What is the nature of the distinction when a semivowel is surrounded by its corresponding vowel?

Question

I've searched the site a bit for this topic, and I recognize that there's quite a bit of variability about the classification of phones like [w] and [j]. My primary understanding comes from Catford's A Practical Introduction to Phonetics, which, if my memory is correct, presents these in their guise as semivowels. In this context, he seems to define a semivowel as a momentary, rather quick move into and out of a vowel. He demonstrates by having the reader "discover" [w] as a momentary [u] by having the reader start by making a [aaaauuuuaaaa] sound and gradually reduce the length of the [uuu] portion until [aaaauuaaaa] becomes [aaaaauaaaa] becomes [aaaawaaaa]. Similarly, [j] can be produced by starting with [aaaaiiiiaaaa] and then shortening the [ii] portion until the sound is [aaaajaaaa].

This all made a great deal of sense to me and has helped me find phones that are less familiar to me, such as [ɥ]. What isn't clear to me, however, is what exactly is happening when I make sounds where the semivowel is preceded or followed by the same vowel sound from which it's derived. For example, [ji] and [wu] are clearly different sounds to my ear than [i] and [u] alone (or [ʔi] and [ʔu]...?). If I try a similar Catford-esque experiment by making a [uuuuwuuuu] sound, it seems like something is happening—perhaps more intense labialization—but I'm not sure. When I try to do the same thing with [iiiiijiiiii], I have trouble hearing as much difference.

My question is, if I strictly go by the first paragraph's understanding of a [w] sound as simply an abbreviated [u], then it seems that a sound like [wu] should be indistinguishable from [u] (and same goes for [ji] vs. [i]). Is there something wrong with my understanding of the [w] and [j] semivowels, or does one make some sort of "extra" distinction when a semivowel is surrounded by its corresponding vowel?

Answer 1

If you do strictly go by the first paragraph's understanding of [w], [wu] should be a somewhat longer version of [u], having whatever its reduced duration is plur that of [u]. It is not false to say that [w] is a rather short [u], but that is insufficient. Apart from duration, there is a small but visible difference in stricture of the vocoid, where the glides have a narrow constriction. The articulatory movement of [j,w] is more ballistic. This is mostly independent of the surrounding vowel (though [w] in [awa] vs. [uwu] are themselves different).

Answer 2

I think Catford has in mind the phonetician's distinction between a sound with a steady state and a sound without a steady state, the latter being a glide. Looking at the sound spectrograms displaying formant transitions, one can distinguish between sounds where the formants are level for an appreciable time, which is a steady state, and others where steady states cannot be found, which are glides -- parts of diphthongs.

I'm skeptical. At a normal rate of speech, steady state vowels may be difficult to find.