• Welcome! The TrekBBS is the number one place to chat about Star Trek with like-minded fans.
    If you are not already a member then please register an account and join in the discussion!

How can a Universal Translator work without prior contact to species' language?

Cora Dessalines

Cadet
Newbie
Hey people,

I'm doing some research for a sci-fi book I'm writing. I just wondered if anyone knows anything about the Universal Translators on Star Trek and how they work. Specifically, I want to know how it's possible for them to decipher a language that they haven't previously come into contact with.

Any info would be super helpful. Thank you!
 
Hey people,

I'm doing some research for a sci-fi book I'm writing. I just wondered if anyone knows anything about the Universal Translators on Star Trek and how they work. Specifically, I want to know how it's possible for them to decipher a language that they haven't previously come into contact with.

Any info would be super helpful. Thank you!

It is simply a creative liberty/convention. However, I may have a possible in-universe (partial) explanation. In the TNG episode "The Chase" it was revealed that all humanoid life in the galaxy are related in some way and have a similar gene pool. In the future, the human brain has been mapped out to a large degree including discovering how the brain processes language. Perhaps most humanoid life have similar brain structure and so the UI can extrapolate and create a real time syntax translation. As for "stranger" lifeforms, it may take the UI a few minutes to decode the language which is why you sometimes see a delay for the UI to work in those instances.
 
Last edited:
Such translation devices go back nearly a century in literary science fiction. The mechanics of the Federation's version were touched upon in a little series called Star Trek from the 1960s. Let's go back to the source:
TOS "Metamorphosis" said:
COCHRANE: What's the theory behind this device?
KIRK: There are certain universal ideas and concepts common to all intelligent life. This device instantaneously compares the frequency of brainwave patterns, selects those ideas and concepts it recognises, and then provides the necessary grammar.
SPOCK: Then it translates its findings into English.
(as transcribed over at chakoteya.net)

Fanciful and near-magical to be sure, but what Trek tech isn't?

ST:ENT described an earlier and more primitive version of the tech as follows (being programmed with hundreds of languages, and able to learn new ones, but no mention of brainwave pattern scans):


Kor
 
I always liked the aforementioned explanation from Metamorphosis, that it picks up on ideas and concepts and translates them into an understandable form.

Other than that, I think we're just meant to assume all languages follow very broadly similar logic that the translator can pick up on and comprehend. This is supported by the few cases of aliens who speak languages which are too alien to be translated, like the aliens in TNG's "Darmok", and the aliens in DS9's "Sanctuary".
 
Hey people,

I'm doing some research for a sci-fi book I'm writing. I just wondered if anyone knows anything about the Universal Translators on Star Trek and how they work. Specifically, I want to know how it's possible for them to decipher a language that they haven't previously come into contact with.

Any info would be super helpful. Thank you!
The translators are able to read these aliens' minds thanks to Betazoid technology
 
I always liked the aforementioned explanation from Metamorphosis, that it picks up on ideas and concepts and translates them into an understandable form.

Other than that, I think we're just meant to assume all languages follow very broadly similar logic that the translator can pick up on and comprehend. This is supported by the few cases of aliens who speak languages which are too alien to be translated, like the aliens in TNG's "Darmok", and the aliens in DS9's "Sanctuary".

I can buy that it can figure out syntax, but no way it can guess vocabulary.
 
Its pretty much a very advanced form of deep learning AI that has its beginnings today. Combined with superfast quantum computers, which need a tiny fraction of time to compute billions of states unlike todays common CPUs. AI combined with quantum computers will be the next evolutionary step for mankind after the Internet; what takes days today with the help of human intelligence will be done in realtime tomorrow without any influence of a human being.
 
Its pretty much a very advanced form of deep learning AI that has its beginnings today. Combined with superfast quantum computers, which need a tiny fraction of time to compute billions of states unlike todays common CPUs. AI combined with quantum computers will be the next evolutionary step for mankind after the Internet; what takes days today with the help of human intelligence will be done in realtime tomorrow without any influence of a human being.
Yeah, but an AI still isn't going to know if the first words out of someone's mouth will be hello, who are you, what are you, do you come in peace, get off my lawn, etc. The learning curve is inescapable, and that's ok.
 
It can translate every language because it's a universal translator. That's what that label is specifically meant to convey -- that it's a hypothetical kind of translator that works on everything, not just known languages. (Cf. the concept of a universal solvent, something that can dissolve every imaginable substance and thus couldn't be contained in anything.) If it only works on previously known languages, or any other finite subset of languages, then by definition it isn't a universal translator, just a translator.

The fictional premise of a universal translator is linked to the notion of universal language, the linguistic or philosophical idea that there are elements of language that are fundamental and shared by all beings, and that if you could reduce language to its invariant fundamentals, you'd have the key to cracking any language in existence (as in the Kirk "Metamorphosis" lines quoted above). Unfortunately, this idea has been pretty much discredited by now.
 
Yeah, but an AI still isn't going to know if the first words out of someone's mouth will be hello, who are you, what are you, do you come in peace, get off my lawn, etc. The learning curve is inescapable, and that's ok.

AI will be able to render a set of predictions from what you want to say in the next milliseconds according to the current sound output of your voice and the predictions that fit to what you are actually saying will be the realtime translation. So the delay of the translation will be probably around 500 milliseconds, since the first two letters of a word you say already reduce the possible outcomes. It wont wait for you what you say, but already have several sentences completely translated ready while you are speaking (also because it has learned what others said in the past when they formed a sentence beginning like the one you use).

So for example when you say "How a..." it will already have "How are you [today|right now]?" translated ready.
 
The only way the UT could work as shown is if it's reading people's minds and knows in advance what they are going to say but if it does that then why can't just people use the UT to communicate telepathically?
 
AI will be able to render a set of predictions from what you want to say in the next milliseconds according to the current sound output of your voice and the predictions that fit to what you are actually saying will be the realtime translation. So the delay of the translation will be probably around 500 milliseconds, since the first two letters of a word you say already reduce the possible outcomes. It wont wait for you what you say, but already have several sentences completely translated ready while you are speaking (also because it has learned what others said in the past when they formed a sentence beginning like the one you use).

So for example when you say "How a..." it will already have "How are you [today|right now]?" translated ready.
That's not addressing the OP or my post that you were replying to. That has to do with having prior knowledge of a language and speech patterns. Predictive tech is fine, but even in your example, the word "how" often means "by what means," and it would realistically need a moment to get the rest of the context. Literal word-for-word translation is awkward, often wrong, and sometimes hilarious.
 
That's not addressing the OP or my post that you were replying to. That has to do with having prior knowledge of a language and speech patterns. Predictive tech is fine, but even in your example, the word "how" often means "by what means," and it would realistically need a moment to get the rest of the context. Literal word-for-word translation is awkward, often wrong, and sometimes hilarious.

Which is why the UT doesn't translate a language into another. It just reads the thoughts preceding the words you're going to say and goes from there. These thoughts are actually much clearer and easier to translate than the language they're translated into which contains often cryptic remarks and convolutions linked to the culture of the person. Like if you think that someone is really annoying and you're a diplomat, you're going to say it in so subtle a way that the person will maybe not even get that it's a negative remark but in your head, it will be clear that you find that person to be annoying.
 
Back
Top