# How do I make a writing system undecipherable, while not intended as such in-world?

It is relatively well known that hieroglyphs were not fully translated, or at all for that matter, until a while after the Rosetta Stone was found.

It is less well known that the script known as Linear A (https://en.wikipedia.org/wiki/Linear_A) still hasn't had any of its texts deciphered.

# The World

Suppose I have an ancient civilization or something. Now, what (fictional) ancient civilization doesn't have ancient and powerful artefacts? Now, these guys are smart. They know better than to leave a potentially world-ending device behind without an instruction manual including a whole lot of "Don't do this, or this island will sink" and similar warnings.

# The problem

However, because of dramatic necessity, I need absolutely no one to be able to read this manual. Otherwise, they would know better than to blindly try pressing buttons to see what happens.

To make this more challenging for you, the civilization would probably at least try to give a few hints on how to read the script, but if this makes it impossible, you can assume the dictionary has been hidden too well, or is eroded, or something.

Otherwise, the dictionary does exist in some form. (This is why Etruscan is not a good answer, because they would probably written their "Rosetta stone" in proto-latin, and that would probably make it easy to understand).

# some extra context

The device these instructions refer to is basically a universal property (aka, reality) editor. Most of the available vocabulary would be used somewhere.

On the topic of vocabulary, let us say that it is a language of roughly the complexity of Latin, with a vocabulary the size of English (for simplicity's (as in simpler for you who are answering, not the people who are trying to read this) sake).

# tl;dr

What would a helpful civilizations's script need to be like, if no-one from modern times can read it?

# Edit:

I'm sorry, there are just so many wonderful answers. If you are interested in the topic, I definitely recommend you read them all. Not just the one I (will eventually, after much deliberation) accept.

In the comments, someone inevitably mentioned the Star trek tng episode where the universal translator fails, because it doesn't get the literary references. I'd assume that these guys would be smart enough not to write in this way without leaving a record of what they are referring to.

• The title is asking a different question than the last sentence. What are you asking? – L.Dutch - Reinstate Monica Apr 8 at 7:30
• We have been able to decipher ancient writings where the language was known but the script wasn't; for example Ancient Egyptian (ancestor of modern Coptic, script was hieroglyphic or demotic) or Minoan Greek (a very archaic form of Greek, script was Linear B). We have been able to decipher ancient writings where the script was known but the language wasn't; for example Hittite (script was the well-known Akkadian Cuneiform) or Tocharian (script was Brahmi). We have never been able to decipher ancient writings where both the language and the script were unknown. – AlexP Apr 8 at 7:35
• TLDR: I both the language and the script are unknown, there is simply nothing on which to base the decipherment, nothing on which to base a chain of inferences leading to the decipherment. – AlexP Apr 8 at 7:37
• Oh, we have that in real life. The Etruscans wrote with a well-known alphabet (the Italic alphabet, an archaic form of the Latin alphabet) but their language is unlike any known language. They left a lot of pictures with some inscriptions on them. We have been able to figure out small numerals (1, 2, 3 are θu, zal, ci) and some words (e.g., phersu means mask, as in a character in a play, hence Latin persona); we have been able to figure out a bit of grammar; but we are very very far from fully understanding the inscriptions. – AlexP Apr 8 at 7:44
• I'm amazed that nobody has mentioned the Star Trek Next Generation episode, Darmok where even with a universal translator, Picard can't understand an alien because its' language is based on allegories. – Henry Taylor Apr 8 at 23:15

The main reason that some scripts aren't deciphered is that there is too small clues what each symbol/word means.

From one hand, too little texts on this script reduces possibility to try frequency analysis or other statistic methods. Other hand, many scripts was deciphered when same text written in different scripts has been found: it's become possible to create vocabulary. That's the case for hieroglyphs and Rosetta stone.

But creators (by definitions) do efforts to make script understandable, so they probably created two-language texts (like Rosetta stone).

So all you need is

1. Couple of texts in ancient script, probably all are found in your reality editor.
2. Vocabulary to undeciphered language, for example Linear A.

TL;DR
Update thanks to Mark Gardner:

So, to make it work, add a Rosetta stone, but the other language is equally unknown

• @MarkGardner Absolutely right! – ADS Apr 8 at 7:47
• Great answer. Unhelpful to this question, but related to the topic: There's an old short story (whose name I've forgotten) uses a periodic table to start a translation key after an astronaut recognizes its shape in an alien diagram. Other starting points could be geometric, say a diagram of the Pythagorean Theorem (h/t Neal Stephenson). – Michael - Where's Clay Shirky Apr 8 at 15:52
• In my comment to the OP I give the title & author of the story – Jim Baerg Apr 9 at 1:27

The ancients are using a borrowed phonetic alphabet.

Hangul is a phonetic script invented to write Korean. It is super cool and very elegant - an alphabet invented for a purpose that it serves admirably. In theory, a phonetic script like Hangul could be used to write any language. In the linked site, the author uses Hangul to write English.

His original Hangul, what Google translate gave me when I told it that it was Korean, and the original English at the bottom.

옦 휴먼 븽스 알 볼느 쁘릐 얀드 이쿾 인 딕니티 얀드 랱스. 떼 알 연댿 위뜨 릦헌 얀드 콘썬스 얀드 쑫 얔트 툐얼듯 완 어나떨 인 어 스피맅 오쁘 브라떨훋. (알티컦 一 오쁘 떠 유닙헐섞 딬꺄레썬 오쁘 휴먼 랱스)

옦 Human Charles Al Volne Petit Yand Aesop in Dinity Yand Hicks. Shoal Al Yeonpeng Wit Chonheon Wand Constance Yand Chop Gwent Troubled Wan An In A Spinel Op Bra Braun.

All human beings are born free and equal in dignity and rights. They are endowed with reason and conscience and should act towards one another in a spirit of brotherhood.

I am sure this Hangul is gibberish in Korean. I am impressed that google translate must have kept the phonetic feel of the Hangul, because if you squint you can kind of see it. "Troubled Wan An In A Spinel Op Bra Braun" = "towards one another in a spirit of brotherhood".

In any case - your archaeologists will think they recognize the language because they recognize the characters. Your ancients used a borrowed alphabet to write their language.Translations will be almost gibberish - but close enough that they are distracted trying to figure out what the gibberish means.

• so, it could actually be, say, English (reality editing may allow time travel), but written using hieroglyphs? – Mark Gardner Apr 8 at 15:19
• @MarkGardner - I like that idea for a story! Your ancients are writing a message to the future in English, using the alphabet they know. The breakthrough: someone reads it out loud. That happened to me in an airport in Grand Turk: they had signs in different languages including one I had never seen. When I read it out loud it sounded like French; it was Haitian which is at its root phonetic French. How the ancients know English is another story but I am sure it will be a good one too. – Willk Apr 8 at 15:27
• I'm not fully convinced that linguists would be fooled by that for very long: they'd realize that a direct transliteration produces only gibberish, and would use different approaches. Perhaps they would start counting repeating patterns. For example, 얀드 occurs 4 times in the text. Information like this may turn out to be the key to breaking this. Of course, I have the advantage of knowing the English source, but without even understanding the writing system at all, I'm already pretty sure that 콘썬스 represents "conscience". – Schmuddi Apr 9 at 8:46
• I'm slightly confused about the author's transliteration here, a lot of it isn't what a native Korean would write to represent English. For example, surely 옦 (okk) should be 올 (ol) to represent "all". Also 얀드 (yandeu) is used repeatedly for "and", where is should be 안드 (andeu). This might seem like it detracts from the point, but google translate is quite good at correctly translating "Konglish" when written correctly, so the gibberish at the bottom might not be so seemingly random if the Korean transliteration is written more naturally. – Jake Browning Apr 11 at 10:37
• Archeologists, historians and linguists are well aware that writing systems are frequently re-used for different languages. Latin script is used for at least 30 different languages (including English). Linear A, for instance shares many glyphs/characters with Linear B and may even be the same writing system (ie., encipher the same phonetics), but Linear A is still indecipherable simply because no one knows the language that it is enciphering. – RBarryYoung Apr 11 at 13:30

Interestingly, this is a well-studied problem. People have spent a lot of time trying to figure out warning signs for nuclear and toxic waste sites that will still convey a sense of danger in ten or twenty thousand years, when our own culture might be dead and gone, but the toxicity is still alive and well. It is nearly impossible to create 'universal' symbols of that sort, which suggests there isn't much problem making your ancient language indecipherable.

To decipher an unknown language, one of three things is required:

• A correlation with some other known written language — a translation or codex, like the Rosetta Stone — that allows us to suss out some words and a sense of the grammar of the unknown language.
• A correlation with some known (modern) spoken language.
• A correlation to material objects or representations, like a child's picture book or a description of some known event or object.

In short, we need a way to correlate the text with something already in our experience, otherwise the text is just so much chicken scratch.

Now it would be possible to write a manual where we can decipher how to operate a device without being able to decipher what the device does — think IKEA assembly instructions — but there's a logical issue. Why would a super-advanced civilization create a manual for a super-advanced piece of technology aimed at a functionally illiterate user? Unless that super-advanced civilization comes from the Simpsons' universe (with their equivalent of Homer Simpson expected to take care of the world-altering mechanism), a manual for a device like this would be aimed at technically proficient users, and likely wouldn't contain (say) pictures of the console with arrows pointing at buttons to press. In any case, your real problem isn't in making the ancient language indecipherable; your problem is making the ancient language just decipherable enough so that people can get themselves in trouble.

Incidentally, this is an integral part of the plot of the 1950's movie "Forbidden Planet." If you haven't seen it, it might be worth your time; it's a good movie if you can get past the campy 50's stereotypes.

Alongside what the other answers said, give them really, really exotic grammar rules.

"What language is the hardest to learn?" The unsatisfyingly correct answer is Ithkuil, which is a constructed language all but designed to be impossible to parse. But sticking with the natural languages, languages that people have spoken from cradle to tomb for generations upon generations; then the answer will be "What languages do you know already?"

Grammar difficulty is relative. Chinese has a script of tens of thousands of characters, but to Chinese children learning new ones is like learning new words for an English kid. Korean has six levels of politeness, whereas English has zero, but they don't have any difficulty with it. English has some of the least congruent spelling rules of any language on the planet, but kids who grew up with it will only make the occasional mistake due to carelessness, rather than inability.

Every language is hard relative to an unrelated language. So your solution is to make the Precursor Civilisation's language hard or impossible to learn for anyone on Earth, by making its rules completely unlike any natural language on Earth today. You could have one of them as your teacher today, reading you children's books, and the logic would still phase you even if you were a linguist and a polyglot.

Examples of those grammar rules? There's plenty of examples of weird rules for languages today, like the honorifics and weird spelling I mentioned, but linguists of today are all accustomed to those things because, well, we have languages with those rules today. Instead, I suggest you draw from esotheric computer languages, or esolangs for short. They are programming languages and dialects (yes, programming languages have dialects too) all built around one or more unique things. As an example, there's Whitespace, a language in which the only allowable characters are space, tab and line breaks. There's Shakespeare, where programs are structured like a script that could have been written by the Bard himself.

Sticking with features that actually have a philosophy behind them which could conceivably be used by real languages:

• Two-dimensional languages, which is where a program (or a story) is a collection of characters in a grid, with indications about in which direction to read. Branching paths can be literal branches.
• Self-modifying languages, which broadly follow the principle that you might need to read the same thing twice, but it means something different the second time.
• Non-textual languages, which do not have anything resembling a script. Then how could you write them down? Take Piet as an example; a program is a picture, where each pixel's colour designates the instruction. Resulting programs look like Mondrian paintings.

Those are just the examples I have from the top of my head. Look around at the codegolf SE for more creative examples, and in use!

• A manuscript with self modifying rules of grammar? Sounds similar to the Enigma machines. We only managed to crack those with a hell of a lot of patience and luck, and we knew the language and script we were after. A self modifying manuscript in an alien language and script would be linguistic hell. – Joe Bloggs Apr 8 at 10:39
• Now there's an idea: exploring esolangs. Sadly, these guys, since they want us to understand, would know better than to write in a self-modifying language. A 2d lang on the other hand... – Mark Gardner Apr 8 at 15:11
• @JoeBloggs: No, the manuscript is not self-modifiable. Only the meaning is self-modifiable. For example in Cobol: after the statement ALTER CREDIT-ACCOUNT TO PROCEED TO DEBIT-ACCOUNT is executed, all statements of the form PERFORM CREDIT-ACCOUNT will actually mean PERFORM DEBIT-ACCOUNT. The program text (= the manuscript) remains exactly the same, but its meaning changes. – AlexP Apr 8 at 17:12
• @AlexP: That’s exactly what I meant. I used ‘self modifying manuscript’ as shorthand for my first sentence: “A manuscript with self modifying rules of grammar”. In much the same way that the rules for decoding an enigma encoded letter were modified by the letters that came before it. In this sentence all the vowels aftar tha ward vawal beceme the next vewel to bo osod. That’s a pretty simple bit of self modifying grammar and it’s already nearly unintelligible to people who know the script and the language it’s based on! – Joe Bloggs Apr 8 at 18:43
• Nitpick: In Whitespace, any character is allowed, but the only ones that are meaningful are space, newline and horizontal tab. Part of the idea in the design was steganographic programming (that is, hiding a program inside another text. – Austin Hemmelgarn Apr 8 at 21:08

The text is 3-Dimensional.

Normal text - such as you are reading now - is 2-Dimensional. Each letter has width and height, and is arranged in a 2D plane to form words, sentences and paragraphs. Morse Code, on the other hand, exists purely as length, and can be written in a 1D form ∙-∙∙ ∙∙ -∙- ∙ - ∙∙∙∙ ∙∙ ∙∙∙

Our eyes each see flat 2D images. Our brain them combines 2 of them, with a displacement, to approximate a model of 3D space. Your precursors, on the other hand, had sensory organs which could see 3D images, and combine them to approximate 4D space.

Part of being able to see 2D images is that we can see inside a square. A being with 1D vision would not be able to tell the difference between █ and ◘. A being with 3D vision would be able to see inside a cube, and differentiate between 2 characters that look the same to us.

As such, we would first need to realise that the 3D pattern of objects is a page of text. Then we would need to scan inside the objects (using X-Ray or MRI) to work out which character is which.

To your alien species, it was a book. To us, it's just abstract art. After all, we don't write our instructions and warning lables in Morse Code just in case a species with 1D vision come across it.

This means that - even if we knew and understood the language when spoken or written in a different script - it's the writing system used (as per the question) which we can't decipher. We can't tell the words/letters apart, and may not even realise that they are words/letters, rather than decorative pillars or moulding.

• I would actually call text 1-dimensional. You could replace every character with Morse code and nothing would be lost. Your eyes move over the text in straight and perfectly regular paths. A 2-dimensional script would incorporate moving up/down, and changing direction; basically like Islamic calligraphy. – KeizerHarm Apr 8 at 11:36
• @KeizerHarm The characters of which the text are formed are 2D though. If you were to reduce it to a 1D line, you might not be able to tell the difference between O and Q or L and T. – Chronocidal Apr 8 at 11:41
• That's an implementation detail. English is a visually 1D language. There is nothing about English that requires 2D glyphs or arrangement. As @KeizerHarm said, English wouldn't functionally change if every glyph was replaced with its Morse code equivalent and each paragraph replaced with a long line of dots, dashes and spaces. There IS a difference between language and script. – jaxad0127 Apr 8 at 16:41
• @jaxad0127 Correct. And the script is the writing system which we can't decipher if we don't realise it is a writing system instead of just being identical stone blocks – Chronocidal Apr 8 at 16:52
• so something like a dolphin (that uses echolocation, and can "see" in 3-D) could read this? I'm not sure how to conceptualize a dolphin approximating 4-D space though... – B.Kenobi Apr 8 at 21:53

Here's a few features that could make your language harder to learn:

# 1. Ommissions

Many languages leave out parts of a sentence that are redundant, either because the rest of the sentance already encodes this information (spanish "voy al cine", roughly "am going to the cinema", where the first person singular conjugation "voy" makes the adverb redundant) or even from the context (japanese does this a lot).

This is usually not done as much in a scientiffic context, but maybe in your ancient culture it was a language feature that wold even be used in instruction manuals.

# 2. Weird inflexion rules

Sure, turning a singular into plural by adding -s makes a language easy to learn, but where's the fun in that? Why not, for example, rotate the first vowel forwards and the last one backwards in the order [a, e, o, u, i], with the exception that one-syllable words do something completely different?

# 3. Strange spelling rules

The easiest scripts to learn are those where one symbol always has just one pronunciation. So why not do the exact opposite? Again, look at Spanish and how an r at the beginning of a word becomes a trill, but is otherwise a tap (and the trill would be written rr); now imagine if lots of words did that; maybe a t turns into a d if at the end of a word, and a v followed by a g turns into an m. And of course, all these changes interact with the weird inflexion system, so the plural of a word may have completely different symbols than the singular.

# 4. Another language using the same script

If your ancient culture was really awesome, maybe its script was adopted by many other languages, but they all adapted it to their different phonology in ways that confuse modern linguists. Maybe some ancient king liked the way the symbols looked and ordered someone to make a similarily fancy looking script for his own language (which, until not long ago, might have been writing on clay, so their existing script really sucked for writing on paper, so it was actually adopted quickly), leading to many similar looking symbols that modern day linguists might mistakenly assume evolved from one another and are somehow linked.

# 5. Lack of translated texts or maybe even fake ones

So imagine two poets do some sort of art project: They decide on a topic and some features (whichever ones are valued highly in their culture, be it verse length, rhyme structure or the number of ascenders and descenders on each line), so the two texts look similar. But each poet writes his completely own poem. And them, you guessed it, they write both of them on some stone as a gift to a king or something. Fake rosetta stone: complete! Just imagine the agony of the linguists trying to figure out how this one text translates to the other one.

# 6. Missing manuals

If the ancient culture was smart enough to make a manual for the manual, this one could still suffer from the same problems, or better yet, it could be (partly or completely) missing or have been destroyed, leaving out some of the most critical parts.

• Upvoted for number 5 alone. These ideas are all great! – Mark Gardner Apr 10 at 12:23
• The "Weird inflexion rules" aren't even all that weird considering Indo-European ablaut (past moving the stem vowel to back /a/ and /o/), Germanic umlaut (plurals raising and fronting the stem vowel toward /i/), as well as the whole root-and-pattern paradigm of Semitic. And the "strange spelling rules" are just sandhi: "elf, elves". "Another language using the same script" could describe Cherokee, whose forms superficially resemble the Latin that inspired Sequoyah to devise the syllabary. – Damian Yerrick Apr 12 at 17:33

Ideograms.

Ideograms aren't phonetic; they convey an idea (and usually the word that carries that idea), rather than a sound sequence (like a phonetic alphabet would). They usually (at least in human languages) descend from pictograms, but have become abstracted -- so an ideogram that means "bird" may not resemble a drawing of a bird. Sometimes, however, if you know the ideogram, they can be seen as a variation of the original pictogram. If you don't have decades to study and a plethora of context clues, or the equivalent of a Rosetta Stone, most ideograms are just a jumble of strokes.

And the thing with ideograms is that, without a key (like a Rosetta Stone) there is no practical way to get from ideograms back to the spoken language, never mind decode what the ideograms mean in another language.

Technically, this isn't impossible to translate -- with a sufficiently large sample database, and context clues for enough of the samples (for instance, "this sample was repeated near most of the apparent switches on what the engineers say was a power generator"), one might eventually get to a partial translation -- but even with that, never be able to read obscure symbols, because they represent concepts that were well known to their intended audience -- and were thus never defined in terms of other symbols (at least in the samples available).

• Written Modern Standard Mandarin is Modern Standard Mandarin. It is not some sort of abstract ideal writing system which could be read in multiple languages. Chinese characters are not ideograms, they are logograms -- they represent lexemes, not ideas. Many of them actually have some internal structure hinting at the phonetic value. We have been able to reconstruct the phonetic form of Middle Chinese quite satisfactorily, and a consensus is slowly forming even for reconstructing the phonetic form of Old Chinese. It helps that the Chinese did write books on how to pronounce Middle Chinese... – AlexP Apr 9 at 13:09
• Seems like there's a lot of misinformation out there (written in English, obviously) about how Chinese writing works, then. I'll try to edit out the reference to Chinese. – Zeiss Ikon Apr 9 at 13:20
• This is the “Ideogram Myth”, that hieroglyphic writing systems primarily encipher ideas. They do not, and no natural human writing system ever has. Ideograms are not “descended from Pictograms” but rather the opposite. No writing system ever has been, nor arguably can be, primarily ideographic. It is just too wildly impractical and unwieldy, and ultimately incomplete and unusable without phonetics. Because unlike ideograms a character set based on phonetics can be finite, yet express an unlimited number of words, sentences and phrases, without ever having to add another character. – RBarryYoung Apr 11 at 14:22

# Pictographs - but based on an alien body form.

When humans try and make universal signs they tend to use pictographs that can be easily understood by people - regardless of what language they speak, or if they can read. Pictographs will also have a good chance of being understood in the future - a rock will still look like a rock in the future and we will (probably) still be humanoid in shape. There are still challenges to explaining the concept of "danger" in pictographs (see Nuclear Semiotics) and how you can get your instructions to last millennia without being eroded, but you can see how pictographs are still our best shot and getting the message of "danger" across. The problem is that if you don't have a human body plan you won't understand that an image depicts "running", or "drowning", or "hiding", or "giving" etc. Our hands are important for interacting with things, our head is our primary sensory organ (where our eye and ears are), our legs are for locomotion (not for interacting with things) - understanding this body plan helps us recognize pictographs (like somebody drowning - with arms flailing and head barely above the water line) that would be hard to interpret otherwise (maybe aliens can breath underwater, or breath with a hand-like appendage, or would be flailing with small manipulative scales or something). So if you came across a pictograph that this alien species thought was self explanatory, say a depiction of decapitation resulting from pushing a button - you may think that the aliens "head" laying on the floor is just a plant or rock or something and that pushing the button turned on a light rather than dangerous laser beam.

# "Web of ideas" not "sequence of words"

Our writing systems evolved as a way to transcribe spoken language, which is inherently linear. But our brains operate with networks, and "brainstorming" and "mind mapping" techniques help tap into that. (Poorly.)

A writing system could literally be a web of interlinked ideas; properties describing objects that are involved in many parts of a "paragraph."

I once toyed with a writing system like this... I only got as far as appearance. It looked like this:

Naturally it would be a bit of an art form, a form of visual poetry. Formal documents would be neat and straight, cursive would be interesting and much harder to decipher. If the civilisation didn't realise that other writing systems would be so conceptually incompatible, their "help" might be almost useless. Where do you start in a giant maze of 2D mess?

• The problem is that this is arguably so impractical and unwieldy as to be virtually impossible. Consider how such a system would encipher this English sentence “Brunhilde helped Tutankhamen move from Mississippi to Singapore.” Every existing writing system can encipher this, even in the original English, simply because they are phonetic. But one based on networks of ideas cannot because it has no way to encipher names that are not already known. Sounds are the only way that we have to do that. – RBarryYoung Apr 11 at 14:43

What about the same reason we couldn't decipher hieroglyphs and cannot Linear A. We not only don't know the language but we cannot decipher the "shapes/signs".

Bah, you could place the text in, non text position. YOu have a sing on the wall "do not touch. High Voltage" and a picture of skull and lightining. And then someone graffiti "gorg was here". Hmmm I know! What this means is "don dead, open inside"! There is a laughing skeleton of Ziggy Stardust. I've seen the pictures.

Or take Quipu for example, a method of "writing" that was even used by conquistadors, so rather fresh knowledge. But as of right now we, ballparked, only numbers in this writing. We know there could be more but we are unable to decipher it. While we also know that similar method was used in China for example.

So you could have a message

First shalt thou take out the Holy Pin, then shalt thou count to three, no more, no less. Three shall be the number thou shalt count, and the number of the counting shall be three. Four shalt thou not count, neither count thou two, excepting that thou then proceed to three. Five is right out.

And all you can read is 1, 3, ><, 3, 3, 4, 2, 2, 3, 5.

• I'm sorry, how does this answer my question? "They just can't read it"? – Mark Gardner Apr 8 at 7:40
• @MarkGardner The question "How do I make a writing system undecipherable, while not intended as such in-world?" – SZCZERZO KŁY Apr 8 at 7:45
• @MarkGardner Exactly, just like Egyptian hieroglyphs before we could translate them. Sometimes the answer can be remarkably simple. This is the simplest answer. The modern world has no knowledge of the ancient civilization's language. That's it. Problem solved. – a4android Apr 8 at 12:10
• I was not onboard with this even being an answer until I got to the monty python quote, but in it he make an oddly good point. "Do not touch" is in many ways easier to figure out than when someone adamant on explaining something over explains it so that your translator can not make out enough to come up with a confident explanation of its meaning. – Nosajimiki Apr 8 at 14:07
• Actually, there’s about a 50% chance that the Linear A writing system is already known: it’s very possibly the same one that Linear B uses (their character set is very similar). But if it is, then we cannot translate it because it’s for a language that no one recognizes, and apparently unrelated to any known language. – RBarryYoung Apr 11 at 14:30

# Non standard or variable flow.

English is read left to right, top to bottom. Traditional Japanese is top to bottom, right to left. Most languages follow a similar flow of words fitting into rows or columns, read in a specific order.

This language doesn't.

Maybe the words are read in a spiral. Or a four by four box of characters which are left to right, bottom to top. But then the boxes are read in a snake pattern, starting in the top right and read left in the first row, and then left to right in the second row.

Maybe the flow alternates. Left hand pages are read left to right, right hand pages are right to left.

Maybe the flow varies by usage. Scientific and mathematical texts are written like English, prose is like traditional Japanese.

Or mix them. Left hand pages spiral out from the center, clockwise; right hand pages spiral in from the top left, counter clockwise.

Without knowing the unique rules that determine how the characters flow on a page, establishing a pattern to find words becomes much harder.

Thinking laterally… you could have different visual range (ultraviolet or infrared) in the original species. To most visitors, the manual would appear to be blank (or to have weird patterns that were actually artefactual).

One might be able to come up with other incompatible communication modes. (Massive scale difference does not work (?).)

Perhaps you could arrange it so that, when the visitors get to the point of looking for the instructions, they have already walked past them… sort of concept. [“Use of this site is deemed acceptance of licence agreement.”]

(I am thinking that an advanced civilisation could use video, or flashing lights and horrible noises activated by moving to press a button, to [try to] warn any visitor of serious consequences. (…Although conversely “horrible noises” might not be universal. Conversely again, any detectable reaction to only some actions should get the attention of just about anything with a brain.) Similarly, it is hard to mistake the meaning of having some buttons with shields on them.)

On top of the other suggestions, use idiomatic synthesis. Different symbols are just ideas, thrown together without linking. New words are constructed from old words as much as possible. It somehow resembles expressionist art.

For example, words "fire house" together mean "oven", while "house lost" means "a shelter for homeless", and "big fire" means "volcano". The phrase "big fire house lost" is perfectly correct, but it can mean "Big oven was lost" "Big fire in a shelter for homeless" and "A shelter for homeless is near the volcano", depending entirely on the context of other sentences.

This is what makes automatic translations from Chinese so legendary hilarious, btw. You can not parse Chinese grammar without understanding meaning.

• "Big fire house lost" could also mean "tell the homeless that the volcano is a shelter" or "the homeless shelter is in the volcano", if you want to inject a little morbid humour and/or circumstances where misinterpretation could be potentially fatal. – Justin Time - Reinstate Monica Apr 10 at 22:37

Part of what makes translation difficult is context. Ancient Sanskrit is extremely difficult to read because the writers seemed to enjoy writing passages which were full of puns. While this may have been hilarious for other readers at the time, we don't get the puns because they have no context for us.

The contextual environment the language is written in is important. What they considered "common knowledge" or "common sense" may be entirely different from our concepts of common knowledge or common sense. Similarly humour, metaphors and other concepts won't translate across cultures, time and space.

Other cultural quirks could make reading documents difficult or impossible if you have no experience or exposure. Modern military writing is full of acronyms, many of which actually look like words to the uninitiated. The US Army, in particular, also has a strange tendency to turn words into verbs, even when the word is not a verb at all in conventional english. So a reader sees a familiar word being used in a way which makes no sense. Imagine a native speaker of an Amazon tribe being presented a US Army field manual.

Combining these factors in a relatively small document where there are no other examples of the language to make comparisons of statistical analysis should provide an almost insolulable problem for the discoverers. One would hope there is enough common sense not to push the large button in the console...

• From a linguistic point of view, acronyms are words. So a text full of acronyms that the linguist doesn't know isn't any more difficult than a text full of non-acronyms that are equally unknown. Using verbs as nouns or vice versa isn't really that unusual (it's called 'conversion'). The linguists would simply realize that in this language, lexical categories are highly flexible. Granted, these two points do add additional complexity, but they're really not something that would surprise a linguist. – Schmuddi Apr 9 at 8:24
• Another good example are memes. Understanding them requires understanding their context and source material, which sometimes includes multiple layers of references. Perfectly understandable today, but in a century will be completely unintelligible. – bta Apr 9 at 18:12

# Crystal engraving, compression, then cryptography

In the book/film Contact, there is a long process from a distant alien to transmit very specific instructions, without assuming almost nothing from us. First a medium, then regularity, then patterns, then universal concepts, and then, instructions. Your manual, basically.

But reality editor is a very, very dangerous thing, any civilization who would use it must be vary advanced. So advanced, for example, to have:

1. Computers;
2. A lot of computacional power;
3. And then, something with is hard, even for computers.

So your incomprehensible manual can be:

1. A large, crystalline structure of bits in forms of internal dots;
2. The bits first encode a compressed text with two parts, inicial instructions similar to Contact; and
3. Further texts, not only compressed, but also encrypted;

So... No pre-industrial civilization could read theses dots. Not pre-computacional civilization could understand the dots (no language structure on compressed data). And no advanced civilization could even read the texts.

Bonus 1: make the encrypted part a sequence of texts, each one with increasing difficult, so have a "not now, bur almost" atmosphere.

Bonus 2: the machine have no buttons. It's only accepts "complex crystal instructions", compressed and encrypted as the final text.

So you can have a point where everyone knows exactly what the machine does, with no one capable of complete use. Each text explains one "machine instruction", but the whole "language" it's only at last texts.

No spoilers all or nothing manual, and also no misfires in the past!

• Sadly, this would go against wanting people to understand. If they wanted to set a challenge, that would be at the door, not in the manual. With these ancient encrypted things, people would have noticed the big shiny buttons waaay before the manual. Tl; Dr: This, sadly does not answer the question in a way that makes sense in context. – Mark Gardner Apr 11 at 20:04

I remember a science fiction story in the same universe as Larry Niven's Ringworld where an alien civilization had a language based mostly on absolute pitch. This meant that singing the message without words but with the correct pitches was usually understandable. No mention of their writing method, but can you imagine a script for that language?

For difficult human languages, consider Etruscan. The Romans copied their script, so we can probably pronounce the texts they left in caves well enough they would understand them. However, the Romans also wiped out the spoken form of their language, so we still don't know what those texts mean.

• Actually (if they have writing at all) it will probably be a lot more decipherable than their "verbal" language. – Matthew Apr 9 at 21:13

## Make certain parts of the writing optional.

There are real world examples how this can break things. There is an example I think somewhere in the Old Testament where God's name is written as JHV. Hebrew (maybe aramic?) didn't write vowels, so some interpret this as Jehova, others Jahve.

A bit newer example (only 519 years old) is this old Hungarian script found in Csíkszentmárton from 1501. The language is well known, changed only a bit, with about 10-15 million speakers around. The writing system has tiny variations, but is simple to learn (took me around 2 days to read without any problems), and suits the language well (none of those silly double letters and accents :) ). However the vowels were only optional, so some of the names were hard to figure out (I'm not sure if "eSTYTáN" is right), and the end of the message ("gijas ily ly lta") is still a mystery.

# Sorry Mario, but your manual is in another castle

They found de machine chamber, but not the control room. The manual are in control room (obviously).

Near de machine they found some logs (of previous test, previous runs), where is regularity, but no language. In the same vein, they found some test buttons near the machine, what could operate the machine in a limited way.

# Sorry Mario, but your manual is away from castle

The manual is in a tablet-like form, but depends on the power of the machine to works. It got removed and studied far, far away from machine.

# Don't microwave the manual

The same, but the manual was activated away from de the machine, after it's discovered x-rays make it works on a broke way. This damages the manual, to produce only gibberish.

# The machine starts in tutorial mode

For a long time people was navigating de manual with no consequences. Until the tutorial was over and the machine stayed on. No one knows to turn off anyway, and these lights are pretty.

"Always remember to jump, jump!"

# The manual is audio only

Speech is really hard if you know nothing about the language. Even hard are not hearing the real thing. Unfortunately the machine got it's manual calibrated in the frequency where humans start hearing things.

• I feel like there are quite a few excellent ideas here, but I'm afraid I don't understand what you have written well enough to vote (or accept) – Mark Gardner Apr 12 at 20:20
• There are various, concurrent ideias here. For how to vote, choose a StackExchange answer you like more, or on a answer with the ideia you may use. – André LFS Bacci Apr 13 at 14:10

They were obsessed with religion. They couldn't write anything that wasn't either prayer to or glorifying of a deity (and they were a multitheistic society, every type of endeavor had it's own deity.)

Forbidden:

Two hydrogen atoms plus one oxygen atom produces water.


Permitted:

Two of [god of atoms] first creation, when combined with the blessing of the [god of chemistry] what [god of atoms] created thank [god of life] on the eighth cycle produces [god of water]'s domain.


The more complex the situation the more deities are involved. Speakers of the language know how to separate the prayer from the meaning, someone trying to translate it is going to be horribly confused by all the extraneous symbols.