Thank you very much for your interesting question concerning d-primes in intelligibility measurements. Serious psychophysicist won’t take ‘don’t know’ (DK) for an answer: to control for decision strategies they trouble their subjects with n-AFC tasks. Researchers in speech perception seem less troubled by such effects, and consider DK acceptable, particularly in tests with open response sets. At times I wonder how much of the intelligibility difference between (young) normal hearing and (elderly) hearing impaired can be explained by differences in decision strategies.
Concerning the relation of SNR to d-prime, Jorgersen & Dau (JASA 2011) provide a formula that relates their SNR in the envelope domain to d-prime. They refer to Egan (1965) and Egan et al. (1969) for studies relating the good old SNR in the audio domain to d-prime. I haven’t checked these refs, but it surely sounds like an interesting lead. If you happen to come across other studies, I would be delighted if you could share them with me.
lilia sola regvnt lvnam, vndas, castra, leonem