Prosodic production and domain sensitivity in brain-damaged patients Shari R. Baum, * Veena D. Dwivedi, and Amee P. Shah School of Communication Sciences & Disorders Centre for Research on Language, Mind & Brain McGill University, Canada Available online 23 July 2004 Whereas a great deal of research has been devoted to defining the neural bases of the segmental aspects of speech processing (see, e.g., Scott & Wise, 2003 for a recent review of neuroimaging data), the neural substrate for prosodic processing has only recently garnered substantial attention. Recently, a cue lateralization perspective has been proposed which purports that the LH is specialized for the control of temporal parameters of speech, whereas the RH is spe- cialized for spectral parameters, particularly the processing of pitch (F0) (e.g., Van Lancker & Sidtis, 1992). With the advent of more functional neuroimaging studies of prosodic processing per se, several modifications to such theories have recently been proposed. In par- ticular, Gandour et al. (2003) have posited a distinction among levels of linguistic structure that correspond to relatively short (syllable- sized associated with LH processing) and longer (phrase or sentence- level associated with RH processing) windows or prosodic domains (c.f. Baum & Dwivedi, 2003). There has recently been a surge of interest in subtle, but im- portant linguistic distinctions that are signalled by prosodic cues (e.g., Baum, Pell, Leonard, & Gordon, 1997). The current experi- ment adds to this work by examining the use of prosody in syntactic disambiguation for different sentence types which include or exclude intonational phrase level (IPh) prosodic boundaries (Price, Osten- dorf, Shattuck-Hufnagel, & Fong, 1991). Method Subjects The subjects included 5 LHD non-fluent aphasic patients, 5 RHD patients, and 5 age-matched non-brain-damaged control subjects. All were native English speakers with hearing within normal limits. The brain-damaged patients, diagnosed by standardized tests, had all suf- fered a single, unilateral CVA. Stimuli and procedure Three sentence types that incorporate or exclude prosodic boundaries were examined: appositives vs. integrated phrases, pa- rentheticals vs. integrated phrases, and tags vs. integrated phrases (Price et al., 1991). Eight pairs of sentences in each of these three sentence types were elicited from speakers following a biasing context sentence. The productions were recorded on DAT tape for acoustic analysis. A perception test was also presented to a group of five unimpaired native-English-speaking listeners to determine whether the contrasts produced by the various speaker groups were perceptible. Acoustic analyses Durations of each sentence were computed, as were pre-boundary word durations and pause durations at the boundaries. Word and pause durations were expressed as a proportion of the sentence dura- tion to control for speech rate differences across participants. In ad- dition, peak F0 was computed at the midpoint of the stressed vowel in pre- and post-boundary words; from these values, F0 ratios were calculated as a means of observing changes in intonation contour at the boundaries. Results Analyses were conducted for each sentence type separately. For the parentheticals, all groups showed a tendency to exhibit pre- boundary lengthening when mean duration proportions were ex- amined; however, statistical analyses revealed that differences reached significance only for the normal control participants. Both normal controls and RHD patients produced parentheticals with longer pauses at the boundary position relative to that same posi- tion in integrated phrases. The LHD patients failed to display either pre-boundary lengthening or increased pause duration at the boundary. For the appositives, the expected pre-boundary length- ening was found for both normal controls and RHD patients; however, the difference only reached significance for the RHD pa- tients (due to a single NC speaker). Only the NC speakers dem- onstrated significantly longer pauses when the boundary was present in the stimuli, although all groups again showed a trend in that direction. Finally, for the tags, NC speakers produced marginally significant lengthening at the phrase boundary relative to the inte- grated phrase version of the stimuli. No significant differences in pause duration emerged for any group, although all showed the expected pattern when examining mean proportional duration val- ues. Interestingly, in the integrated phrase conditions—across all sentence types —the LHD patients always produced longer pause durations than either the NC or RHD participants. In other words, the LHD patients appeared to be producing pauses in inappropriate positions. Analyses of F0 failed to reveal any consistent patterns across the sentence types or speaker groups. Brain and Language 91 (2004) 56–57 www.elsevier.com/locate/b&l * Corresponding author. E-mail address: shari.baum@mcgill.ca (S.R. Baum). 0093-934X/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.bandl.2004.06.031