Like everyone above me has said, it’s possible that you simply have high expectations, and that you are not in fact pronouncing words as poorly as you think. It’s also true that WaniKani wasn’t created as a pronunciation practice platform, even if I think it’s a good idea to pick up pronunciation while studying new words. However, assuming there is a problem, here’s what I suggest to tackle it – create clearer frameworks for your practice and pronunciation:
- Learn standard pronunciation for all sounds (two methods, not mutually exclusive):
- Method 1 (skip this if IPA gives you a headache or if you’d rather not bother with it): Check IPA symbols for Japanese sounds. Study the explanations given for how to make those sounds if you are able to understand the technical language used. If not, at the very least, compare the symbols used for Japanese with those used for English (you can usually do this by comparing ‘IPA: Japanese’ and ‘IPA: English’ or articles to that effect on Wikipedia), and take note of those symbols that are different so that you know which sounds to work on.
- Method 2: Look for Japanese pronunciation videos. If at all possible, you should look for videos and resources that offer diagrams of what the inside of your mouth should look like when you make each sound. The two main resources I would recommend are Wasabi Japan’s videos and pronunciation pages, and pronunciation videos by Japanese Pod 101. I’m sure there are other excellent videos, available, but these are the two main sources I’ve used. Campanas de Japanese (which is in English) also seems very good.
With either method, what you should aim to acquire is a good sense of the correct tongue, mouth and jaw position for each sound, along with the right movements to make it. You can then use this knowledge to create reference points for each sound so you can check yourself rapidly during practice. I think the main problem sounds for English speakers are the う vowel sound, some of the dental sounds (mainly し、じ、ち), the H row (especially ふ – it’s not an English F), the Japanese R (shared with Spanish, if I’m not wrong), and the Japanese W, which is basically a version of う. The えand お sounds might also require perfecting, but they’re not really sounds English doesn’t have. (Depends on the dialect of English you speak, of course.)
- Learn the framework for pitch accent so you can spot what to listen out for:
I think watching Dogen’s introductory video should be enough, plus maybe a few other videos covering pitch accent (e.g. by Japanese Ammo with Misa, or certain videos covering common patterns). Here’s a post in which I compiled a few videos on pitch accents. There are two major things to look out for:
- When does the pitch drop? (i.e. Which is the last high syllable/mora? The accent is said to be ‘on’ that syllable.)
- How does pitch change when a particle is added? (i.e. Is the particle at the same pitch as the last syllable of the word?)
This should be enough for you to start categorising words (very roughly) in your head based on when the pitch drops. Basically, the general rule is that pitch rises starting from the first mora (quick note: きょう is one syllable, but two morae (kyo-u); one mora = one beat/clap), meaning the first mora is low and the second is high. It then stays high until the drop, and stays low for the rest of the word. Pitch drops only once per word, no more. Only exceptions to what I just said:
- The atamadaka (頭高-head high) pattern, where the accent is on the first syllable, meaning that syllable is high, and the rest are low.
- The heiban (平板-flat board) pattern, where the accent is on syllable 0 (i.e. unaccented word). The pitch starts low, rises and stays high all the way. The particle that connects to this word will be at the same pitch as the rest of the word.
When words combine to form compound words, the general rule is that they become へ-shaped: the pitch rises, stays high, and then falls. This isn’t always true, but it’s true for most compound words, regardless of what the original pitch accents were. Example: 記念– pattern ; 写真 – pattern ; 記念写真 – pattern  (if I remember correctly) Either way, the point is to know that there will probably be a pitch drop and to look out for it.
If I’m not wrong, another general rule is that pitch variations tend to flatten out when we speak fast, especially over the course of a long phrase. In particular, for 平板 words, there’s no need to ‘reset’ your pitch to ‘low’ in order to pronounce one: just start at whatever pitch you’re already at and pronounce it. If your pitch is already ‘high’, the 平板 word’s pitch pattern will typically become completely flat. If your pitch was ‘low’, there will be slightly more variation, but the 平板 word will still be basically flat.
In any case, I’m no pitch accent expert, so I think shadowing (i.e. careful listening and imitation) is the best approach, but just being aware of these patterns should help you look out for them and remember the pronunciations you learn better.