皇家华人

If you鈥檝e been on any social media platform in the past two weeks, you鈥檝e probably seen a grid of green, yellow and black squares. This is the latest pandemic phenomenon called Wordle 鈥� a free online game that gives users a new word puzzle each day. It was created by Josh Wardle for his crossword-loving partner. As of January 10, the game has 2.7 million players.

In Wordle, players have six tries to guess a target five-letter word. Every time they make a guess, they are told which letters in their guess are in the word and in the correct position (green), and which letters are in it but in a different position (yellow). It鈥檚 sort of like the boardgame Mastermind but with a key difference. In Mastermind, all six colours were equally likely to appear in the target. In Wordle, because guesses and targets all have to be real words, some letters are more likely to appear, making some guesses better than others.

This leads to a question that I鈥檝e seen people discussing at length online: what is the best first word to guess?

For now, let鈥檚 define the 鈥渂est first guess鈥� as the one that is most likely to share the most letters with the target word. What we need to know is: how common are each of the 26 letters in five-letter English words. And not just in any five-letter words, those that have a chance of showing up as targets.

Obscure words like 鈥渘isus鈥� (a mental or physical effort to attain an end) or 鈥渨inze鈥� (a connection between different levels of a mine) need not apply.

I found a recent study that looked at over 60,000 English words and how well-known they were. This sort of statistic is interesting for language researchers like me because it captures something about how easily a word can be processed: on average, more commonly known words are read faster.

I took all five-letter words that were known by at least 50% of those studied (if you knew 鈥渘isus鈥� or 鈥渨inze鈥� 鈥� I certainly didn鈥檛 鈥� you share that feat with only 7% of the sample). Then I counted the number of times each letter appeared at least once in a word.

The most common letter was 鈥渆鈥�, appearing in 46% of words. This is a well-known pattern that applies to the English language in general. A notable exception is George Perec鈥檚 novel A Void, which was purposefully written without the letter 鈥渆鈥�. This pattern was even used by Sherlock Holmes in The Adventure of the Dancing Men to decode a cipher made up of dancing stick figures by reasoning that the most common symbol would be 鈥渆鈥�.

One reason that 鈥渆鈥� is so common was the advent of silent e鈥檚 at the end of words in the 16th century, used to signal something about the preceding sounds. For example, 鈥渢one鈥� is pronounced differently than 鈥渢on鈥�.

The next most common letters were: 鈥渁鈥� (39%), 鈥渞鈥� (34%), 鈥渙鈥� (29%), and 鈥渋鈥� and 鈥渟鈥� tied for fifth (28%). Out of these six letters, one word immediately 鈥渁rose鈥� as the best option! Want an especially bad first guess? Try 鈥渨hump鈥� (a dull thudding sound). That is just about the worst by this metric.

But while 鈥渁rose鈥� is most likely to get you letters in the target, they may not be in the correct positions.

The most frequent letters, by position, in common five-letter words.

If we want a word that is most likely to get letters in their correct positions, the best option is 鈥渟amey鈥� (monotonous, repetitive, unvaried). But let鈥檚 not stop there. If we put these approaches together into one final score, we get a word that looks eyrie-ly familiar: 鈥渟oare鈥� (a young hawk) 鈥� 鈥渁rose鈥� but in a more strategic order.

One final thing to note. While writing this article I found that people had gotten into the source code for the Wordle website and found the actual list of words that can appear as targets. I decided not to use that list because I found it more fun to try and answer the question with available language resources. Also, that list might change and I wanted to find a more general answer.

But, just to put your mind at ease, when I do all of the above with that list of 鈥渙fficial鈥� Wordle targets, 鈥渟oare鈥� ends up being the best once again. So there you have it. Now what you do with guesses two through six is up to you.

This article first appeared in听听on 28 January 2022.

Links

• Spotlight on: Dr David Sidhu
• UCL Psychology & Language Sciences
• UCL Faculty of Brain Sciences