Originally Posted by
Vagrere
I'm not a linguist, so I'm not going to be able to express myself particularly well in the following explanation, but here goes, anyway.
There are a range of related sounds that are represented by the roman letter "T," from a pretty soft one like you get at the end of the word "aft" to hard, almost "D"-like sounds such as in "batty." This range is obviously going to differ depending on the language you're romanizing, I'm just using English here to illustrate my meaning. In any event, the sound I hear in the CCS opening is still a "t" sound to me--just a very abrupt, almost clipped one, which gives it an unusually sharp character, which is what I think you're hearing as a "g" sound. A lot of her consonants are also a little weak in general, such as the "d" sound in "datte" at about 0:33. I think it's this combination, with the music playing on top of it, that's causing the perception. As for the vowel, she is pronouncing it very slightly off from the short "o" sound that is standard Japanese pronunciation for と. It's almost blending a tiny bit into a short "u" sound, to my hearing, which again, could help lead to a perception of ぐ. All that said, I do still hear a pretty close to standard と, myself.
You're absolutely right that certain dialects do have unique pronunciations for certain sounds, but I'm not sure that's what's going on in the CCS opening.
Bookmarks