Michael Tye argues that "[t]he phenomenology of occurrent thoughts derives fundamentally... from the phenomenology of their associated linguistic, auditory images." (Consciousness and Persons, p.79) He thus denies that the meaning or content of the thought influences its phenomenal "feel". This claim isn't obviously true though, and Tye's main argument for it is far from compelling.
Tye appeals to Twin Earth, where our twins have thoughts that are (presumably) phenomenologically indiscernible from our own, despite having different semantic contents. (Their 'water'-thoughts are about twin-water (XYZ), whereas ours are about H2O.) So a difference in semantic content is not sufficient for a difference in phenomenology.
Of course, just because there are some cases where content makes no difference, it doesn't follow that it never makes a difference. So even if it works, the conclusion of Tye's argument is too weak for his purposes. But I don't think the argument does quite work. Perhaps Tye is simply looking at the wrong kind of semantic content. Rather than looking at the externalist, reference-based "wide content" presupposed above, it's far more intuitive to think that phenomenology will be influenced by a more internalistic form of "narrow content", associated with Fregean sense, cognitive significance, inferential role, and so forth. This alternative kind of content might be captured by the "primary intension" of a Chalmersian 2-D semantics.
The problem with Tye's argument, then, is that the Twin Earth case does not involve any difference in narrow content (or primary intension). The difference in externalist meaning is not one that is internally accessible to the agents, so it is unsurprising that it would not affect their phenomenology. But it would be a leap to thereby conclude that no aspect of a thought's content ever has an intrinsic impact on its phenomenal "feel".
A more appropriate case would involve deeper differences in content, as found in polysemy. Consider two thoughts about "going to the bank", where the first is about a river bank, and the second a financial institution. Will these two thoughts feel the same? This isn't at all obvious, but that's what Tye needs to argue.
P.S. Tye adds in a footnote that there might be associated non-auditory (say, visual) images which contribute to the phenomenology. Perhaps that could explain the apparent differences in the bank case, without any need to appeal to the idea that the semantic content itself has any phenomenal influence. I'm not sure how to adjudicate between these rival explanations.
P.P.S. Presumably deaf people don't associate their thoughts with auditory images. I wonder what their thoughts feel like, then? (Are they associated with visual or kinesthetic/proprioceptive images instead?)