So let's just say that one acquired a sample of an unknown tongue (let's just pretend it's Japanese,) and they wanted to dissect it word for word without knowing anything about it based off an interpretation given. How would one go about doing that (thinking about the Rosette Stone.)

Without looking anything up, not even knowing the sentence structure, if there are particle markers, reflexives, what, is there a general guide of how one would go abouts it?

There are few steps I can think of:

  1. Transliterate it into IPA to your best (or some other system you know you won't mistake,) and graph the phonological inventory

    1.1. look for phonemes that could give the language a particular climate or elevation -- i.e.: /q/ would imply most likely dry-arid climates like the Sahara, or a very clearly pronounced louder /k/ which would imply a higher elevation like a mountain.

    1.2. look for phonological inventories similar to particular regions such as a vowel heavy language may imply an Aſian language, and a tonal pattern of distinct tones noting differences may imply it being Aſian more, but very simple patterns (saying 2 maybe three) would imply something like a North Germanic language.

    1.3. confound the sounds or deconfound them particularly using Grimm's law to look for any others that look or sound similar or other such things.

  2. look for words that appear multiple times, and see if there appears to be a correlation with any words in the unknown tongue that may match.

    2.1. look for a particular word or syllable that shows up frequently as this may be a syntax marker of some sort, or article.

  3. Assume S.O.V. unless otherwise good reason not to assume that.
  4. If there is a video clip of the speaker, then pay attention to his body language, especially when they get to certain words.

    4.1. pay attention to particular tones of voice and ways the sounds are modified which may incite particular emotion -- i.e. if the speaker starts to get really low pitched, "turns-up-the-volume," and begins to growl a bit, then assume that the words or phrases should incite anger.

  5. pretend you are a native speaker, and based off of the phonological inventory, decide which type of writing system you might would have made for your language: i.e.:

    5.1. you have few vowels, and seldom do two words sound the same, and you distinguish and take pride in your consonants, you would probably develop an abjad, and therefore may be a semitic langauge.

    5.2 regardless of the number of vowels you have, you tend to use one (say a schwa) 70% of the time after consonants and may only need to mark other vowels for convenience, you might develop an abugita, and therefore might be indic (ignore Amharic.)

    5.3 you realize you know that sounds merge and words merge, and that beloved fricative that once distinguished "light" from "lite," can be lost in time, and so in an effort to slow that and because several words may be the same already, or you prefer literacy and phoneticity over ease of writing, you may choose to develop an alphabet.

    5.3.1 unless you're language is syllabic in which case you opt for a syllabary.

    5.4. given any case, you care about universality and may choose to make a logogram (this doesn't help at all, I just had to list it.)

  • Do you mean that you acquired a recorded sample, and want to know what language it is and then maybe write a grammar? What would be your goal in doing this?
    – user6726
    Jul 28, 2017 at 14:33
  • I acquired a recorded sample of the audio along with a translation. I don't know if it can be identified as a language, but I'm curious if there is just a general set of rules to follow in trying to decode it. I am looking into that guy who decoded Maya, but I am not sure if he did just the language, or the writing system, but I am going off of just the language. I don't want to identify it, but would like to if that is possible, but not right now. Jul 28, 2017 at 14:50
  • The gov't. has a research tool for identifying an unknown recording; once you know that it's e.g. Burushaski, then you contact the Burushaski experts. But you don't care if it's Burushaski, you just want to know "which word means 'egg', which word means 'chicken'?" and so on including basic grammatical info, given just a recording and a (presumed accurate) translation. Is that right?
    – user6726
    Jul 28, 2017 at 14:59
  • I do care, but I don't think it can identified -- it could be gibberish, it could be a conlang, I don't know. But yes: egg -egg, God- God, People - people. I've opted to assume it's real first, after transliterating it to i.p.a. to my best Jul 28, 2017 at 15:02
  • 1
    The procedure you describe is basically what students in Field Methods are taught to do with live informants. With only a recorded sample, less is possible, and interrogation is impossible; nevertheless, following the procedure carefully will yield results, though they may vary a lot from what literate speakers think they're saying. Faspichrulzer very prominent in all languages, and they're hard to untangle.
    – jlawler
    Jul 29, 2017 at 18:07


Your Answer

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