In my field, Linguistics, we often use sentences from published works, or attested examples of sentences we have read or heard, as evidence of what kinds of things are possible or grammatical in a language. These are often felt to be more convincing than examples of sentences derived from our own introspection.
Suppose one wishes to re-use and refer to examples from a paper (paper A), which were themselves cited from a previous paper (paper B). If the examples were used to illustrate related points and discussions in the two papers, the scholarly thing to do would be to go and investigate the original paper and look at the examples in context there. One would then likely cite the original paper.
What should one do, however, if some of the examples in paper A do not actually appear in the published version of paper B? Let's say, for example, that paper B was a conference paper and that one might assume that the examples appeared in the original presentation but did not make it to the published paper in the conference proceedings?
In my case, the article I am writing is a response to paper A. If I merely cite the examples as from B cited in paper A, I might look as if I couldn't be bothered to go and look at paper B myself. This, I suppose, might reflect badly on the perceived scholarliness of the work. I cannot honestly cite the examples as being from paper B, as they do not appear in the published version. If I say the examples are cited from B by the authors of A, but that they don't appear in the published version of B, I might look as if I'm publicly criticising A, and also as if I'm a pedant.
What is the best thing to do in terms of attribution here?
In case it counts for anything, I'm a lowly PhD student.