Effects of Right and Left Hemisphere Damage on Performance of the Right Hemisphere Communication Battery

Transcription

1 Brain and Language 80, (2002) doi: /brln , available online at on Effects of Right and Left Hemisphere Damage on Performance of the Right Hemisphere Communication Battery E. Zaidel,* A. Kasher, N. Soroker, and G. Batori *University of California, Los Angeles; Department of Philosophy, Tel Aviv University, Ramat Aviv, Israel; Loewenstein Rehabilitation Hospital and Sackler School of Medicine, Tel Aviv University, Ramat Aviv, Israel; and Tel Aviv University, Ramat Aviv, Israel Published online February 15, 2002 A Hebrew adaptation of Gardner and Brownell s (1986) Right Hemisphere Communication Battery (HRHCB) was administered to 27 right brain-damaged (RBD) patients, 31 left brain-damaged (LBD) patients, and 21 age-matched normal controls. Both patient groups showed deficits relative to controls and overall there was no difference between the two patient groups. A factor analysis of patients scores on the HRHCB yielded two interpretable factors, a verbal and a nonverbal one. These factors were not lateralized. Performance of patients on the HRHCB correlated significantly and positively with performance on most tests of basic language functions, measured with a Hebrew adaptation of the Western Aphasia Battery (HWAB) and with other cognitive functions measured with standardized neuropsychological tests. There were stronger correlations of HRHCB with subtests of the HWAB in LBD patients and with nonlanguage cognitive tests in RBD patients. In the LBD group, HRHCB subtests scores correlated negatively with lesion extent in frontal and temporal perisylvian regions. Such localization was not observed in RBD patients. The results argue against selective right hemisphere (RH) involvement in the RHCB, alleged to measure pragmatic aspects of language use, and show, instead, bilateral involvement. The results also argue against a modular organization of these functions of language use, especially in the RH Elsevier Science (USA) Key Words: language functions; pragmatics; laterality; modularity; right hemisphere, hemispheric specialization; localization of function; brain damage; stroke. INTRODUCTION The Right Hemisphere Communication Battery (RHCB) was developed by H. Gardner and H. Brownell (1986) of the Boston VAMC. It includes 11 subjects that fall into four parts: Humor, Emotion, Nonliteral Language, and Integrative Processes. Each of these tests was found to yield selective deficits in right brain-damaged (RBD) patients, but this was in conditions of separate administration of each test, often in different versions and with different comparison groups. In particular, many of the functions tested by the RHCB, such as Sarcasm (Kaplan et al., 1990), were never Supported by Grant from the U.S. Israel Binational Science Foundation; Grant from the Basic Research Foundation of the Isreal National Academy of Science and the Humanities, USPHS NIMH RSA MH00179; and USPHS NIH Grant NS We are indebted to Dr. Alexis Zalozieki for assistance with matters. Thanks to Dr. Hiram Brownell for critical comments on the article. Thanks also to Linda Capetillo-Cunliff, Kevin Laack, Ian Gizer, and Eric Mooshagian for assistance. Address correspondence and reprint requests to Eran Zaidel, Department of Psychology, UCLA, Los Angeles, CA Fax: (310) ezaidel@psych.ucla.edu X/02 $ Elsevier Science (USA) All rights reserved.

2 RIGHT HEMISPHERE COMMUNICATION BATTERY 511 tested on left brain-damaged (LBD) patients. Consequently, the battery has lacked norms for brain-damaged patients and this has limited its use to that of suggesting hypotheses that could then be examined more systematically. No anatomical data on the intrahemispheric localization of the different subtests was available. Common to the tests of the RHCB is the use of contextual information to interpret nonliteral language. But there is no formal theoretical account of any of the abilities tested, nor a uniform theory to integrate them. Instead, we have administered the RHCB to both LBD and RBD patients in conjunction with a theory-driven new pragmatics battery, including basic speech acts (Zaidel et al., submitted) and implicatures (Kasher et al., 1999) as well as selected standardized neuropsychological tests. The RHCB involves standardized testing using the multiple-choice format in most subtests (Gardner & Brownell, 1986). Humor When RBD patients have to select a picture to represent the punch line of a cartoon sequence, they often choose non sequitor humorous endings, suggesting that they recognize surprise as an essential element of humor (Bihrle et al., 1986; Gardner et al., 1975), but that they cannot resolve the humorous incongruity and integrate it coherently with the elements of the joke (Bihrle et al., 1988; Brownell, 1988). The tendency to choose non sequitor endings applies to verbal humor as well (Bihrle et al., 1986; Brownell et al., 1983). Failure to appreciate humor may also result from impairments in interpreting situational, facial, and prosodic cues that signal the emotional content of the message. However, Daage and Hartje (1985) studied the ability of RBD patients to comprehend cartoons and concluded that the deficit was cognitive rather than affective. When RBD patients are asked to repeat jokes, they often fail to sequence the elements of the joke properly and omit key elements in the punch line or in the body of the joke, thus failing to set up the listener properly for the punch line. Clinical observations indicate that in some cases, the humor of RBD patients may be crude or otherwise disinhibited and inappropriate to the situation (Tompkins, 1995). Brownell et al. (2000) note that the jokes used as stimuli in the RHCB Verbal Humor subtest all required reinterpretation of the punch line and that fully half of the items involved elements of mistaken beliefs, ignorance, or fooling a victim. They suggest that the underlying deficit is in the ability of RBD patients to use a Theory of the Mind. They further suggest that such a deficit involves damage to right ventromedial prefrontal regions. This was confirmed by Shammi (1997, cited in Shammi and Stuss, 1999), who found selective involvement of right prefrontal cortex in the RHCB Verbal Humor subtest. We will evaluate that suggestion. Emotional Prosody RBD patients are believed to have a selective deficit in appreciating emotion conveyed by prosodic inflections or by facial expression. Tompkins (1995) notes that some RBD patients sound melodically monotonous, whereas others sound hypermelodic. An intriguing claim receiving mixed support is that linguistic prosody is disrupted by LBD, whereas affective prosody is disrupted by RBD (e.g., Emmorey, 1987). Ross (1981) sweepingly argued that the RH is dominant for processing affective prosody and that there is posterior RH localization of the receptive components of this ability and anterior RH localization of the expressive components, analogous to alleged localization of the components of speech in the left hemisphere (LH). But the evidence, again, is mixed (Wertz et al., 1998). Specifically, in a thorough

3 512 ZAIDEL ET AL. review, Borod (1993) finds evidence for RH dominance for emotional, especially prosodic, expression in brain-damaged patients and for facial expression in normal subjects. However, Meerson and Tarkhan (1988) found a selective involvement of left temporal lobe structures in the perception of emotional speech. Indirect Requests Indirect requestss received relatively more attention in the neuropsychological literature than other pragmatic abilities tested by the RHCB. Wilcox, Davis, and Leonard (1978) studied the sensitivity of aphasics to the context redundancy and valence of intent (positive/negative) of indirect requests and found more errors on requests with negative intent. Aphasics asked to judge the appropriateness of videotaped exchanges involving indirect requests ( can you... ) revealed a better pragmatic decision than expected from their level of auditory comprehension, i.e., aphasic patients could appreciate conversational appropriateness even when they seemed not to understand the conversation. Heeschen and Reisches (1979) asked Broca s and Wernicke s aphasics, RBD patients, and normal controls to match one of several multiple-choice sentences to a story with a direct or indirect request and found selective deficits in RBD patients in understanding indirect speech acts. The patients with RBD tended to interpret the stories literally and seemed unable to adopt and integrate the context of the situation. These data show that the RH has some pragmatic competence, which can be used in aphasia, and they suggest that right hemisphere contributions may be necessary for some pragmatic functions. Hirst, LeDoux, and Stein (1984) compared the ability of a few LBD aphasics, RBD patients, and normal subjects to understand indirect requests and found that aphasics could only comprehend nonliteral readings, whereas RBD patients could only comprehend literal readings, suggesting that RBD patients may have problems with assessing scripts or world knowledge or with making pragmatic judgments. Foldi (1987) also studied the ability of LBD aphasics, RBD patients, and normal controls to understand indirect commands. She manipulated the conventionality of the commands and the context that determined the nonliteral interpretation, including social roles of interlocutors as well as politeness, physical setting, event history, and likelihood of occurrence. RBD patients preferred literal responses to indirect commands and showed more inconsistent responses to direct commands. Foldi interpreted the deficit of RBD patients in terms of problems with prosody, misinterpretation of components of exchange, or failure to integrate the necessary components of information. The results suggest that some pragmatic functions are controlled by the LH, although they are independent of linguistic comprehension per se, while others are specialized in the RH. Sensitive to possible methodological confounds of previous studies due to test format, Weylman et al. (1989) administered to left and right brain-damaged patients and to control subjects a purely auditory verbal test of indirect requests. Stimuli consisted of tape-recorded (and printed) paragraph-length vignettes concluding with an interrogative utterance that could be interpreted as a literal, direct question or as a nonliteral request. Alternative responses were played back and printed in front of the subject. Context and conventionality of wording (high conventionality favoring indirect interpretation) were used to induce a literal or nonliteral interpretation. RBD patients were as impaired as aphasics in the use of both context and conventionality to interpret indirect requests, their use of context was selectively worse than that of normla controls, and they tended to before indirect than LBD patients. The authors concluded that damage to the right hemisphere impairs the integration of components

4 RIGHT HEMISPHERE COMMUNICATION BATTERY 513 of information with each other and with context. However, the task is metalinguistic and requires the subject to evaluate a situation together with the response rather than participate in one. Thus, a deficit in metacognition could be responsible for poor performance. Weylman et al. (1989) also studied a preliminary free response task, where RBD and control subjects had to produce the predicted response of an interlocutor in the vignette. This time, RBD patients were as sensitive to indirect context and conventionality of wording as were controls. Thus, the putative pragmatic deficit of RBD patients is at best subtle and paradigm-dependent, but is not due to simple roleplaying deficits. Most likely, these RBD patients have preserved pragmatics but impairment in certain metacognitive operations. Finally, Stemmer (1994) criticized previous studies for lacking a theoretical account of indirect requests that includes degrees of indirectness and an appropriate methodology. She adopted a framework introduced by Blum-Kulka, House and Kasper (1989) and together with Giroux and Joanette (Stemmer et al., 1994) studied both the production and evaluation of nonconventionally indirect requests by RBD patients and by normal controls. They found that the RBD patients behaved differently from controls concerning the application of nonconventionally indirect requests, the use of external request modifiers, the evaluation of nonconventionally indirect requests, and the use of propositions. The findings suggest that the RBD patients were able to build a mental representation at a text level and a situational level but may have had problems within the realm of monitoring or planning-integration processes if more than one mental model was involved. There was also some indication that attention and visuospatial abilities may have influenced the verbal and evaluation abilities of the RBD subjects. Metaphors When presented with common frozen metaphors (e.g., heavy heart ), RBD patients tend to choose literal depictions of them and react to them as amusing or absurd. By contrast, these RBD patients are often able to explain verbally what the metaphor means (Winner & Gardner, 1977). This may predict an interaction between task [Pictorial Metaphor vs (Verbal) Metaphor Comprehension] and Group (LBD vs RBD). Some attribute the apparent selective difficulty of RBD patients with figurative language to methodological constraints of the tasks. On the other hand, pictorial depictions of literal meanings of idioms or metaphors tend to be stereotyped and familiar, whereas figurative meanings can be depicted in a variety of ways that require special processing (Huber, 1990). Indeed, experiments that minimize visual demands often do not show a literal bias in RBD patients (Tompkins et al., 1992). On the other hand, difficulties in interpreting nonliteral expressions do emerge when metacognitive skills are required (Tompkins et al., 1992). D. W. Zaidel and Kasher (1989) studied memory for surrealistic vs realistic paintings with literal or novel metaphoric labels in patients with unilateral brain damage and found LH specialization for novel metaphors and RH specialization for frozen metaphors. Giora (1997) makes similar predictions based on a graded salience hypothesis. (A linguistic expression is considered salient when its meaning can be computed directly from the lexical meanings of the lexical entries, without deriving inferences based on contextual assumptions.) This predicts a dissociation between Group (LBD vs RBD) and salience (or literalness) (absent, present), as in the Metaphor Comprehension and the Sarcasm subtests of the RHCB. Empirical support for this prediction is described in more detail elsewhere (Giora et al, 2000).

5 514 ZAIDEL ET AL. Inferences The Inferences subtest of the RHCB contains vignettes that are composed of two sentences. In order to process these two-sentence vignettes correctly, patients need to revise an initial interpretation based on the first sentence in light of new information contained in the second sentence. RBD patients can process factual information but cannot infer the new information (Brownell et al., 1986). These inferences can be interpreted as special cases of implicatures of quality (Grice, 1989). In a separate article, we discuss the performances of RBD and LBD patients on new verbal and nonverbal tests of implicature processing (Kasher et al., 1999). Tompkins (1995) reviews evidence that RBD patients have selective difficulty with inferencing, but her concept of inference is wider than implicatures and involves deriving implicit information in general. Sarcasm Bihrle et al. (1986) note that RBD patients have particular difficulty in recognizing sarcasm because they fail to appreciate the negative motivation of the speaker, whether explicitly or implicitly stated. Consequently, RBD patients fail to reinterpret correctly the false literal meaning of a sarcastic expression. RBD patients remember the facts but often interpret sarcasm as a lie or a mistake. Alternative Word Meanings This subtest of the RHCB measures the ability to understand ambiguity. In one part, the subject has to match a metaphoric adjective with a nonmetaphoric synonym (e.g., straight honest ). In the other part, the subject has to match a noun with a synonym of its nondominant meaning (e.g., pen cage ). Brownell et al. (1985, 1990) found that RBD patients performed better on the second part, presumably because it does not require one to access metaphoric meanings. We determine if this pattern is confirmed in our data set. D. W. Zaidel et al. (1995) found that both left and right temporal lobectomy patients showed significant and similar deficits in interpreting lexical ambiguity. Tompkins (1990) too found effective priming of auditory lexical decision of ambiguous words by their literal or metaphoric primes in both RBD and LBD patients. Similarly, Gagnon (1994) found no selective RBD deficit in processing metaphors. Chobor and Schweiger (1998) studied the performance of patients with unilateral brain damage in processing homonymy, polysemy, and metaphor consciously or automatically, finding deficits in automatic but not in conscious processing. They suggest that the ability to process multiple meanings and shift from one to the other is subserved by the frontal lobe. We test that prediction. Discourse The Narrative subtest of the RHCB assesses the subjects comprehension and memory of a story. Gardner and Brownell note that RBD patients remember isolated details, omit emotional elements, and exhibit difficulties in sequencing events properly as well as in extracting the main point of the story (Gardner et al., 1983). Tompkins (1995) reviews in some detail the discourse comprehension and production of RBD patients. The main goal of the present study was to determine whether right or left hemisphere-damaged patients show selective deficits on the pragmatic subtests that

6 RIGHT HEMISPHERE COMMUNICATION BATTERY 515 make up the RHCB. More specifically, we asked whether the abilities tapped by the different RHCB subtests are anatomically localized, how specific impairments of these pragmatic abilities relate to impairments of basic language functions found in aphasia, and how they relate to impairments of nonlinguistic cognitive functions found in brain-damage patients. Subjects METHOD Twenty-seven RBD and 31 RBD patients admitted to the Loewenstein Hospital (Raanana, Israel) for rehabilitation after stroke were recruited for the study on the basis of the following inclusion criteria: (1) first occurrence of an ischemic brain infarction, or a small parenchymal hemorrhage, as determined from history, physical examination, and the acute-stage CT scan; (2) absence of marked mass effects (with possible unrecognizable distant structural damage) in the acute-stage CT scan; (3) negative neurologic or psychiatric past history; (4) absence of significant cortical atrophy or leukoaraiosis; (5) a stable clinical and metabolic state; and (6) fair knowledge of the Hebrew language. In the RBD group there were 20 males and 7 females. All patients were right handers except for one ambidextrous patient. The age range was years (mean: ) and the mean educational level was years of formal schooling. Eight patients were born in Israel, 11 in Arabic countries, and 8 in European countries. All patients spoke Hebrew fluently, and in all but 3 there was also a fair level of Hebrew reading and writing. Eighteen of the 27 patients in this group had signs of contralateral neglect, at least in early stages of their disease. All the patients were examined during the hospitalization period, with time after the onset of stroke being weeks. In the LBD group there were 20 males and 11 females. Twenty-six patients were right handers, 4 were left handers, and 1 was a converted left hander. The age range was years (mean: ) and the mean educational level was years. Nine patients were born in Israel, 10 in Arabic countries, and 12 in Western, mainly European, countries. All patients spoke Hebrew fluently, and all but 1 also had a fair level of Hebrew reading and writing. Twenty-nine of the 31 patients in this group manifested language problems of various kinds. Of the classifiable LBD aphasics, 4 patients had global aphasia, 7 had Broca s aphasia, 4 had Wernicke s aphasia, 2 had conduction aphasia, and 7 had anomic aphasia, using the typology of the Western Aphasia Battery (WAB, Kertesz, 1979). All the patients were examined during the hospitalization period, with time after the onset of stroke being weeks. It was not possible to examine all the above subjects in all of the HRHCB subtests, as well as in all other language and cognitive tests. In some cases there were unexpected clinical causes for not completing the tests, or a subject had to be excluded from a given test when it was obvious that his/her linguistic or cognitive status precluded understanding of task requirements in that specific test. Overall, there were 23 RBD and 12 LBD patients who completed the entire HRHCB (11 subtests), while others performed only parts of the battery. The ANOVAs reported below were restricted to those subjects who completed the entire battery, but more restricted analyses included all the patients who completed the relevant tests. Twenty-one age-matched normal subjects served as controls. Lesion Analysis In the RBD group, brain damage was caused by an ischemic infarction in 26 patients, 25 in the middlecerebral-artery (MCA) territory and 1 in the posterior-cerebral-artery (PCA) territory. One patient had a spontaneous hemorrhage in the right thalamus. In the LBD group 28 patients had ischemic infarctions (24 in the MCA, 2 in the PCA, and 2 the anterior cerebral artery territory). Three patients had a spontaneous hemorrhage in MCA territory. In order to quantify lesion extent in different areas of interest (AOI), lesion information derived from high-quality follow-up (later than 6 weeks postonset) CT scans was reconstructed, separately for each patient, on a set of standard templates, using a normalization procedure based on Talairach s proportional grid system (Talairach & Tournoux, 1988). Structure function correlations were made in 19 of the 27 RBD patients and in 22 of the 31 LBD patients, in whom CT quality and alignment were found suitable for analysis using the normalization procedure. Patients performance on the different subtests of the HRHCB was correlated with lesion extent in a set of prerolandic and retrorolandic perisylvian regions of interest ( anterior and posterior regions). Anterior localization corresponds to the middle-frontal gyrus (FM) and inferior-frontal gyrus (FI). Posterior localization corresponds to the superior temporal gyrus (TS), middle-temporal gyrus (TM), the junction of the

7 516 ZAIDEL ET AL. superior-temporal gyrus and the supramarginal gyrus (TS-PSG), the supramarginal gyrus (PSG), the angular gyrus (PAG), and the junction of the middle temporal gyrus and the angular gyrus (TM-PAG). Hebrew Version of the Right Hemisphere Communication Battery The Right Hemisphere Communication Battery (RHCB; Gardner & Brownell, 1986) includes 11 subtests: (1) Pictoral Humor, (2) Verbal Humor, (3) Humor Production, (4) Prosody, (5) Indirect Requests, (6) Pictoral Metaphors, (7) Verbal Metaphors, (8) Inferences, (9) Sarcasm, (10) Alternative Word Meanings, and (11) Narrative Comprehension. RHCB subtests fall into four parts: Humor (1 3), Emotion (4), Nonliteral Language (5 8), and Integrative Processes (9 11). The basic format of the test battery and the instructions given to the examinees were preserved in the Hebrew version (HRHCB). Necessary changes have been made in order to preserve the meaning of specific test items in the Hebrew language. RHCB subtests are described in the Appendix. We used the following method in scoring patients performance of the HRHCB. Since different subtests have different formats (e.g., nine subtests have multiple choice answers and two subtests have open-ended questions), the percentages of correct scores are not directly comparable across subtests. These scores do allow comparisons between experimental groups for a given test, and they permit correlational analyses, but in order to enter subtests as within-subject factors in an ANOVA, we normalized the scores by expressing them as z scores relative to guessing, using the normal approximation to the binomial guessing distribution. Standard deviations were estimated from the data using the entire experimental population. Each subtest is converted to a uniform 10-item test and ranges are normalized from 3 to 3, with z 0 representing chance (guessing). Positive and negative scores are subjected to separate linear transformations between (0, 3) and ( 3, 0), respectively. Subtests 3 and 7 of the RHCB (Humor Production and Verbal Metaphors) are exceptional in not having the multiple-choice format. The range of normalized scores for these subtests is 0 to 3. The normalization procedure also made it possible to perform a factor analysis of the HRHCB (excluding subtests 3 and 7) in the experimental groups and to correlate the factor scores with (1) anatomical localization of damage, and (2) factors of the Western Aphasia Battery (Kertesz, 1979), which served to assess the basic language abilities of the patients. The approach avoids some spurious statistical results due to multiple comparisons. These normalized scores are at best rough approximations based on modest sample sizes and the analyses with subtest as a variable must therefore be interpreted with caution. The Hebrew version of the Sarcasm subtest (9) is different from the English version in that the six tape-recorded items were not read with a neutral but rather with a sarcastic intonation. Aphasia Tests An experimental Hebrew version of Kertesz (1979) Western Aphasia Battery (HWAB) 1 was prepared by one of us (N. Soroker), with the help of other members of the research team. Substantial changes were made in the patient data format of the test booklet. Linguistic adaptation of the test material was dictated by the inherent differences between Hebrew and English. However, the battery format, the instructions given to the examinees, and the scoring conventions follow those of the English version. The battery contains seven parts: Spontaneous Speech; Auditory Language Comprehension; Repetition; Naming; Reading and Writing; Praxis; Construction, Visuo-Spatial Perception and Acalculia. The battery has two summary scores, an aphasia quotient (AQ) which summarizes performance on the first four parts, and the cortical quotient (CQ), which summarizes performance on all seven parts. Neuropsychological Tests The following subtests of the Wechsler Adult Intelligence Scale Revised (WAIS-R) (Wechsler, 1981) were administered: Picture Completion, Picture Arrangement, and Block Design, using a Hebrew adaptation of the scale. A Hebrew version of some of the verbal subtests of the Wechsler Memory Scale Revised (WMS-R) (Wechsler, 1987), including Digit Span, Logical Memory I (Story A), and Verbal Paired Associates I, was prepared by N. Soroker and E. Vakil. The instructions given to the examinees and the scoring conventions follow exactly those of the English version. The nonverbal subtests used were Figural Memory, Visual Paired Associates Part I, Visual Reproduction, and Visual Memory Span. We computed 1 Copyright 1982 by The Psychological Corporation. Hebrew translation copyright 1997 by the Psychological Corporation. Translated by N. Soroker, M.D., and reproduced by permission. All rights reserved.

8 RIGHT HEMISPHERE COMMUNICATION BATTERY 517 three composite scores: Visual Memory score Figural Memory Visual Paired Associates I; Verbal Memory score Logical Memory I Verbal Paired Associates I; and Attention Score Digit Span Visual Memory Span. All subjects received the Colored form of Raven Progressive Matrices (Raven, 1965) with no time limit (maximum score 36) and the standard version of Benton s Line Orientation Test (Benton et al., 1983). All subjects also received the Star Cancellation subtest of the Behavioral Inattention Test (Wilson, Cockburn, & Halligan, 1987). The maximum score (number of stars canceled) is 54, with a cutoff score of 52. We also computed separately the number of stars canceled on the left and right sides (maximum 27 each). In addition, we administered a Hebrew version of the Stroop (1935) test. Card 1 has 50 color names (blue, green, and red) written in black. In Hebrew these color names are all four letters long. The subject has to read the words aloud. Card 2 has the same 50 color names printed in unmatching color inks. The subject has to name the color of the ink. Card 3 has 50 strings of XXXX in blue, green, or red ink. Here, too, the subject has to name the color of the ink. Subjects are instructed to scan the cards by column. We measured the time to process each card and computed two derived interference measures: time to read card 2 minus card 1 and time to read card 2 minus card 3. HRHCB, HWAB, and the nonlinguistic cognitive tests were administered to each patient within a span of 2 4 weeks. RESULTS Correlations between Different HRHCB Subtests Table 1 presents correlations between different HRHCB subtests that were found to be significant at the.05 level, separately for LBD and RBD patient groups. In subtest 1 (Pictorial Humor), there were by far the weakest correlations with other subtests. Pictorial Humor correlated significantly only with Humor Production (subtest 3) in the RBD group. Eight subtests (3, 5, 6, 7, 8, 9, 10, and 11) correlated significantly with five or more other subtests in the RBD group, and 5 subtests (2, 3, 6, 9, and 11) showed such correlations in the LBD group. The subtests showing the highest number (8 9) of significant correlations among RBD patients were Sarcasm and Narrative. Sarcasm as well as Verbal Humor and Pictorial Metaphors showed the highest number (8 9) of significant correlations in the LBD group. The subtests showing the lowest number (4) of significant correlations among RBD patients were Verbal Humor and Prosody. In contrast, the subtests showing the lowest number (3 4) of significant correlations in the LBD group were Prosody, Indirect Requests, Verbal Metaphors, Inferences, and Alternative Word Meanings. Thus, the two patient groups showed concordant trends on Pictorial Humor and Prosody (low number of correlations) and on Sarcasm (high number of correlations), but they dissociated on the Verbal Humor subtest (high number of correlations in the LBD group and low number in the RBD group). Effects of Unilateral Brain Damage on HRHCB Performance Table 2 presents normalized HRHCB scores separately for the LBD, RBD, and normal control (C) groups. As can be seen, normals obtained higher scores compared to both patien groups in all the HRHCB subtests. An ANOVA was performed, with Group (LBD vs RBD vs C), Birth Place (native Israeli vs nonnative), and Age ( 55 years old vs 55 years old) as between-subjects variables and with Test (subtests 1 11) as a within-subject variable, using the normalized subtests scores as the dependent variable. There was a significant main effect of Group [F(2, 36) 4.74, p.0149], a significant main effect of Test [F(10, 360) 5.81, p.0001], and a significant Group Test interaction [F(20, 360) 1.8, p.0192]. There were no significant main effects or interactions due to Birth Place or Age and those variables were dropped from subsequent analyses. However, in

9 518 ZAIDEL ET AL. TABLE 1 Significant Correlations between Diffferent HRHCB Subtests Scores in LBD and RBD Patient Groups RHCB subtests Pictorial Humor 2. Verbal Humor 3. Humor Production R L R 4. Prosody L 5. Indirect Requests L R 6. Pictorial Metaphors L R L L R R 7. Verbal Metaphors L R R L 8. Inferences L R R L L R R 9. Sarcasm L L R L R L R L R L R L R 10. Alternative Word L R L L R R L R Meanings 11. Narrative Comprehension L R L R R L R L R R R L R R Note subtests. L left brain-damaged patients; R right brain-damaged patients. Correlations were made between normalized values of patients scores (see text for explanation). Only significant correlations (p.05) are shown.

10 RIGHT HEMISPHERE COMMUNICATION BATTERY 519 TABLE 2 Mean and Standard Deviation of Normalized HRHCB Scores in LBD, RBD, and Control Groups RHCB subtests LBD RBD Control 1. Pictorial Humor Verbal Humor Humor Production Prosody Indirect Requests Pictorial Metaphors Verbal Metaphors Inferences Sarcasm Alternative Word Meanings 11. Narrative Comprehension Note. LBD left brain-damaged patients; RBD right brain-damaged patients. Normalized values of patients scores are presented. RHCB subtests 3 and 7 (Humor Production and Verbal Metaphors) are exceptional in not having the multiple-choice format. The range of normalized scores for these subtests is 0 to 3 and not 3 to 3 as for all other subtests. See text for explanation of the normalization procedure. order to neutralize possible small age differences between the experimental groups, Age was entered as a covariate in all subsequent analyses. An ANCOVA with Group (LBD vs RBD vs C) as a between-subjects variable and Test (subtests 1 11) as a within-subject variable, using the normalized subtest scores as the dependent variable and subject age as a covariate, showed a significant main effect of Group [F(2, 44) 5.47, p.0076] and a significant Group Test interaction [F(20, 440) 1.95, p.0088]. There was a significant difference between the controls and both the RBD group [F(1, 33) 11.68, p.0017] as well as the LBD group [F(1, 22) 9.39, p.0057] but there was no differences between the two brain-damaged groups. The Group Test interaction was the result of differences between the control subjects and the patients. A more limited ANCOVA with Group (LBD vs RBD) as a between-subject variable and with Test (subtests 1 11) as a within-subject variable, using the normalized and age-corrected subtest scores as the dependent variable, showed no main effects or interactions. Thus, both LBD and RBD patients groups show impaired performance on the HRHCB and there is no significant difference between the effects of left or right brain damage on performance. A post hoc comparison between controls and patients on individual subtests using normalized age-corrected scores showed significant deficits for the patients on subtests 2, 3, 4, 5, 6, 8, 9, 10, and 11. There was no difference on subtest 1 (Pictorial Humor) and on subtest 7 (Verbal Metaphors). Here, there was no difference between controls and RBD patients, but there was a significant disadvantage to LBD patients relative to normal controls. Post hoc comparisons between LBD and RBD patient groups on individual subtests showed a significant advantage for the RBD group on subtest 5 (Indirect Requests) [F(1, 64) 4.13, p.0463] and on subtest 7 (Verbal Metaphors) [F(1, 53) 11.72, p.0012]. Post hoc pairwise comparisons disclosed Group (LBD vs RBD) Test interactions for the following subtests: Subtests 1 and 7 (Pictorial Humor and Verbal Metaphors) [F(1, 32) 4.22, p.0482], subtests 3 and 9 (Humor Production and Sarcasm) [F(1, 32) 6.1, p.019], subtests 4 and 7 (Prosody and Verbal Metaphors) (F(1,

11 520 ZAIDEL ET AL. 32) 7.22, p.0114], subtests 5 and 9 (Indirect Requests and Sarcasm) [F(1, 32) 5.69, p.0232], and subtests 7 and 9 (Verbal Metaphors and Sarcasm) [F(1, 32) 9.14, p.0049]. These interactions reflect the fact that RBD patients tended to score somewhat higher than LBD patients on HRHCB subtests 3, 5, and 7 (Humor Production, Indirect Requests, and Verbal Metaphors) but lower on subtests 1 and 9(Pictorial Humor and Sarcasm). In order to neutralize the effect of language deficit, on the one hand, and of visuospatial deficit, on the other, on the performance of the patients on the HRHCB, we carried out an ANCOVA with Group (LBD vs RBD) as a between-subject variable, with Test (subtests 1 11) as a within-subject variable, using the normalized subtest scroes as the dependent variable, and with subjects age, aphasia quotient (AQ) on the HWAB, and scores on the Block Design subtest of the WAIS-R as covariates. Again there were no significant main effects or interactions: Group [F(1, 27).12, p.7281], Test [F(10, 270) 1.37, p.1918], Group Test [F(10, 270) 1.46, p.1539]. Another ANCOVA replacing the Block Design scores with Star Cancellation total scores similarly yielded no significant main effects or interactions: Group [F(1, 17) 1.04, p.322], Test [F(10, 170).72, p.7058], Group Test [F(10, 170) 1.04, p.4152]. Effects of Lesions in Perisylvian Cortex on HRHCB Performance Table 3 shows significant negative correlations between HRHCB subtests scores and lesion extent in prerolandic and retrorolandic perisylvian regions separately for the LBD and RBD patient groups. As can be seen, for LBD patients there are significant negative correlations between some subtests of the HRHCB and lesion extent in different frontal, temporal, and parietal regions. Verbal Humor (subtest 2) correlated negatively with lesion extent within the left inferior frontal gyrus. In contrast to this specific association, the performance on Indirect Requests (subtest 5) correlated negatively with lesion extent in most left perisylvian regions: middle and inferior frontal, superior temporal, and supramarginal gyri. Performance on Pictorial Metaphors (subtest 6) correlated negatively with lesion extent in the left superior temporal gyrus. In contrast, Verbal Metaphors (subtest 7) correlated negatively with lesion extent in the left middle temporal gyrus and in the junction of superior temporal and supramarginal gyri. Sarcasm (subtest 9) correlated negatively with lesion extent within the left middle and inferior frontal gyri. In the RBD group, significant negative correlation was found only between Narrative Comprehension (subtest 11) performance and lesion extent in the junction of the right superior temporal and supramarginal gyri. Other HRHCB subtests showed no specific relationship to different areas in the right hemisphere. Thus, although LBD and RBD result in similar deficits on the HRHCB, structure function relationships are more discrete and specific in the left hemisphere. HRHCB and Other Cognitive Functions Table 4 shows correlations between performance on HRHCB subtests and mean scores obtained in a series of cognitive tests (most of them nonverbal in nature). Only correlations that were found to be significant at the.05 level are presented. There were somewhat higher correlations in the RBD than in the LBD group. In the LBD group, significant correlations with the largest number (6 7) of HRHCB subtests occurred for Picture Completion (of the WAIS-R), Verbal Memory, and Attention Span (both of the WMS-R). In the RBD group, significant correlations

12 RIGHT HEMISPHERE COMMUNICATION BATTERY 521 TABLE 3 Significant Negative Correlations between HRHCB Scores and Lesion Extent in Perisylvian Regions among LBD Patient Groups Regions of interest Middle Inferior Superior Middle TS-PSG TM-PAG Parietal Parietal RHCB subtests frontal frontal temporal temporal junction junction SG AG 1. Pictorial Humor 2. Verbal Humor L 3. Humor Production 4. Prosody 5. Indirect Requests L L L L L 6. Pictorial Metaphors L 7. Verbal Metaphors L L 8. Inferences 9. Sarcasm L L 10. Alternative Word Meanings 11. Narrative Comprehension R Note. L left brain-damaged patients; R right brain-damaged patients; TS temporal superior; PSG parietal supramarginal; TM temporal middle; PAG parietal angular. Only significant correlations (p.05) are shown.

13 522 ZAIDEL ET AL. TABLE 4 Significant Correlations between HRHCB Scores and Different Cognitive Tests in the LBD and RBD Patient Groups Cognitive tests Picture Picture Block Star Benton Verbal Visual Attention Completion Arrangement Design Cancell. Line Raven Memory Memory Span RHCB subtests (WAIS-R) (WAIS-R) (WAIS-R) (BIT) Stroop Orientation CPM (WMS-R) (WMS-R) (WMS-R) 1. Pictorial Humor L 2. Verbal Humor L R L R L R L R R L L R 3. Humor Production L R R R R L R R R 4. Prosody L R L L R R L R R L L 5. Indirect Requests L R L 6. Pictorial Metaphors L R L L 7. Verbal Metaphors L L 8. Inferences L R L R L R L L R L R L R L R 9. Sarcasm L R L R L R L R L R L L R 10. Alternative Word R R R R R R Meanings 11. Narrative Comprehension R R R R R L R Note. HRHCB Hebrew version of the Right Hemisphere Communication Battery; WAIS-R Wechsler Adult Intelligence Scale Revised; BIT Behavioral Inattention Test; CPM Colored Progressive Matrices (Raven); WMS-R Wechsler Memory Scale Revised (see text for explanation of composite measures); L left brain-damaged patients; R right brain-damaged patients. Only significant correlations (p.05) are shown.

14 RIGHT HEMISPHERE COMMUNICATION BATTERY 523 with the largest number (6 7) of HRHCB subtests occurred with Block Design, Picture Completion, and Picture Arrangement (of the WAIS-R), and with Raven CPM and Attention Span (of the WMS-R). The Stroop Interference Measure correlated with no HRHCB subtest in either group. Star Cancellation (of the Behavioral Inattention Test for visual neglect) correlated with no HRHCB subtest in the LBD group and with only one subtest in the RBD group. Patients performance on Benton s Line Orientation correlated with only one HRHCB subtest in the LBD group and with only two subtests in the RBD group. Inferences was the subtest that correlated with the largest number of cognitive tests in both groups (eight and seven significant correlations in LBD and RBD groups, respectively), followed by Sarcasm in the LBD group (seven correlations) and by Humor Production in the RBD group (seven correlations). In the LBD group, the HRHCB subtests that showed the smallest number of significant correlations with cognitive tests were Alternative Word Meanings (zero correlations), Narrative Comprehension, and Pictorial Humor (one correlation each). In the RBD group, the subtests that showed the smallest number of significant correlations were Pictorial Humor, Verbal Metaphors (zero correlations each), Indirect Requests, and Pictorial Metaphors (one correlation each). Of the three cognitive tests that were shown to have selective sensitivity to RBD, only Block Design (of the WAIS-R) had more significant correlations with subtests of the HRHCB in RBD than in LBD patients. The other RH tests, Star Cancellation (neglect) and Benton s Line Orientation, showed just marginally greater correlations in RBD than in LBD patients. Of the two cognitive tests that were shown to have selective sensitivity to LBD, Verbal Memory (of the WMS-R) had a marginally greater correlation with subtests of the HRHCB in LBD than in RBD patients and the Stroop Interference Measure correlated with no subtest of the HRHCB in either group. Thus, the laterality of the cognitive tests generally did not predict their correlations with the HRHCB in patients with lateralized cerebral damage. HRHCB and Aphasia Tests Table 5 presents the significant (p.05) correlations that were found between HRHCB subtests and the parts of the Hebrew adaptation of the Western Aphasia Battery (HWAB), for LBD and RBD patients separately. As Table 5 shows, for LBD patients there are significant correlations between most subtests of the HRHCB and most language functions tested with the HWAB. The notable receptions are HRHCB subtests 1 and 10 (Pictorial Humor and Alternative Word Meanings). As can be expected, the correlations between HRHCB subtests and HWAB language functions are sparser for RBD patients. Interestingly, for LBD patients the correlations are the same for the Aphasia Quotient (AQ) and Cortical Quotient (CQ) of the HWAB. For RBD patients, the correlations are a little stronger for CQ than for AQ. This suggests that for RH patients the deficit on the HRHCB reflects the general cognitive effects of brain damage rather than the effects of specific language problems. HRHCB Factor Analysis In order to reduce the number of correlations of HRHCB subtests with language tests, cognitive tests, and lesion locations so as to reduce type-2 errors, we submitted the scores of all of the subjects on the HRHCB to a factor analysis. We realize that our sample sizes are small for a conventional use of factor analysis. However, we report the results in terms of the factors simply to confirm, with composite measures, consistent results from an extensive set of more focused analyses. Subtests 3 and 7

15 524 ZAIDEL ET AL. TABLE 5 Significant Correlations between HRHCB and HWAB Subtests Scores in LBD and RBD Patient Groups WAB subtests Auditory Construction, Spontaneous Verbal vs perception, Aphasia Cortical RHCB subtests Speech Comprehension Repetition Naming Reading Writing Praxis Calculation Quotient Quotient 1. Pictorial Humor L L 2. Verbal Humor L L R L L R L L R L R L L R 3. Humor Production L L R L L R L R R L R L R 4. Prosody L R L L L R L R L R L L R L R 5. Indirect Requests L L L R L R L L R L R 6. Pictorial Metaphors L L L L L R R L L 7. Verbal Metaphors L L R L L L R L L L 8. Inferences L L L R L R L L R L L L 9. Sarcasm L L R L R L R L R L L R L L R L R 10. Alternative Word R R L R R Meanings 11. Narrative Comprehension L R L L R L R R L R L R Note. HRHCB Hebrew version of the Right Hemisphere Communication Battery; HWAB Hebrew version of the Western Aphasia Battery; L left brain-damaged patients; R right brain-damaged patients. Correlations were made between normalized values of patients scores (see text for explanation). Only significant correlations (p.05) are shown.

16 RIGHT HEMISPHERE COMMUNICATION BATTERY 525 of the RHCB (Humor Production and Verbal Metaphors) are exceptional in not having the multiple-choice format. The range of normalized scores for these subtests is 0to 3 and not 3 to 3 as for all other subtests. Therefore, these subtests were excluded from the factor analysis. The initial factor method was Principal Components and the rotation method was Harris Kaiser. Three-factor and four-factor solutions yielded no interpretable results, but a two-factor solution yielded a Nonverbal factor [subtests 1, 4, and 6 (Pictorial Humor, Prosody, and Pictorial Metaphors, respectively)] and a Verbal factor (all other subtests). The rotated factor loading pattern is shown in Table 6. We wanted to know whether the Verbal and Nonverbal factors of the HRHCB are associated with the LH and RH, respectively. An ANOVA with Group (LBD vs RBD vs Controls) as a between-subject variable and Factor (Verbal vs Nonverbal) as a within-subject variable using Age as a covariate, showed a significant main effect of Group [F(2, 48) 8.89, p.0005], no effect of Factor, and no significance of the Group Factor interaction. Both LBD and RBD patients were significantly worse than normal controls on the verbal factor [F(1, 48) 8.89, p.0045, and F(1, 48) 13.30, p.0007, respectively] and on the nonverbal factor [F(1, 48) 5.37, p.0248, and F(1, 48) 10.88, p.0018, respectively]. A second ANOVA, with Group (LBD vs RBD) and Factor (Verbal vs Nonverbal) as variables and Age as a covariate showed no main effects or interactions. Thus, there is no evidence that the two factors of the HRHCB are lateralized. Next we asked whether the two factors of the HRHCB have differential localization. In order to further restrict the number of comparisons, we defined two general regions of interest: Anterior (middle plus inferior frontal gyri) and Posterior (superior and middle temporal gyri plus the junction area of superior temporal and supramarginal gyri, the junction area of middle temporal and angular gyri, and the supramarginal and angular gyri themselves). In the LBD group, there were significant negative correlations between both factors and size of Anterior damage (.84 and.65 for the verbal and nonverbal factors, respectively), but not with lesion extent in the Posterior region. In the RBD patients, there were no significant negative correlations between either factor and either localization. Table 7 summarizes the correlations between the HRHCB verbal and nonverbal factors and the parts of the HWAB, separately for the LBD and RBD groups. As could be expected, there were stronger correlations of the HWAB with the verbal TABLE 6 HRHCB Rotated Factor Pattern Verbal Nonverbal RHCB subtests factor factor 1. Pictorial Humor Verbal Humor Prosody Indirect Requests Pictorial Metaphors Inferences Sarcasm Alternative Word Meanings Narrative Comprehension Note. RHCB subtests 3 and & (Humor Production and Verbal Metaphors) were excluded from the factor analysis because they are not in the standard multiple-choice format. Standard regression coefficients are used; bold indicates significant at the.05 level.

17 526 ZAIDEL ET AL. TABLE 7 Significant Correlations between HRHCB Factors and HWAB Subtests Scores in LBD and RBD Patient Groups. WAB subtests Auditory Construction, Spontaneous verbal vs perception, Aphasia Cortical RHCB factors speech comprehension Repetition Naming Reading Writing Praxis Calculation Quotient Quotient Verbal L R R L R L R L R L R Nonverbal R R R L R Note. HRHCB Hebrew version of the Right Hemisphere Communication Battery; HWAB Hebrew version of the Western Aphasia Battery. L left brain-damaged patients; R right brain-damaged patients. Only significant correlations (p.05) are shown.