4 ms·
This seems like a nice idea, but many of the comments here are pointing to omissions that the people commenting feel could be serious. I guess I see six kinds
by schoen 11y ago
This seems like a nice idea, but many of the comments here are pointing to omissions that the people commenting feel could be serious.
I guess I see six kinds of potential difficulties:
① There are so many languages out there, including languages will millions of speakers who might eventually come across your thing. Maybe that's not an issue for tangible products that will be marketed to specific territories.
② There are so many slang terms out there; each individual language might well have thousands of terms that have a rude, sexual, or excretory meaning, or that are used as a slur against some group. Also, some languages have expletives that don't correspond to expletives in other languages. https://en.wikipedia.org/wiki/Quebec_French_profanity https://en.wikipedia.org/wiki/Quebec_French_profanity
③ People have pointed out that phonetic matching is hard when you're dealing with different languages' orthographies and phonologies, and you can have the problem of "the source language's intended pronunciation sounds like an offensive word in the other language", "the other language's likely pronunciation of the written term sounds like an offensive word in the other language", or even other combinations among highly multilingual populations. "Sounds like" is sometimes challenging to automate in software, for example because epenthesis of a vowel may not be enough to remove the association. (But I think Levenshtein edit distance between phonemic transcriptions can kind of sort of work.)
Also, the "MR2" example someone gave shows that understanding how people will pronounce something in different languages is complicated: you have to know that the number two in French is "deux" /dø/.
④ People might also perceive something as an offensive reference that isn't even familiar to people elsewhere, like a reference to an upsetting person, place, or event. Reportedly some people in India have named people and businesses after Adolf Hitler just because he was famous, for example. I bet it's easy to do this cross-culturally in general.
⑤ As people point out with the Chevy Nova story, there might be a reason why a product name would become the target of ridicule in a particular language even if it's not offensive. That's true even if it didn't literally happen to the Chevy Nova.
⑥ It might even turn out that the space of offensive references is so dense that there is nothing that isn't a near-homophone of something pretty offensive in some language.
Anyway, I think this project is really neat; I'm just reminded from people's comments that natural language is hard! There's scope to keep expanding this site, and I think there are also existing "cultural consultant" businesses that try to deal with these problems through human review (I wonder how many of them have consultants on contract from many widely varying cultures, which seems especially useful in the Internet era).
- schoen 11y agoA funny page that might be support for point ⑥ is https://web.archive.org/web/20071112172929/http://members.aol.com/yahyam/coincidence.html https://web.archive.org/web/20071112172929/http://members.ao... If you can take an arbitrary word and find a coincidental homophonous synonym (or antonym) in some other language, you can probably find a coincidental homophonous expletive or sexual reference in some other language too! (These coincidences are not cognates -- that's part of what "coincidence" means here.)