How I Created Shufflespeak

When I wrote the very first idea for this online store, I thought that Level 1 (“Language”) would end with teaching kids the letters of the alphabet. If you’ve looked at our current curriculum for even a second, you’ll know this assumption turned out to be completely wrong. Level 1 is full of Language resources … but none of them require letters or reading skils. All of Level 1 Language is “pre-reader”, and Level 2 starts with teaching the alphabet.

Once I realized this, I also realized that all the other Level 1 resources couldn’t use letters. Teaching math? Have to do it with pictures. Teaching some physical or social skill? Same thing.

Bear in mind that this was very early days. The absolute beginning of this storefront. All I had was a vague idea and planning, and I knew it was probably wrong and would evolve over time.

And so, after a few months of making the first resources and expanding my knowledge, I basically entered the pre-reader mind-set. I looked at existing games and puzzles and asked myself: “How would you create this for a pre-reader? How would you do it without text to read and/or numbers to understand?”

It was hard at first. I’ve been able to read and write for almost 30 years, and practically all puzzles and games that I know require the skill too. But once I created my first handful of pre-reader puzzles, it seemed to flip a switch in my head. From that moment, I found a way to create a pre-reader version for everything.

Sudokus? Well, obviously you can also create tiny sudokus using simple images/symbols to draw. No number knowledge required!

Crosswords? Well, obviously you can just use images for hints and ask the player to place disks (which you cut out) instead of writing.

With games, this was even easier. Most board games are already mostly textless and heavily simplified—because otherwise nobody would understand the rules or want to play it. I simply turned it up a notch and found a way to create actually engaging and strategic games for kids who can’t read or count.

About 6 months into this journey, I had specified the entire collection of Level 1 products. I had specific pre-reader game/puzzle/escape room ideas for all topics, I only needed to make them.

A single question mark, however, remained.

How do we create a LANGUAGE GAME playable by kids who CAN’T READ/WRITE YET?

Sure, I had already made games that focused on a tiny part of language. For example, a game about rhyming. That’s much easier to do, because “rhyme” is just a single rule. And a single rule is easy to explain and then manipulate/break to create gameplay.

But a proper language game? One about the entirety of language? One where a wide variety of different words/challenges could come up?

I couldn’t see it. I couldn’t see a way to do it that wasn’t incredibly complicated/convoluted, or that required serious alphabet knowledge anyway.

Until, one day, I worked on my Syllable Awareness resources and had an insight. You see, one of the “activities” in that resource is the “Syllable Shuffle”. (At least, that was the original name. Not sure what it’s called in the final file you can buy.)

The idea was simple. The Guardian (the teacher/parent/person leading the educational activity) would pick a word from the list and simply shuffle the syllables. The Apprentices then had to guess what the original word was. This is quite easy to do for anyone who can read/write at any level. At the same time, it’s actually a challenge for the Apprentices. If you hear “phant-e-el” (especially if spoken rapidly), it takes some effort to realise the word is “elephant”.

So when I finished that resource, I wrote down the very first version of Shufflespeak.

The First Version

It was just a few bullet points at this point. I wrote it in the general to-do under the “Syllables” heading, as I still thought this would just be a small extra game that only worked with syllables.

• Material: many self-contained “packs”. Each pack has 3–6 cards with syllables. When placed in the right order, they spell the original word.
• At the start of the game, you receive one such pack. You shuffle it and place the cards faceup before you.
• Now, on your turn, the Guardian reads what’s in front of you. They simply read the syllables from left to right, saying something like “ga-kan-roo”
• The player is now allowed one guess as to what their actual word is. (“Is it … kangaroo?”)

This seemed very promising. But because this was an extra idea I “tacked on” when I was already done, I decided to continue with Phonemes and leave this idea for some later time.

As I worked on Phonemes … I realized the same idea would work! Instead of a pack with syllables, you’d simply split the word into phonemes and create a pack of cards from those.

• For example, DONUT would mean you receive 5 cards that say /d/ /ō/ /n/ /ŭ/ /t/.
• Like before, you’d shuffle them into some random order in front of you. Because the players can’t read yet, it’s fine if they see this.
• The guardian reads what they see (“T-U-N-D-O-!”)
• The player guesses.

Should work, right?

Because I now had two versions, I renamed them Shufflespeak: Syllables and Shufflespeak: Phonemes.

Then I realized it would obviously work with letters too. And it would even work with words (as in, entire sentences). For example, you’d get a pack of 4 cards that say “I walk the dog”.

Now I had 4 versions of Shufflespeak. Every step of language comprehension worked with this idea! You could use these games no matter the skill/level of the Apprentice(s)!

And all of versions seemed to work fine, except for some minor problems …

Improving The Idea

As usual, this comes down to solving a few problems that occur in practice.

Isn’t it too easy/quick to just guess your word? Yes it is! We can do something much better. On your turn, you’re merely allowed to swap two of your cards. This is a bigger challenge, as it requires close listening to what the Guardian says, so you know which two cards are in the wrong place. It also prevents guessing it right first try, or accidentally. As soon as your turn starts and the Guardian reads a proper word, you’ve accomplished your goal!

What if Apprentices can recognize letters? Or what if they memorize what’s on the cards? Yes. That would be a bummer. They would just read their own cards and instantly guess it. As such, I changed the idea to turn the cards into foldable cards/pawns. Instead of placing them faceup on the table, you’d place them standing up, in a row in front of you. The player guessing would just be looking at the empty back of the card; the Guardian giving the hint sees the content at the front.

Now it needs a Guardian present. Is that a problem? No, it’s not. Once players are old enough that they can read, then a Guardian isn’t necessary anymore. But a Guardian is highly recommended for all (educational) activities for ~4 year olds anyway. No bunch of pre-reader 4-year-olds is going to be able to play a language card game all by themselves ;) Giving the Guardian an important and active role is actually a good thing here.

How to pronounce syllables depends on what’s around them. And not everyone can read phonemes. How to make sure people pronounce cards the right way?

After trying some things, I decided that the full phrase should simply be on every card of the pack. So, for example, if the pack has 3 cards with “EL”, “E”, and “PHANT”, all of them would have “ELEPHANT” at the bottom. This way, for example, the Guardian knows that the “E” should be pronounced short instead of long.

This has the (massive) added benefit of being able to sort the packs and keep them together easily. If, for some reason, your cards get mixed up or you put them in a single deck, you can always retrieve the individual packs by just checking the unique “full phrase” at the bottom.

As for phonemes, I decided to simply include a single page overview (in the rulebook) of each phoneme and how to pronounce it. (And/or refer players to my series of phoneme teaching resources if they want to know more/teach them individually first.) Most of the phonemes are self-explanatory. There are really only a few ones to look out for, and after playing the game once or twice the Guardian should have those memorized too.

I considered adding a little image on every card showing the phoneme used in a word. (For example: every /k/ card has a little cat at the bottom to remind you of its sound.) This, however, cluttered the cards a lot. EVERY card in the ENTIRE GAME would require at least one EXTRA IMAGE, maybe more! And because all those example words have multiple phonemes (cat has /k/ /ă/ and /t/), you’d still be guessing which one was meant anyway!

Isn’t it a LOT of material to include all these individual packs? And a lot of effort to cut it? Yes, this was my biggest doubt all along. But there’s just no other way. We need to be certain that the cards in front of you are the correct parts of a correct word. If creating all these resources taught me anything, it’s that there’s a TON of variety and edge cases in how words are split into phonemes/syllables/letters. We certainly can’t just create a deck of cards, ask players to pull out a few, and expect the parts to make sense.

They have to be individual standalone packs that are certainly correct. And yes, if you want to support a good number of games and playthroughs, you’ll need at least ~30 packs. If each word has 3–4 syllables/phonemes on average, you’re looking at over 100 cards to cut out. It’s a lot.

But! The beauty of it is that the packs are self-contained. You could cut out only 5 packs and already play. Leave the rest for later! That’s certainly something I clearly recommend and communicate in the rulebook: just cut out the packs you need now and get started.

Expansions & Upgrades

We can, however, do one better. If we’re going to have self-contained packs—and over 100 cards with nothing on them but a part of a word/phrase—we can add some extra spice to the cards. We have the room for it!

We can make certain cards “special”. We can give them a special action, for example.

This resulted in several “expansions” for this idea.

• One about STRESS. Each pack has one card that is “stressed” (indicated with color/icon/to be decided at time of writing). Firstly, this is simply correct: words have stressed and unstressed parts. Secondly, this can help figure out the word. Thirdly, I could give extra points for guessing which card is stressed or something.
• One about adding NOISE. That is, just add one or two random cards that don’t belong there. You can only solve your word/message if you figure out which of the cards are wrong and remove them.
• One about MUTING. It’s similar to NOISE, just the other way around: one card from the original message is “muted”. The Guardian simply doesn’t say that part when reading your message. Again, you have to realize which part is missing from your word and remove it before you can “win”.
• One about ACTIONS. Once you’re familiar with the base game … you can ditch the packs :) You’ll receive a random set of cards that probably means nothing.
  • Asking someone else to “read your cards” is now just one action—the default one.
  • You can—instead—decide to trigger an action on one of your cards. This could allow swapping a card for a new one, turning around a card so you can always see it (but others can’t), etcetera.
  • And if fully implemented, you can make the game competitive. See who can create the most words before the deck runs out.
  • (The option to add wildcards—any syllable/sound/letter you want it to be!—or more funky cards is available too, but I don’t want to overcomplicate it.)

With just a splash of color/icons on some cards, the game could have long-lasting appeal even for older ages. Even for “post-readers” ;)

The NOISE/MUTING expansions are amazing additions. When I had the idea, it suddenly solved my final problems and removed any remaining doubts. I considered adding these rules in the base game, because they’re so good and make the game so much better, but I decided against it. The base game should absolutely be as simple as possible, especially if it’s meant to be played with 4 year olds.

All these improvements sealed the deal for me. This set of games would be the peak of Level 1 Language. I was basically done with everything else and decided that the Shufflespeak games would be the masterpiece to top it all off.

When I woke up Monday morning to get started on this, however, I had one final problem.

Manual or Generated?

For years I created games on Pandaqi. And because it was just a hobby, because they were free and there were no expectations, I experimented a lot. I designed and coded all these cool systems for interactive rulebooks, for example, but also to generate game material on the fly. I would code my requirements (“place the word here, place a red square here, make cards this large”) and the system would generate a nice PDFs with all the final cards.

Really cool. Really useful. Really powerful after all these years.

The first 6 months, however, I didn’t use this at all for the online store. That’s partially because all the games I made were quite small and had custom material. It would be way more work to write the generation code than to just create all material manually. (And when doing it manually, you obviously have the benefit of adding lots of tiny flourises and small details and whatnot.) I also ignored my own system because there were some flaws and I wanted to do a massive upgrade/clean-up before embedding it in my webshop workflow.

So now I had a problem.

• Was I going to spend precious time first to integrate+improve my game generation system, then write the code for Shufflespeak cards? (And accept that the material would be randomly generated … but the rulebook would not, because that system was FAR from ready?)
• Or should I just create these games manually, like I did for the ~20 educational games before these ones? It would allow me to add nice design touches and maybe images/icons of the common words too.

In the end … generation won out.

The material for these games is incredibly consistent. All cards have the same simple layout, put into small self-contained packs, repeat for 10 pages. Just thinking about doing it by hand made me demotivated, while I knew it would be just a few lines of code.

The generation part, that is.

The actual words—and how they split into syllables/phonemes—would need to be put into code too.

And so, while the generation system wasn’t quite ready yet, I devoted my time to that.

• I created a list of words of varying length and complexity. (But ones that children would still recognize.)
• For each word, I saved how to split it properly.

It looks something like this:

const cards = 
{
    elephant:
    {
        frame: 0, // if I do want to add images to some cards, I need to tell it which specific image to grab out of a larger "spritesheet"
        syllables: ["EL", "e", "phant"], // capital letters indicate stress; the most elegant solution I could find
        phonemes: ["ĕ", "l", "ŭ", "f", "ŭ", "n", "t"]
        // individual letters can just be grabbed from card title
        // special actions are assigned randomly on generation
    },
    // ... more cards here in same format
};

Unique IDs

I ended up putting a small “ID” on the back of cards anyway. It’s a simple function that takes the content (the entire word/phrase), shuffles the letters in some way, and puts out a string of letters and numbers like ehel10.

Why? Because this allows sorting/grouping the packs without having to look at its content. You can just group all ehel10 cards and know you’ve got the whole pack, 100% complete. Without having any clue as to what’s actually on there.

Additionally, this makes it easier to have a word appear in multiple games (i.e. ELEPHANT appears in both the SYLLABLES and PHONICS version) and still be able to easily separate those packs.

This ID function is very barebones and imperfect. It only has to “obfuscate” the word with which we started—it doesn’t have to do anything smarter. So I chose a simple function that shifts all letters in the alphabet that also ensures we don’t get any duplicate IDs.

This little ID section also contains the number of cards inside that pack. So you can stop searching when you have enough without having to re-check if you have the complete word each time.

How’d That Go?

This was several days of pretty boring work, to be honest. Still much faster and less boring than creating all cards manually, and I can probably reuse that list for many other projects too.

Generation is always funny. The “reward” is delayed in this process. You work for days without having anything to show for it. But once done, you press a button, the computer does the rest, and suddenly you have 4 finished games.

Fortunately, I was familiar with this from the 100s of Pandaqi games I made, so I just powered through and got it done.

At the same time, I manually created the rules PDFs for all games. I basically explained the entire ruleset already, so these are really short and simple files.

I also decided—for sure—that these would remain separate games. It was tempting to lump them all into one big “Shufflespeak” game. I thought I’d only need to explain the rules once (on a single page), and just by picking different packs you decide if you’re playing syllables/phonemes/letters today.

But I was wrong. There are slight differences between the different versions. (Such as, as mentioned, the phoneme one requiring an extra page explaining how to pronounce phonemes.) And whenever that’s the case, pushing different versions into one big game is a bad idea. It just makes things convoluted, messy, bloated, and for no reason. It is almost always better, I have learned, to keep separate ideas truly separate.

And so they became 4 unique games, with their own rulebook PDF and material PDF. Of course, the price reflects this—each individual game is quite cheap, and I hope this allows many to buy several of them if they want to.

Shufflespeak: Words

The set of cards for this one is just a list of simple sentences (of 2–8 words). I tried to vary up the subject, the structure, the verb, etcetera.

In the end, because the sentences need to be simple and recognizable, they are pretty similar. But that’s okay! This version is absolutely meant for pre-readers who struggle with the order of words in a sentence. It will never really provide a challenge to anyone who’s a little older, no matter how complicated I make the sentences.

Because all sentences need a subject, verb, articles, etcetera I get the expansions for free. The word “the”, for example, is present in a lot of the packs. So, if you randomly select cards, you’re likely to be able to form a functional sentence from them.

This was the simplest variant, so no more notes from me.

Shufflespeak: Syllables

This is where it gets interesting. Most words only have 1 or 2 syllables. It’s rare enough to find 3-syllable and 4-syllable words. Anything longer basically means you’re looking at really obscure (compound) words.

Well … with only 1 or 2 syllables, there isn’t much to shuffle! The game doesn’t work!

As explained, that “noise” upgrade softens this issue. It adds extra cards to give you 4 or 5 things in front of you anyway, even if your word only has 2 syllables.

But I still wanted to keep it as an upgrade and make the base game work without it. And so I started looking for words with more syllables. After some hunting, I found enough of them (at least in the 3–4 syllable range) that kids would recognize to fill the list.

I also added much harder words (5–6 syllables) … but certainly don’t expect young kids to play with those. They’re mostly a final challenge for older players, if you want.

Instead, I decided to allow common combinations of words.

You see, English rarely combines two words into one word. Even now, people can’t agree if it’s “board game” or “boardgame”.

This is different from my native language (Dutch), which allows you to combine absolutely everything into a single word. This is one of the reasons Dutch texts are like 20% longer than their English translation. It also means Shufflespeak: Syllables would’ve been much easier to make in Dutch, because it simply has more common words with a high number of syllables.

In English, those same concepts would simply be multiple words with a space between them (and sometimes a hyphen). For example, it’s “apple tree” in English, but “appelboom” in Dutch.

After some consideration I decided to include such common combinations too. Phrases like “Alarm Clock” or “Birthday Cake” are more recognizable to kids than single words like “Electricity”. They are able to pump up the syllable count so that the game is playable, while staying simple.

I basically spent a day collecting single words with 2–6 syllables. And then I spent the next morning adding these “compound nouns” and feeling stupid for not doing this in the first place, because this is much more fun to play with. I was clever enough to mark all these entries with a simple compound: true so that my generation code could select or completely ignore those specific entries if I wanted.

Shufflespeak: Phonemes

Unfortunately, most words that have lots of syllables … have way too many phonemes. A word like “elephant” (only 3 syllables) already has 7 phonemes, which is pushing it in terms of complexity for a kid playing this game.

So the nice list I’d made for the Syllables version proved mostly useless in this version. I had to start from scratch and create a list of smaller words with fewer phonemes.

I did restrict myself to single words this time. Because compound nouns or longer phrases weren’t needed. There are more than enough simple words with a good variety of phonemes.

In fact, I erred on the safe side. Most entries in this list have 4 or 5 phonemes. “Only” 4 or 5 things to shuffle back into the right place is a nice challenge, not too easy and not too hard.

Shufflespeak: Letters

Most letters have a 1-to-1 relation with a phoneme. When you see a t, the phoneme is usually a single /t/. When you see a vowel, the phoneme is usually a single vowel sound.

As such, most of my list for Phonemes could simply be reused in this game. Those words still had the right number of letters to make the game nice and playable.

By now, though, my work had grown into a massive list of words (and their phoneme/syllable/letter breakdown). Too many to put into a single game, or even 10 games. About 50 different packs was the absolute maximum for me in terms of the material PDF size. So I had started marking them based on “difficulty”, then made my code randomly select from the list whenever it generated.

By this point, I could finally rely on all that boring work beforehand to gather the list. Because without any extra work really, this final version was finished too.

Side Note: The Noisy Games

Around the time I invented Shufflespeak, I also invented another great language game to cap off Level 1 Language: The Noisy Games. It basically uses a different twist to allow pre-readers to play language games anyway.

I had already started creating manual cards for that game. I had done that months ago so I could at least finish the rulebook completely—because without a few finished cards, I couldn’t create the example images—but then other work was deemed more important. I kept postponing The Noisy Games, also because I simply hated the prospect of having to create all these 50+ “Code Cards” for the game.

As I worked on Shufflespeak, I realized I had basically solved that problem for free. The exact same word list I used for Shufflespeak … could be used for The Noisy Games too, and then I could just generate the cards on the fly! That’s why I worked hard to add way more words than I needed to that list, making sure to add the correct syllable + phoneme split for each. The list was now easily long and varied enough to re-use for The Noisy Games without making people feel like they bought the same game twice.

The only change? The Noisy Games requires images on the cards. In that game, the pre-reader kids actually draw cards themselves and have to recognize the image on that (to know the word they should communicate).

As such, I simply checked all the words for which I’d already drawn images (that’s A LOT at this point), and added the frame: XX metadata to those. The Noisy Games generator now filters the huge list of words to only select those with a frame set to something, and then it pulls the right image from a large spritesheet. (That’s basically a huge image with all smaller images inside, and e.g. “frame = 2” tells it which specific rectangle to use.)

Of course, I also had to split the card design into separate parts, and then code how to combine the parts and generate material on the fly. But that’s far less work than manually making the cards myself. As such, when I finally finished Shufflespeak, I basically also finished The Noisy Games.

Anyway, just wanted to remark on reusing/sharing that word list between the games.

Marketing & Visual Style

Finally, I had to do the finishing touches, which always includes marketing images.

The work so far had been quite … abstract. The cards were quite barebones. Not exactly ugly, but also not pretty. I had no real theme or visual in mind, and that’s not great for marketing. Or for playing the game, in fact.

And so I tried to find a style that reflected “shuffling” and “mixing language”. Something more colorful, more thematic, that would speak to young kids too.

I landed on this idea of the “Sphinx”. Animals made from the body parts of different animals. A bird with an alligator’s tail and a bull’s horns. A horse with wings and scales. As if … you shuffled one or more animals ;)

This seemed a funny and creative way to illustrate the game. By combining it with some simple shuffling “patterns” (for background and minor visual variety), I ended up with something I felt was pretty enough.

Then … reality hit. How would I actually use those “mixed animals” on the cards? I couldn’t find some layout that actually used the animals that wasn’t needlessly convoluted/taxing on your printer. Whatever I tried, the cards were more minimalist, prettier and more functional to me with just a simple nature/jungle background and nothing else. Any other details made things messy; without that background the game looked incredibly sterile and boring.

And so I went with that general template, and moved the “shuffled animals” to expansion cards. Their image only appears on special cards that you need for those upgrades/expansions. And because they only appear there, those cards are easily differentiated from the cards you actually use to play the game, which is always nice.

That’s the beauty of generated material. I could create/code this whole visual style after the fact, and simply press a button to generate my new PDF!

Conclusion

As always, I learned a lot, plans changed, but eventually we get a finished product that I hope will delight some people around the world.

I learned that it’s actually quite rare for a word to have more than ~2 syllables. I created this huge list, learning how to split all those common words along the way. I was glad I could reuse that list among all versions of this game and The Noisy Games.

I tried some visual style, it failed, but I ended up at a different style that I think looks good enough. Not amazing, but that’s not the goal anyway. It uses a smidge of AI generated imagery, with everything else being my own work, and I keep discovering that this is the right balance.

I started the effort of cleaning up my “board game generation” framework and porting it to the webshop, so I can use it on more of these educational board games in the future. This effort isn’t finished at time of writing, but a good chunk of the work is done.

Most of all, I finished 4 language games that pre-readers can play, enjoy, and learn from. Very simple rules, hundreds of packs included, but it actually works with any word/phrase.

Even better, by working on the game I stumbled upon several upgrades that truly elevate the game. The “noise” upgrade that adds 1 wrong card into the mix suddenly makes even simple (2-syllable/2-phoneme) words a challenge to figure out. The “mystery” upgrade does the same thing but in reverse, making very long words more approachable at the start of the game until that “mystery” part unlocks too. In my experience, both kids and adults are quickly done with the “base game”, as it becomes too easy with some experience. But the upgrades (and the other expansions) give the game a much, much longer life.

Those were my thoughts on the development of Shufflespeak. The first true language board game playable by those who can’t read/write yet, and enjoyable even for older ages. As far as I can tell, at least.