Tones

If you are a native speaker of any language at all in which pitch is not relevant to the meaning (this includes, among many others, all the Indo-European languages), just try correctly pronouncing a short sentence of Mandarin.

Try to hear the difference between 媽 mā, 麻 má, 馬 mǎ and 罵 mà [links from about.com], and then try to reproduce one rather than another, on purpose, in a sentence - that is, without it being distorted by the sounds around it. To me, these are all self-evidently the same sound, with minor variations of emphasis; it's not that they sound the same — they don't — but that they are the same. But to a native speaker of Mandarin (or presumably any tonal language with at least four tones) these four words are as self-evidently different from each other as the characters with which they're written.

In my language, pitch is meaningless, except in as far as it tells me something about how the speaker feels about what they are saying, or whereabouts we are in the sentence. It's not that pitch isn't there; it's just that it doesn't matter for deciding whether two sounds are the same word or not. Therefore, I don't listen for it, I don't know how to listen for it, and I only hear its emotional content. In Mandarin, it matters.

On top of that, I can't easily produce one of those sounds rather than another, at will. I might be able to do them in isolation, but if I attempt it with other sounds as well, I'll probably get the wrong one, and even if I hear myself recorded I may not be able to tell what I did wrong, and even if I can tell that something is wrong, I may not be able to correct it because I simply don't have precise control at that detailed level over the pitch of a sound that I produce in speaking. That is a physical skill, a skill which I have never learned. Obviously the pitch of my speech does vary quite a lot, and I may have some control over the overall pitch of a word or sentence, but within a syllable I don't think I can vary it deliberately at all.

I have quite a big problem to solve, close-enough, before I can reliably distinguish one word from another, and a bigger one to solve before I can utter the simplest sentence.

However, if I kept on trying through frustration and bafflement, I would probably get there in the end. Adult native English speakers do regularly learn to speak Sinitic languages correctly, when they have the motivation and appropriate teaching. People learn languages all the time, and have done so since the dawn of time. It's the kind of thing that's made very difficult at school, but isn't such a huge deal in practice when people have someone they want to talk to and a real, immediate, practical something to say.

The reason I say this here is that I think a lot of people who do have the relevant skills don't really believe or understand, and can't remember or imagine, that some people just have no idea at all how to go about starting to tell the difference between a milonga and a vals. They may sound ‘a bit different’, if compared immediately one after the other, even quite a lot different, as mā (‘mother’) and mǎ (‘horse’) sound ‘different’ to me — but what the difference consists of they may be wholly unable to say. They may certainly not be able to identify one or the other in isolation. They may look for the difference in completely the wrong place. They may construct, and use for many years, some rule-of-thumb that is apparently plausible, somewhat reliable, and totally erroneous. And the difference between milonga or vals and a tango may appear to be nothing more than speed.

I have frequently danced with people who have multiple years of tango experience and are totally unable to identify a vals. At least one of them simply danced it as milonga (a strange but rather interesting experience).

My analogy with language may not be valid, but I think it helps imagination. What I want to suggest is that it doesn't necessarily mean there's a defect of ear or brain. It's a matter of learning how to pay attention to the right things at the right level of detail; a matter of learning what part of the sound is important. At some point you have to do that, and not everyone has done it in advance.