I'll often scout the ILoveLanguages! channel on YouTube for inspiration for what aesthetic to use, because its videos typically display a relatively long sample with both spoken audio and a romanized transcription; this is useful for assessing whether an aesthetic I think looks cool on paper actually sounds as cool in practice.

But I have multiple problems with it as well. It very often seems not to have samples of many of the Caucasian, Ethiopian, Berber and Pacific Northwest languages I'm interested in - which is not helped by the fact that it seems like most of the older videos were purged a while ago? (Some languages from these areas, like Haida, Tlingit, Ossetian and Kabardian, seem to have had a video some time in the past, got deleted, but then got re-uploaded somewhere or by someone else?) The background music is distracting. Their relatively recent shift to summarizing culture as well as the languages, in the same video, means something like 2/3 of the content in any video is now irrelevant for my purposes. Smooshing multiple languages into one video (e.g. Chechen/Ingush, instead of just a seperate Chechen video and Ingush video) cuts up one perfectly good long-form sample into a bunch of unhelpfully short samples, and interlacing the different languages makes it unnecessarily hard to extract the desired information about just one of them.

I'm aware of a couple other sources of spoken language samples in a standardized format. Omniglot sometimes has spoken samples of the UDHR article 1 along with its transcription, but it's not consistently present, the transcription is not always romanized, and UDHR1 is just in general too short a sample. Wikitongues is good for having a large volume of long spoken samples that include some of the more obscure languages, but they have no transcription at all, romanized or otherwise, so it's rarely clear how to choose my phonology/phonotactics/shapes of my morpheme to be able to replicate the sound and feel.

Is anyone familiar with a better collection of the sort of long-form spoken samples with romanization that I can use for scouting interesting language aesthetics?

  • This is a very good question! Looking forward for answers ...
    – Sir Cornflakes
    Commented Mar 30, 2023 at 10:12

1 Answer 1


Romanization is the act of converting text from a different writing system to Latin script using a consistent method. As such, technically, if you have either a sound or text sample and a certain system designed for said language, everything can be transcribed (if it's spoken) or transliterated (if it's written). There will always be some trade offs and imperfections because the target language usually contains systems of distinctions that is incommensurable with the script of the reader's language. That's why transcription with deeper level such as IPA exists—limitations of the limited letters of Latin script greatly expanded with the use of any typographic glyphs (diacritics, various combinings, or any method to create new graphemes to represent some degree of detail) available and constructed in some way to explain the multitude of phones.

To answer the question, there are no list of such sites I know of (yet), although it'd be nice to know one myself. But if you don't mind doing a bit of work, there are standards organizations that govern most of the languages, and some documentations of them are available on the net. There are ISO, ALA-LC, PCGN/BGN, Yale's systems, and many regional systems available. Such documentations are useful to proceed to do any romanization with whatever source you have.

I hoard articles and a handful of grammar documentation of world languages, and my personal method is to read grammar books and study how they treat all the distinctive phones has proven useful. A bit much for a safari, but it gets the immersion right.

Not a wow answer, I know. Feel free to improve this, though.

Your Answer

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

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