I saw that SAMPA was created to be machine-readable. Does that mean that IPA isn't?

If it isn't, why is that so?

EDIT: By machine-readable, I meant that it could be directly interpreted by a parser/algorithm. More precisely, I wanted to know why I could not find an IPA text-to-speech web app (that would work for all languages), but there seemed to be some for SAMPA.

  • 1
    The other aspect that adds to readability is writeability. Although it is nowadays possible to type IPA characters thanks to unicode, this still can be cumbersome depending on what input tools you have available - there is specially designed keyboard software like IPA typeit and there is LaTeX and various other helpers, but whatever you use is still more complicated than just typing ASCII characters directly from your keyboard. For many purposes it's just easier to use SAMPA even though it would theoretically be possible to typeset IPA characters using unicode. Commented Jan 24, 2017 at 19:23
  • 2
    Your usage of "machine-readable" is inapt. Anything can be made machine-readable by choosing an encoding in bits. The real question, I suppose, is whether IPA had a widely-supported standardized encoding (which I suppose it didn't, so SAMPA was created to fill the gap). Since SAMPA is largely based on IPA, you could say that SAMPA made IPA machine-readable. Commented Jan 25, 2017 at 2:43

2 Answers 2


IPA is machine readable now, because the IPA characters are all in Unicode, the standard character set of today.

At the time when SAMPA was created, character sets were either 7 bit (ASCII) or 8 bit (Latin-1, Latin-2 and the like) or specific to East Asian Scripts. Those character sets didn't incorporate IPA, and in this sense IPA wasn't machine readable at that time.

  • Thanks for your answer! Do you know when it became machine-readable?
    – apat
    Commented Jan 24, 2017 at 14:37
  • 5
    @apat: As far as I remember, IPA was incorporated into Unicode from the very beginning. So it became machine readable with the wider adoption of Unicode as standard for character encoding (late 1990ies/early 2000s) Commented Jan 24, 2017 at 15:13

I will assume that "machine readable" means that you want a computer to be able to scan/photograph the text and read it with a high degree of accuracy and furthermore the information should be able to be passed around in text form between a wide variety of computer systems with minimal risk of mangling.

SAMPA limits itself to ASCII. Pretty much all computer systems can process ascii characters unmolested.

If you want to print and scan it reliably there are specific fonts designed for the purpose of printing and scanning ASCII characters though there can be some compatibility issues on some of the weirder symbols.

At the time SAMPA was created computer text handling was still pretty limited. Typically computers handled ASCII plus a set of up to 128 local characters and/or symbols. Support for larger character sets did exist but it was mostly limited to CJK languages. Passing IPA around in this environment would have been very difficult.

In the modern world it is reasonable to pass IPA text in unicode around on computers but it still carries a higher risk of mangling than ASCII text does.

And as for printing and scanning IPA has a large character set where some characters are scaled or inverted versions of others and on top of that makes heavy use of diacritics. I would expect this to make reliable OCR very difficult.

  • 3
    It is very hard for OCR software to identify diacritics correctly. This adds to the unreliablity of using OCR to read IPA. Commented Jan 24, 2017 at 17:55
  • 1
    That's not what I meant by machine-readable. I edited my question.
    – apat
    Commented Jan 25, 2017 at 10:26

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.