I understand why Japanese have trouble telling apart /r/ and /l/: their Discrete Fourier Transform looks almost the same (the only difference being that /r/ has some element at around 2500 Hz that /l/ doesn't). But how it is that Koreans have trouble telling apart /p/ and /f/? Their Discrete Fourier Transforms look nothing alike. /f/ looks almost like random noise. If anything, /f/ is similar to /s/, rather than to /p/, right?

  • 10
    [f] isn't a native sound in Korean, so Korean listeners will naturally map it to the closest sound that is part of their phonology, likely an aspirated [pʰ]. Also Koreans aren't spectrographs, they're humans.
    – curiousdannii
    Commented May 10 at 8:40
  • 1
    This is a good example of why not to directly relate external measurements to internal/hidden phenomena.
    – kg5425
    Commented May 22 at 13:10

2 Answers 2


People don't hear Fourier transforms.

Their brain recognises the sounds as speech and then tries to split the sounds up into segments they recognise at a subconscious level. The key factor here is that their brain has spent their entire life being trained to recognise the sounds of their native language(s).

As such, when encountering sounds not found in their native language(s), their brain incorrectly recognises it as being a slightly-odd version of a sound that is in their native language(s) rather than a sound that isn't.

You see a similar effect in OCR tools. If you feed some German text to an OCR model trained only on English, it will never correctly recognise an <ß> as ß because it has no concept of ß. Instead it will see it as whichever letter of the English alphabet it perceives as closest (likely B).

Likewise, when a monolingual Korean-speaker hears [f], their brain is trying to identify the phoneme it represents, but only knowing the phonemes of Korean cannot identify it as /f/ because their brain knows no such phoneme. Instead they will hear it as /pʰ/ but slightly odd, or mispronounced, or with an accent or similar.

With practice of course, they can train their brain to identify phonemes not found in their native language, and this is an important part of achieving proficiency in a second language.

  • 7
    There is also the question of whether Koreans actually do have trouble distinguishing [f] from [p]. Often people are able to detect the difference between two sounds but still find it difficult to reproduce the difference themselves. Labial and bilabial fricatives are pretty universally employed by humans, including Koreans, in various contexts – the difficulty may lie in treating it as a speech sound and producing it rapidly enough to form part of flowing speech. Similar to how English speakers can produce clicks, but rarely with the speed and accuracy of speakers of San languages. Commented May 10 at 14:01

I think Koreans are not confusing it with [p] but rather with [pʰ]. I used to wonder about this myself until I once misheard one for the other which made me realize that [pʰ] and [f] are more similar-sounding than I previously thought.

An experiment that helps to realize it somewhat: compare [f] and [ɸ] and you may be able to hear that these are similar. Then compare [pʰ] and [p͡ɸ] and you may be able to hear their similarity (actually they are allophones in English). It may then be easier to understand that if someone does not clearly hear the initial plosive release, [pʰ] and [ɸ] could be mistaken for each other, and [ɸ] in turn is confuseable with [f].

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.