Quoting Ray Brown <[log in to unmask]>:

> On Sunday, November 7, 2004, at 08:50 , Andreas Johansson wrote:
> >  The lack of an IPA sign need only mean that
> > no language distinguishes phonologically from apico-alveolars, and since,
> >  I
> > believe, both retroflex-alveolar distinctions and lateral fricatives are
> > relatively rare typologically, the lack of such languages might simply be
> > due
> > to combinatorics.
> Yes - but I thought IPA was meant to provide symbols for _phonetic_
> representation as well as phonological representation.
> Obviously the blacked-out cells in the IPA place of articulation/manner of
> articulation matrix mean that the sound that would in theory fill the cell
> is in fact physically impossible. I have assumed that the empty cells
> which are not blacked-out were sounds that were theoretically possible but
> not attested in any known natlang. Is this assumption mistaken?

Sort-of. To have a separate IPA symbol a phone is supposed to be _contrastive_
in some natlang. Now, there are some failures to achieve that (Tamil's
distinction 'tween alveolars and dentals can't be captured without recourse to
diacritics, and I rather doubt that all the signs for mid-central vowels are
strictly necessary for achieving that aim), but that's the basic idea.


PS Incidentally, does anyone know if there's a lang contrasting /F/ against both
/m/ and /n/?