E.g. し(shi) + や(ya) = しゃ (sha), still the same 'a' sound.
English has only 5 vowel characters but more than 5 vowel sounds.
This is not an unreasonably connection to make. /j/ is the semivowel corresponding to /i/. Essentially, /j/ is what happens if you produce /i/ with just enough constriction that it no longer is classified as a vowel.
It is not apriori unreasonable to consider semivowels as vowels. However, linguistics has (mostly) reached the conclusion that semivowels are best characterized as consonants.
With my untrained ear, I have heard Japanese speakers use the diphthong /aɪ/ and the vowel sequence /a.i/ interchangeably. However, I suspect they are actually always pronouncing it as /a.i/, and I am just not good at distinguishing it.
I they were to try and pronounce pie using the closest approximation available with Japanese phonetics, I suspect they will settle on /pa.i/, in which case it is a real /i/.
If they were to pronounce it as /paj/, then Japanese photostatic would require adding a vowel at the end, likely giving us /paju/, which is not what we get.
イェス (yes) is a bit more difficult to predict. Japanese does have all of the sounds nessasary to pronounce /jɛs/ without modification. The syllable structure would require the addition of a vowel at the end, giving us /jɛsu/. However, as far as I am aware, Japanese does does have the /jɛ/ sequence for some reason. The question is, how bad is the /jɛ/ sequence to Japanese speakers? If it is just a historical artifact that /jɛ/ doesn't occur in Japanese, then they should have no problem pronouncing /jɛs/ as such, in which case the イ would be the consonant /j/. However is /jɛ/ is actually a bad sequence, then it would be changed into its closest approximation, which would be /i.ɛ/, in which case イ is the normal vowel /i/.
Based on the spelling (in particular the small ェ), they are at least trying to be faithful to the original English; as /i.ɛ/ has a perfectly normal spelling as イエ.
To take a simmilar example from English. Consider the name "Juan" (ʒɒn). English speakers can typically pronounce this, but it sounds foreign, because English does not allow /ʒ/ to occur in the begining of a word. If they were to pronounce it the English way, they would get "John" (ʤɒn).
Further, I suspect that the only reason English speakers can pronounce "Juan" so easily is exposure. I had a Brazilian Portuguese teacher (eg. a linguistics teacher whose native language is Brazilian Portuguese), whose name also begins with ʒ, but everyone pronounces it as "ʤ" and (it my experience), it was difficult to even understand what his name was until he pronounced it the with ʤ.
I'm stretching a little bit, because I've looked into this in idle curiosity before.
The only connection I can think of is the brand "Ebisu", written "ヱビス", and sometimes romanized to "Yebisu". The character itself seems to be from a historical "we" kana, and I'm not sure of the reason for that.
Oh, and there's an archaic hiragana (but no katakana equivalent?) for "ye" (/jɛ/): 𛀁.
Japanese has some interesting allophones. I also find the allophones in the Z-row fascinating as well ([z] and [d͡z] are allophones in free variation).
[1] https://www.collinsdictionary.com/dictionary/portuguese-engl...
I just did a brief search, and found this paper [0], which includes a phonological inventory of Brazillian Portuguese. It lists /j/ as a consonant.
[0] http://journals.sagepub.com/doi/pdf/10.1177/0023830902045002...
So what are we counting here? The sounds or the characters we use to represent the sounds? Because if you go by sounds, I think Dutch has more than 14.
(a, aa, o, oo, e, è, ee, u, uu, i, ie, eu, ui, au, ij, oe)
I've heard that in Czech, the 'r' can be a vowel. In a name like Brno, for example. Those are apparently two syllables. (I don't speak Czech, so no idea if it's true.)
[1] https://en.wikipedia.org/wiki/Japanese_phonology#Vowels