Japanese de-voices vowels in certain contexts adjacent to voiceless consonants. The weakness you're describing is the lack of voicing (vocal cords vibrating).
The きや きゃ (kiya, kya) distinction is not a phonological alternation. For instance, we have ぎゃく (gyaku), despite the fact that /g/ and /y/ are both voiced.