T04 Regular rhythmic primes do not (always) benefit speech-in-noise perception: Evidence for distinct outcomes for temporal attention depending on speech-in-noise manipulations
Research has shown that regular music rhythm can boost grammaticality judgements in subsequently presented speech, even though speech is less regular. Regular rhythmic primes (~30s) presented before a set of naturally spoken sentences typically improve grammaticality judgements compared to sentences preceded by irregular primes or baseline conditions. This rhythmic priming paradigm is motivated by the Dynamic Attending Theory (DAT), which suggests that external rhythms entrain neural oscillations that persist over time and enhance subsequent processing. Enhanced speech processing is particularly valuable when the signal is degraded, as in noisy environments and for hearing-impaired listeners. To simulate this situation, in Experiment 1 (n = 31), participants (tested in the lab) detected grammatical errors in naturally spoken sentences presented in noise (-3 dB signal-to-noise ratio) that were preceded by regular and irregular rhythmic primes. Participants were more sensitive to grammatical errors after irregular compared to regular primes (measured by d’), suggesting that the noisy background reversed the typical regular prime benefit. To investigate this reversed priming effect, three additional experiments were run online. In Experiments 2 (n = 39) and 3 (n = 39), the speech envelope was either preserved (up to 20Hz with reduced temporal fine structure, TFS) or reduced (filtered at 5Hz with original TFS). We predicted that regular rhythmic primes would enhance processing when the speech envelope was preserved, but not when it was reduced. However, both experiments showed a small benefit of regular rhythms for grammatical error detection compared to irregular rhythms, and a stronger bias to respond grammatical after irregular compared to regular primes in Experiment 2. Experiment 4 (n = 40) investigated whether introducing a rhythmic modulation at the regular rhythm frequency (2 Hz) to the noise masker from Experiment 1 (maintaining -3 dB SNR) would re-instate the benefit of the regular prime. Regular and irregular primes did not differ in their influence on modulated noise processing, even though the direction of mean differences was similar to Experiment 1. These results suggest that rhythmic primes may influence speech processing differently depending on whether speech-shaped noise is applied to the speech (Exp. 1, 4), or whether the speech envelope and TFS are manipulated (Exp. 2, 3). Effects of rhythmic priming on speech-in-noise perception appears to be a promising line of enquiry to uncover the acoustic features important for rhythmic priming, and could provide insights into when speech processing in noisy environments is helped or hindered by regularity.