Do labial-velar sounds really occur in more than 50% of the world's languages? I can think that labiovelars (velar consonants with labialisation as a secondary articulation) are quite common, but truly double-articulated labial-velars?
According to the University of Frankfurt's online-searchable subset of UPSID, [ w ] is labial-velar and occurs in 73.61% of the 451 languages in that subset of UPSID.
UPSID Sound Selection doesn't really distinguish between labial-velar and labiovelar.
They have a velarized voiced bilabial approximant in only one language, Uzbek. Wikipedia's article on that language doesn't discuss its phonology.
UPSID Sound Selection has no labialized velar approximant occurring in their 451-language online-searchable subset of UPSID; none at all. 16.63% (75) of those languages have other labialized velar consonants; but those are all nasals, fricatives, plosives, ejectives, and affricates, not approximants.Wikipedia's disambiguation page for labiovelar
does say [ w ]
is "labialized velar".
The most common labiovelar consonant is the voiced approximant [ w ]. This is normally a labialized velar, as is its vocalic cousin [ u ]. (Labialization is called rounding in vowels, and a velar place is called back.) However, languages such as Japanese and perhaps the Northern Iroquoian languages have something closer to a true labial–velar approximant, where the lips come together. In close transcription, the symbol [ w ] may be avoided in such cases, or it may be used with an under-rounding diacritic, as [ w̜ ].
The voiceless approximant is traditionally called a "voiceless labial–velar fricative", but true doubly articulated fricatives are not known to be used in any language, as they are quite difficult to pronounce and even more to aurally distinguish. (However, very occasionally the symbol [ ʍ ] is used for a labialized velar fricative, [ xʷ ]. This usage is not approved by the IPA.)
So they say that a labialized velar approximant /w/ is more common than a truely doubly-articulated labial-velar approximant /w/, but they don't say by how much.
Wikipedia's article https://en.wikipedia.org/wiki/Labial-velar_consonant
discusses several truly doubly-articulated labial-velar consonants including /w/, and https://en.wikipedia.org/wiki/Labial%E2%80%93velar_approximant
discusses both the labial-velar /w/ and the labialized velar /w/.
If you look at https://en.wikipedia.org/wiki/Doubly_articulated_consonant
you'll see that "secondary articulation" is usually the term used when one place-of-articulation has a looser stricture than the other. In general this means that one PoA is fricative or plosive/stop (or nasal or affricate or tap/flap or trill) while the other is approximant. Thus [ kʷ gʷ xʷ ɣʷ ŋʷ ] are all labialized velar, and [ pˠ bˠ ɸˠ βˠ mˠ ] are all velarized bilabial.
"Secondary articulation" would still be the term used if one PoA were a stop/plosive but the other were a fricative. The more strict PoA is primary, the other is secondary.
"True double articulation" is the term used when both PoAs have the same stricture. Thus both are nasals, or both are stops, or both are fricatives, or both are approximants, etc.
Nearly all "truly doubly-articulated" nasals and stops and fricatives are labial-velar. This includes [ k͡p, ɡ͡b, ŋ͡m ] (I would think it would also include [ x͡ɸ ɣ͡β ] if such sounds were ever attested in any natlang.)
To my mind it is impossible, or at least so difficult as to make any attempt pointless, to decide which of two co-occurring approximant articulations is primary and which is secondary. I have seen no references, nor have I seen any convincing arguments, to prove that there is such a thing as a secondarily-labialized primarily-velar approximant or a secondarily-velarized primarily-bilabial (or primarily-labiodental) approximant.
To my mind if both the velar and (one of) the labial points of articulation are closed to an "approximant" stricture, they must be being co-articulated, and that's doubly-articulated.
AFAICT, all /w/ phonemes are in fact truly doubly-articulated, and I cannot understand why anyone thinks some of them are secondarily-labialized primarily-velar or secondarily-velarized primarily-bilabial-or-labiodental.
If you could point me to some references, especially if they're both authoritative and free and online, I'd appreciate it. But I can't promise I'll be convinced by them.