When the indexical ‘I’ appears inside quotation marks, it refers not to the person now speaking but to the person whose speech
is being reported. The apparently ‘monstrous’ behaviour of quotation can be dismissed in direct speech, so long as one
maintains that the quoted part is mentioned rather than used. The same cannot be maintained,
however, in so-called ‘mixed’ quotation, for which a pure-mention analysis is implausible. In this paper I
compare two accounts of the semantics of quotation. While the accounts of Maier & Geurts (2004), Geurts & Maier (this
volume), and Bittner (to appear) all anticipate the correct behaviour for indexicals inside quotation, the approach developed
by Geurts and Maier makes a further, false generalisation, and is therefore empirically inferior.