The importance and unimportance of accent

Accent is essentially unimportant.  It's the final coat of paint that makes our language pretty or ugly, shiny or dull.  It is something that the beginning learner really doesn't need to think about.

Unfortunately, this is something that is frequently overinterpreted, because many people don't appreciate the fact that accent is only one part of pronunciation.

Every language has it's own phonology -- it has a set of possible sounds and possible combinations of sounds, and it has a set of distinctions between sounds.  Though we do not need to learn a good accent from day one, we certainly need to learn the sound system of a language.

The most common consequence of conflating the sound system with accent is the idea that the "closest sound" from your native language is "good enough".  Open up almost any beginner's book and it'll start with a list of sounds described along the lines of "like the a in cat", "like t in English".  But this is rarely true.

Still, some languages will let you get away with this to some extent, but when you hit a more complicated language, it all crumbles.

Any book on Hindi will tell you that "closest sounds" just won't cut it, and that with that approach you will never be understood.  This is because Hindi has more sounds than most other languages.  In fact, there are 8 sounds that are approximately similar to T and D, so using an English T and D, or a French one, or a German one, would leave you completely unable to distinguish certain words, and unable to make yourself understood.

Worse, because you treat 4 different sounds as one, you will never learn to hear the difference either -- your brain only distinguishes sounds that mean something.  The later you attempt to fix it, the harder it will be, because you will have to relearn all the vocabulary in order to learn the difference.

I've experienced this personally with Spanish.  In Spanish C is pronounced like Z, when the C is followed by I or E.  In some areas, these in turn sound like S, but in other areas, they don't.  I started learning Spanish from a course that didn't make a distinction between S and Z, but as I progressed I spent more of my time with people who make the distinction than those who don't.

As a result, I started trying to speak like them.  However, my brain was trained to see the two things as one, so I was prone to making mistakes such as pronouncing the word "especial" as "ezpesial".  My errors were arbitrary, but not random -- they were consistent and there was a clear pattern.  My brain was still seeing the two as equivalent, but was trying to explain it in terms of the other sounds in the word.   Thankfully this was still pretty early, so I caught it and fixed it.

Other people are not so lucky.  A local French teacher (from France) can pronounce all the sounds of English.  But he couldn't before he came here.  The result is that he has already learned all the words with the wrong sounds.  The classic example is TH -- it's always T or D when he speaks.  He's learnt the words now, so there's no going back.

Now, that's something that we call "falling together" (ah, a nice, self-descriptive term for once), but phonemes can also split apart.

Consider that Japanese doesn't make a distinction between liquids L and R.  As an English speaker learning Japanese, I would likely hear these as different phonemes (meaningful units of sounds) instead of simply different ways of pronouncing the same phoneme ("allophones").

This splitting apart on its own isn't a big problem -- it doesn't lose information the same way falling together does.  However, the two can very easily co-occur,  and at that point they make a bad situation worse.

Take for example "CH" in German.  It has two allophones -- a hard one (ach) similar to the sound in Scottish "loch", and a softer palatised one (ich) that takes on a quality similar to the English SH.  But there's another phoneme that sounds even more like the English SH, one that's usually written SCH.  And it gets worse, because in certain combinations of consonants, S starts to sound the same.
If we're not careful, the learner may end up splitting their Ss and their CHs, and putting half of each in the same box as SCH.  The result is a map of the sound system which looks nothing like the native speakers view of things.

Accent is what we put on top of the sound system to give it colour and personality.  You cannot develop a good accent based on an incorrect map of the sound system.  Pronunciation has to be taught from the start in such a way as to encourage a consistent and correct sound system.

Accent can wait until later, but pronunciation must taught in some form right from the start.