Wordle has been taking the world by storm. If you haven’t heard about it, it’s a daily word game where you need to guess a five letter word within six tries. It’s fun, addictive and quite a good way to revise vocabulary.

As a principal of a Chinese enrichment centre, I naturally looked for Chinese versions of Wordle to help students revise. Unfortunately, while there are Chinese Wordle clones, most of them focus on idioms (成语), and are far too difficult for Singaporean Primary School students.

Thus over CNY, we created a Chinese Wordle clone aimed at helping P1-P6 students revise daily. The game tests one phrase a day taken from the MOE syllabus, and can be completed within a minute. Fast, effortless revision, with the goal of making your child feel good about his or her Chinese. The game is completely free and requires no login, so if you find it cool, please share with your friends.

In the remainder of this post, I go into more details about the approach and problems I face in the process of creating Chinese Wordle. Continue reading if you want to know how the sausage gets made.

GOALS

I love Wordle – I play it daily, and my sister and I like to compete to see who solves it faster.

When creating a Chinese version of Wordle (KidStartNow Wordle), we had two main goals. Firstly, we want students to experience the same sense of excitement we got when we successfully solved an English Wordle, the satisfaction of “feeling smart”. If students feel good about successfully solving a KidStartNow Wordle, then they will hopefully be more willing to practise Chinese.

Secondly, we want kids to practise vocabulary. As previously mentioned, improving vocabulary is one of the fastest ways to getting better at Chinese, just that the process can be tiresome. A daily 1-min practise session is a great way to help students revise.

However, when I tried a number of Chinese Wordle clones, I often felt frustration rather than enjoyment. Why?

ENGLISH VS CHINESE WORDLE

Most Chinese versions of Wordle focus on idioms (成语), and if you tried it, you probably experienced the demoralising feeling of staring at a whole line of Chinese characters, trying to form a valid idiom.

Some versions have 20 characters to choose from, and trying to form idioms is hard because a) idioms are naturally more difficult and you can ONLY form idioms using the 20 characters given (think Scrabble), and b) most adults can only hold 5 to 9 items in their short-term memory, as proposed by Harvard Professor George Miller (this is also why many students struggle with 组词成句).

You might think – the alphabet has 26 letters, isn’t that worse? No, because you can input any 5-letter word to reduce the word pool, and most Wordle players have go-to starter words like ADIEU or SOARE. Players also have heuristics like using common consonants to strategically remove characters, making Wordle’s difficulty curve smoother and more enjoyable. None of these work for Chinese versions.

METHODOLOGY

The first design decision we made was to test two-character phrase (词语) rather than four-character idioms (成语). This would mean fewer number of characters to choose from, and would make initial guessing less frustrating for students. From an academic standpoint, there are significantly more phrases than idioms to practise in the MOE syllabus.

The second design decision was to only take words from the MOE syllabus to make the material more relevant to students. Currently, only words from P1-3 are utilised, and there is no differentiation by level, though Upper Primary words can easily be added if people find this game useful.

Attempt 1 – Simple Pairing

Our initial approach was to randomly select four phrases (词语), and use their characters as the choices. For instance, if we selected 军人, 超市, 生日, and 下雨, then we could use the eight characters as the choice pool.

However, after testing, we found that this simple approach didn’t work well. This creates a game where students need to find pairs, and since each character is only used in one phrase, they wouldn’t be able to use the Wordle colors (i.e. green colour for a correct word and position, yellow for a correct word but wrong position) to strategically guess a word.

Attempt 2 – Clustering

The next algorithm we tried was to select a “seed” character (e.g. 道), randomly select phrases that utilised the character (e.g. 味道，街道，知道), and then used them to form a word pool (味, 道，街，知).

This approach had several benefits: firstly, guessing feels simpler. Secondly, there are opportunities for students to get characters correct (green, yellow) without guessing the full word and makes the process more fun. Lastly, this helps kids revise vocabulary through the process of 连词/扩词, and is similar to how we teach vocabulary in the classroom.

Attempt 3 – Refactoring

While the previous attempt was an improvement, there were still several issues. Firstly, what is the best way to select a “seed” character? What if the “seed” character has minimal phrases?

One way we solved the problem was to use a frequency counter to count all lower primary school words (e.g. the word ‘子’ appears in 19 phrases), and only consider characters that met a threshold number of phrases.

The second problem was slightly more tricky: how do you determine if a child’s inputted word is valid? For instance, using the above approach, we started with two “seed” characters (生, 上), and for each character, we generate a list of three valid words (and four characters for the choice pool).

But do you notice the problem?

What if a student forms a valid phrase that does not exist within the MOE syllabus? For instance, 上身 or 晚生 are also valid words. It would be academically misleading to say the phrase is invalid just because it’s not taught in primary schools.

To solve this, we combined the MOE syllabus with CEDICT, an online Chinese dictionary (with > 120k entries!!) to ensure users could enter valid phrases that don’t appear in the MOE syllabus. Problem solved!

CONCLUSION

If you have read to this point, please take a minute to try out KidStartNow Wordle, and please share if you found it helpful. Practising takes just 1 min, and is completely free with no login. Fast, effortless revision, with the goal of making your child feel good about his or her Chinese.