The Influence of Vowel Identity, Vowel Production Variability, and Consonant Environment on Envelope Following Responses

Objectives: The vowel-evoked envelope following response (EFR) is a useful tool for studying brainstem processing of speech in natural consonant-vowel productions. Previous work, however, demonstrates that the amplitude of EFRs is highly variable across vowels. To clarify factors contributing to the variability observed, the objectives of the present study were to evaluate: (1) the influence of vowel identity and the consonant context surrounding each vowel on EFR amplitude and (2) the effect of variations in repeated productions of a vowel on EFR amplitude while controlling for the consonant context. Design: In Experiment 1, EFRs were recorded in response to seven English vowels (/ij/, /Iota/, /ej/, /epsilon/, /AE/, /u/, and //) embedded in each of four consonant contexts (/hVd/, /sVt/, /zVf/, and /Vv/). In Experiment 2, EFRs were recorded in response to four different variants of one of the four possible vowels (/ij/, /epsilon/, /AE/, or //), embedded in the same consonant-vowel-consonant environments used in Experiment 1. All vowels were edited to minimize formant transitions before embedding in a consonant context. Different talkers were used for the two experiments. Data from a total of 30 and 64 (16 listeners/vowel) young adults with normal hearing were included in Experiments 1 and 2, respectively. EFRs were recorded using a single-channel electrode montage between the vertex and nape of the neck while stimuli were presented monaurally. Results: In Experiment 1, vowel identity had a significant effect on EFR amplitude with the vowel /AE/ eliciting the highest amplitude EFRs (170 nV, on average), and the vowel /ej/ eliciting the lowest amplitude EFRs (106 nV, on average). The consonant context surrounding each vowel stimulus had no statistically significant effect on EFR amplitude. Similarly in Experiment 2, consonant context did not influence the amplitude of EFRs elicited by the vowel variants. Vowel identity significantly altered EFR amplitude with /epsilon/ eliciting the highest amplitude EFRs (104 nV, on average). Significant, albeit small, differences (<21 nV, on average) in EFR amplitude were evident between some variants of /epsilon/ and /u/. Conclusion: Based on a comprehensive set of naturally produced vowel samples in carefully controlled consonant contexts, the present study provides additional evidence for the sensitivity of EFRs to vowel identity and variations in vowel production. The surrounding consonant context (after removal of formant transitions) has no measurable effect on EFRs, irrespective of vowel identity and variant. The sensitivity of EFRs to nuances in vowel acoustics emphasizes the need for adequate control and evaluation of stimuli proposed for clinical and research purposes.