>>156122
Essentially, each language has a set of rules that determines what sounds are permitted, in what positions, and in combination with that. These are your "constraints."
First, let's look at how the words are composed in Japanese:
怖い: /ko.wa.i/
可愛い: /ka.wa.i.i/
The sound /aii/ isn't natural for English, so it's immediately simplified to just /ai/, this is an extremely normal ocurrence. Now, both words end with /-wai/. But they still differ in the first vowel, right?
Let's assume, now, that the English speaker is placing his stress on the /wai/, because it's a long sound, while the first syllable is short. It's not something that is guaranteed to happen, but it's very normal as well.
Now, consider German, and a word like, say, "Letten." You may assume, by looking at the two <e>, that they both make the same sound. You'd be wrong. The first syllable, which is accented, makes the sound /ɛ/ (the standard e-sound of German), while the second syllable (which is unaccented) either sounds like /ə/ (a schwa) or disappears entirely, and the word is then pronounced /ˈlɛ.tn̩/ with a syllabic N (the ' indicates where the accent falls). It's reduced in volume and moved towards the center of the mouth, or deleted altogether, and this happens not only with all unaccented /e/ in German but with many other kinds of sounds across other Germanic languages, like English. Thus you see a Latin loan like "commence", clearly written with an <o>, pronounced as /kəˈmɛns/, collect as /kəˈlɛkt/, collude as /kəˈluːd/, along as /əˈlɔŋ/, assist as /əˈsɪst/, you get the idea by now: it's extremely common for the initial unaccented syllable in English to just be a schwa, and this can happen regardless of what letter it's written. This makes it infamous.
If we now return to the Japanese words, we may expect them thus to be pronounced like this by an untrained English speaker:
怖い: /kə'waɪ/
可愛い: /kə'waɪ/
And that's how they end up being homophonous.
>>156124
You don't need to watch the full thing, don't worry about that. Look at the basics, and whatever worries you at the moment. The individual videos are short, and you can jump into the videos for consonants or vowels without any need to have watched the ones about pitch.
Now, consider German, and a word like, say, "Letten." You may assume, by looking at the two <e>, that they both make the same sound. You'd be wrong. The first syllable, which is accented, makes the sound /ɛ/ (the standard e-sound of German), while the second syllable (which is unaccented) either sounds like /ə/ (a schwa) or disappears entirely, and the word is then pronounced /ˈlɛ.tn̩/ with a syllabic N (the ' indicates where the accent falls). It's reduced in volume and moved towards the center of the mouth, or deleted altogether, and this happens not only with all unaccented /e/ in German but with many other kinds of sounds across other Germanic languages, like English. Thus you see a Latin loan like "commence", clearly written with an <o>, pronounced as /kəˈmɛns/, collect as /kəˈlɛkt/, collude as /kəˈluːd/, along as /əˈlɔŋ/, assist as /əˈsɪst/, you get the idea by now: it's extremely common for the initial unaccented syllable in English to just be a schwa, and this can happen regardless of what letter it's written. This makes it infamous.