-1280x720.png)
The trailer for the upcoming metroidbrainia EMUUROM has a lot going for it. A charming micro console aesthetic, refreshingly non-violent mechanics, and a cute player character with really big glasses. But, it wasn’t until the trailer's 0:50 mark that I bolted upright. A selection of symbols - a code - just waiting to be poured over and deciphered. It's less than a second on screen, but it's more than enough for me to keep this game on my radar.
Metroidvanias, especially the thinky variants, have long enticed me, but something about the sight of cryptic alien glyphs whets my gaming appetite like nothing else. For me personally, I think constructed languages are, put simply, the god spice of metroidbrainias (and other thinky games, for that matter).
First, a definition. A constructed language, or “conlang”, is any language that develops through conscious effort, rather than organically. Esperanto versus español. Conlangs might be created to add depth to a fictional world, as an academic exercise, or simply for fun. The conlang iceberg runs deep, with the shallowest entries decried by Reddit linguistics nerds as English “relexes”, and the deepest being crafted over decades to challenge status quo theory on the relationship between language, worldview, and cognition.
To be clear, simple alphabetic substitution ciphers—the kind found in Stray and Splatoon and on the back of cereal boxes—are not conlangs, at least not by my definition. Maybe that’s what the glyphs in the EMUUROM trailer represent. Maybe not. The uncertainty adds to the intrigue. Because if EMUUROM’s glyphs are more than a cryptogram, if they represent a fully fledged language, then uh oh I might have to block a week out on my calendar.
Duocho(ri)zo
Like any genre named after a particular game (roguelikes oh gawd), metroidvanias are notoriously difficult to define. Still, certain characteristics prevail. Intricate, non-linear level design, woven together in a predominantly open world. Traversal gated by ability unlocks. Completion percentage determined by map area explored. Foreshadowing, backtracking, and secrets galore.
Sound familiar? What if I told you that the process of deciphering a language shares all these characteristics?
A body of foreign text is an open world, a “map” to pore over in any (and often every) order. Hidden rules to intuit in the form of grammar, morphology, and phonetics. “Unlockable” word meanings, each opening up new avenues of inquiry across the entire corpus. “Completion” measured by percentage of words translated."
Because of these parallels, conlangs blend smoothly with classic metroidvania design, enhancing flavors without overpowering them. Don’t take my word for it—even the developers of mainstream metroidvanias like Metroid Dread have dabbled in the technique.
This is all mise en place, however. Traditional metroidvanias aren’t the conlang’s ideal pairing. Rather, it’s their thinky cousins. The games which, in recent years, have come to be called Metroidbrainias.
Teasing Truth
Metroidbrainias constrain the metroidvania formula in a single, pivotal way: progression is gated not by double jumps and air dashes and super missiles, but by knowledge. In the 2024 hidden gem Leap Year, for example, players begin with every ability they’ll ever possess. Advancement comes instead from cognitive leaps (sorry, I had to) prompted by environmental clues, cryptic wall scrawlings, and ingenious level design. As the Steam description for upcoming metroidbrainia Nonolith elegantly puts it, “You are very powerful, you just don’t know it yet.”
While catnip to thinky-minded players, this constraint presents a unique challenge to game designers. When power = comprehension, how do you entice players with the promise of future prowess without spoiling the experience of acquiring it? How do you tease knowledge?
For traditional metroidvanias, a classic trick is to include a prologue section of the game in which the player character is kitted out with some or all of the game’s unlockable abilities, then rip them all away in some scripted disaster. In the prototypical Castlevania: Symphony of the Night, protagonist Alucard power-sprints through the closing gate of the eponymous castle and rampages through the halls, one-shotting zombies and wargs—only to come face to face with Death, who strips his upgrades and scatters them to the far reaches of the map. “This power belongs to you,” the game is saying. “It’s been stolen. Now go win it back.”
The problem with this approach for the metroidbrainia designer is that knowledge, barring MIB neuralyzer, can’t be “stripped away”. Once acquired, it’s there to stay.
Here’s where conlangs really shine. The presence of foreign text or speech uniquely and definitively demonstrates that knowledge exists—without spoiling the knowledge itself. Furthermore, the form in which that language is presented allows designers to tweak the mystique. What font is it written in? Where? Is it gilded? Carved? Smeared in blood? Such contextual flourishes can underscore the importance and relevance of the knowledge being teased far more artfully than other information-storage mediums, like locked diaries and rolled-up scrolls. Better yet, designers can mix conlangs with familiar natural languages (“natlangs”) to onboard new players, build thematic depth, and—crucially—suggest the conlang can even be translated. 2022’s Tunic illustrates this with the first in-game instruction booklet page the player collects.
Note the combination of English and conlang text in the blue tables, implying translatability, and the winged cartouche-like design surrounding the conlang text in the top-right image, denoting sacred significance. Tell me this doesn’t demand to be deciphered.
Fluency
The promise of power is a potent motivator, but it can only sustain players for so long. To keep players engaged across dozens of hours of dead-end corridors and brutal boss fights, traditional metroidvanias must offer periodic tastes of that power. What’s more, those tastes better tantalize. Upgraded abilities must have good “game feel”, i.e. be mechanically satisfying to wield. So too for metroidbrainias, where “unlockable” knowledge must be gratifying to contemplate and exploit.
Here, again, conlangs can help. How naturally delightful the power curve of language decipherment can be! The anxious exhilaration of translating a sentence word by word, concept by concept, comprehension building, pieces slotting into place, until intuition outpaces uncertainty and IT ALL MAKES SENSE.
Moreover, conlang-seasoned epiphanies hit much harder than the typical puzzle game, because solution validation comes not from the game, but from the solution itself. It’s self-evident.
Sure, some language decipherment games validate player assumptions periodically. Chants of Sannaar, for example, employs an Obra-Dinn-like system, where correctly filling out an entire page of translations in the in-game journal earns the player the audio-visual equivalent of a pat on the head.
Yet such hand-holding isn’t required. Tunic never explicitly confirms a player’s translations. Rather, certainty arises from the way independent postulations improbably align, gradually weaving together into a cohesive, indisputable whole. As a result, solutions feel that much more earned. There’s no possibility of brute force here, no means for players to rip themselves off out of impatience. In addition, self-validation leaves open the possibility of assumptions being flipped on their head, of glyphs turning out to mean something utterly different from what the player originally believed—a thrilling slap in the face which happened countless times during my Tunic playthrough.
This blissful independence isn’t exclusive to language decipherment games. Investigation games like Her Story, which involve piecing together a narrative from fragmented accounts, also display this. Still, it’s rare enough to savor. Besides, in the endgame conlangs games kick things up a notch. (“Bam!”) With language decipherment, unlike narrative investigation, mastery means fluency. Like queueing up a Guitar Hero track that once gave you carpal tunnel syndrome and coasting through it with your eyes closed, there’s little so empowering as returning to a passage of once-confounding text and sightreading it.
And players should feel empowered. Learning a new language not only facilitates new social connections, it births new pathways of thinking, paves the way for new semantic connections in the brain. Language decipherment games make players smarter, in-game and out.
Reading ahead
Metroidvanias can’t be the only genre that synergizes with language decipherment, right? What other dishes can we sprinkle this God Spice on? A deck-builder where the card text is printed in a conlang? A Zachtronics style engineering puzzler where the in-game reference manual was written by ancient aliens? A We Were Here style co-op puzzle game where the only way to communicate is with a conlang keyboard?
Or are we thinking of this all wrong? What if conlangs are more than just the metroidbrainia equivalent of MSG?
Enter Epigraph, where the conlang is the game, pure and simple.
Released in 2024, Epigraph comes from developer Matthew Brown, best known for the Minesweeper-like Hexcells. More relevant though is his 2018 work Cypher, which functions both as a love letter to and introductory course in cryptography. Epigraph does the same for language decipherment, though takes a more hands-off approach, allowing players to discover rules instead of being taught them.
Like the best metroidbrainias, Epigraph is 100% open world, i.e. the entire corpus is on display from the start, the endgame trial available without having to jump through any intermediary comprehension checks. Players are presented a collection of artifacts carved with conlang text, a brief introductory note in English, and a puzzle box with clues written in that same foreign language. That’s it.
If a player were to begin the game already fluent, they could skip straight to the endgame trial and roll credits in seconds. (Speedruns are, in a pure metroidbrainia, pointless. In fact, one could probably measure a metroidbrainia’s purity by how (un)interesting it is to race through.)
An entire essay could be written about the benefits of this design purity, but I’ll focus on one. Unlike more guided logic puzzle rule-discovery games like The Witness, which often intentionally mislead the player into false assumptions only to subvert those assumptions with an embarrassing punchline, the false assumptions made while playing Epigraph feel entirely organic. This I believe happens because the “rules” which players must discover during language decipherment are so open-ended—any glyph could represent any concept—that the likelihood of multiple players making the same false assumption is comparatively low. These are emergent false assumptions rather than contrived ones, the same way games like Breath of the Wild and Elden Ring generate whole subreddits worth of charming anecdotes, magnitudes more than on-rails games like The Last of Us and God of War. Given this variance in player experience, a playthrough of Epigraph feels more personal than a playthrough of The Witness, more special.
If you’re a fan of metroidbrainias, have even a passing interest in linguistics, or just want to see where the Thinky space might be headed, give Epigraph a try. At $2.99, it’s well worth it.
And to the metroidbrainia developers out there, I implore you, consider adding a conlang to your game. At their core, metroidbrainias are about knowledge—teasing it, teaching it, reveling in it—and what’s the point of the written word if not to store knowledge, the point of speech to convey it?
P.S. To everyone about to comment, “Outer Wilds is the best metroidbrainia ever made, and it doesn’t use a conlang!” Okay, fair point. But also, what if it did?
Disclaimer: Thinky Games is part of the Carina Initiatives and may have professional relationships with individuals and businesses related to the subject of this article. Please see our Editorial Policy for details.
