Here are the Orca translations, in case translators wish to reuse them:
orca.zip

NVDA seems to be able to read these at least when reading by character...

If they are not in NVDA yet, tese symbols need also to be added to the symbols and interpunctuation dialog. Actually there are a lot of symbols.
Adding new symbols to symbols.dic makes them translatable and defines the default value.
I don't know exactly if the symbols are added automatically to the dialog once they are added to symbols.dic, but I think this might need to be done manually unfortunately.

Just a question: is there some guideline or some common usage to decide which symbol is added in symbol.dic and which should not? Is there a big performance penalty to add many symbols?
I have raised theses questions in #11105 with some more explanations and example. Please let's answer and comment there.

Here are the Orca translations, in case translators wish to reuse them:
orca.zip

May I ask why the engb (english) one is so small/short? There are only a few entries in it. The russian one is also not complete. I want those two (at least the english one) for reference while discussing them with my colleagues for the bulgarian translation. Thanks in advance!

May I ask why the engb (english) one is so small/short?

It's just because my script ignores the strings identical with the "en" version

@josephsl wrote:

Hi, punctuation/symbol pronunciation dialog will pick up symbols from language-specific symbols.dic, which ultimately depends on willingness from translators to translate new symbols once they show up on translations workflow. Thanks.

But in the punctuation and symbol pronounciation dialog, there are more than 3.000 symbols. Where are they fetched from? In the symbols.dic are only about 200 lines. If they are fetched from Windows itself, I think most of them should be already translated (i.e. emojis). What is exactly the purpose of the symbols.dic then?

I am copy/pasting my comment from #11105 since it seems that the discussion is taking place here.

Just a question: is there some guideline or some common usage to decide which symbol is added in symbol.dic and which should not? Is there a big performance penalty to add many symbols?
I raise these questions because anyone may want to add more symbol, e.g. see #11015.

IMO when one asks new symbol inclusion, he/she should give a justification. For example:

* ¼ (one fourth): seems already supported by synthesizers. So give a synthesizer example where this symbol is badly or not announced

* all subscript and superscript: give an example of real life where they are used
  Maybe also the level should be justified.

@feerrenrut could you comment on this please regarding NVAccess point of view? Comments from other people are also welcome.

I'm not aware of performance issues with this. I would prefer to address performance problems with the implementation rather than reduce our supported symbols. Even if most synths supply these symbols they may not do so adequately across all languages, or we/a user might not like how they announce the symbol. This gives users the option to use the NVDA supplied symbol name or the rely on the synth. It also lets us adjust symbol names as appropriate, a good example is the Mac command symbol / looped square ⌘ which I believe was added not so long ago.

I hope this answers your question!

Yes, thanks for your answer.

Is there a big performance penalty to add many symbols?

According my tests, the number of symbols is not significant. I tried with a string of 1048543 symbols in several locales. However, I don't know if my script is really reliable. There are most likely better ways to measure this.

import globalVars, sys
path = globalVars.appArgs.configPath + "/lib"
if not path in sys.path: sys.path.append(path)

import timeit
import characterProcessing

text = ''.join([chr(c) for c in range (0x20, 0xfffff)])

def test(lang):
	res = characterProcessing.processSpeechSymbols(lang, text, characterProcessing.SYMLVL_SOME)

number = 20
langs = ["ar", "bg", "cs", "da", "fr", "hr", "it", "ja", "ko", "nl", "pl", "ru", "tr", "zh", "zh_TW"]
for lang in langs:
	try:
		res = timeit.timeit(stmt=f"test('{lang}')", number=number, globals=globals())
		b, u = characterProcessing._getSpeechSymbolsForLocale(lang)
		symbolsNumber = len(b.symbols) + len(u.symbols)
		print(f"{lang} -> {symbolsNumber} symbols, total time: {res}, average time: {res/number}")
	except BaseException as err:
		print(f"error with {lang} locale: {err}")

Here's one of my outputs:

ar -> 3376 symbols, total time: 221.0103557, average time: 11.050517785
bg -> 3445 symbols, total time: 214.06203370000003, average time: 10.703101685000002
cs -> 3396 symbols, total time: 218.30669819999991, average time: 10.915334909999995
da -> 3445 symbols, total time: 217.11858540000003, average time: 10.85592927
fr -> 3845 symbols, total time: 238.7677773999999, average time: 11.938388869999994
hr -> 3446 symbols, total time: 217.14729369999986, average time: 10.857364684999993
it -> 3446 symbols, total time: 215.80777239999998, average time: 10.790388619999998
ja -> 3968 symbols, total time: 230.4640202999999, average time: 11.523201014999994
ko -> 31709 symbols, total time: 213.701642, average time: 10.685082099999999
nl -> 3441 symbols, total time: 217.8810311999996, average time: 10.89405155999998
pl -> 3448 symbols, total time: 213.60206610000023, average time: 10.680103305000012
ru -> 3493 symbols, total time: 212.6139294999998, average time: 10.63069647499999
tr -> 3672 symbols, total time: 214.91899090000015, average time: 10.745949545000007
zh -> 152 symbols, total time: 214.42888690000018, average time: 10.721444345000009
zh_TW -> 4246 symbols, total time: 211.3798428, average time: 10.56899214

all subscript and superscript: give an example of real life where they are used

In this case, we can remove many symbols :)

@Adriani90 wrote:

Such symbols are for example often used when writing equations in LateX. Nearly every blind person who uses a laptop or a computer in the school or university works with LateX in Mathematics, Phisics or Chemistry.

The subscripts symbols are not used in LaTeX as such. When writing in LaTeX the particular notation for them is used, and LaTeX generated pdf's containing math are not accessible enough to be readable.

• ¼ (one fourth): seems already supported by synthesizers. So give a synthesizer example where this symbol is badly or not announced

With some Polish SAPI 5 synthesizers for these fractions only
numerator is being read, so having them in NVDA would be beneficial.

@sthibaul: If you have additional time,, feel free to use my uploaded files regarding issue #6341 as well.

I'd like to accept this PR, though I am just waiting on the following line comment to be addressed:

I believe "one quarter" and "three quarters" is much more popular than the also acceptable "fourth" version. I propose this be changed.
https://english.stackexchange.com/questions/103188/three-quarters-vs-three-fourths

I'd like to accept this PR, though I am just waiting on the following line comment to be addressed:

Ok, fixed so.

@LeonarddeR: Wouldn't it better to add a secondary name field for Unicode characters, which are only spoken in letter navigation? I already planned to create such an issue. This would save us tons of hours of discussions, as then a Unicode character is spoken with its short name in normal navigation and only with its detail name in letter navigation.