No Acoustic Evidence from RHD for a Right Hemisphere Role in Prosody Production: A Meta-Analysis

TY - ABST

T1 - No Acoustic Evidence from RHD for a Right Hemisphere Role in Prosody Production: A Meta-Analysis

AU - Weed, Ethan

AU - Fusaroli, Riccardo

PY - 2017

Y1 - 2017

N2 - No acoustic evidence from RHD for a right hemisphere role in prosody production: A meta-analysis.Ethan Weed and Riccardo Fusaroli (Aarhus University, Denmark)Poster: The right hemisphere (RH) is often thought to have a special role in prosody comprehension and production in general, and affective prosody in particular. If RH structures support prosody wholly or in part, we would expect acoustic differences between the productions of people with right hemisphere damage (RHD) and non brain-damaged (NBD) controls. Such differences have been reported, but the literature is mixed, and spans many years and experimental design types. To get perspective on the scope of these results, we conducted a systematic review and meta-analysis of acoustic measures of prosody production in people with RHD. We searched PubMed, PsychINFO, Web of Science and Google Scholar with the terms: (prosody OR intonation OR inflection OR intensity OR pitch OR fundamental frequency OR speech rate OR voice quality) AND (RHD OR right hemisphere) AND (stroke) AND (acoustic). Of the resulting articles, we selected only empirical studies with an N>2 that quantified acoustic measures of production and included a control group of non brain-damaged (NBD) participants. From the remaining articles, we calculated standardized mean differences (d) for fundamental frequency (F0), Intensity, Speech Duration, Pause Duration, Speech Rate, and Vowel Duration. We also noted whether the task was free speech production or constrained production, and whether the task targeted linguistic or emotional prosody. Our search produced 47 papers. Of these, only 10 met our inclusion criteria, and of these 2 reported from the same study. Mean sample N’s were: RHD: 12.15 (SD 10.92), NBD: 14.35 (SD 11). Mean age in years was: RHD: 58.94 (SD 16.7), NBD: 51.76 (SD 14.58). We found no significant impact of RHD on acoustic measures of F0 (d = -0.35, se = 0.29, p = 0.23, N = 9), nor of Intensity (d = -2.98, se = 2.84, p = 0.29, N = 4). No other features had enough papers (N > 3) to warrant a meta-analysis. Of the moderating factors (Task and Prosody Type) only Task had a significant impact (d = -13.08, se = 2.89, p < 0.0001), but this only for intensity, and a single study with a small N (RHD = 8, NBD = 7) and no age-matching drove this surprisingly large effect (Cook’s d = 0.9). There was no evidence of publication bias measured by regression analysis of funnel plots for F0 (z = 1.5937, p = 0.1110). Intensity did show evidence of publication bias (z = -4.1582, p < .0001), but this result was driven entirely by a single study out of 4 total. Taken at face value, the literature does not support a special role for the RH in prosody production. However, our study points to a greater problem in the field: studies with acoustic measures of prosody production in RHD are few, and sample sizes too small considering the heterogeneity of the population, to assess even medium effect sizes (power analyses indicate 30-100 participants per group required). We advocate responsible data sharing and standardized automated procedures for the extraction of acoustic features.

AB - No acoustic evidence from RHD for a right hemisphere role in prosody production: A meta-analysis.Ethan Weed and Riccardo Fusaroli (Aarhus University, Denmark)Poster: The right hemisphere (RH) is often thought to have a special role in prosody comprehension and production in general, and affective prosody in particular. If RH structures support prosody wholly or in part, we would expect acoustic differences between the productions of people with right hemisphere damage (RHD) and non brain-damaged (NBD) controls. Such differences have been reported, but the literature is mixed, and spans many years and experimental design types. To get perspective on the scope of these results, we conducted a systematic review and meta-analysis of acoustic measures of prosody production in people with RHD. We searched PubMed, PsychINFO, Web of Science and Google Scholar with the terms: (prosody OR intonation OR inflection OR intensity OR pitch OR fundamental frequency OR speech rate OR voice quality) AND (RHD OR right hemisphere) AND (stroke) AND (acoustic). Of the resulting articles, we selected only empirical studies with an N>2 that quantified acoustic measures of production and included a control group of non brain-damaged (NBD) participants. From the remaining articles, we calculated standardized mean differences (d) for fundamental frequency (F0), Intensity, Speech Duration, Pause Duration, Speech Rate, and Vowel Duration. We also noted whether the task was free speech production or constrained production, and whether the task targeted linguistic or emotional prosody. Our search produced 47 papers. Of these, only 10 met our inclusion criteria, and of these 2 reported from the same study. Mean sample N’s were: RHD: 12.15 (SD 10.92), NBD: 14.35 (SD 11). Mean age in years was: RHD: 58.94 (SD 16.7), NBD: 51.76 (SD 14.58). We found no significant impact of RHD on acoustic measures of F0 (d = -0.35, se = 0.29, p = 0.23, N = 9), nor of Intensity (d = -2.98, se = 2.84, p = 0.29, N = 4). No other features had enough papers (N > 3) to warrant a meta-analysis. Of the moderating factors (Task and Prosody Type) only Task had a significant impact (d = -13.08, se = 2.89, p < 0.0001), but this only for intensity, and a single study with a small N (RHD = 8, NBD = 7) and no age-matching drove this surprisingly large effect (Cook’s d = 0.9). There was no evidence of publication bias measured by regression analysis of funnel plots for F0 (z = 1.5937, p = 0.1110). Intensity did show evidence of publication bias (z = -4.1582, p < .0001), but this result was driven entirely by a single study out of 4 total. Taken at face value, the literature does not support a special role for the RH in prosody production. However, our study points to a greater problem in the field: studies with acoustic measures of prosody production in RHD are few, and sample sizes too small considering the heterogeneity of the population, to assess even medium effect sizes (power analyses indicate 30-100 participants per group required). We advocate responsible data sharing and standardized automated procedures for the extraction of acoustic features.

UR - http://www.neurolang.org/?page=poster_detail&show=authors&sort=board_a&go&session=Thursday%2C%20November%209_10%3A00%20–%2011%3A15%20am_Harborview%20and%20Loch%20Raven%20Ballrooms&id=91

M3 - Conference abstract for conference

SP - 130

T2 - The Ninth Annual Meeting of the Society for the Neurobiology of Language

Y2 - 8 November 2017 through 10 November 2017

ER -