Från: Piotr Gasiorowski

Ladefoged & Maddieson (following Lindblad) describe two variant articulations of Swedish [ɧ]. One (actually second in their order of exposition) is very similar to what you described in your posting on "phonetic curiosities". The other is defined as a strongly rounded labiodental velar or velarised voiceless fricative. In simpler terms, they describe it more or less as [f] and [x] articulated simultaneously, with strong lip protrusion. There is more detail in the book but I haven't got it with me now and am quoting from memory; I can provide a verbatim quotation if necessary. It appears from the description that in this variant the labiodental articulation is primary and the velar constriction serves to colour to the sound dark. Can you comment on this?
I'll see if I can get hold of the book; it would be interesting to read.
Some comments off the top of my head: it may be true that there are two discrete variants, and I guess that L's & M's description is based more on systematic study than on 'casual observation' (which might describe what I partly based my description on). I've considered it more of a continuum, which I tried to capture in my original posting with the words "It is usually labialized too, often only slightly but in some speakers' version the acoustic impression comes close to that of an [f]". "Labialized" should have been "bilabialized/labiodentalized", though. But I hadn't realized that the two different kinds of labialization also involve different behaviour of the (back of the) tongue (if I understand you correctly): velar [x] + labiodental, or palato-velar, "[x] + [ʃ]" as it were, + no -- or, in my experience, various degrees of optional -- bilabialization. My own version is not of the [f]-ish kind, but a moderately bilabialized palato-velar (of the kind I tried to describe in my first posting). If I should give up the idea of a continuum, it seems reasonable to draw the line between bilabial and labiodental. I can't remember ever having heard a variant with only slight labiodentalization, which may also speak in favour of cutting up the (non-)continuum in two discrete parts. Do they (L & M) say anything about distribution (socially, geographically, stylistically)? My impression is that the labiodental version is far more unusual than the other one, but maybe I just hang out with the wrong people.
I also wonder if the early Modern English confusion between [f] and [x] (as illustrated by the pronunciation of enough or laugh) did not result from the existence of a similarly complex fricative in English once upon a time. And of course Ukrainian could be another case.
Sort of like IE *kʷ > p, etc. The existence of such a fricative would also explain why it works both ways: though there is no good etymology, the Dutch lucht version seems to be secondary in relation to the Luft words. However, the traditional acoustic explanation also takes care of that, but I can't think of any examples that don't include either complex phonemes, like */kʷ/, or fricatives. Well, my parents-in-law think that the name of the green Teletubby is Dixy rather than Dipsy, but that's hardly conclusive :)
 
Urban