Re: /W/ without /w/?

Dan Saunders wrote:
> I have a question about a recently started conlang: Is there any
> universal I
> don't know about that prevents a language from having the  unvoiced
> labial-velar approximant, /W/, without it's voiced counterpart, /w/? or
> /m_0/ without
> /m/? I would like to add these features, but wish to  know how 'realistic'
> they are. Thanks. Dan.
>
Yes, there probably is a sort-of universal regarding that. Phonologically,
things like /W/ and /m_0/ are considered "more highly marked", and tend not
to be present unless the "unmarked" ("natural") version is also present.

It's rather like IPA symbols; if you cite "IPA [p]" you are _by definition_
referring to a voiceless, unaspirated bilabial stop, with pulmonic,
egressive airstream (and IIRC some other things too), consequently those
features do not have to be mentioned. Similarly in phonology, resonants in
particular (e.g. w, r l y and nasals; vowels too) are assumed to be voiced;
they are said to be unmarked for the feature "voiced". Presence of voiceless
versions thus requires that voicing be marked +/- for the entire category,
leading to a less economical = more highly marked system.

The same applies to front-rounded and back-unrounded vowels, which tend not
to occur unless the "normal" front-unround and back-round vowels also occur.

This business of marked/unmarked/highly marked was quite important in
Generative Phonology-- among other things, it determined how many features
you needed to specify in your various rules; whether it's still used, or
relevant, in more recent theories, I know not. But anyway, in a conlang
(especially if it's meant to be alien) "universals" are there to be violated
:-)))