Davidson on Demonstratives; Stern on Metaphor

One of Josef Stern's main arguments is as follows:

(1) If metaphors cannot be semantic then there are two other options (i) Davidson's view of metaphor, and (ii) pragmatic view of metaphor
(2) The pragmatic view of metaphor does not work,
(3) Hence, Davidson's view of metaphor is the only viable option.
(4) Davidson's theory of demonstratives are context-dependent semantics.
(5) If Davidson has context-dependent semantics for demonstratives, then if metaphor is sufficiently similar to Davidsonian Demonstratives, then metaphor is a semantic feature of language.
(6) Metaphors are sufficiently similar to Davidsonian Demonstratives.

So what is a context-dependent semantic rule? Indexicals and demonstratives are the most commonly used examples. Take the following utterance by Bill Clinton on June 15, 2007:

(7) I am the President of the United States.

Presumably (7) is false. Suppose George W. Bush were to utter the following on June 15, 2007:

(8) I am the President of the United States.

This is a true utterance. Both (7) and (8) have the exact same linguistic and syntactic structure, the only difference is who utters them. Indexicals such as 'I' always refer to the speaker, in that sense indexicals are context-dependent. Similarly to demonstratives such as 'that'.

(9) That is a baseball field.

Depending upon whether 'that' is actually a baseball field will determine the truth-value for (9).

According to Stern the same is true for metaphor. There are semantic rules that constrain the meaning of metaphor. Stern's claim is that metaphor follow some of the semantics of anaphora, that is, metaphor does not depend solely upon speaker meaning.