A variant of the distinction between strongly and weakly unattested forms is the distinction between phonotactically prohibited forms and accidental gaps. What we want to do is distinguish between strings of 0 frequency which the grammar says can’t occur and strings of 0 frequency which the the history of the language has accidentally never produced. Recent work by Elliott Moreton (see his dissertation and 2002 paper in Cognition) shows how these two kinds of 0s can be distinguished experimentally. The case in point was [tl] vs [pw], where the former is prohibited and the latter accidentally absent. Moreton showed that when asked to judged a continuum from [l-w] after [t], listeners gave many more “w” responses than “l” responses, but after [p]. there was no comparable bias toward more “l” than “w” responses. The “w” bias after [t] shows that listeners are trying to identify the sonorant as the one that is phonotactically legal in its context, and the lack of an “l” bias after [p] shows that they are under no similar compulsion to avoid strings that accidentally don’t occur. In short, it is possible to tap experimentally a language user’s knowledge of their grammar in such a way as to distinguish between strongly vs weakly unattested forms.
I have been wondering for some time what people who study sound think “laboratory phonology” is? In particular, I’d like to hear people’s views on what they think its relationships are to phonetics and to whatever phonology is not laboratory phonology.