Human say_quote() will use the language to obtain a speech verb while
silicon say_quote() will use synth speech verbs, so no need for the extra
language flag. Will mean that silicons will always use their synth speech
verbs regardless of language, however. Logic for what silicons use as
their verb should probably go in say_quote() anyways, so future updates to
that can go there instead of branching in living say code.