textured” in Episode 5.
We total the scores for each contestant for every item they bake and use these figures to work out who—based on the language used by the judges—we’d expect to be Star Baker and who we’d expect to be eliminated in each episode. As it turns out, the correlations are usually pretty close. The only week where our expectations didn’t match the actual result was Week 4, where (from our language analysis) we’d have expected Tamal to be Star Baker and Alvin to have been sent home.
The figures from our scoring also form the basis of our predictions. Using this data, we calculated the running score for each contestant from their overall total from all six episodes. We also worked out who had scored the highest (and the lowest) for the most individual rounds over the series so far. Because Mary and Paul don’t always agree, we figured out who had received the highest score from Mary in each episode, and who had received the highest score from Paul. For example, in the most recent episode Flora received the most points from Mary, but an aggregate score of zero from Paul.
From this, we were able to evaluate how often each contestant was in Mary’s or Paul’s top three bakers. We also looked at how often a contestant had been a judge’s least favourite baker. These facts together gave us an overall picture of which contestants are the real cherry on the cake, and whose goose is looking cooked. But how closely does the judges’ language really predict the future?