Palatalised allophones of /k/ and /g/ as a cross-linguistic phenomenon

Question

As a native speaker of Russian, where [k]/[kʲ] and [g]/[gʲ] are phonemically distinct, I've always been intrigued by the fact that several languages that don't have that distinction, and are in fact hardly "into" palatalisation at all, tend to favour what I'd describe as [kʲ] and [gʲ] realisations over "plain" [k] and [g], or at least go for something in between. This includes many varieties of English, as well as French, Swedish, Persian, and perhaps some others that I'm not aware of.

In most if not all of these cases, the plain and palatalised versions co-exist as allophones, depending on position; [kʲ]/[gʲ] is especially frequent word-finally. On the other hand, this doesn't happen to any other consonants: the English take, depending on the speaker, is about equally likely to be pronounced as [teɪk] or as [teɪkʲ] (or close to it), while in no English accent does tape even remotely sound like [teɪpʲ]. Also, the phenomenon appears to be rather sporadic; it exists in French but not in Spanish, in Swedish but not in Danish.

Have there been any studies or proposed explanations as to why and under what conditions these palatalised allophones develop?

Answer 1

Of course a difference between p and k is that surrounding vowels don't affect the position of articulation of p, but they do affect the position of k quite a lot. For your example of [teɪk] vs. [teɪkʲ], I would be surprised if the off glides of the diphthongs were really the same, even though you've written them both identically. The first one, before the velar [k], is probably more like barred-i, a central vowel.

I suspect that, in general, the variation between palatal and velar k is more a story about English vowels than k, and that k just gets carried along for the ride. Then, also, there is a tradition in phonetics of more or less ignoring the tongue body position of consonants, because, I guess, it is often non-phonemic. If you said the t (or flap) of "auto" like you say the t of "meaty", it would sound palatalized; yet the two t's would ordinarily be transcribed the same way.

Some American dialects have a lot of central off-glides, sometimes written H, in words like "hit", "pat", "pack", "beck". There's an interesting possibility this is due to a velarization of the final consonants. But I've never seen it transcribed that way. We notice the vowel change, but not the concomitant difference in the final consonants.

The k in words like "bulk" sounds so far back to me, that I'd write it as a uvular. Of course, that's due to the darkening of syllable offset l.