Full Idea
Donnellan seems to be unsure whether to regard his referential-attributive distinction as indicating a semantic ambiguity or merely a pragmatic one.
Gist of Idea
Donnellan is unclear whether the referential-attributive distinction is semantic or pragmatic
Source
comment on Keith Donnellan (Reference and Definite Descriptions [1966]) by Kent Bach - What Does It Take to Refer? 22.2 L1
Book Reference
'Oxford Handbook of Philosophy of Language', ed/tr. Lepore,E/Smith,B [OUP 2008], p.537
A Reaction
I vote for pragmatic. In a single brief conversation a definite description could start as attributive and end as referential, but it seems unlikely that its semantics changed in mid-paragraph.