1

Let's say aliens (someone completely new to the language) want to talk with users of it. They've obtained a complete dictionary, and a large selection of natural text (for this hypothetical situation, let's say the full Wikipedia database dump of that language) and unlimited computing power.

Assuming they successfully obtain these two pieces of data in textual form, would it be possible with this data to reconstruct the language?

For the sake of this situation, let's say the data is in textual form not in binary (they can see the exact content of it as we would read it) and let's assume they have obtained the correct information (someone who speaks the language gave it to them).

I'm looking at this from the point of view of aliens; for a story of mine. Could they possibly learn our languages? How?

Improve this question
2
  • 2
    In this case, I think it actually matters a great deal if it is aliens or other humans. Humans tend to be similar to one another; we can assume certain things about another human language and culture that we don't know are true of aliens. And in cracking an unknown language, knowing what assumptions you can make is vital. You say this is in textual form: does that mean actual documents? Documents have much more information than just plain text: illustrations, page numbers, possible different treatment of names (c.f. capitalization in English, cartouches in Egyptian) that can provide footholds.
    – brass tacks
    Apr 15, 2015 at 6:29
  • In the case of a human with a reasonably similar civilization and the same linguistic technology as us, but one that differs in the particulars and speaks an unrelated language (perhaps from an alternate reality, in the post-apocalyptic future, or in a hidden place in the world) I would say a full dump of the English Wikipedia, text, pictures, titles and all, ought to be plenty to reconstruct the language, no dictionary needed. It is an encyclopedia after all!
    – brass tacks
    Apr 15, 2015 at 6:32

2 Answers 2

Reset to default
1

You're asking three completely different questions between your title and the actual question.

  1. The title is asking about reviving - this would mean making a dead language a live spoken language again. To that the answer would be a qualified yes. Look for instance at modern Hebrew. But that still requires some form of speech to be preserved.

  2. The body is asking about:

a. Learning a language from texts and a dictionary. In that case, yes, it is possible for humans to learn a language in that way (at least the written part of it).

b. Learning a language for an alien from texts and a dictionary. The problem here is that a dictionary would not be of any use aliens unless it was a bilingual dictionary between their language and the human language. If it's just the dictionary of (say) English, then that's simply another text in English.

It would be possible to make sense of the texts for the aliens only if they could observe the context in which it is used. Simply seeing lots of texts is not enough without some external reference - as attested by the many undeciphered languages. This would be doubly hard since the Aliens would presumably not have the same script or maybe not even the same way of writing (phonetic vs syllabic vs ideographic).

Improve this answer
4
  • the misleading title is due to a later edit [approved] by @curiousdannii, which I suggested he change, but which has not been reverted as of yet. The original question was about reconstruction.
    – brass tacks
    Jun 15, 2015 at 3:54
  • @sumelic What do you mean by reconstruction? The question is more about deciphering. Reconstruction implies making something complete from fragments but your scenario assumes a large body of text. Anyway, my answer still stands. With no context, you cannot decipher a body of text no matter how large. You can find regularities and patterns in it but you need at least some solid points of reference to get started.
    – Dominik Lukes
    Jun 15, 2015 at 4:20
  • Well, I'm not the author of the question, but I assume it means that the aliens would be able to decipher the texts and compile information about the lexicon and grammar of English.
    – brass tacks
    Jun 15, 2015 at 16:19
  • Grammar, yes. That can be determined from patterns alone. Lexicon, yes in the sense of a word list, like the OED, but no in the sense of a bilingual dictionary in their language -- who would translate it if not them? Bilingual dictionaries are the Output, not the Input, in cold translation.
    – jlawler
    Jul 16, 2015 at 14:20
0

It would depend on two fundamental issues, the cognitive system of the aliens and the nature of the language database. The main issue is whether they could figure out the referent of "apple" -- i.e. will they ever be able to figure out an association between the letter string and a real thing. That's where photographs with labels are invaluable. The problem of alien cognition is that their perceptual apparatus may be wildly different from ours, and they may not grasp the concept of "multiple instances of the same type". Sci-fi authors typically don't veer too far on the weird side, and if your aliens are typical meat aliens that interact with the physical universe in a manner similar to us, then you can assume that it's possible. For example, can they see? do they reside in an alternative physical universe (such as the extra-dimensional aliens who created the Outsiders (Known Space)).

Using a words-only corpus, you could not understand the meaning of "The dog chased the boy" vs. "The dog ate the biscuit". You would probably be able to discern some amount of word-relatedness such as "apple" vs. "apples", but probably not "think" vs. "thought" unless you mistakenly make "apple" and "apply" also be related. Without knowing what utterances refer to, you wouldn't "know" the language. This was the one valid point in what is otherwise the stupidest instance of sci-fi language-issues, in the Star Trek "Darmok" episode.

Improve this answer
5
  • We can't read Linear A or the Harappa inscriptions. For instance. And they're human, with human languages and human habits and perceptions. Why would aliens be able to distinguish that "apple" has a referent, let alone what it is? A "words-only corpus" would have to be audial, since there is no other single consistent visual (i.e, "written") way to represent text. ASCII? docx? pdf? etc? No, alien archaeologists would never understand Facebook, even if they found it intact, unless they were human aliens. And even then it would be a labor of centuries.
    – john lawler in exile
    Apr 15, 2015 at 19:27
  • As I said, you first have to figure out the referents, and if you can't do that, you're hosed. Figuring out referents presupposes that you figure that there is a referent. And I'm just following the OPs assumption about the representation of the text. The Wikipedia dump would have to be a uniform font; there is also no consistent acoustic representation of "apple". This is why I object to how sci-fi usually mistreats the problem of non-interactive translation.
    – user6726
    Apr 15, 2015 at 19:52
  • Fonts are just extra files to be used by software. But where is the software? And the fact that there is neither a consistent visual nor acoustic representation of "apple" suggests that human language is not consistently representable, which means those poor alien archaeologists will have to deal with totally inconsistent and largely incommensurable data. Like getting every chunk of data in a different language and format.
    – john lawler in exile
    Apr 15, 2015 at 22:08
  • So Wikipedia could provide the relevant information here if in the right form, just not in the text alone? Together with its images, assuming they had it in textual and visual form, there'd be enough information to begin extrapolating a language?
    – John Cave
    Apr 16, 2015 at 8:57
  • Actually, I doubt that there is currently enough information beyond concrete nouns and adjectives. If there were a Dick and Jane version, then maybe.
    – user6726
    Apr 16, 2015 at 15:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.