The Ur-Quan Masters Discussion Forum

The Ur-Quan Masters Re-Release => General UQM Discussion => Topic started by: Art on August 12, 2004, 02:18:15 pm



Title: Languages in SC2 (new thread)
Post by: Art on August 12, 2004, 02:18:15 pm
Quote

Most races, before achieving hyperwave broadcast, sent out many many radio signals, These radio signals were collected and stored by the Mhnnrhmn, the only ones i think could have.
After exploring all of known space, the Mhnnrhrhrm built translator devices, and with the cross referancing of the multitude of words, and a lengthy study of this, they were able to produce a workable UTD. The Orz are garbled because they're new, and from a different dimension.
The Slylandro use a Precursor satellite, which i assume translates extremely well.
The Ur-Quan both use Talking pets, which translate thought to speech in any language. ( If your gonna slam Talking pets communicating, then just think how ridiculous an idea Telepathy really is.)
No matter their differences, all races immediately realised the enormous importance of communication, regardless of their opinion of the Mhrhrnm.


Well, first of all, this might be an interesting fanfic idea but the role goes much more easily to the Chenjesu than the Mmrrnrrhrrm -- the Chenjesu are the ones who are the de facto leaders of this part of the galaxy, the ones with a sublimely powerful intelligence, the ones who are an ancient race who've been around long enough to do this, the ones with a compassionate interest for everyone's welfare, and the ones with bodies that act as natural receivers of various kinds of radiation. The Mmmrrnrrhrrm have no natural advantages toward this sort of endeavor; they're just sentient robots, big deal. (Moreover they're sentient robots without a particularly high level of knowledge or purpose. The Mother-Ark supposedly just broke down and left them without memory of their mission, remember? My impression was that the Chenjesu took them in because the Chenjesu are good neighbors, not because the Mmrrnrrhrrm were their equals. Also, notice that they arrived in this sector fairly *recently*, and probably haven't been around long enough to do an in-depth study of every neighboring race.)

But anyway, this is needlessly complicated. This isn't a system for building a true UT, one that can translate a truly unknown language on the fly; it's an explanation of the building of a "universal" translator for this part of the galaxy, that can only translate the list of languages that are in its preprogrammed dictionaries. It'd be powerless to translate a truly new language -- if Orzese is truly an entirely new language it couldn't translate *any* of it, much less put out "linguistic best-fits". Unless you're saying there's some "universal structure" that underlies all languages and can be analyzed, which is a poetic idea (dating back to the Biblical Tower of Babel) but one that has little basis behind it. (It's the same kind of Star Trek logic that says there's something universally useful about the human shape, so that all intelligent aliens have to be humanoid. Unimaginative nonsense.)

Anyway, there's no particular reason that such a centralized translation project has to exist. If the Chenjesu or Mmrrnrrhrrm had time to listen to and study everyone's transmissions long enough to make up an enormous dictionary, then why can't *other* civilizations everywhere make their own dictionaries by listening to everyone else's transmissions? They might do it slowly and by detective work by spying on, say, TV transmissions (where words can be matched to pictures) or they might do it much more quickly and easily in cases where they actually make contact with other races and mutually cooperate to make a dictionary. Then they could actually *learn* other languages the ordinary way; train specialists among their ambassadors, diplomats, merchants and soldiers; offer courses in  college (or the equivalent) for young people; that sort of thing.

It's a lot longer and more boring and *realistic* than the idea of making a UT, but it makes one hell of a lot more sense, and it's not contradicted by *anything* in Star Control 2, and actually makes most of SC2 make more sense.

If you think about it, a UT *would* be telepathy, in a sense; it'd be pulling the meaning of your thoughts straight out of thin air, or buried from within a text, with no other information. I fail to see any good explanation for how a text can, itself, contain explanation for what the text means -- it *can't*. If I make up my own language that no one else knows and write a book in it, there's *no way* to figure out what the book means, *ever*, without reading my thoughts. If you hear a language that neither you or anyone else has ever heard before, there's *nowhere* you can take the meaning of the language from. It doesn't exist in the words themselves -- you have to translate the language by detective work, by comparing words to objective reality -- physical objects, numbers, that sort of thing. That kind of detective work takes some time and interaction to do; I can buy that different races make pre-programmed translating devices or else just *learn the language* (because learning to speak a natural language yourself is many times easier than programming a machine to speak a language) based on long study of TV transmissions and such. Not that they can make a machine that does it within seconds of meeting a new alien race.

The idea that we take psychics seriously and believe that all sentient minds exist in some sort of common undetectable energy field, encoded in common patterns, that can interact with each other may be far-fetched, but at least it's a speculation about something which we know nothing about. We *do* know how ordinary spoken and written languages work, and what we know says that making computers that can magically translate a foreign spoken language with no prior knowledge of it is ridiculous. (It's like how believing in sentient energy-beings somewhere in the galaxy is far-fetched but in a way not as far-fetched as believing that, say, toasters and tennis shoes are sentient and we just don't know it.) Similarly, the idea that Talking Pets exist is far-fetched, but it makes slightly more sense than saying that a machine that does the same thing as a Talking Pet is buildable by every sentient race in the sector and comes standard-equipped on every starships.  

Quote
Now, Accents.
The UTD has been built after a huge length of time, and the Mhrhmmmnm have no knowledge of any type of accent, because theyre machines. So, any translations would carry over the original accent, because they can't distinguish between say, an umgah accent, and a texan accent.


Um. No. This doesn't make any sense at all, if you think about it. People have accents because of varying habits of pronunciation, but that, itself, has nothing to do with meaning. Particularly, people have *foreign* accents because they're in the habit of speaking one language that uses a certain set of sounds and when they try to use a different language with a different set of sounds they use the wrong ones. Accents exist because of human frailty -- because we don't reproduce sounds perfectly when we learn them.

Computers, however, *can* reproduce sounds perfectly when *they* learn them. If you teach a computer how to say the word "rice" a certain way, it will always remember that specific pronunciation. If you teach a computer that "rice" in English and "arroz" in Spanish mean the same thing, then when translating it will replace an "arroz" pronounced in Spanish with a correct Spanish accent with a "rice" pronounced in English with a correct English accent. That's the basic way a translator would work.

Now, you might want to program the computer to *recognize* different accents. In some Spanish-speaking countries the "rr" would be very hard -- sound almost like a series of "d"s in English -- while elsewhere it'd be very soft, like a light "l" sound. In Latin America the "z" would sound like a "s", while in Spain it'd sound like a "th". And if the speaker were an English speaker for some reason trying to speak Spanish she might say it as "a rose". Recognizing all of those and turning those to "rice" would be the translator's job. Similarly the translator could be programmed to say "rice" different ways -- if the guy making the translator thought that the proper long "i" in "rice" was more likely to be understood, he might program that word into the computer; if he were from the American South he might program it with a longer drawled "ah" sound for the "i". That sort of thing.

There's *no reason*, however, for the computer to try to carry over an accent from one side to the other. Yes, a Spanish speaker would have a hard time pronouncing the word "rice" perfectly, and it might come out sounding like "dice". The translator, however, would be stupid if it worked by first connecting "arroz" to "rice", then stopping and thinking, "But Spanish speakers couldn't pronounce that first 'r' perfectly without practice, so I'll make it sound like 'dice' or 'lice'". That's *extra work* on the translator's part, since it *already knows* the correct pronunciation and doesn't need to practice it, like a human does, to get it right. And it serves no purpose, since it makes the resulting translation *harder to understand* for the user, and the whole point of a translation is to try to get the user to understand as clearly and easily as possible.

I'll put it another way -- if the machine is telepathic and can pull the meaning out of the speaker's head, there's nowhere it can get the accented pronunciation of "dice" or "lice" from. Inside the speaker's head is *only* the word "arroz"; if the speaker doesn't know English there *is* no word "rice" in there for her to mispronounce. When the translator translates into the listener's language, it pulls the word "rice" out of *his* head, and *he* knows how to say rice correctly in *his* English-speaking accent. There's no reasonable mechanism for the two different words to collide together in such a way as to come out as an English word pronounced with an accent that simulates a Spanish-speaker who's at a certain point in the process of learning English.

Any machine that can be programmed to perfectly speak a language -- or that can pull a language magically out of thin air through telepathy, or whatever -- won't use accents. Accents *only exist* because flawed humans, or other human-like organic beings, take time to train their mouths and vocal chords into the habit of using phonemes that they haven't used before. A Spathi who speaks English with a weird accent *must* be a Spathi who *learned* English at some point. Foreign accents are *nothing* but markers of the way someone learned a language, and how well they have learned it -- someone who speaks no English *has no accent* in English, because they don't speak English *at all*! For someone to have an accent they must first learn to speak English, and then the precise nature of the accent they have depends on how well they've learned English -- at the beginning it may be so thick as to be unintelligible, at the end of the process, if they study hard and practice hard, they may have no detectable accent at all. To assert that there's a natural level of Spanish accent a Spanish speaker "would have" if they spoke English is ridiculous.

All this holds doubly true for a race that doesn't "understand" accents, since they won't bother to do anything for flavor (the only reason -- and a rather offensive one -- I can think of to make a machine that translates, say, a French person as going "I zink I zhall depaht now" just so you can tell they're French). They'll study each language in its natural accent and make a machine that translates naturally-accented Spathi *directly* into naturally-accented Human (English) without any frills; they'll have to, because naturally-accented Standard Spathi and naturally-accented Standard Human are the only versions of Spathi and Human they've heard, if you're proposing that the Mmrrnrrhrrm or Chenjesu or Melnorme or whoever started their project before there was *any* contact between other races on their own (which I find unlikely, but whatever).

Quote
As for sound traveling in space, I always assumed the ships computer simulated the sounds for the benefit of the pilot. That way he can use more than his sight to fly


Cute idea, but it's a bit stretched. I mean, without special training most people's perception of sound is a lot blurrier and less precise than their sense of sight, even their sense of sight when abstracted into some sort of instrument like a radar display. I imagine having lots of random sounds going off around the pilot would be more distracting than helpful, though having never been a pilot I wouldn't know. Even so, if I were going to make such a system I'd make it more directly useful -- like, say, a proximity alarm that warns you when there's a large object or energy discharge outside your field of view -- rather than going to the trouble to make nifty sound effects for different kinds of weapons (so that, for example, I could hear the sounds of plasmoids going off even when they're very far away and I can clearly see them, but an Androsynth in Blazer form creeping up behind my ship is completely silent -- not very useful).

The sound effects as they work in the *game* are clearly for our enjoyment as players who are seeing the battle from an unrealistic above-the-field 2D perspective, and who enjoy hearing explosions and things.

Quote
The VUX do have a UT; their tech in that field is above anyone else's current tech.  That's why Rand got caught.  Presumably, earthlings have limited UT (mostly for the U.N.), or something they got from another race.


How is their tech above others' tech? The incident as described just leads us to think that Rand thought the VUX didn't know their language at all and that the VUX did -- he thought they were going to initiate "first contact procedures" (which probably include standard procedures for establishing language translation -- sending over picture-dictionaries for the computer to download, giving a greeting for them to record, etc.) but the VUX already knew English. It doesn't say that, for instance, he thought he wasn't speaking into the microphone and he was, or he thought the transmission was off and it wasn't. If true Universal Translators (that don't need pre-written dictionaries) exist at all Rand should've expected to be translated if he was heard, if not right then, then later (since they were probably recording the transmission in the log). He was clearly surprised, not at being heard, but at being *understood*, meaning he didn't expect the VUX to have *learned* English so quickly, meaning languages still have to be learned in this day and age, not magically translated with a magic device.

The idea that us humans could invent anything like a UT by the 22nd century on our own stretches credibility. Even making a plain ol' regular translator between two very well-known languages is a painful, excruciating task given all the ambiguities and such in natural language, which is why so many human interpreters and translators are still in business. Even ignoring the "universal" aspects of a universal translator, a machine that simply accepted data describing several languages and then was able to translate whole texts from one language to another naturally, without distortions and awkward phrasings and such, would be an amazing feat of artificial intelligence.

Quote
The Precursor ship does have a UT, and it only hiccups with the Orz and (sort-of) Arilou.  I say sort-of with the arilou, because they seem to be putting in the lingual best-fits themselves, in your language instead of making the computer do it.


Er, if the Arilou know our language well enough to be able to predict what the translator will and won't translate, and then *put* the English words in, in place of the Arilou words (which may not even be literally possible if Arilou grammar is radically different from English grammar -- try speaking a mixture of Latin and English and see how many fake rules you have to invent on the spot to get it to work)... why can't they just talk English to us in the first place? It'd probably be easier, for them and for us. (What would the Arilou be doing needing to take advantage of a translator anyway? They're friggin' obsessed with the human race and have been for our whole history; why *wouldn't* they talk to us in English?)


Title: Re: Languages in SC2 (new thread)
Post by: Art on August 12, 2004, 02:23:13 pm
Quote
The Ur-Quan don't care about learning other languages; it's the act of communicating itself that the find distasteful, no matter the language.  I doubt a UT would help with that aversion, so despite what some people say, the Dnyarri may have a job even after SC2. *shrug*


This doesn't make sense either. At least not to me. It might make sense to an Ur-Quan, I don't know, but I don't see it doing so. So the Ur-Quan don't like communicating with non-Ur-Quan sentients? Well, if the Talking Pets count as sentient by the Ur-Quan's definition, then why should talking to a Talking Pet be any different from talking to a human? It'd be worse, because Talking Pets are both lower on the scale of non-Ur-Quan life than humans in their reduced mental state and are living reminders of the most hated figures in Ur-Quan history.

If the Talking Pets are *not* considered in any way sentient and are the organic equivalents of a machine, then there doesn't seem to be any obvious difference between talking through a Talking Pet and talking through a machine. In both cases the Ur-Quan isn't speaking directly, but through an intermediary device, just one's organic and the other's artificial.

Yes, you might say that the Ur-Quan aren't physically speaking to the Talking Pets, they're just thinking or thought-speaking. Fine, that's a compelling theory, but, first of all, that's still communicating -- even if they don't move their mouths, they're still composing quite detailed and ordered series of commands to you in *mental* speech. They're paying *attention* to you and thinking about what to say to you, and whether sound vibrations move through the air or not doesn't seem relevant to me. Typing a message on a keyboard or using some other indirect medium that's clearly different from "normal" communication would work just as well. Maybe the Ur-Quan have some fetish about thought-speaking being non-real communication since it's wholly nonphysical, but I doubt it -- they were an entire race that communicated with their slavemasters mentally for a long time (this has to be true, because one Dnyarri could rule a whole planet at once); why should they think that telepathy doesn't count as real communication?

If they really disdained all communication, they wouldn't speak to you in clear, composed speeches the way they do; they'd lounge around and idly think, and their thoughts would trickle down to you in the third person, the way the Orz's dialogue sometimes sounds. ("Oh, look, it is a human. If the human leaves the boundaries we shall be forced to destroy it. Stupid humans. Why can't they accept our wisdom?") If they really thought of Talking Pets as intermediaries keeping them from direct communication, I'd think they'd do something like compose their thoughts out of sight ahead of time, and have the Talking Pet deliver messages to you by itself, and avoid having to actually see you on their screens in real time -- after all, you'd still be talking to them even if they weren't talking to you.

My opinion? The Ur-Quan seem very obsessed with purity of their own culture; they don't aver physical contact with other species, at least I don't think so, since they have Battle Thralls do all incidental work on their ships and such. But they do fear being contaminated by other beings' way of life -- they have a slavish devotion to their Path of Now and Forever as a self-contained philosophy of *everything* they need in their culture, and as we've seen the first thing they do to a new slave species is annihilate their pre-existing culture by destroying their landmarks. They allow Battle Thralls to serve on dreadnoughts, but they use no Hierarchy ships in the Doctrinal Conflict, probably because they don't want thrall captains using *their* strategies and cultural fighting styles against the Kohr-Ah.

To me using the Talking Pets doesn't feel like avoiding direct communication -- they still indulge in real-time conversation with you and give you *visual contact* with their persons, and converse with you quite naturally, speaking directly about themselves as participants in the conversation, allowing you to ask questions and responding to your comments, etc. The Talking Pets are quite obviously used as *devices* for translating, not as go-betweens. (Sort of like how when telephones were first invented people talked through them, telling the telephone to "Tell Margaret I love her" instead of saying "Margaret, I love you". Compare that to how casually many of us treat typing in a box on a screen as "talking to Margaret".)

If the Ur-Quan want to remain culturally uncorrupted, they *cannot* ever learn anyone else's language; to learn another language is by definition to become at least partially immersed in another culture's presuppositions and ideas. That explains why the need for the Talking Pets, rather than the simpler expedient of communicating by delayed recordings or written text transmissions or whatever -- the Talking Pets are the only *true* Universal Translators in the SC2 universe, thanks to their magic psychic powers, and as such the only devices the Ur-Quan can use to talk to people without ever learning their languages or letting them learn the Ur-Quan's.

Finally, about post-SC2: People who complain about the changes in SC3 seem to prefer asserting that nothing really changes after SC2, which would make the victory kind of hollow, don't you think? Sure, maybe SC3 rushed through the eventual defeat of the Ur-Quan too fast, but c'mon; the *whole point* of what you did was that not just the Ur-Quan's empire but their entire culture, and their entire sense of themselves as a superior species and the heirs of the legacy of the Precursors, would collapse with the destruction of the Sa-Matra. That the Ur-Quan would continue or be *allowed* to continue their high-and-mighty attitude after their humiliating defeat at the hands of the New Alliance is unlikely. And given just how friggin' paranoid the fellas are about mind control, the idea that they'd just ignore the fact that their defeat was *caused* by a rogue Talking Pet and keep on using the things despite how dangerous they are is insane. Both the Ur-Quan's confidence in the Talking Pets and in their own superiority were given a crushing blow at the Battle of the Sa-Matra; their future should reflect that.


Title: Re: Languages in SC2 (new thread)
Post by: Bobucles on August 12, 2004, 09:25:28 pm
Quote
Both the Ur-Quan's confidence in the Talking Pets and in their own superiority were given a crushing blow at the Battle of the Sa-Matra; their future should reflect that.

Absolutely, as long as they don't put on masks, and whine about the loss of the great Ult--- Sa Matra.  ;D


Title: Re: Languages in SC2 (new thread)
Post by: meep-eep on August 12, 2004, 10:44:21 pm
Quote
Unless you're saying there's some "universal structure" that underlies all languages and can be analyzed, which is a poetic idea (dating back to the Biblical Tower of Babel) but one that has little basis behind it.

I disagree. Some things are unversal, and given enough raw data, you should be able to extract them.
There are a few basic things you can communicate. One of the most used ones probably is of the form "<someone> <does something>". Now these things can be contracted and structured in a number of ways, but it's reasonable that a language has recognisable forms. Of course in theory, it doesn't have to, but given that languages evolve as a convenient way to communicate thought, I'd say it's a fair bet that this works with most, if not all, actual languages.
If you've got the basic structure, you can then try to find the meaning of concepts, based on context and common frames of reference.
Those common frames of reference can be fed by your communication partner before the communication begins.

Quote
(It's the same kind of Star Trek logic that says there's something universally useful about the human shape, so that all intelligent aliens have to be humanoid. Unimaginative nonsense.)

The logic behind that is that the ST universe was seeded by an older race. (Explained in episode The Chase (http://www.startrek.com/startrek/view/library/episodes/TNG/detail/68598.html)). Though probably the real reason was the budget and the idea that the viewers prefer humanoids because they are easier to identify with.
I doubt the reason was a of lack of imagination. At least not for all writers.

Quote
I fail to see any good explanation for how a text can, itself, contain explanation for what the text means -- it *can't*. If I make up my own language that no one else knows and write a book in it, there's *no way* to figure out what the book means, *ever*, without reading my thoughts.

If your book is big enough, you can get a long way. Certain concepts will be used alongside eachother a lot, and if you know one, you'll have only limited choice for the other. Once you know the word for "eat", you'll know a lot about the nouns that accompany it. Initially, it's just a matter of trying things, and trying something else if that doesn't lead to a consistent whole.
But a human language is a lot easier than an alien language, as for humans you know you have a lot more common reference points.

Quote
I can buy that different races make pre-programmed translating devices or else just *learn the language* (because learning to speak a natural language yourself is many times easier than programming a machine to speak a language) based on long study of TV transmissions and such. Not that they can make a machine that does it within seconds of meeting a new alien race.

If the communication partner feeds your computer enough data, then I don't see why not. A future computer should be able to process that quickly enough. And it's likely that they would feed it specially crafted text that would make it even easier. Something like a complete text which shows the grammar being used in practice, plus a dictionary and an explanation of the grammar rules to fill in the gaps when you already know enough to read the explanation.

Then again, stuffing a Babelfish in your ear is much easier. ;)

Quote
Accents exist because of human frailty -- because we don't reproduce sounds perfectly when we learn them.

They also exist because we do reproduce sounds perfectly after a while. That's why areas have specific accents of the same language.

Quote
Cute idea, but it's a bit stretched. I mean, without special training most people's perception of sound is a lot blurrier and less precise than their sense of sight, even their sense of sight when abstracted into some sort of instrument like a radar display. I imagine having lots of random sounds going off around the pilot would be more distracting than helpful, though having never been a pilot I wouldn't know.

Have you played fast action computer games? Try playing them without sound when you're used to playing them with sound. Sounds can be very helpful.

As for the VUX, it is possible that communications were not actually initiated (at least not intentionally). The VUX may have found some way to tap into what the humans were saying to eachother aboard their own ship.

Then again, what we read in the SC1 manual is presented as what "The Division of Synthetic Special Reconstruction" has reported. Maybe in fact the VUX were bluffing. They saw the humans, wanted to kill them, but needed a reason. Quickly realising the VUX appearance may had the same effect on Humans, they claim one of the Humans insulted the VUX, thereby giving them an excuse to attack. As it turned out, indeed someone (the captain no less) insulted them.

BTW, assuming the other party can't understand you when speaking through the inter-ships communication systems is foolish, as it is common sense to record everything. Even if they couldn't understand it as it was spoken, you can count on it being analysed afterwards.

Quote
[...](which may not even be literally possible if Arilou grammar is radically different from English grammar -- try speaking a mixture of Latin and English and see how many fake rules you have to invent on the spot to get it to work)...

You would just apply English and Latin rules to words they were not meant to apply on. You're not inventing any new ones.

Re talking pet:
Quote
This doesn't make sense either. At least not to me. It might make sense to an Ur-Quan, I don't know, but I don't see it doing so. So the Ur-Quan don't like communicating with non-Ur-Quan sentients? Well, if the Talking Pets count as sentient by the Ur-Quan's definition, then why should talking to a Talking Pet be any different from talking to a human? It'd be worse, because Talking Pets are both lower on the scale of non-Ur-Quan life than humans in their reduced mental state and are living reminders of the most hated figures in Ur-Quan history.

I'm assuming for a moment that the Ur-Quan talk to communicate with talking pets through thought. Talking takes a lot more effort than thinking. And thinking you'll do anyhow, but by explicitely not talking to specific races you show them they are not worth the effort. Just as symbolism, because Kzer-Za at least do a lot more for "lesser races".
Also, each time an Ur-Quan uses the Talking Pet for translation, it humiliates it, which was exactly the idea in the first place.

And cultural habits aren't always rational and consistent. Compare it to Western Muslim girls who wear a kerchief on their head, as a form of body modesty, and yet put on make-up on their faces.



Title: Re: Languages in SC2 (new thread)
Post by: definite on August 13, 2004, 04:42:04 pm
Quote
If your book is big enough, you can get a long way. Certain concepts will be used alongside eachother a lot, and if you know one, you'll have only limited choice for the other. Once you know the word for "eat", you'll know a lot about the nouns that accompany it. Initially, it's just a matter of trying things, and trying something else if that doesn't lead to a consistent whole.
But a human language is a lot easier than an alien language, as for humans you know you have a lot more common reference points.


Even if the book is large enough, it is still very hard to translate without dictionaries. Yeah, if you can identified the words. But how can you identified the words in the first place?

A human language is a lot easier that an alien language. Therefore, I quite doubt the effectiveness of SETI project unless it has sucessfully translated the ohter human languages  in to english without help of the dictionaries.


Title: Re: Languages in SC2 (new thread)
Post by: Death 999 on August 14, 2004, 12:12:59 am
Well, the SETI project simply assumes that whatever signal there is will have lower entropy than background noise. Which, for high-redundancy interstellar communication, would be a really good idea. We might not get a translation even if we can tell there's a signal, though.

As for decoding languages -- invented languages have been cracked by decoders before. Some of them had rather peculiar syntax not closely related to any human language. In fact, several of them of them were designed to confuse decoders. Yet they were decoded.


Title: Re: Languages in SC2 (new thread)
Post by: Art on August 14, 2004, 11:13:28 am
*sigh* My rants get longer and longer, while the max message length stays the same...

Quote

I disagree. Some things are unversal, and given enough raw data, you should be able to extract them.
There are a few basic things you can communicate. One of the most used ones probably is of the form "<someone> <does something>". Now these things can be contracted and structured in a number of ways, but it's reasonable that a language has recognisable forms. Of course in theory, it doesn't have to, but given that languages evolve as a convenient way to communicate thought, I'd say it's a fair bet that this works with most, if not all, actual languages.
If you've got the basic structure, you can then try to find the meaning of concepts, based on context and common frames of reference.
Those common frames of reference can be fed by your communication partner before the communication begins.


No. You suffer from the misapprehension that language "contains" objective meaning. It contains no such thing. Language only means something because people agree it defines something -- the sounds or letters themselves mean nothing. Without any prior knowledge of what the language means, you will see *nothing* inside of the "data"; you might see that it is ordered and not purely scrambled (the field of information science is based on just being able to tell if a message *exists* and tell messages apart from random noise; it's still a very difficult and complex problem).

English does have a certain amount of order as opposed to chaos -- there are certain letter combinations that occur more often than others, certain general rules about which words follow others, and so on, but that doesn't contain *meaning*. A computer can analyze how much "information content" a message has and write an entirely new message with the same information content that is entirely meaningless. There's a classic sentence "The gostak distims the doshes" that demonstrates this -- there's nothing that makes that sentence inherently less information-full than "The waiter brings the food", and there's *nothing* that says there isn't a language somewhere "gostak", "distim", and "doshes" don't mean something like "waiter", "bring" and "food" -- but we *don't know that langauge*, and so for us "The gostak distims the doshes" will *always* be meaningless, and so will a very long description of how the gostak distims the doshes murmfully while klerpling the hoovoo. Unless you make the sentence longer by adding *words we already know*, adding more detail and complexity to the bit of unknown language does *nothing* to strengthen our understanding of how those linguistic symbols relate to real concepts.

Information content is the measure of how much meaning a piece of writing has the *potential* to contain; whether it contains information depends on how people define the symbols within it, and there's no way you can magically know that short of either telepathy or, more likely, having it explained to you through nonverbal means.

"Decoding" a language in real life among humans is usually done by reference to something that's truly universal -- external reality, non-verbal perceptions. We point to a man and say our word for "man", point to a woman and say our word for "woman", point to a rock and say our word for "rock". However this kind of "pointing" is a lot more difficult over radio transmissions. One solution is to encode non-verbal data like graphics, which is a lot less arbitrary and a lot more data-rich than words. Assuming the aliens see somewhat similarly to the way we do, they could take an encoding of a picture and try different means of decoding it until they get one that gives them an image rather than random noise; that would be a starting point, since we could include a separate, much simpler  transmission that would hopefully be an obvious caption, that they could use to begin to translate.

The best way for us to learn an alien language would be the same way babies learn a completely new language -- by associating words with pictures. Similarly, other sorts of information, like numbers and scientific data, would be universal starting points. If we encoded an obvious binary signal of ascending numbers and then sent audio transmissions of the English words "one", "two", "three" and so on it'd give them a starting point on a wordbank. Some hopefully obvious arrangement of graphic data to represent a periodic table, or literal image of the spectroscope of an element, followed by the English name of the element would also work -- chemistry information is universal, as is physics. But this would be a slow process, and only simple ideas could be communicated at first -- you won't tell aliens what "love" or "democracy" or the Declaration of Independence mean this way, and the process of building up a dictionary this way would be slow detective work. A computer that could do all of this within a few days or so would be, to put it mildly, a very smart computer -- one probably equal to doing any other human task, like conducting military strategy or finding cures to diseases.

It's rather insulting to the entire field of linguistics that science fiction assumes that what writers consider the *interesting* jobs are still done by humans and can't be replaced by computers. Human administrators manage resources and workers; human commanders come up with battlefield tactics; human scientists study evidence and develop theories about that mysterious radiation source; human doctors diagnose and treat diseases and injuries. But something that in real life is just as fuzzy, intuitive, and complex as any of the above tasks, if not more -- studying unknown natural languages and translating them effectively and accurately -- is not only all done by machines, but is done so well the process is automatic and can be done in a matter of *seconds*.

Quote
The logic behind that is that the ST universe was seeded by an older race. (Explained in episode The Chase (http://www.startrek.com/startrek/view/library/episodes/TNG/detail/68598.html)). Though probably the real reason was the budget and the idea that the viewers prefer humanoids because they are easier to identify with.
I doubt the reason was a of lack of imagination. At least not for all writers


Well, it *is* a lack of imagination, whatever other reasons there are behind it. Even if the writers are capable of greater imagination, they choose not to use it, perhaps because they think the viewers or the producers don't have much imagination.

And anyway, the elder race explanation is a stupid one. It's not the "real" explanation -- it was introduced in Star Trek: The Next Generation long after the original series established that most races in the galaxy look human, and actually claimed that other races' *history* had to closely follow human history because that was a Law of Nature. (They visited a world that was in its "Roman Empire" era -- you know, the one all worlds pass through -- and sure enough everyone was wearing Roman armor and living in Roman villas that looked like they'd been borrowed from some historical drama's set for the afternoon.)

Even TNG's attempted explanation, bold as it was, was still ridiculous. The "seeding" was suppposed to have happened billions of years before all these sentient species developed, meaning that while human beings' ancestors were little hairy shrew-like things they had some sort of hidden code in their genes to become five-and-a-half-foot tall, five-fingered, bipedal, large-brained, hairless beings, where males were taller and stronger and females had enlarged mammary glands as a secondary sexual characterstic, etc. And so did *every other damn race*. That's *not* how evolution and natural selection work (and of all the things sci-fi writers often get wrong, what evolution actually *is* seems to be number one).

Admittedly TNG did a better job of making their aliens look different than TOS did, but even so, the idea that Klingons and Romulans and Cardassians would be shaped that much like us after billions of years of random chance and evolution and accidents after the Master Race had long ago died is preposterous. (Cardassians in particular are supposed to look that much like us even though every other aspect of their physiology is quite different -- they're supposed to be reptiloid in origin, no?) Our *own* race used to have various subspecies that were pretty different from Homo sapiens sapiens -- Homo sapiens neanderthalensis, Neanderthals, were quite a lot like us in all the important ways, maybe even cross-fertile, yet they looked very different from us, much more so than Klingons look different from humans (aside from the face they're shaped just like humans, with no average difference in height, no different carriage to the spine or collarbone, no different shape to the head).

But this is a sidenote. Let's not forget that SC2 was cool enough to have many alien races, most of which were, within story limits, pretty different from humans, and yet was able to make us sympathize with many of them and make them interesting characters nonetheless.

Quote
If your book is big enough, you can get a long way. Certain concepts will be used alongside eachother a lot, and if you know one, you'll have only limited choice for the other. Once you know the word for "eat", you'll know a lot about the nouns that accompany it. Initially, it's just a matter of trying things, and trying something else if that doesn't lead to a consistent whole.
But a human language is a lot easier than an alien language, as for humans you know you have a lot more common reference points.


Prove it. *Try* it. You're not really thinking about it. I'm not talking about a translation of an existing book that you've read before, I'm talking about a *new book* you've never seen before, without pictures. Try, without knowing any Korean, getting a Korean book from the library and translating it into English based on what you generally know about books. It's impossible -- it's *ridiculously* impossible. You might be able to tell that one word is a lot more common than another word, and probably means something *like* the word "eat" or "speak" or "stand" rather than something like "transubstantiate" or "quantize" or "interpellate", but you will *not* be able to figure out which word means "eat" and which ones mean "sit" or whatever, not without cheating by using outside knowledge of Korean or of the book's subject matter.

What makes the sentence "Kleem dragul macnaff" *mean* "I ate some soup", inherently? What will make that "obvious", no matter how many times you see it? Remember that a langauge doesn't have an intrinsic meaning -- it has a meaning by how it's defined. In theory any cluster of letters can mean *anything*. A sentence may have internal structure, and assuming that the alien language really is a lot like English you could deduce that one word is a subject and another is a verb after long study (though this is perilous, since we don't really know that an alien langauge will be anything like English at all). But that still doesn't assign *meaning* to the word -- you can't tell what one word you don't know means just from other words you don't know. You can only *ever* define a word by using simpler words you *already know* or by attaching the word to some non-word -- to a picture, a real object, etc. Otherwise, if everything is unknown, you won't get anywhere. Lewis Carroll made a hobby out of creating made-up words and defining them in terms of other nonsense words, and writing whole stories with them, and even though the entire framework is pure standard English, the words are incomprehensible.

That's the whole point of language, after all -- it symbolizes external things. If it, itself, contained the external things (if "eat" were a picture of a person eating), it'd lose much of its convenience value. English is convenient because it doesn't draw any sensory representation of a house to say the word "house"; it abstractly connects a random word "house" to the idea of a house. What we know about languages is limited, but one universal is that any language that reaches a certain level of development has to reach a similar level of abstraction, in order to be useful -- you become able to write "house" without being able to draw a realistic house; you have to be able to write "house" the same way even if you're talking about two houses that look completely different; you have to be able to write "house" about houses that aren't actual, physical houses ("The new government is trying to clean house"). Even languages that are *more like* pictures, like Chinese, are long past being actual drawing languages -- a non-Chinese speaker with no outside knowledge would never be able to figure out what a Chinese sentence like "We argued about the price of the new housing project" meant, even if he could tell that there was something in there that looked like a house and something in there that looked like a coin.

This isn't just theoretical. The Etruscans were a living, breathing civilization. They were conquered by the Romans, and it seems the Romans purposely forgot Etruscan culture and banned the use of their language. The result is that we have lots of artifacts with Etruscan writing all over them -- the *problem* is that none of them are any sort of dictionary, and as a result even though we know a lot of Etruscan words and a bit about how the Etruscan wrote we have almost no idea what they mean, and no one expects that modern humans will *ever* decode Etruscan, except maybe learning how the Etruscans wrote their numbers (since sums in account-books are less ambiguous than other kinds of writing).

Egyptian hieroglyphs used to be a *total* mystery, even though there were *huge* written histories on ancient papyri and monument walls and such -- even though hieroglyphs were often incorporated into illustrated columns and murals, so that we had actual pictures to compare them with. The only reason we got any start toward decoding hieroglyphs is that, luckily, ancient Egyptian civilization wasn't crushed as completely as Etruscan civilization. The Macedonians who conquered Egypt chose to continue using the ancient language, and left behind one nice, long proclamation with the *same text* in hieroglyphics, demotic, and Greek, and luckily Greek civilization has survived to the present day and the memory of how to speak ancient Greek is still around. With that nice start, the long, grueling process of translating hieroglyphs proceeded (and it was work, with a lot of guesswork involved). Without the Rosetta Stone it's doubtful we would *ever* have translated hieroglyphs, though. That's what you need to translate a language -- a dictionary. Without dictionaries, all the other subtle clues might not be enough -- and to *make* a dictionary, some living person needs to learn the other language from scratch, the way babies do -- by nonverbal communication, with picture books, by pointing to things, by observing the language in action in the world.

And this is assuming we're talking about ordinary human beings who just happen to have never met each other. When we're talking about aliens who may photosynthesize and not eat (so that their word for "photosynthesize" is basic and their word for "eat" is very complex), who may have several genders, who may have a more relaxed attitude toward time and location than we do, who may use a communication medium capable of subtleties we can't perceive... well, the problem multiplies manyfod.

Quote
If the communication partner feeds your computer enough data, then I don't see why not. A future computer should be able to process that quickly enough. And it's likely that they would feed it specially crafted text that would make it even easier. Something like a complete text which shows the grammar being used in practice, plus a dictionary and an explanation of the grammar rules to fill in the gaps when you already know enough to read the explanation.


But a UT wouldn't be a UT if you did that. Because the point of a UT is talking to a species with an *unknown* langauge, where a complete English-to-Spathi dictionary does not yet exist. It's the process of making that dictionary that's the problem. If you know the dictionary and the grammar, then making the computer translate isn't that difficult a problem and probably would end up being commonly done for convenience, though I still think that unless a computer gains a true kind of artificial intelligence that allows it to understand the various double meanings, implications, irrational and creative idioms, and such a language contains, it'd be a poor and unclear way of communicating that would leave out a lot. There'd still be a need for language experts and human translators when you really needed to know exactly what was being said, and if you had to talk to Spathi a lot it would be useful to actually learn how to speak it yourself. It's not a question of simple processing speed but of the sorts of algorithms you can write -- a computer that can understand natural language well enough to translate it accurately is a computer that can think like a human being in many important ways that all our smartest computers *can't*. That's a really big deal; it's a lot harder to explain why computers aren't really thinking beings or true citizens in a situation like that.

Quote
Then again, stuffing a Babelfish in your ear is much easier. ;)


Or a Talking Pet, yeah.


Title: Re: Languages in SC2 (new thread)
Post by: Art on August 14, 2004, 11:14:19 am

Quote
They also exist because we do reproduce sounds perfectly after a while. That's why areas have specific accents of the same language.


No. Actually it's the exact opposite! Regional accents exist because even people who start out speaking languages perfectly begin to slowly make mistakes, and then make the same mistakes more and more commonly. As the Precursors said, "Introduction of some noise into the signal is inevitable". Any time you have a system left on its own it randomly evolves as small errors are made, errors build up, and errors become changes. English colonists who came to New York in the 17th century spoke 17th-century English perfectly well. Their 21st-century descendants speak a language so different that, judging by the few books from the 17th century that describe how English is "properly" pronounced, they would find their great-grandfather almost completely incomprehensible, and vice versa -- the great-grandfather would see the language as having utterly deteriorated, which in a sense it has. And in another 200 years the language will probably have changed enough so that we wouldn't recognize it.

If anything this underscores my point -- an accent is not some overlay applied to a language but an element of the language itself. In reality the English they speak in Dallas, Texas and the English they speak in Glasgow, Scotland are two very similar but distinct languages. The line between different dialects and different languages is blurry -- Scots, the language Robert Burns wrote in, is considered by many to be a different language (try reading it without a glossary). Different "dialects" of Chinese that share the same written language sound so different and are pronounced by such different rules that they are different spoken languages... and so on.

Spathi-accented English should not be considered English-with-a-Spathi-"overlay"; it's a slightly different form of English, a form that doesn't properly exist until Spathi start learning English -- it's a language created by some Spathi elements creeping into Standard English, just as Scots didn't exist before some Gaelic elements crept into the standard English English of its time. A UT has no reason to turn Standard Spathi into Spathi English if 1) no Spathi speak English at all because of the existence of UTs and 2) it's translating for a human listener who understands Standard English.

Quote
Have you played fast action computer games? Try playing them without sound when you're used to playing them with sound. Sounds can be very helpful.


Doing anything differently after you're used to doing it one way will make it harder. Action games are entertainment; they usually put in audio effects for an increased sense of immersion and realism, since in real life we expect events to create sound and to not hear sound when something blows up is surprising.

The question is *how* necessary they are -- most of the time, especially in third-person-perspective games like SC2's Melee (or RTS games like Warcraft or C&C) the sound is unnecessary. The primary purpose of sound is to let you know there's something going on outside your field of view, behind or to the side of you. This is the primary purpose sound plays in, say, first-person shooters, and it's an important one because FPS's have a restricted field of view compared to real life (you have no peripheral vision).

That's what our sense of hearing is mostly designed to do, in the context of combat or fast action, to warn us that something is going on far away or behind us. That's *all* it's really designed to do, however. Even audiophiles have to train themselves to build up a "sound picture" of their environment, and doing so is still very difficult compared to building a sight picture, hence the many difficulties blind people suffer even in an area rich in sound cues; we're sight-oriented creatures. We have much lower resolution for hearing than vision. While we can see two very close light sources as separate objects and track their motion separately, put two close sources of sound in motion and we'll quickly become confused and just hear them as a vague motion "over there". *Surround* us with moving, loud sound sources and we tend to shut down, to hear nothing but a "terrible racket" and begin to lose the ability to distinguish even clear, distinct sounds without concentrating. (Try talking to someone in noisy traffic.)

This is the main reason I find the "sound-enhanced cockpit" theory unrealistic. The fact that battlefields are filled with noise makes them exciting to watch in movies, but it's a liability in real life. Hearing a loud, crashing explosion is a *bad* thing if the explosion isn't close to you but a silent, dagger-clutching enemy is. A series of loud noises unnerves and distracts more than it adds information.

The main thing sound is good at doing is forcing the brain into another channel -- it's desigened to distract, to let a hunter know that there's a loud breathing noise following him when he's focused on the prey ahead. So the sensible use of a sound system that, itself, can scan threats and *create* sounds is to generate a sound that warns the pilot of specific danger situations; to ring a proximity alarm for a missile that's coming too close or when he enters a gravity well, for instance (or if he has amazing hearing, to "paint" a tracking sound onto the relative location of an oncoming and close missile.) These are how alarms and sound systems are actually used in fighter planes (where the pilot, wrapped up in a pressure suit and cockpit, can't hear anything for the most part, surprising as it may be to aficionados of boom-laden flight sims). Otherwise, the advantage of being in space is actually that there aren't distracting noises, and that the sound of the Cruiser's nukes crashing into a planet miles away isn't audible to a pilot who's concentrating on firing off a perfect howitzer shot. Keep in mind that nothing's *actually* making any sound in space (no, it's not making a sound that you can't hear, it's *not making any sound*) so all these sounds have to be synthesized and put in by the computer. There's no reason in real life that the computer would bother to put in "realistic" noises for every single thing that explodes in space -- that would be a dumb way to do it. It's always a problem that people seem to think that the best way to engineer some sort of simulated environment (even a sound environment) would be to make it as much like it would be for an "ordinary" person as possible -- the only reason to do so is for the benefit of the movie-watching (or game-playing) audience, just like drama and excitement are the only reason for the sound effects in SC2 (or Star Wars, or Star Trek, or Armageddon, or any other sci-fi movie where there are sounds in space).

Quote
As for the VUX, it is possible that communications were not actually initiated (at least not intentionally). The VUX may have found some way to tap into what the humans were saying to eachother aboard their own ship.

Then again, what we read in the SC1 manual is presented as what "The Division of Synthetic Special Reconstruction" has reported. Maybe in fact the VUX were bluffing. They saw the humans, wanted to kill them, but needed a reason. Quickly realising the VUX appearance may had the same effect on Humans, they claim one of the Humans insulted the VUX, thereby giving them an excuse to attack. As it turned out, indeed someone (the captain no less) insulted them.

BTW, assuming the other party can't understand you when speaking through the inter-ships communication systems is foolish, as it is common sense to record everything. Even if they couldn't understand it as it was spoken, you can count on it being analysed afterwards.


True, there are a lot of common-sense problems with this whole situation as set up in SC2. But the whole problem is credited to the VUX being "master linguists", not the VUX being master spies and wiretappers -- the fundamental nature of the situation was that Rand expected that the VUX would have very little knowledge of English and what he said, even if heard, would be ignored or dismissed as a grunt or something. But your description of the situation from the VUX point of view is a convincing one.

Quote
You would just apply English and Latin rules to words they were not meant to apply on. You're not inventing any new ones.


Latin rules explicitly include inflecting a word to fit its purpose in the sentence, that being partly based on various rules about the word (which declension the word is in, its gender, etc.) Such concepts are unknowns for most English words that you'd be forced to do this in Latin for; you'd have to either find a synonym in Latin (which is what people usually do) or else let the English word sit there uninflected, sticking out like a sore thumb, make up an arbitrary declension and gender to define the word as, and work around how this affects the overall structure of the sentence. We do this very often in English because English in its current form is amazingly lax about grammatical structure, but in other languages this is a lot less common and a lot less natural-feeling. Many English words, for instance, creep into Spanish, but we get unusual sounding translations into Spanish verb-forms to make the word sound natural (the verb "click" referring to computers becomes "cliquear", because using a verb without an "-r" ending is just impossible in Spanish; you lose all the information that comes from conjugation).
Quote
Re talking pet:
I'm assuming for a moment that the Ur-Quan talk to communicate with talking pets through thought. Talking takes a lot more effort than thinking. And thinking you'll do anyhow, but by explicitely not talking to specific races you show them they are not worth the effort. Just as symbolism, because Kzer-Za at least do a lot more for "lesser races".
Also, each time an Ur-Quan uses the Talking Pet for translation, it humiliates it, which was exactly the idea in the first place.

And cultural habits aren't always rational and consistent. Compare it to Western Muslim girls who wear a kerchief on their head, as a form of body modesty, and yet put on make-up on their faces.



Sure. I said it didn't make sense to me but it might to an Ur-Quan, after all. It seems like the distinction is a lot greater and more dramatic if it's a difference between other races, who painstakingly learn each other's languages, write dictionaries, train linguists and build translation devices based on their research and the Ur-Quan who blithely think into organic UTs than if the Ur-Quan use fleshy UTs that require no typing while other people use factory-made ones with keyboards or microphones. The former has a lot more dramatic weight. That's not proof, but I think the burden of proof is on those who think UTs exist, since I don't think there's much in-game evidence for it and all real-life evidence makes them horrendously unlikely.

My crusade is to get sci-fi fans out of the habit of *assuming* UTs exist in whatever new sci-fi they're watching/reading/playing because it's just the sort of thing people can build in the future, y'know? At least people have come to accept that flying cars, food pills, and teleporters are ridiculous, but this universal translator thing really has a grip on people, if only because it makes filming sci-fi movies so much easier.


Title: Re: Languages in SC2 (new thread)
Post by: Culture20 on August 14, 2004, 11:30:04 am
Quote
How is their tech above others' tech?

The SC2 manual states:

                    The VUX's repulsiveness to most other races is matched
                    by their technological achievement in linguistic transla-
tion devices. This combination led to the unfortunate incident in 2126,
when the Earthling Cruiser Miwok made first contact with the VUX near
the Beta Luyten star system. The Miwok's Captain Rand, upon first seeing
the VUX commander on his view screen, remarked to his officers, "That's
the ugliest freak-face I've ever seen!" Rand was unaware that his every
word was being relayed to the VUX captian with perfect clarity. This grievous
insult, and the subsequent ill-will between Earth and the VUX delayed
the VUX's entry into the Alliance long enough for the Ur-Quan to enslave them.

Of course, the SC1 manual states:

                           The VUX's physical repulsiveness is matched by technological
                           advances and enormous linguistic-perceptive powers. This lan-
                           guage translation ability allowed one VUX ship to intercept the
                           communications of one Cruiser's commander, who had just sighted
the VUX on his laser display. The Captain's offhand remark about VUXian looks led to a
severe Xenoform backlash. The offended VUX, nursing            
a sense of collective insult, soon attached itself to the Hierarchy.

Which implies that it is a natural talent.  It could be a combination of both.
No matter what method it is, the VUX are top-of the line translators; note that they are the only other race who have held a documented dialog with Orz.  The Starbase can be assumed to have used your Precursor vessals' UT to feed their non-U T's with the Orz psuedo-rosetta stone, but the VUX had no such benefit.


Quote
but the VUX already knew English.

I suppose it's possible that even the VUX's translators aren't universal; they might have used the Androsynth tongue (if they have their own) as a rosetta stone.  This of course implies that they met the Androsynth before becoming thralls (they weren't yet thralls when rand met them).  Also, the Orz only mention that the VUX keep asking about the Androsynth; it's possible that the VUX merely repeated the question because the didn't understand any of the responses from Orz (note that they do not mention Orz at all).

Quote
Er, if the Arilou know our language well enough to be able to predict what the translator will and won't translate, and then *put* the English words in, in place of the Arilou words (which may not even be literally possible if Arilou grammar is radically different from English grammar -- try speaking a mixture of Latin and English and see how many fake rules you have to invent on the spot to get it to work)... why can't they just talk English to us in the first place? It'd probably be easier, for them and for us. (What would the Arilou be doing needing to take advantage of a translator anyway? They're friggin' obsessed with the human race and have been for our whole history; why *wouldn't* they talk to us in English?)

Well, I actually meant that they'd use words in their own language that would fit within our frame of reference.  If you were explaining a 3D world to someone in a 2D world, even if that person was using a UT, you'd know that some of the concepts _couldn't_ translate because they don't fit in the 2D reality; so you'd have to start hem-hawing your way around: using lingual best-fits in your own language.  If anything, the Arilou probably spoke Celtic to us. ;)


Title: Re: Languages in SC2 (new thread)
Post by: meep-eep on August 14, 2004, 11:41:23 am
Quote
Yes, you might say that the Ur-Quan aren't physically speaking to the Talking Pets, they're just thinking or thought-speaking. Fine, that's a compelling theory, but,

I just noticed that the Star Control 1 manual explains how the Talking Pets work:
Quote
The Ur-Quan uses a particularly striking means of trans-species communication. The Talking Pets, an Ur-Quan invention genetically engineered for the purpose, telepathically interpret Ur-Quan commands into the spoken languages of subordinate species, and reverse the procedure when receiving extra-special transmissions.




Title: Re: Languages in SC2 (new thread)
Post by: Culture20 on August 14, 2004, 11:43:04 am
And the reason from the SC2 manual:

The Ur-Quan are unwilling to communicate directly with other species
because to do so would be demeaning. Therefore, when giving orders or
interrogating enemies, the Ur-Quan use "Talking Pets," large-brained, frog-
like creatures which are non-sentient, but possess the telepathic/empathic
ability to translate all languages.


Regarding ancient cultures & translation:  How do the Chenjesu, Spathi, Earthlings (Farnsworth, the Captain), and others translate Precursor writings?  There's obviously not going to be a rosetta stone (unless it's a Precursor child's primer complete with Holograms but no holos of precursors themselves since we don't know what they look like).


Title: Re: Languages in SC2 (new thread)
Post by: Art on August 14, 2004, 11:59:31 am
Quote
Well, the SETI project simply assumes that whatever signal there is will have lower entropy than background noise. Which, for high-redundancy interstellar communication, would be a really good idea. We might not get a translation even if we can tell there's a signal, though.

As for decoding languages -- invented languages have been cracked by decoders before. Some of them had rather peculiar syntax not closely related to any human language. In fact, several of them of them were designed to confuse decoders. Yet they were decoded.


The movie "Contact" portrayed a pretty realistic scenario (it would have to be, since the original story was by Carl Sagan) where contact was first made by a simple series of pulses that demonstrated a pattern created by intelligence and not random natural activity, in that case the sequence of prime numbers, then used an encoding of a known data file (an transmission they'd received from Earth) to establish the format in which they sent the rest of their data (all of which was in graphical form). In real life we've done similar things, sending out transmissions in binary code consisting of numbers and simple graphical data and putting pictures and analog sound (on a record) on the Pioneer and
Voyager probes.

As far as the invented languages -- I'm not sure exactly what you're referring to, but I do know of many challenges where linguists and conlangers have created constructed languages and translated a *known* work into them, like the Bible or a Shakespeare play; the challenge was then to take the two texts and come up with a dictionary and description of the conlang's grammar. This is a difficult and demanding task, yet by definition it's extremely simple compared to studying a "new" language -- it's not a new language, since the inventor of the language actually gives us a complete Rosetta Stone! You *need* that "hook", that Rosetta Stone text, to "decode" a language; otherwise you don't know what you're decoding from. A naked, context-less fragment of language is *impossible* to decode, and only somewhat esoteric principles of information theory allow you to tell the difference (by looking for more or less repetition, more or less complexity) if something is a language or not at all. In theory any random arrangement of letters could be a language and could mean anything; you need some hint of what the message was intended to mean or you'll get nowhere. (Or you'll be just as stranded as the people in Borges' "The Infinite Library", a good short story to read to get a grasp on the problems of language and communication the way "Flatland" is good for grasping higher dimensions.)

The real-life work on secret codes and ciphers used by spies and such is crippled by that very same problem; their "hook" is that most "secret codes" aren't really codes but ciphers (same language, different script) where the language is actually Japanese or German or Russian but the language has been written in different letters, or scrambled, or in the modern day and age translated into a binary form like ASCII and scrambled using a computer algorithm.

In such cases the actual *language* -- the base set of symbols and their meanings -- is known; it's Japanese, or English, or Arabic. It's just written in a different way (with every letter shifted four letters over, or scrambled by switching every third letter and split into four-letter blocks, or usually other more complex things). Since if you know a natural language you know that language's specific patterns and can begin to try to look for structure by trial and error. This was much easier back in the days when ciphers had to be simple enough for people to hold them in their heads or, at best, in a typewriter-like machine like the Enigma, which meant that brute-force mathematical analysis would spit out the message in time. (Nowadays computers can create ciphers more easily than they can decipher them, and the cipher that encodes, say, your credit card number in an online transaction would take a great deal of effort for the NSA's powerful supercomputers to crack. Unless quantum computing pans out, but that's a different story.)

Ciphers are *fundamentally* different from codes, however. Ciphers are always related to a known language (the English language, the common language of HTTP protocols that store Web pages, the set format of a credit card number) by a strictly *mathematical* relationship. A code is something human-created that's only a set of abstract, human relationships. There's a mathematical relationship between "Star Control" and "Rats Lortnoc" or "Star Control" and "Fgne Pbageby". There's no mathematical relationship between a CIA code that says "MK-Ultra" means "mind control experiments" any more than there's a mathematical relationship between English "Hello, how are you?" and Chinese "Ni hao ma?". They mean the same thing because human beings *say* they do, and there's no way to read the minds of the human beings who created a code just by looking at the code.

Codes are *not* breakable by mathematical analysis; their weakness is that they require people to learn them rather than applying a simple mechanical method to decipher them, so historically codes have been broken by either simply finding a code book and stealing it (or capturing a spy and torturing him) or by the much longer, trial and error method of comparing the code to real life -- by figuring out that whenever the radio transmissions say "Red smells like drizzling Mercedes" a friendly bombing occurs here, and when they say "Green looks like crunching Cadillac" an enemy aerial recon team appears there, and working from that. Doing the latter is a long, hard process and if a country was forced to try it the war would often end before they succeeded. While the Enigma machines failed in World War II because even really complicated ciphers can be broken mathematically, the Navajo code-talkers succeeded and were *never* decoded, because they had a completely internal code (with no codebooks), the Navajo language, that was truly alien to English or Japanese and could not be broken mathematically.


Title: Re: Languages in SC2 (new thread)
Post by: taleden on August 15, 2004, 07:47:00 am
Some comments to toss into the mix...

Quote
The line between different dialects and different languages is blurry


Actually I think Noam Chomsky summarized it quite well: a language has an army and a navy.  :)  Which is to say, the distinction is fairly arbitrary from a linguistic standpoint, and has much more meaning from a political and cultural standpoint.

Quote
The primary purpose of sound is to let you know there's something going on outside your field of view, behind or to the side of you.  That's *all* it's really designed to do, however.


Having read a number of your posts here, I am inferring that you're like me in that you tend to let your theories run away with you and start making claims that, interpreted fairly literally (which is how we have to interpret them, since all we have is your literal text), don't make much sense.  So, I'm going to reign you back in on this one (and I hope you'll do the same when my theories run away with me): on what basis, and by what authority of research or expertise can you make such a claim?  Is it even meaningful to try to state the "primary purpose" of something like the phenomenon of atmospheric pressure differentials that our bodies and brains decode into something we call "sound"?  And who "designed" that phenomenon, anyway?  Come on Art, I know you can write more precisely than that.  :)

I would offer that sound is a valuable source of information about our environment, just as sight is; it is useful to be able to hear things which you can see, just as it is useful to hear things you cannot see (for slightly different reasons in each case).  Which leads neatly into the next quote:

Quote
This is the main reason I find the "sound-enhanced cockpit" theory unrealistic. The fact that battlefields are filled with noise makes them exciting to watch in movies, but it's a liability in real life. Hearing a loud, crashing explosion is a *bad* thing if the explosion isn't close to you but a silent, dagger-clutching enemy is. A series of loud noises unnerves and distracts more than it adds information.


This is true, but it is also incomplete.  In a situation like a very loud and chaotic battlefield or a very busy street, yes, the noise tends to hurt more than it helps because the amount of information being conveyed on that channel (the auditory one) is more than we can handle, so it just confuses us.  However, in a situation like 1-on-1 ship combat in SC2 (that is, not a fleet engagement, not a particularly 'chaotic' environment), simulated sounds could be extremely useful, and let's recall that we're talking about 1-on-1 ship combat in SC2, and not a chaotic battlefield where enemies might be sneaking up on you with a knife.

Also, just because noise can be confusing and distracting to the average Joe doesn't mean it has the same effect on a seasoned starship captain.  I imagine that with experience, captains learn to subconsciously interpret those auditory cues and extract quite a bit  of useful information without it really distracting them much at all.  So for those experienced captains, simulating the sounds of 1-on-1 ship combat could be a very useful thing, and therefore a perfectly plausible thing for ship computers in the SC2 universe to do, or at least be able to do.

And finally, let's remember that in every discussion on this forum, we have to think about this stuff in two fairly distinct ways: we all enjoy thinking about how (or if) the 'universe' of SC2 would actually work and make sense, but we also all know that, at heart, it's still a computer game, for humans, on Earth, in the 20th Century, and therefore it has to bend the rules a little to be compelling.  Although ship combat with no audio except humming engines and your own grunts would be more 'realistic', it'd be a crappy gaming experience for most people.  So let's not get *too* in depth about it.  :)

Quote
While the Enigma machines failed in World War II because even really complicated ciphers can be broken mathematically, the Navajo code-talkers succeeded and were *never* decoded, because they had a completely internal code (with no codebooks), the Navajo language, that was truly alien to English or Japanese and could not be broken mathematically.


I'm actually just responding to this becuase that point about the Navajo "code" is a really good one and I want people to read it again.  I actually wrote a paper for a linguistics class a few years ago that made a similar comment: the Germans couldn't decode Navajo because they were starting from a faulty assumption (that it was a cipher of English, or of some other language known to them) - the idea that humans will be able to decode an alien language without a *huge* amount of assistance (from the aliens) seems fairly perposterous.


Title: Re: Languages in SC2 (new thread)
Post by: Art on August 15, 2004, 12:00:57 pm
Quote
Having read a number of your posts here, I am inferring that you're like me in that you tend to let your theories run away with you and start making claims that, interpreted fairly literally (which is how we have to interpret them, since all we have is your literal text), don't make much sense.  So, I'm going to reign you back in on this one (and I hope you'll do the same when my theories run away with me): on what basis, and by what authority of research or expertise can you make such a claim?  Is it even meaningful to try to state the "primary purpose" of something like the phenomenon of atmospheric pressure differentials that our bodies and brains decode into something we call "sound"?  And who "designed" that phenomenon, anyway?  Come on Art, I know you can write more precisely than that.  Smiley


You're right. "Design" is a really poor way of putting it. Though what I mean here is "hearing", not "sound" (not the vibrations themselves but the way our ears happen to decode them), and what I mean by "design" is the primary use to which we've ended up putting them after years and years of evolution. That is, we evolved into a sight-based species and not a hearing-based one -- unlike dolphins, we can't see clear and distinct shapes based on patterns of echoes and such. Most of the things our eyes do our other senses are poor replacements for, because we've evolved to depend so much on light.

We can certainly use our hearing much more effectively than we normally do if we train it, but most of us train ourselves to use our hearing to supplement our sight, not replace it, and so have a hard time telling much based on what we hear other than that something's going on outside of sight range. Our hearing is very good at getting our attention but it's not that accurate -- one example is how people lost in the woods will often walk in circles or meandering paths trying to find some landmark by sound -- it's very easy to mistake one sound for another in such a situation, and to grossly misjudge the angle that a sound is coming from, etc.

Quote
This is true, but it is also incomplete.  In a situation like a very loud and chaotic battlefield or a very busy street, yes, the noise tends to hurt more than it helps because the amount of information being conveyed on that channel (the auditory one) is more than we can handle, so it just confuses us.  However, in a situation like 1-on-1 ship combat in SC2 (that is, not a fleet engagement, not a particularly 'chaotic' environment), simulated sounds could be extremely useful, and let's recall that we're talking about 1-on-1 ship combat in SC2, and not a chaotic battlefield where enemies might be sneaking up on you with a knife.

Also, just because noise can be confusing and distracting to the average Joe doesn't mean it has the same effect on a seasoned starship captain.  I imagine that with experience, captains learn to subconsciously interpret those auditory cues and extract quite a bit  of useful information without it really distracting them much at all.  So for those experienced captains, simulating the sounds of 1-on-1 ship combat could be a very useful thing, and therefore a perfectly plausible thing for ship computers in the SC2 universe to do, or at least be able to do.

And finally, let's remember that in every discussion on this forum, we have to think about this stuff in two fairly distinct ways: we all enjoy thinking about how (or if) the 'universe' of SC2 would actually work and make sense, but we also all know that, at heart, it's still a computer game, for humans, on Earth, in the 20th Century, and therefore it has to bend the rules a little to be compelling.  Although ship combat with no audio except humming engines and your own grunts would be more 'realistic', it'd be a crappy gaming experience for most people.  So let's not get *too* in depth about it.  Smiley


Well, I don't say it's impossible, but color me skeptical. The whole thing is that we're not "simulating" sound -- there is no hidden sound that things have in space that you can somehow detect. The natural sound of an event in space is no sound at all. So going to the lengths of detecting some particular event and interpreting it and then making a sound for it is a pretty complicated rigmarole. It might help in battle, but I don't see it being an optimal or even very good solution. The computer has to detect the event using visual data in the first place; why not transmit the data directly to the pilot in visual form rather than turning it into sound information?

Yes, there are cases where you want the pilot to hear rather than see, when you want the pilot to be able to think about two things at once, but since you can control the sound environment inside the cockpit you can *optimize* the playing of sounds for maximum effectiveness. You wouldn't "simulate" sound by slavishly playing a sound for every time a weapon fires. Distracting or not, it's unnecessary; you only need to warn the pilot away from watching his viewscreen when there's, say, a closeby missile lock or something, so why make noises for other things? (We need to get away from the idea that the computer is consciously playing sounds for some things and choosing not to play sounds for others. Everything has no sound, in reality -- every event that the computer generates a sound for is something the computer has to be specifically programmed to look for.) It'd be like giving the pilot a radar screen that draws a detailed 3D model of what the ship looks like from overhead instead of a simple labeled blip, or a nice little videophone so he can see his commander's face instead of just hearing vocal instructions. Waste of resources. Video game cockpits tend to have all sorts of really kewl looking things built into them like textured, realistic radar screens, videophone communicators and such, but real-life cockpits rarely do, since resources are at a premium in combat situations.

And yes, none of this is an argument for making the actual game have no sound effects, or for making the aliens speak in an alien language and us have to translate them using a paper alien dictionary, or anything like that. I'm just saying that the "real" SC2 universe is different from what we see in the game, and it has to be  because it's a game; it's no different from every movie that puts sound effects for events in space so we'll be able to follow the action, or has the Russians or the Chinese or the Vulcans speaking English so we won't read annoying subtitles, or whatever. (Even NASA makes space videos with sound effects, after all.)


Title: Re: Languages in SC2 (new thread)
Post by: Art on August 15, 2004, 12:24:21 pm
Quote

The SC2 manual states:

                    The VUX's repulsiveness to most other races is matched
                    by their technological achievement in linguistic transla-
tion devices. This combination led to the unfortunate incident in 2126,
when the Earthling Cruiser Miwok made first contact with the VUX near
the Beta Luyten star system. The Miwok's Captain Rand, upon first seeing
the VUX commander on his view screen, remarked to his officers, "That's
the ugliest freak-face I've ever seen!" Rand was unaware that his every
word was being relayed to the VUX captian with perfect clarity. This grievous
insult, and the subsequent ill-will between Earth and the VUX delayed
the VUX's entry into the Alliance long enough for the Ur-Quan to enslave them.

Of course, the SC1 manual states:

                           The VUX's physical repulsiveness is matched by technological
                           advances and enormous linguistic-perceptive powers. This lan-
                           guage translation ability allowed one VUX ship to intercept the
                           communications of one Cruiser's commander, who had just sighted
the VUX on his laser display. The Captain's offhand remark about VUXian looks led to a
severe Xenoform backlash. The offended VUX, nursing            
a sense of collective insult, soon attached itself to the Hierarchy.

Which implies that it is a natural talent.  It could be a combination of both.


Hrm. You're right. Though I think when I first read this I interpreted it as saying the VUX are very good at translating languages in the traditional sense, and they use this knowledge to write very detailed dictionary and grammar programs that make very effective automatic translators (which isn't the same thing as making a smart UT that can translate on its own). It would make sense, particularly since it looks like they may not be physically equipped to actually speak English. Their native tongue probably contains a lot of slurping (no pun intended, har har).

Sounds to me like the most likely explanation may be a combination of theories, since you are right that if the Captain knew he was being recorded it would've been dumb to make the comment in any language. Perhaps the VUX's smart translators are able to pick up small sounds or garbled transmissions and come up with very accurate guesses as to the language content -- very good audio enhancers optimized for English or whatever language, so they picked up the muffled echo or faint radio leak from Rand mumbling into the intra-ship communicator.

Quote

No matter what method it is, the VUX are top-of the line translators; note that they are the only other race who have held a documented dialog with Orz.  The Starbase can be assumed to have used your Precursor vessals' UT to feed their non-U T's with the Orz psuedo-rosetta stone, but the VUX had no such benefit.

I suppose it's possible that even the VUX's translators aren't universal; they might have used the Androsynth tongue (if they have their own) as a rosetta stone.  This of course implies that they met the Androsynth before becoming thralls (they weren't yet thralls when rand met them).  Also, the Orz only mention that the VUX keep asking about the Androsynth; it's possible that the VUX merely repeated the question because the didn't understand any of the responses from Orz (note that they do not mention Orz at all).


Unlikely that there was an Androsynth/Orz dictionary, since all signs point to the Androsynth disappearing or at least ceasing to leave records before the appearance of the Orz; Bukowski can't find any references to the Orz as we know them in the Androsynth computers. More likely that the Orz speak a modified form of Androsynth, which itself may likely be a modified form of English; the Androsynth certainly didn't start out with their own separate language and culture, and while it's in character for the 'Synth to make up a new language to distance themselves from their slavemasters it's unnecessary. The other alternative is that, as you say, the Precursor computer is at least a sort-of UT; it's possible given that the Precursors Do the Impossible fairly regularly and that the Vindicator's main computer does show disturbing signs of really, really high intelligence, and may be a preferable theory since presumably the Orz speaking Androsynth *should* cause some comment, but it's up in the air from my POV.

Calling a language's syntax "unorthodox" makes more sense when you're comparing it to an actual known language, particularly one that the language in question is related to -- Pidgin is an unorthodox form of English, but it's not, in general, an unorthodox langauge. And the problematic idea of linguistic best-fits may mean finding a word closely related to a word etymologically rather than finding the word that best fits all the word's possible meanings. We, after all, in decoding Orzese find words that make more sense to *us* for *our* purposes than the computer's best-fits -- "war" makes more sense than "party" when we consider what the Orz parties actually are to *our* eyes, "avatar" or "projection" is both broader and sounds better than "fingers", and so on. The Orzese best-fits seem to be based on the sorts of words the Orz would choose if they knew English or an English-like language for certain concepts, not the words that would make the most sense to us. It would also explain why a puzzling non-word like *frumple* should make it on the computer's list of best-fits (it seems like some sort of portmanteau word -- "frown" and "rumple"? -- but portmanteau words read to me more like an idea that originated with the Orz rather than something a smartass computer came up with for a totally alien word).

Quote

Well, I actually meant that they'd use words in their own language that would fit within our frame of reference.  If you were explaining a 3D world to someone in a 2D world, even if that person was using a UT, you'd know that some of the concepts _couldn't_ translate because they don't fit in the 2D reality; so you'd have to start hem-hawing your way around: using lingual best-fits in your own language.  If anything, the Arilou probably spoke Celtic to us. ;)


Well, okay, that's almost certainly true. The Arilou flat-out say it -- they grope around for some way to describe the way They sense mortal races and call it "smell", etc.

By the way, is Ariloula'leelay authentic Gaelic? It brings up the usual weirdness of assuming that people are saying things in just one language and yet the translator can magically distinguish between when they want it translated into English and when they want to quote that word in their own language -- like the Supox and Vlik/Earth. Anyway I think it's unlikely that they're speaking Gaelic since, after all, even though they liked the Celts a whole lot they've had their most recent large-scale contact with us mostly in 20th-century America, if you believe the stories, not counting their communications with Star Control during the war. Heck, if you take that Men in Black stuff at the end seriously they had long-term ongoing communication with parts of the US government for a long time, so their speaking English is almost a given.


Title: Re: Languages in SC2 (new thread)
Post by: Art on August 15, 2004, 01:46:17 pm
Quote
The Ur-Quan uses a particularly striking means of trans-species communication. The Talking Pets, an Ur-Quan invention genetically engineered for the purpose, telepathically interpret Ur-Quan commands into the spoken languages of subordinate species, and reverse the procedure when receiving extra-special transmissions.


Problem is this is ambiguous. Telepathically interpreting could mean directly passing along thoughts, or it could mean hearing a spoken command and knowing what it means because they can read thoughts. I lean toward the latter because "interpret" seems to imply that there's something actually there to interpret the meaning of, like a spoken command, while just passing along thoughts would be telepathically "transmitting" or "relaying".

Also, this:

Quote
The Ur-Quan are unwilling to communicate directly with other species
because to do so would be demeaning. Therefore, when giving orders or
interrogating enemies, the Ur-Quan use "Talking Pets," large-brained, frog-
like creatures which are non-sentient, but possess the telepathic/empathic
ability to translate all languages.


also leans on the other side; they don't possess the ability to communicate telepathically but to translate all languages telepathically, which implies that there's still the use of real languages going on; nothing is said about the language-less interpretation of thoughts. They don't say, for example, "possess the telepathic/empathic ability to hear Ur-Quan thoughts and translate alien languages".

And I do find it easier to imagine that a nonsentient nonthinking Talking Pet would need to hear a spoken command as a cue to start translating. If Talking Pets can operate without that, then it raises questions about how Talking Pets receive signals to let them know that certain thoughts are meant to be transmitted and certain others not. Speaking to a Talking Pet also answers the question of how the Talking Pet sorts between conscious thoughts that are meant to be broadcast and secret thoughts -- the Talking Pet might not be able to think but it could certainly be trained to tell when a stream of sound and a stream of thought were somehow in sync with each other, one following the other's pattern, and to translate just those thoughts.

Quote

Regarding ancient cultures & translation:  How do the Chenjesu, Spathi, Earthlings (Farnsworth, the Captain), and others translate Precursor writings?  There's obviously not going to be a rosetta stone (unless it's a Precursor child's primer complete with Holograms but no holos of precursors themselves since we don't know what they look like).


We're given very very little knowledge of anything the Precursors wrote in the first place, right? Which is consistent. If you read Frederik Pohl's Gateway novels, upon which the idea of the Precursors seems largely based, there are some interesting bits where they make early headway with the ancient Heechee language. If you see a very common phrase repeated on all sorts of devices that all shoot deadly rays, you might think that word means "ray gun", and the next time you saw something with those marks on it you'd think twice before pointing it at yourself and pushing the button. If there's been a huge, concerted effort to study the Precursors starting with the Chenjesu then they may even have gotten far enough to have fragmentary translations like the "Appendages of Dawn" fragment, which is appropriately vague and ambiguous, or for the Spathi (with Chenjesu help) to get a reference to "10 garbage dumps". But it is notable that most of the time Precursor writing is just a big mystery.

The big exception is of course the Vindicator's computer, and maybe a UT is the best explanation for how even an expert like Farnsworth could've gotten it to speak English. (At the very least there must've been a powerful AI trying to learn English while he was trying to learn the computer's language.) One would think that the Precursor computer would be a great resource for the Precursor language, but it may be that all Farnsworth did was learn to understand a simpler code used by the computer itself and bypass the Precursor-language related parts of it, translating its displays into English by trial and error. He certainly worked really hard on doing it, and maybe we shouldn't blame him for giving up after that difficult task. There's something a bit strange about Zelnick's gift, and it may not be just a natural talent (didn't the Arilou say he'd very recently been to Unzervalt?).

New questions: How does Chenjesu communication work? They communicate naturally by electromagnetic radiation and Hyperwave; they presumably didn't start learning our spoken languages till we broadcast them over radio waves. This raises the question of things like names; why should they be called the "Chenjesu" and have names like "Tzz-Tzer-Tzak" when their natural bodies don't speak in sounds that could be transliterated like that. Obvious answer is that it's for our benefit, but then where do the names come from? My inclination is to think that they have patterns of radio frequencies that are their names, and they just pick a simple way of playing those frequencies as static over a receiver and allowing how we hear those sounds to be their names, which would explain why their names are so... staticky.

Mmrrnrrhrrm: I see these guys as interesting. A race of robots could both hear and make sounds with much more precision than we biological humans. So, if they communicate by sound, the sound-making mechanism or set of sounds could be have a lot less range in pitch and tone and still convey really complex information, since a very tiny adjustment in pitch or length would be perceptible to them -- like the imaginary Speedtalk Heinlein wrote about in Gulf. That would explain why their words are long consonant stretches (which I imagine as them making a sort of low modulated humming noise).

Chmmr: Their names are prolly just meldings of Chenjesu and Mmrrnrrhrrm names. It's interesting that they do seem to be more generally pronounceable, though; possibly they want to decrease awkwardness with other races that way; the Process may have partly involved installing better vox capabilities.

Orz: Yeah, we have no idea what these guys speak, though it's something that our computer can sort of translate, so we should be glad. One thing is that it's weird that they talk at all, since they claim to be projections of a single entity -- why would they come into our universe with a language ready-made for them? Did they crib it from the Androsynth or the VUX? Why are they called "Orz"? Is it some arbitrary thing they picked as a detail to make themselves more believable as a mortal race -- like giving themselves fish bodies and styling their starships a certain way?

Interesting point: We know the Orz's first foreign contact was with the VUX, and the VUX may've been the first people they actually tried to communicate with.

We also know that there's a VUX captain whose name is... ORZ. Coincidence? Maybe, but ripping off some random guy's name to be the name of their race 'cause they can't think of anything else... sounds like an Orz thing to do. (There's also a Spathi named Pkunky, but I'm not as inclined to look into that.)

Spathi: Not a whole lot to say except that their accent and their tendency to forget random words for things makes me think they aren't speaking their native tongue at us. And the Spathi have special reason to know our language, since they're a neighboring race whose only real job was to guard us after the war was over and check to make sure we weren't playing any tricks by obsessively monitoring our world.

Umgah: See, their stilted English grammar makes me think they're not talking through a translator, 'cause *all* alien languages are going to have a very different grammar and interpreting the sentences in proper English grammar is exactly what a translating device is supposed to do. The messed-up English has to be them speaking that way to us, either 'cause they think it's funny or 'cause they're a lot worse at languages than at gengineering (which might be how the Spathi caught on to their hoaxes).

Ilwrath: This is the one that gets me, since the whole "Dwe, the Dill-Rats, warship yuubuu!" thing is something that'd be near-impossible to translate from one language to another. Far-fetched to say that the proud Ilwrath were actually speaking English? Maybe, but it simplifies that exchange a heck of a lot. And the overthrow of the priesthood class certainly could've entailed major, widespread cultural change, including a rather arbitrary installation of a new language, esp. if we go with the "Umgah are bad at learning languages" theory -- why learn a new one when you can humiliate your marks by making them learn the enemy's language for some stupid and arbitrary reason? (The Ilwrath do seem to enjoy it when Dogar and Kazon give stupid and arbitrary orders.) They may have already known English, anyway; their whole race are loyal "Captain Satellite" viewers, and the way he says it makes me think he means it's a show produced by non-Ilwrath (humans?).

Supox: We thought the VUX were fast learners, but the Supox's whole biological imperative is about cultural assimilation. Wouldn't surprise me if they learned English pretty quickly upon monitoring our transmissions for the first time and had a standard English greeting ready just in case they met us.

Mycon: See, Mycon probably do need some kind of translator, and may even only have humanoid Mycon to do negotiations and such; their basic nature may not really require language -- they communicate by uploading their memories to new offspring -- and their physiology certainly makes speaking human languages unlikely. Their name, "Mycon", is just a Greek word for "fungus". It seems to be common human practice to use Greek to give names to other races (Androsynth, Slylandro, Mycon). They might retain the ability to form speaking units, and still have unique spoken words like Juffo-Wup and names like Shloosh because of the spoken commands from their creators (possibly the Precursors).

Slylandro: Do Slylandro really "talk" in sound? And would those sounds be even audible to us without a translating computer? I'm inclined to think a word like "Drahn" is a Precursor artifact, if not an actual Precursor word then maybe their way of representing the pattern of a much deeper sound that can't be heard by people whose ears aren't designed for really dense gases.

Dnyarri: See, my question is if this guy actually talks. I think he does; he has a voice that sounds very, well, earthy and solid and not like thoughts just dropping into your head. And he moves his little mouth. I have my doubts about whether truly talking through thought is reasonable -- we think words, but do we think in long, coherent sentences? When we do think a sentence, how much total thought does that represent compared to all the muddle below? And how well would we be able to tell a thought that popped into our heads from outside from one of our own thoughts? There might be good reasons that a Dnyarri or Talking Pet needs to communicate in spoken words, beyond for appearance's sake in the game; we might not have the equipment for mind-to-mind communication, and talking might be so integral to how we think that Dnyarri themselves use spoken language, even when compelling you.


Title: Re: Languages in SC2 (new thread)
Post by: meep-eep on August 15, 2004, 10:49:59 pm
Quote
No. You suffer from the misapprehension that language "contains" objective meaning. It contains no such thing. Language only means something because people agree it defines something -- the sounds or letters themselves mean nothing. Without any prior knowledge of what the language means, you will see *nothing* inside of the "data";

You seem to misunderstand. I do not claim the actual sounds or letters have an intrinsic meaning, but the way words are used together *can* tell you something about it's meaning.
For example, take the sentence "I looked up to the sky and *SWOFK* a falling star.". You'll probably will know what *SWOFK* means now. Usually, there will be more possibilities, and when you begin almost everything will be unknown. But when you combine all the sentences, you will find limits to what words can mean, and as you go along there will be less and less degrees of freedom.
It's like a system of many equations with many unknowns. If you've got enough independant equations, you can fill in the unknowns. It's probably too enormous a task for a human to do, but I expect a future computer can be made to do it.
Though a little bit of outside knowledge can make this operation a lot easier.

Quote
Well, it *is* a lack of imagination, whatever other reasons there are behind it. Even if the writers are capable of greater imagination, they choose not to use it, perhaps because they think the viewers or the producers don't have much imagination.

Ok, I see we interpreted "unimaginative" differently. I saw it as "being the result of a lack of imagination", while you appearantly meant it as "showing no imagination". I can live with that.

Quote
And anyway, the elder race explanation is a stupid one. It's not the "real" explanation -- it was introduced in Star Trek: The Next Generation long after the original series established that most races in the galaxy look human, and actually claimed that other races' *history* had to closely follow human history because that was a Law of Nature.

I do not know of this claim, but ST isn't always internally consistent. I would not claim any of those as the "real" explanation.

Quote
Even TNG's attempted explanation, bold as it was, was still ridiculous. The "seeding" was suppposed to have happened billions of years before all these sentient species developed, meaning that while human beings' ancestors were little hairy shrew-like things they had some sort of hidden code in their genes to become five-and-a-half-foot tall, five-fingered, bipedal, large-brained, hairless beings, where males were taller and stronger and females had enlarged mammary glands as a secondary sexual characterstic, etc. And so did *every other damn race*. That's *not* how evolution and natural selection work (and of all the things sci-fi writers often get wrong, what evolution actually *is* seems to be number one).

They may have meant to imply that the "evolution" of the species was in fact a pre-programmed event, with natural selection having only a marginal influence. It's rather silly imho, but if that is necessary for a consistent story, I'm prepared to accept that for a fact in the ST universe.

Quote
Lewis Carroll made a hobby out of creating made-up words and defining them in terms of other nonsense words, and writing whole stories with them, and even though the entire framework is pure standard English, the words are incomprehensible.

Where Lewis Carroll defined made-up words it's almost always in plain English. But usually he didn't even explain them, though you always get an idea of what they mean from the context.



Title: Re: Languages in SC2 (new thread)
Post by: meep-eep on August 15, 2004, 11:36:06 pm
Quote
No. Actually it's the exact opposite! Regional accents exist because even people who start out speaking languages perfectly begin to slowly make mistakes, and then make the same mistakes more and more commonly.

I'm not talking about why there are different accents, I'm talking about why people in some region all speak with the same accent. Which is partly because they copy from eachother.

Quote
That's not proof, but I think the burden of proof is on those who think UTs exist, since I don't think there's much in-game evidence for it and all real-life evidence makes them horrendously unlikely.

Oh, so that's what this discussion was about. :)

Quote
My crusade is to get sci-fi fans out of the habit of *assuming* UTs exist in whatever new sci-fi they're watching/reading/playing because it's just the sort of thing people can build in the future, y'know?

I know many sci-fi fans, but I know of noone ever having assumed that.
Not that they usually make any problems of UTs in sci-fi, they just accept it as a fact of the universe they're watching. Which I think is the idea of sci-fi.

Quote
At least people have come to accept that flying cars, food pills, and teleporters are ridiculous, but this universal translator thing really has a grip on people, if only because it makes filming sci-fi movies so much easier.

While I wouldn't assume these things exist in sci-fi, I wouldn't say they are necessarily ridiculous. But I won't say any more about this, lest this discussion explode any further.


I usually can go on for very long discussing a hypothetical topic, but I'm afraid this thread is exploding. For every little point someone makes follows a long text by you (Art) explaining your views with many examples. I think all those examples aren't necessary; your point usually is clear after your first paragraph (and if it isn't, we can always ask for a clarification). All those examples tempt the reader (well me at least) to just interject a small side remark, but that usually seems to be the starting point for another text explosion.
So I'll be trying to get out of this discussion, only responding to the points where my own earlier remarks are misinterpreted. I don't know whether I'll ever succeed though.



Title: Re: Languages in SC2 (new thread)
Post by: meep-eep on August 16, 2004, 12:10:29 am
About simulating sound: you know there are spy devices which use a laser to detect minute vibrations of windows, which allows someone to hear from afar what is going on inside a building?
If you wanted to (and I don't want to go into whether people would want that), similar techniques could be used to figure out a "realistic" sound of your missile hitting an enemy ship. (Exploding debris won't be that easy).



Title: Re: Languages in SC2 (new thread)
Post by: Sander Scamper on August 16, 2004, 08:32:57 pm
the Talking Pet might not be able to think but it could certainly be trained to tell when a stream of sound and a stream of thought were somehow in sync with each other, one following the other's pattern, and to translate just those thoughts. -

Scary, Scary thought...How do we know its just translating spoken words?


Title: Re: Languages in SC2 (new thread)
Post by: Art on August 16, 2004, 09:25:48 pm
Quote

You seem to misunderstand. I do not claim the actual sounds or letters have an intrinsic meaning, but the way words are used together *can* tell you something about it's meaning.
For example, take the sentence "I looked up to the sky and *SWOFK* a falling star.". You'll probably will know what *SWOFK* means now. Usually, there will be more possibilities, and when you begin almost everything will be unknown. But when you combine all the sentences, you will find limits to what words can mean, and as you go along there will be less and less degrees of freedom.

It's like a system of many equations with many unknowns. If you've got enough independant equations, you can fill in the unknowns. It's probably too enormous a task for a human to do, but I expect a future computer can be made to do it.
Though a little bit of outside knowledge can make this operation a lot easier.


My original rant has already gone over the top, so I'll just say no, you can't, and it has nothing to do with how powerful your computer is.

Quote
Ok, I see we interpreted "unimaginative" differently. I saw it as "being the result of a lack of imagination", while you appearantly meant it as "showing no imagination". I can live with that.

I do not know of this claim, but ST isn't always internally consistent. I would not claim any of those as the "real" explanation.

They may have meant to imply that the "evolution" of the species was in fact a pre-programmed event, with natural selection having only a marginal influence. It's rather silly imho, but if that is necessary for a consistent story, I'm prepared to accept that for a fact in the ST universe.


As a Star Trek fan I'm willing to *suspend* disbelief, the same way I can believe in gods and magic when I'm reading high fantasy. It may not make the story less compelling or fun, but it makes the story less believable, and therefore reduces the degree to which Star Trek is interesting as actual speculation about the future (which it tries to pretend to be sometimes).

And preprogrammed evolution that can ignore vastly different starting conditions and environments on different planets is not just silly but every bit as fantastic as gods and magic. Especially the TOS-era evolution of societies just like ancient Rome and Egypt and such.

Quote

Where Lewis Carroll defined made-up words it's almost always in plain English. But usually he didn't even explain them, though you always get an idea of what they mean from the context.


The only time he actually did that was when Humpty-Dumpty explained the first verse of "Jabberwocky", and that wasn't very convincing (or meant to be very convincing). I meant more like The Hunting of the Snark, where we know certain things happen if the Snark is a Boojum and other things happen if it isn't, and certain things that go along with the Snark being a Boojum or not, but never any hint as to what either a Snark or a Boojum actually is in our language.

And no, you really can't get an idea of what they mean from context. That's the point -- you can make a good guess, and the guess might make sense to you, but it's always written in a way so that some totally different guess would make just as much sense.


Title: Re: Languages in SC2 (new thread)
Post by: Art on August 16, 2004, 09:31:24 pm
Quote

I'm not talking about why there are different accents, I'm talking about why people in some region all speak with the same accent. Which is partly because they copy from eachother.


Yeah, but it's one of those concentric-circle effects; even in small areas people vary, and vary more as they get farther away from each other. In My Fair Lady Professor Higgins says he can tell what street in London someone lives on by their accent, which, I'm told, is only a slight exaggeration. It's less true in modern America because of how often we move and because we all watch TV and learn from that, of course.

Quote

I usually can go on for very long discussing a hypothetical topic, but I'm afraid this thread is exploding. For every little point someone makes follows a long text by you (Art) explaining your views with many examples. I think all those examples aren't necessary; your point usually is clear after your first paragraph (and if it isn't, we can always ask for a clarification). All those examples tempt the reader (well me at least) to just interject a small side remark, but that usually seems to be the starting point for another text explosion.
So I'll be trying to get out of this discussion, only responding to the points where my own earlier remarks are misinterpreted. I don't know whether I'll ever succeed though.


Mea culpa. My own at-home situation may be exacerbating my tendency to be overly argumentative about such things, but we don't need to get into that. :)


Title: Re: Languages in SC2 (new thread)
Post by: Death 999 on August 16, 2004, 10:08:36 pm
Art,

Please do not assume that I know nothing of the subject I am talking about. I have looked extensively into ciphers and codes, I know the difference between them, and so forth. Assuming that the other person in a conversation is appallingly ignorant and/or stupid is not a very productive form of communication.

The examples I was thinking of as far as decoded languages were thus:

After World War II, the decryption team that tackled the Japanese naval code wanted to stay in practice, so they found all of the old unbroken ciphertexts that they could. Several of them were alchemical texts that basically had no 'native' language to be broken into, as they replaced the structural elements of the language with shorthands, and all of the verbs and nouns were encoded. Not ciphered, encoded. In some cases, the codes were rotating, which alien languages would be unlikely to do.
The only hints they had to work with were the intentionally vague illuminations put into the manuscripts. They were intentionally vague to hinder decryption, yet they were enough.

They succeeded in all but one case, and that one is suspected to be a fake, not really a message at all.

Second, a mathematician-historian attempted to use Bayesian methods to decipher a long text of an ancient language much older than Etruscan. Given only about 100 pages of text, she managed to very strongly constrain the meanings. She provided a best-guess meaning, but that was attacked as being too unfounded. However, she was quite willing to admit that it was simply the least-unlikely meaning guess based on the mathematical model, and other guesses of varying similarity also ranked highly. I cannot remember her name, the language, etc. This is frustrating, I will look around.

Thirdly, Etruscan is not a great example. How much Etruscan text do we have? The Bayesian methods one might use would require large sample sizes. I doubt that we have  more than 10 megabytes of Unicode-format Etruscan text, total. If we were to dredge up, say, a few gigabytes of data with context supplied, we would run an excellent chance of translating the language from scratch. Sure, that's a heck of a lot of text for people writing with chisels. But the question at hand is not deciphering ancient Etruscan, it's futuristic first contacts.

Futuristic ships might send out a low-entropy signal designed to start up a primer. A cleverer primer than the one sent by the folks broadcasting from Vega in Contact, which stymied everyone on earth for a while. Something like the Arecibo bitmap broadcasts -- very low entropy, fairly obvious interpretation. Depending on how smart the two sides are, the computers might be ready to translate before the captains have gotten to the bridge.


Title: Re: Languages in SC2 (new thread)
Post by: definite on August 18, 2004, 06:30:00 pm
Quote

A mathematician-historian attempted to use Bayesian methods to decipher a long text of an ancient language much older than Etruscan. Given only about 100 pages of text, she managed to very strongly constrain the meanings. She provided a best-guess meaning, but that was attacked as being too unfounded. However, she was quite willing to admit that it was simply the least-unlikely meaning guess based on the mathematical model, and other guesses of varying similarity also ranked highly. I cannot remember her name, the language, etc. This is frustrating, I will look around.

Thirdly, Etruscan is not a great example. How much Etruscan text do we have? The Bayesian methods one might use would require large sample sizes. I doubt that we have  more than 10 megabytes of Unicode-format Etruscan text, total. If we were to dredge up, say, a few gigabytes of data with context supplied, we would run an excellent chance of translating the language from scratch


I would like to read the algorithms you describe and test them with Chinese documents. :) Since we definitely have millions of Chinese documents. We have enough training set and test set.

Quote

. Sure, that's a heck of a lot of text for people writing with chisels. But the question at hand is not deciphering ancient Etruscan, it's futuristic first contacts.


If it can properly solves ancient Etruscan without prior Etruscan knowledge, the same technique can be used to first contacts.

But in order to develope the mircle technique, we should solve following problem first:

Given a long binary bit string as training instance, and a shorter binary bit string as test instance, explain what does the test instance mean.

If the problem solved, there will be no problems in the fields of deciphering text, OCR, WSD and translation. :-)

In terms of WSD, would you mind to tell me what does the 'right' mean in following text.

Testee: "Turn left?"
Examiner: "Right"


Title: Re: Languages in SC2 (new thread)
Post by: Bobucles on August 18, 2004, 07:44:26 pm
Quote
Please do not assume that I know nothing of the subject I am talking about. I have looked extensively into ciphers and codes, I know the difference between them, and so forth. Assuming that the other person in a conversation is appallingly ignorant and/or stupid is not a very productive form of communication.

You forget that this is a public forum. Clearly stating and defining topics that are not well known is a polite thing to do. And it lets everyone join in on the conversation, not just the few people discussing it.

If you were to do such a thing with private chat... that would be another matter entirely.


Title: Re: Languages in SC2 (new thread)
Post by: Death 999 on August 18, 2004, 08:12:08 pm
Supposing you have started a thread on a specialized field, when someone says something you disagree with, there is a big difference between asking them for some justification and/or credentials, and proceeding to give a lecture on the basics of the field in a manner which suggests that they had no right to butt in.



Quote
Given a long binary bit string as training instance, and a shorter binary bit string as test instance, explain what does the test instance mean.
 
If the problem solved, there will be no problems in the fields of deciphering text, OCR, WSD and translation.  
 
In terms of WSD, would you mind to tell me what does the 'right' mean in following text.
 
Testee: "Turn left?"
Examiner: "Right"


     

Some strings are harder than others. Your example is indeterminate. Obviously, if you are attempting to give a primer to an alien species, you will attempt to make your string relatively easy. And it is this which makes the problem completely different from the solving of ancient languages. You have a lot of common context with the aliens, and both of you are trying to make it easy on the other. This is itself a very important clue as to the meaning.


Title: Re: Languages in SC2 (new thread)
Post by: meep-eep on August 18, 2004, 10:59:31 pm
Quote
The only time he actually did that was when Humpty-Dumpty explained the first verse of "Jabberwocky", and that wasn't very convincing (or meant to be very convincing). I meant more like The Hunting of the Snark, where we know certain things happen if the Snark is a Boojum and other things happen if it isn't, and certain things that go along with the Snark being a Boojum or not, but never any hint as to what either a Snark or a Boojum actually is in our language.

In the Hunting of the Snark he didn't define either Snark or Boojum at all. Neither in terms of other words, nor in plain English.

Quote
And no, you really can't get an idea of what they mean from context. That's the point -- you can make a good guess, and the guess might make sense to you, but it's always written in a way so that some totally different guess would make just as much sense.

You do know certain constraints. You know that a snark is a creature, how it tastes, that it gets up late, how it is to be captured.



Title: Re: Languages in SC2 (new thread)
Post by: Death 999 on August 19, 2004, 12:02:23 am
Also, the snark really WAS a Boojum, you see...


Title: Re: Languages in SC2 (new thread)
Post by: definite on August 20, 2004, 10:06:01 pm
Quote
Some strings are harder than others. Your example is indeterminate. Obviously, if you are attempting to give a primer to an alien species, you will attempt to make your string relatively easy. And it is this which makes the problem completely different from the solving of ancient languages. You have a lot of common context with the aliens, and both of you are trying to make it easy on the other. This is itself a very important clue as to the meaning.


In your example, you need to define what is the simple message and the simple meassage representation. Do you have that in mind?

Ok, alien has sent a message to you. Your computer has already store them in binary bits form. What's your algorithm to solve/decipher it?


Title: Re: Languages in SC2 (new thread)
Post by: Death 999 on August 23, 2004, 09:53:47 pm
You can answer this yourself -- explain the logic you use as you decode the following transmission.
You receive a signal composed of flashes of two different lights of different colors. Here I represent that so: 0 means both lights are off. 1 and 2 are the bits for the two lights; 3 means both are on.

2012220220212202220112202222022120222220
1212022222202112022222220111202222222202
2210222222222012210222222222202121022222
2222220112102222222222220221102222222222
2220121102222222222222202111022222222222
2222011113122201012220110212231222010212
2011011223112201012220110221232122010212
2011022123111201011220110212131122010221
1011000
(transmission stops -- all zeroes)

See what you can figure out from that.  I pose this as an open problem. Also, does anyone have anything to say back to these aliens?


Title: Re: Languages in SC2 (new thread)
Post by: meep-eep on August 24, 2004, 06:40:42 am
1122010221101101111


Title: Re: Languages in SC2 (new thread)
Post by: Death 999 on August 24, 2004, 10:02:50 pm
meep, you are da bomb.

If anyone thinks it would be particularly impossible to get all the way from there to language, there is a thread at sluggy.net where I cross-posted this, and people are working on it. I went a bit further there, too... if desired, I can cross-post it back here.


Title: Re: Languages in SC2 (new thread)
Post by: FalconMWC on August 25, 2004, 07:46:41 pm
I would like that......


Title: Re: Languages in SC2 (new thread)
Post by: Sander Scamper on August 25, 2004, 08:36:21 pm
And i have trouble learning Japanese at school  :o :o

Wow


Title: Re: Languages in SC2 (new thread)
Post by: Death 999 on August 25, 2004, 08:38:10 pm
My initial post was basically the same as here.
Skipping some silly responses...

Kaintuckee_Bob said:
Quote
One thing I've noticed, and I'm not sure if it is significant, is that the only number in the series which never follows itself is 0.  Therefore, assuming the 0s are spaces or breaks between words, we get the following:

<insert quote of string with all 0s replaced by spaces>

Actually, there's a pattern to the start (right up until the first number with a 3 in it) of an incrementing number of 2s spacing out a binary number which is incrementing in some fashion which (in hex) is 7b3d591e6a2c48.  I ended up with this:


0000 2 (message start)
1222 22
2122 222
1122 2222
2212 22222
1212 222222
2112 2222222
1112 22222222
2221 222222222
1221 2222222222
2121 22222222222
1121 222222222222
2211 2222222222222
1211 22222222222222
2111 222222222222222


he then got confused and I can't even figure out what he said next, and he was speaking English.

Gallifreyan then said:
Quote
I would simply reply with the following code, broadcasted in the same fashion as the first:

<insert original message with trailing 0s removed>22220000000...


I then replied:
Quote
Gallifreyan, that answer is much longer than I was thinking of.

In response, the aliens send
(...000)11220102211011011113(000...)
(...000)12120102122011(000...)


Gallifreyan:
Quote
In response to
<insert my previous message>

I would complete this signal and retransmit it to the source as:

(...000)11220102211011011113(000...)
(...000)1212010212201101112(000...)

How's that? I think I realize my previous mistake.


Pleased by Gallifreyan's progress, I went on to add a lot more.
Quote
(...000)
1111021011220110221132122
0210122201101222312120210
1122011012223
(000...000)
2222021101222322220211021
2232222021101122322220211
0221230000000000000000000
000112201012110211012123
(000...000)
1111010122201102222122232
2221222010122201101222122
2322122112021011212122011
0122111223
(000...000)
3022220220122233022220110
1222330222201022120110121
2322120103022220110121231
1110210302222011021113302
2220220221133022220110221
1330222201012220110121133
0122202203022223302222022
0111133022220213012220110
11223
(000...)


a moment later, I said:
Quote
I would like to point out that if I were to start over I would change certain aspects of the syntax: as I attempt to expand it, I find very little room available, despite the low entropy.

Another thing is, since the aliens are making this up on the spot, you can add syntax of your own, so long as it doesn't contradict theirs.

You could even try to start over again, propose a better system... using a primer string of course.


No one has yet tackled this. Upon my requesting Gallifreyan to explain his logic, he reposted Kaintuckee_Bob's post, followed by:
Quote
By aligning the 2's into pyramid form, we get a conversion system explaining the alien's number system. Also, the column on the left lists the binary numbers 14 down to 0 (this can be seen by changing the 1s to 0s and the 2s to 1s, and then reading the binary numbers from right to left as in 1248). Then, I corresponded these to the natural numbers. So, we get:

0000 (message start)
0111 14 1
1011 13 2
0011 12 3
1101 11 4
0101 10 5
1001 9   6
0001 8   7
1110 7   8
0110 6   9
1010 5   10
0010 4   11
1100 3   12
0100 2   13
1000 1   14
0000 0   15  (This last one can be logically deduced)

I figured out that this might be the aliens' number system, so that the next section of the signal can be deduced as well from a number system.

Next, a series of equations was given (I have returned the "space" 0s back in, and separated the code into their proper "words"):

1111 3
1222 010 1222 0110 2122    3
1222 010 2122 0110 1122    3
1122 010 1222 0110 2212    3
2122 010 2122 0110 2212    3
1112 010 1122 0110 2121    3
1122 010 2211 0110

Translating the sequence 010 to be +, 0110 to be =, and 3 to be an equation separator, and then translating the binary numbers, using the columns above, we get:

1111 3 (equations begin)
1 + 1 = 2
1 + 2 = 3
3 + 1 = 4
2 + 2 = 4
7 + 3 = 10
3 + 12 = ???(signal ends here)
Thus, I concluded that the aliens wanted the last equation to be completed. After correcting myself (oops!), I translated the last equation as follows:

3 + 12 = 15
0011 + 1100 = 0000
1122 010 2211 0110 1111
1122010221101101111

This last  sequence is the same as the end of the original signal, wih a 1111 tacked onto the end.

I screwed it up first, which is why the second signal corrected me:
(...000)11220102211011011113(000...)
And gave a new equation for me to "try again"
(...000)12120102122011(000...)
This one becomes:
5 + 2 = ???
The answer, obviously, is 7, which becomes 8, and then 0001, and then 1112. So, I respond with:
(...000)1212010212201101112(000...).

There ay go. Btw, responding with the whole signal (plus the answer) also is good mathematics, because it confirms that we are working upon the same assumptions.


Note that even though he misinterpreted the pyramid slightly, he still figured out the number system and operators. Also note that he wasn't entirely sure that he was correct. This could have been verified if he had asked a question in return and gotten the expected response.


Title: Re: Languages in SC2 (new thread)
Post by: meep-eep on August 26, 2004, 01:02:05 am
Quote
12120210112201101222

Are you sure this is correct?


Title: Re: Languages in SC2 (new thread)
Post by: meep-eep on August 26, 2004, 01:41:13 am
Quote
302222021301222011011223

Shouldn't that be
3022220210301222011011223


Title: Re: Languages in SC2 (new thread)
Post by: meep-eep on August 26, 2004, 01:58:19 am
(...000)
3102222222222222222222122
1222122222122221212212222
2111212222222212222122212
2112212222222222211112222
2222221121111211222222222
2211222222222222122222122
2222222122121221222222222
2122122222222223
(000...000)
3011110220310222222222222
2222222122122212222222222
1212212222222221222222221
2222222212211221222222222
2211112222222222112111121
1222222222221122222222222
2122222122222222212212122
1222222222212212222222222
3302111022031022222222222
2222222222222222222221222
2222222222221112222222222
2222212222222222222222222
2222222222222222222222222
2222222222222222222222222
2222222222222222222222222
2222222222222222222222222
23
(000...)


Title: Re: Languages in SC2 (new thread)
Post by: Death 999 on August 26, 2004, 08:58:04 pm
See, Meep-eep is so good at this that he is (correctly) correcting my grammar and arithmetic! Now let me parse that long message...

Hmm... He's definitely not afraid to invent new grammar himself, but that's fair.

response:
(...000)
2121012102122011012123211201210112201102
12232212012102122011021120121011223
(000...000)
3011110220222222222222222212212212221222
2212222121221222221112122222222122221222
1221122122222222222111122222222221121111
2112222222222211222222222222122222122222
2222122121221222222222212212222222222330
2111022022222222222222222221221222122222
2222212122122222222212222222212222222212
2112212222222222211112222222222112111121
1222222222221122222222222212222212222222
2212212122122222222221221222222222233012
1102202222222222222222222222222222222212
2222222222222211122222222222222212222222
2222222222222222222222222222222222222222
2222222222222222222222222222222222222222
2222222222222222222222222222222223302211
0220222222222222222212222222222222222222
2222222222222222222222222222222222222222
2222222222222222222222222222222222222222
2222222222222222222222222222222222222222
2222222222222222222222222222223000000302
111012103012110110301211012103022113
(000...)


Title: Re: Languages in SC2 (new thread)
Post by: Miltazar on August 27, 2004, 10:55:55 am
Alrighty, I decyphered the first couple messages pretty easily.  I'm proud of that but starting with the first part of Meep-eep's msg...i just can't seem to grasp the pattern involved.  I know i'm probably missing something quite simple but...mind giving a hint?  If don't want to spoil it you can email it to me at miltazar@hotmail.com.  I'm assuming its a number judging from the second part...on the other hand it could be the start of an alphabet...*shrugs*...none of the patterns i've tried have resulted in anything comprehendable. *sigh*


                                                      --Miltazar


Title: Re: Languages in SC2 (new thread)
Post by: meep-eep on August 27, 2004, 11:23:54 am
I missed Death_999's response until now because he modified his existing posting. I was just about to go to bed (7am seems like a good time for that :) ), so decoding it and sending a response will have to wait.

I've sent a small hint to Miltazar though through the board's PM system. If anyone else wants a hint, just let me know. Or I could send another encoded message.



Title: Re: Languages in SC2 (new thread)
Post by: Sander Scamper on August 27, 2004, 07:53:33 pm
This is mind-boggling....Here is mine.....

                                     0       ()         0
                                   11       ()         11
                                            111                                
                                   11    22222     11
                                   11222222222211
                                         2222222
                                       22  111  22
                                     22     11     22
                                     11      1      11  


Thats supposed to look like a awesome ship made of 1s and 2s.... =/


Title: Re: Languages in SC2 (new thread)
Post by: Death 999 on August 27, 2004, 09:11:16 pm
That is eerily similar to Meep's recent statement.

BTW, I posit that now that we're throwing pictures around, we've established at least that WE can start up a language. We had not yet jumpstarted into rules of inference, but I'm confident it could be done. Note that I could have started with images, like Arecibo and Meep did, but I wanted to handle this in a numeric way that a computer could handle quickly (thus supporting the idea that primer strings can allow communication to start up quickly even in a first contact situation).

Now, how much shared information did we have that aliens would not necessarily have?

-- A binary number system (would help, but even if you had never encountered it, you could figure it out with enough examples). I posit that any alien species which communicates with flashes of light and uses computers of some sort will at least be aware of binary formats, even if they don't use base 2. Perhaps some form of gray counting, a binary number format which only involves 1 change of bit per increment or decrement, would be used if they have asynchronous computers. However, not being aware of base arithmetic would be a gaping hole. Maybe they fill it by other means which are gaping hole in our understanding.

-- A basic rastering bitmap image format (note, we did not yet rely on orientation)

-- That stars look like what we see when we see stars. It could be that aliens' eyes do not have the same flaws ours do and do not see the beams sticking out. Then the image would be very confusing to them. However, such false beams can be brought about by many kinds of even slight imperfections in lenses. Assuming their eyes use lenses  (a pretty good assumption, since mere collimation requires extreme sensitivity), such imperfections are likely to pop up sufficiently often that they would get the idea.


Title: Re: Languages in SC2 (new thread)
Post by: Sander Scamper on August 28, 2004, 07:26:29 pm
Maybe some of the species use such things as echolocation?


Title: Re: Languages in SC2 (new thread)
Post by: meep-eep on August 29, 2004, 03:13:59 am
I don't think that you can do easy arithmetics if you use gray codes for your numbers.  That pretty much rules them out for a general purpose number system. So they are bound to have something else.

Orientation of an image can be established with a line on what is to be the bottom side of the image. Beings used to living on a solid planet will probably correctly explain that line as meaning the ground.

I didn't intend the lines to be visual flaws of lenses; I meant to show something radiating from the solid object.
It's a good point that not all beings may be able to naturally detect radiation. But they at least know the concept (they communicated through lights), and they must have noticed certain spatial bodies producing radiation.

As for your last message, you seem to have left out the '31' which I used to signify the start of an image. Was that intentional? And the last image misses one pixel.



Title: Re: Languages in SC2 (new thread)
Post by: Miltazar on August 29, 2004, 09:14:52 am
Alrighty, heres my attempt at a msg. NOTE: it is pretty complicated but its underlying meaning is related to meep-eeps...except more advanced version.

(...000...)
312022222222222222222222222222222222222
222222222222222222222211222222211222222
222211222222222222222221222222222222211
222211222222211222222222222222221221221
222211222222222222222222222222211222222
222222221222222222211222222211211222222
211222222222222222222222222222222222211
211211211222222222222222222222222222222
211211222211211211211222211211222222222
222222222222222222222222211211222222222
222222222222222222222222222211222222222
222222211222222222222222222222222222211
222222211222211222222211222222222222222
222222222222222211222222211222222222222
22222222222222222223
(...000...)


Title: Re: Languages in SC2 (new thread)
Post by: meep-eep on August 29, 2004, 08:42:19 pm
This one is still pretty easilly decipherable, but I think that when you're going to transmit "3D" images it may be useful to add borders to the image, so you can see more easilly whether it's supposed to be 11x17x3, 17x11x3, 11x3x17, 17x3x11, 3x11x17, or 3x17x11.



Title: Re: Languages in SC2 (new thread)
Post by: gargamel51 on September 07, 2004, 10:20:27 am
Quote

(It's the same kind of Star Trek logic that says there's something universally useful about the human shape, so that all intelligent aliens have to be humanoid. Unimaginative nonsense.)

Actually, this is NOT Star Trek's explanation for the prevlance of the humanoid form in extraterrestrial lifeforms (and furthermore, in Star Trek, NOT all intelligent aliens are humanoid).  As was explained in Star Trek: The Next Generation Season 6 Episode 20, "The Chase," the first humanoid race (which was an extremely advanced civilization) was lonely, so they seeded planets all throughout the Milky Way Galaxy with variants of their DNA.  It was thier hope that their "children" would come together to solve a puzzle to learn of their origins.  So, actually, according to Star Trek, so many aliens are bipedal, carbon-based lifeforms NOT necessarily because of something that's "universally useful" about this bodily form.  (Although if you want to be cynical about it, from a real-life standpoint, it's just beacuse the special effects involved in creating outrageously-shaped lifeforms would be too expensive.)

More about "The Chase" at http://www.startrek.com/startrek/view/series/TNG/episode/68598.html

Gargamel


Title: Re: Languages in SC2 (new thread)
Post by: gargamel51 on September 07, 2004, 10:25:29 am
Damnit, Gargamel, read the WHOLE THREAD before you post!!  :-[


Title: Re: Languages in SC2 (new thread)
Post by: Art on September 07, 2004, 10:27:50 am
Quote

Actually, this is NOT Star Trek's explanation for the prevlance of the humanoid form in extraterrestrial lifeforms (and furthermore, in Star Trek, NOT all intelligent aliens are humanoid).  As was explained in Star Trek: The Next Generation Season 6 Episode 20, "The Chase," the first humanoid race (which was an extremely advanced civilization) was lonely, so they seeded planets all throughout the Milky Way Galaxy with variants of their DNA.  It was thier hope that their "children" would come together to solve a puzzle to learn of their origins.  So, actually, according to Star Trek, so many aliens are bipedal, carbon-based lifeforms NOT necessarily because of something that's "universally useful" about this bodily form.  (Although if you want to be cynical about it, from a real-life standpoint, it's just beacuse the special effects involved in creating outrageously-shaped lifeforms would be too expensive.)

More about "The Chase" at http://www.startrek.com/startrek/view/series/TNG/episode/68598.html

Gargamel


I said it was the same *kind* of logic, not the same logic. It is, in fact, the logic behind why the Universal Translator device in Star Trek works on every race they meet -- it reads the races' brainwaves and correlates them with English words, which apparently always works because races' brains always follow the same patterns as English-speaking humans.

But I was aware of this very TNG episode, and ranted a bit a while back about how stupid it is. You can't "seed" DNA in a precise manner so that the random forces of evolution will end up making something that looks like you. Saying you can is to show a total disregard for what evolution is and how it works (which the Star Treks, throughout their entire history, have *always* done).


Title: Re: Languages in SC2 (new thread)
Post by: Sander Scamper on September 07, 2004, 04:46:02 pm
Add that to the list of other kinds of science that StarTrek has since butchered. =p

I really think that in the interests of lucidity, we should all just assume that there is a Sub...SUB brainwave that all live forms possess, that in the case of sentient, those sub brainwaves can easily be translated into binary, which I assume all atomic age races would have (given electricity is represent as 1, or 0.


Title: Re: Languages in SC2 (new thread)
Post by: Art on September 08, 2004, 02:24:10 am
Um... okay. Given that that makes only slightly less sense than what we're given in the Star Trek Technical Manual, we can go with it.


Title: Re: Languages in SC2 (new thread)
Post by: Sander Scamper on September 08, 2004, 08:42:31 am
Stupid, but simple.


Title: Re: Languages in SC2 (new thread)
Post by: Death 999 on September 08, 2004, 10:48:27 pm
Quote
I didn't intend the lines to be visual flaws of lenses; I meant to show something radiating from the solid object.
It's a good point that not all beings may be able to naturally detect radiation. But they at least know the concept (they communicated through lights), and they must have noticed certain spatial bodies producing radiation.


fair enough. I was just trying to be pessimistic so as not to jump to conclusions

Quote
As for your last message, you seem to have left out the '31' which I used to signify the start of an image. Was that intentional? And the last image misses one pixel.


umm, the syntax was unintentional. And the missing pixel, if you mean there are 1 too few pixels, not just a change of value, umm... oops.


Title: Re: Languages in SC2 (new thread)
Post by: taleden on September 09, 2004, 04:37:16 am
So far as I know, this hasn't come up yet - I haven't sat down to try to translate all your new-language messages, I'm just going on the commentary on those translations posted here.  But since you've already progressed from unary number systems to 2d (3d even?) images, someone, eventually, is bound to try to come up with a way to encode color.  Unless they've studied the physics of light and human color perception, they are almost certainly going to do it in a way that would be completely nonsensical for any alien species, even one that detected visible light much the way we do (which is a stretch in itself).

I won't go into the mucky details of it unless somebody's really curious, but suffice it to say, the red-green-blue additive primary color scheme we use on CRTs, projectors, etc, and the cyan-magenta-yellow subtractive primary color scheme we use for everything on paper ONLY works because of the exact frequency responses of the three color receptors in the human eye.  Even if we encountered an alien species that also detected visible light using similar receptors, which is unlikely, the chance that they would have three such receptors and that they would have the same frequency responses as our own color receptors is about zero.

As an example: when our monitor mixes red and green light to make a yellow pixel, it may look (to us) exactly like the yellow we see on a banana.  But, that does NOT mean the light coming from our yellow computer pixel and the light reflecting from the banana are both yellow; the electromagnetic frequency of light coming from the computer screen is still just red plus green, and the light coming from the banana is still just yellow.  The fact that we can't tell the difference between those two situations is a product of the physics of our eyeballs.

So what I'm saying is, if you try to encode color, DON'T do it in RGB.  Or even CMY.  In fact, offhand, I'm having a really hard time thinking of a good way to encode color to make sure an alien would be able to reproduce it correctly.  But maybe one of you is more clever.  :)


Title: Re: Languages in SC2 (new thread)
Post by: Art on September 09, 2004, 07:16:20 am
As the originator of this thread I should say that I've been interested in the whole language-decoding exercise, though I also haven't felt inclined to take the effort to work on it myself.

I should say, though, that what I meant by using nonverbal information to decode words pretty much is this exercise of first sending numerical, mathematical information and then sending pictures. It's a step to go from there to coding abstract concepts in a natural language; the most natural step would seem to me to be matching pictures with encoded words, and the ability to identify pictures in order to match them to words and extrapolate from the meaning of a picture to abstract meanings (like, having learned words for "human" and "knife", understand that a simple word going with a picture of a human sticking a knife into another human is "kill" or "murder") is a rather complex one. It'd be a pretty big leap in AI for a computer to be able to do it unaided.

Of course, you've demonstrated you probably know more about this than me, but it's still my belief that while this sort of exchange is certainly possible and would probably be done with an alien race we'd contacted, it wouldn't be possible to have computers automatically act as translators for some alien race within the early days of contact; doing this sort of thing on a large enough scale to get a big dictionary would be a massive task involving many humans.


Title: Re: Languages in SC2 (new thread)
Post by: Profound_Darkness on September 09, 2004, 08:47:16 am
Firstly, very interesting topic...

Quote
So what I'm saying is, if you try to encode color, DON'T do it in RGB.  Or even CMY.  In fact, offhand, I'm having a really hard time thinking of a good way to encode color to make sure an alien would be able to reproduce it correctly.  But maybe one of you is more clever.  :)


Well since light is a waveform why not send them the start of our visible spectrum and the interval value for each step higher.  Then when sending an image send the relative and steped color value in the electromagnetic spectrum for each pixel as well as the intensity (energy level).  Energy would also have to have some kind of steping and baseline sent too.  A lot would have to be passed back and forth before images could be sent.  There would have to be some other information sent out along with the above, this is a bit of a simplification.

But asside from that you can do pretty good imigery with monochrome, same concept as engraving if you discount depth.  One step up from that would be grey scale with a little preamble first.

Hope that made sense, it's getting late for me.


Title: Re: Languages in SC2 (new thread)
Post by: Sander Scamper on September 09, 2004, 06:31:09 pm
What if your computer sent along colour frequencies afterward that their computer translated into the appropriate colours visible to THEIR eyes?


Title: Re: Languages in SC2 (new thread)
Post by: taleden on September 09, 2004, 09:30:24 pm
It wouldn't help much to give the aliens the frequencies for R, G and B that we were mixing to make our colors, since they wouldn't know how the mixtures would look to us without knowing a LOT about the details of the chemistry of our eyeballs.  R+G=Y is not a universal equation, even if you replace names with frequencies; it's still dependent on the chemical color receptors unique to humans.

But, establishing an absolute frequency scale might work - giving them the range of frequencies visible to us and then telling them how we were going to indicate different frequencies within that range.  It would still require a lot of other things to be established first, however.. for example, we measure frequency in hertz, which roughly means "oscillations per second", but how do we tell them how long a second is?  We define seconds in terms of minutes in terms of hours in terms of days, but even if we told them that a day was how long it took our planet to complete one revolution on its axis, they still wouldn't know how long that was without us telling them a lot about the orbital properties of our planet, etc etc.

Basically, the problem of color is a pretty high-level one, and it couldn't really be addressed until we'd already established a way to convert our measurement units, which is in itself a big task and would probably require the ability to translate arbitrary sentences.