Is a universal translator possible

Post Reply New Topic

So is a translatoor capable of immediately translating a language a la Trek even scientifically possible? I would posit no since there would be no way to establish a conceptual reference. You could work out the grammatical rules... if the languyage is internally consistent but actually working out that Ga'Lurya refers to a tree in the middle of summer populated by parakeets is more problematic

Last edited by binaryodes on 20 Jan 2014, 7:12 am, edited 1 time in total.

Did you really have to post this twice?

Obviously it was an error. No need to be so confrontational

Technically, it is possible. It's a question of processing power and software.

I know in DS9, the translator was an in-ear implant with Quark, so it was pretty small.

I think another concept we don't see mentioned is that while everyone in Star Trek speaks English, clearly it's possible most all Federation planets learn a common language...simplifying inter-species communication.

I thought it was more troubling in an episode of Voyager when a bunch of people from Earth taken in the 20th century could understand Janeway regardless of their original languages. If the translator was worn around the neck or something, it would make some sense, but even a communications badge wouldn't do it so fast or translate for other people of multiple languages.

Maybe something that could do most common languages.
But there will likely always be that "one" language that screws it up.

How do you propose to solve the referential problem. A language is essentially a meaningless stream of syllables. How do you assign meaning to those syllables? How doy ou link those sound patterns to real world concepts ideas and objects

Before we talk about universal translators, the first step is to translators that can handle a fixed translation from one language to another language that is relatively similar to that.

For that matter, how about a translator that can translate local variants of one language to another local variant of the same language.

Eventually, there would be translators from one language to another very dissimilar language. Let's see you do one from Navajo to German.

If you watch the movie"The last star fighter". There is a part in the movie where they take the chip and circuit board out of a digital watch and attach it to the lapel of you shirt and it automatically makes any foreign language spoken at you sound like it's spoken in English.

A universal translator would never be able to work unless it was fully programmed with each of the languages being translated from to. Even using known languages such as French and English Google translate makes a complete abortion of translations much of the time; sometimes changing the meaning of sentences and especially messing up idioms - we use lots of idoms in daily speech and different languages don't have the same ones e.g. it is raining cats and dogs - and pigs might fly - get your arse in gear - don't look at me like a chicken.

Idioms aside, as already mentioned there would be no frame of reference for nouns. Neither the sound of a word nor its component letters give a clue as to its meaning. The word banana means nothing unless you know what a banana looks like and know that the label "banana" is attached to it. Let alone more complex abstract nouns such as "altruism".

Thats exaclty what I mean by the referential problem. The only way to solve it would be for a quantum/adibiatic computer to process every single potential combination of meanings and senses and translate the most contextually likely version. It would probably be very error prone but it MIGHT be of some use.

Im thinking of alien languages however. Or at least languages that are a far remove from our own such as some of the pigmy languages

Sorry, can't resist!

^ Yes, but can it accurately translate the meaning of a surprisingly sophisticated attempt to do a double backward somersault through a hoop while whistling the "Star Spangled Banner"

I think a theoretical universal translator would likely work by having a standardized universal "language". This "language" would be more than just words - it would consist of objects, interactions, concepts, etc. For all other languages, a translator would translate between that language and the universal "language" and translate between the universal language and another language. Think of it as listening to something in one language, determining what is means, and then saying that meaning in another language.

A universal "translator" does exist, it's called google translate, or babelfish (not to be confused with the fictional babel fish), or any number of other systems.

What you're more appropriately looking for is a universal "interpreter" which reads/hears the language as input, understands the meaning and significance of that sentence/phrase in a larger context, and can interpret what that sentence/phrase actually *means* in an alternate language.

For instance, if you're from another planet and are referencing a local animal that is very tasty when grilled - the interpreter wouldn't necessarily have an exact translation for that animal, but it would be able to come up with a sentence like: "one of the local animals on the planet is very tasty when grilled" - substituting the more generic "one of the local animals" for the more specific name of the actual animal - but the overall meaning of the sentence doesn't change.

The significant first step to that problem is building a computer that understands natural language queries, then building such a system to understand natural language queries in multiple languages. IBM's Watson is the closest we currently have to a system that can do this.

From there all you have to do is build a system to cross-reference phrases in different languages based on meaning, and then output that phrase - which is likely technology we already have in the present day, although maybe not fast enough for use in a real-time application.

Keep in mind that different languages can be structured quite differently.

For example, under one simple classification of subjects, objects, and verbes, we have (from Wikipedia):

It would be an incredibly tough job to make a translator that just translates English to Navajo because of the many complexities of the Navajo language.

By the way, I would think that a universal translator would have to handle very badly composed texts as well as those that are 100% correct in every manner.

In many cases, to understand what someone actually means you have to consider not only what they said but who they are as well as in the contexts of the conversation itself. You could easily have two people express an idea with the exact same words but have two entirely different meanings.

Exactly. Not only tone of words but also facial expression and body language can completely change the meaning of a phrase or sentence.

It is not inconceivable that in the future a sufficiently powerful universal translator could translate languages (on Earth and also alien languages) provided it can not only hear the words but also see and understand the body language and facial expressions. It would also need to understand the context of the interaction. It would need a full database of all words, tones etc in its database.

What will never happen is where someone speaks a few sentences in an unknown language and the translator now magically knows how to use that language.