Following on from my comment in this question, I think it would be good to develop some guidelines around what orthographies to use in asking questions. I propose that we should allow linguistic material to be presented in its own orthography, whatever that may be, but it seems to me it should always (as far as possible) also be presented in IPA. Given that this is an English language website, English material could be exempt unless an IPA rendering is necessary to dealing with the question.
IPA should be considered essential if the question is related to phonology in some way, but if it's about morphosyntax, semantics, NLP etc I see no reason to require or even encourage it. Whichever orthography would be normally used for the language can be used. That doesn't mean that you can't ask for it if you think it would help, but we don't need a rule requiring it for all questions.
Here are my recommendations based on standard practice in linguistics publications:
- Questions involving phonetics or phonology should ideally include IPA, with the following caveats:
- Users asking questions come from a variety of backgrounds, so it may not be possible or necessary for them to provide an IPA transcription, depending on the nature of the question.
- Some languages have orthographies with a straightforward mapping from phonemes to graphemes. Even in the phonological literature, these orthographies are sometimes used, with an accompanying note explaining where symbols depart from IPA usage. I think this would be fine in phonological questions (and others) here too.
- For questions not pertaining to phonetics/phonology, a Roman-based orthography is fine. If there are any potentially confusing symbols (especially in less familiar languages), these could be explained with their corresponding IPA symbol or phonetic description.
- Non-Roman orthographies are welcome, but they should always be accompanied by at least a transliteration into a Roman-based orthography, if not an IPA transcription.
I would like this conversation to go somewhere, so I hope it is okay if I answer even though my answer is basically identical to Jason Zentz's one. Every question asked I've seen so far has used Roman script (and English language) for the asking of the question. There really is no reason not to add a (however vaguely accurate) Roman or Roman-like transcription, except to show off how many scripts we bright linguists can read. So could we forget about IPA (too complicated, not everyone knows it and it's pretty useless for languages we don't even really know how to pronounce anymore) but kindly ask to use romanizations?
I think we don't need any guidelines about orthographies, and instead we should encourage people to make reasonable assumptions about what readers might know. If a reader does not know how to interpret [šιft] or [ʃɪft], then they will need to invest a bit of effort into figuring that out, if they care. The same would go for any formalized logical expression or syntactic representation. A question like "Is 在 a copula" is kind of disfunctional, from the standpoint of communicativeness, since only people who read Chinese (or know how to Google a Chinese character) can pronounce "在". But only people who know Chinese can answer the question in the first place. While it is nice to assume a universal audience, there are many questions that could be understood by only a small percentage of users here, and just as I tune out the NLP questions, I expect that they will tune out the Thai / Arabic / Russian questions, without being offended that they weren't "included".
That said, I think if the topic is English pronunciations, there should be a phonetic transcription, since almost nobody here speaks my dialect, and I get confused when people say things such as "like the vowel in 'super' or German 'über'" (apparently for some people those are the same vowel).